Transient power angle stability mode identification method and system based on relative power angle deviation track

Abstract

The invention discloses a transient power angle stability mode identification method and system based on a relative power angle deviation trajectory, and the method comprises the steps: calculating a relative power angle curve of each unit in a power system relative to the inertia center of the system, and obtaining the relative power angle deviation trajectory of each unit through zero offset correction; determining a grouping moment according to the relative power angle deviation range and variance; carrying out per-unit processing on adjacent gaps in the relative power angle deviation value sorting sequence at the grouping moment, analyzing a significant gap, and dividing a leading group and a lagging group by using the significant gap to obtain a plurality of candidate modes; and determining the candidate mode corresponding to the maximum difference value of inertia center offset trajectories of the preceding group and the lagging group of each candidate mode as the dominant stable mode in a time period from the disturbance initial moment to the grouping moment. The method can be suitable for a complex power system with strong nonlinearity and multi-mode coupling characteristics, can quickly and accurately identify the dominant stability mode, and provides reliable support for transient stability evaluation and active defense control.

Description

Technical Field

[0001] This invention relates to a method and system for identifying transient power angle stability patterns in power systems, and more particularly to a method and system for identifying transient power angle stability patterns based on relative power angle offset trajectories. Background Technology

[0002] The transient power angle stability problem of power systems exhibits characteristics such as strong nonlinearity, multi-mode coupling, and inter-group insynchrony. Under large disturbances, the dynamic response of generators shows a superposition structure of "dominant and secondary" multi-modes. Therefore, accurately identifying the dominant stability mode is the foundation for safety and stability analysis and control decision-making.

[0003] However, existing dominant stable pattern recognition methods have the following main drawbacks:

[0004] (1) There is a bottleneck in computational efficiency: Although some methods (such as enumeration) are highly accurate, they require evaluation of all grouping schemes one by one. As the scale of the units and the system dimensions increase, the amount of computation increases exponentially, and the efficiency of application in large power grids is extremely low.

[0005] (2) Criterion defects and poor adaptability: Some analysis methods (such as the same synchronization method) cannot cope with the significant differences between synchronous machines and virtual synchronous machines in terms of physical characteristics and dynamic response mechanisms, resulting in measurement distortion and easy to cause misjudgment of the dominant mode.

[0006] (3) Single dimension for determining the grouping time: Current methods usually rely on the maximum gap criterion to determine the grouping time. In essence, they only utilize two-dimensional local information and fail to fully explore the global statistical features in high-dimensional dynamic trajectories, resulting in insufficient ability to capture the critical moment of group trajectory separation during the overall instability of the system.

[0007] (4) Insufficient adaptability of hybrid systems: With the high proportion of new energy and power electronic equipment connected, existing methods generally adopt absolute quantity criteria (such as absolute power angle difference), which are difficult to adapt to the strong nonlinearity and multi-mode coupling characteristics of hybrid systems of synchronous machines and virtual synchronous machines, and the accuracy of dominant mode recognition has decreased significantly.

[0008] In summary, with the large-scale integration of high-proportion new energy sources and various power electronic devices, the transient power angle stability problem of the power system has become more complex, the identification of dominant modes has become more difficult, and existing methods are unable to effectively address the issue. Summary of the Invention

[0009] Purpose of the invention: This invention provides a method and system for identifying transient power angle stability modes based on relative power angle offset trajectories. It can transform the differences in dynamic response characteristics of various units in a power system into a standardized measure of relative deviation, and accurately identify the dominant stability mode after a power system disturbance and its corresponding dynamic clustering structure of units, providing an accurate basis for power system safety and stability analysis and control decisions.

[0010] Technical solution: The transient power angle stability pattern recognition method based on relative power angle offset trajectory described in this invention includes the following steps:

[0011] (1) Calculate the relative power angle curves of each unit in the power system with respect to the system inertia center, wherein the unit includes each synchronous machine and a virtual synchronous machine;

[0012] (2) Using the relative power angle value at the initial moment of the disturbance as a reference, the relative power angle offset trajectory of each unit is calculated through zero bias correction;

[0013] (3) Calculate the relative power angle offset range at each time point, and determine the time when the range first exceeds the preset threshold as the grouping time point; if the range at all times does not exceed the preset threshold, calculate the relative power angle offset variance at each time point, and determine the time with the largest variance as the grouping time point.

[0014] (4) Sort the relative power angle offset values ​​of each unit at the grouping time, calculate the adjacent gaps of the sorted offset sequence, and normalize them based on the maximum gap; regard all the per-unit values ​​that exceed the preset threshold as significant gaps, divide the leading group and lagging group based on the significant gaps, and obtain several candidate modes.

[0015] (5) During the period from the initial time of the disturbance to the time of grouping, calculate the maximum difference between the inertia center offset trajectories of the leading group and the lagging group of each candidate mode, and determine the corresponding candidate mode as the dominant stable mode.

[0016] This invention fully considers the initial power angle differences of each unit (including traditional synchronous machines and virtual synchronous machines). Using the relative power angle value at the initial moment of the disturbance as a benchmark, it calculates the relative power angle offset trajectory of each unit. Through dual criteria of range or variance and their corresponding thresholds, it dynamically identifies the critical moment of system-wide trajectory separation, avoiding misjudgment of the dominant pattern. Furthermore, this invention utilizes the maximum gap at this critical moment as a benchmark for per-unit processing to adaptively adjust the sensitivity to gap fluctuations under different disturbance intensities.

[0017] Optionally, the physical rotor angle of the synchronous machine is regarded as its power angle, and the virtual rotor angle in the controller of the virtual synchronous machine is regarded as its power angle.

[0018] Optionally, step (1) includes:

[0019] (11) Based on the power angle curves of each synchronous machine and the virtual synchronous machine in the power system after the disturbance, calculate the power angle curve of the system's center of inertia, as shown in the following formula:

[0020]

[0021] Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This is the end time of the measurement or simulation data;

[0022] (12) Calculate the relative power angle curves of each unit, as shown in the following formula:

[0023] Δδ j (t)=δ j (t)-δ c (t), t0≤t≤t end

[0024] Where, Δδ j (t) represents the relative power angle of the j-th generator unit at time t.

[0025] Optionally, the formula for calculating the relative work angle offset trajectory in step (2) is as follows:

[0026] Δδ off,j (t)=Δδ j (t)-Δδ j (t0), t0≤t≤t end

[0027] Where, Δδ off,j (t) represents the relative power angle offset of the j-th generating unit at time t relative to the initial time t0 of the disturbance; Δδ j (t0) represents the relative power angle reference value of the j-th unit at the initial time t0 of the disturbance.

[0028] By calculating the power angle curve of each unit relative to the system's center of inertia and subtracting the value at the initial moment of the disturbance to obtain the offset trajectory, misjudgments caused by the translation of the power angle reference frame and the initial deviation can be effectively avoided.

[0029] Step (3) combines the dual criteria of range and variance to dynamically capture the critical moment of the separation of the overall trajectory of the system, and filters candidate patterns based on multiple significant gaps at that moment, which can effectively prevent the omission of potential patterns.

[0030] Optionally, in step (3), the calculation formula for the relative power angle offset range at each time point is as follows:

[0031]

[0032] Wherein, the R off (t) represents the range of relative work angle deviation at each time point; Δδoff,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0033] Optionally, in step (3), the formula for calculating the variance of the relative work angle offset at each time point is as follows:

[0034]

[0035] in, Δδ represents the variance of the relative work angle offset at time t. off,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0036] Optionally, step (4) includes:

[0037] (41) At the time of grouping, the relative power angle offset values ​​of each unit are arranged in descending order to obtain the sorted offset sequence: Δδ ord,1 ≥Δδ ord,2 ≥…≥Δδ ord,N And record the corresponding original unit number mapping relationship: ID(i)=j,stΔδ ord,i =Δδ off,j (t c ); where Δδ ord,i The relative power angle offset of the i-th unit after sorting is represented by ID(i); ID(i) represents the number of the i-th unit in the original data after sorting; Δδ off,j (t c ) indicates that the j-th unit is at the grouping time t c The relative power angle offset value relative to the initial time t0 of the disturbance;

[0038] (42) Calculate the gap between adjacent items in the sorted offset sequence, and normalize each gap based on the maximum gap, as shown in the following formula:

[0039] g ord,i =Δδ ord,i -Δδ ord,i+1 ,1≤i <N

[0040]

[0041] Among them, g ord,i This represents the gap between the i-th unit and the (i+1)-th unit after sorting; g ord,max Indicates the maximum gap; This represents the per-unit gap between the i-th unit and the (i+1)-th unit after sorting.

[0042] (43) Judge each per-unit gap: if it is greater than the preset per-unit gap threshold β, it is a significant gap;

[0043] For each significant gap corresponding to index k, the first k sorted units are divided into the leading group, and the remaining Nk units are divided into the lagging group. This division result is used as a candidate pattern, as shown in the following formula:

[0044]

[0045] GS k ={ID(1),ID(2),…,ID(k)},k∈K

[0046] GA k ={ID(k+1),…,ID(N)},k∈K

[0047] Where K represents the set of indices corresponding to significant gaps; GS k GA represents the leading group of units based on a significant gap k partition; k This represents the set of lag group units partitioned based on significant gap k.

[0048] Optionally, step (5) includes:

[0049] (51) From the initial disturbance time t0 to the grouping time T c Within the defined time period, the inertia center offset trajectories of the leading and lagging groups of each candidate mode are calculated using the following formula:

[0050]

[0051] Where, Δδ off,S,k (t) represents the relative work angle offset of the leading group's inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode;

[0052] (52) Based on the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode, the maximum difference D is calculated. off,max,k Calculate the global maximum value D off,max As shown in the formula below:

[0053]

[0054] (53) D off,max The corresponding kth * Candidate mode As the dominant stable mode.

[0055] This invention identifies the dominant stable mode by quantifying the maximum difference in the offset trajectories of the two groups of inertia centers, which can effectively reduce the risk of misjudgment.

[0056] The present invention also provides a transient power angle stability pattern recognition system based on relative power angle offset trajectory, comprising:

[0057] The relative power angle calculation module is used to calculate the relative power angle curve of each unit relative to the system's center of inertia.

[0058] The offset trajectory calculation module is used to generate a relative power angle offset trajectory through zero offset correction;

[0059] The grouping time calculation module is used to calculate and determine the grouping time based on the relative power angle offset range or variance at each time point.

[0060] The candidate pattern filtering module is used to divide the leading group and the lagging group based on the per-unit gap between two adjacent units in the sorted sequence of relative power angle offset values ​​of each unit and its preset threshold, and obtain several candidate patterns.

[0061] The dominant mode recognition module is used to calculate the global maximum value of the inertia center offset trajectory under each candidate mode, and determine the corresponding candidate mode as the dominant stable mode.

[0062] Optionally, the relative work angle calculation module includes:

[0063] The center of inertia calculation unit is used to calculate the power angle curve of the system's center of inertia according to the following formula:

[0064]

[0065] Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This is the end time of the measurement or simulation data;

[0066] The relative power angle calculation unit is used to calculate the relative power angle curve of each unit according to the following formula:

[0067] Δδ j (t)=δ j (t)-δ c (t), t0≤t≤t end

[0068] Where, Δδj (t) represents the relative power angle of the j-th generator unit at time t.

[0069] Optionally, the candidate pattern filtering module includes:

[0070] The sorting unit is used to sort the relative power angle offset values ​​of each unit at the time of grouping.

[0071] The boundary division unit is used to calculate the gap between two adjacent items in the sorted offset sequence, and to normalize each gap based on the maximum gap, and to regard those exceeding a preset threshold as significant gaps; the leading group and the lagging group are divided by the significant gaps to obtain several candidate patterns.

[0072] Optionally, the dominant pattern recognition module performs the following process when determining the dominant stable pattern:

[0073] Within the time period from the initial disturbance to the grouping time, the inertia center offset trajectories of each candidate mode leading and lagging groups are calculated according to the following formula:

[0074] t0≤t≤t c And k∈K

[0075] t0≤t≤t c And k∈K

[0076] Where, Δδ off,S,k (t) represents the relative work angle offset of the leading group's inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode;

[0077] Based on the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode... off,max,k Calculate the global maximum value D off,max As shown in the formula below:

[0078]

[0079] D off,max The corresponding kth * Candidate mode As the dominant stable mode.

[0080] Beneficial effects: Compared with existing technologies, this invention proposes an innovative framework of "relative power angle offset trajectory - adaptive grouping screening - dominant mode quantization identification", which significantly improves the ability to capture inter-group out-of-sync characteristics in multi-mode coupling scenarios and provides reliable support for rapid and accurate assessment and active defense control of transient stability of complex power systems. Attached Figure Description

[0081] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0082] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0083] like Figure 1 As shown, the transient power angle stability pattern recognition method based on relative power angle offset trajectory has the following steps.

[0084] Step 1: Calculate the relative work angle.

[0085] Step 11: Obtain the power angle curves of each synchronous machine and the virtual synchronous machine in the power system after the disturbance in advance. For the synchronous machine, its "physical rotor angle" is regarded as the power angle of the machine; for the virtual synchronous machine, the "virtual rotor angle" in its controller is regarded as the power angle of the machine.

[0086] Calculate the power angle curve of the system's center of inertia using the following formula:

[0087]

[0088] Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This refers to the end time of the measurement or simulation data.

[0089] Step 12: Calculate the relative power angle curve of each unit relative to the system center of inertia using the following formula:

[0090] Δδ j (t)=δ j (t)-δ c (t), t0≤t≤t end

[0091] Where, Δδ j (t) represents the relative power angle of the j-th generator unit at time t.

[0092] Step 2: Offset Trajectory Calculation

[0093] Using the relative power angle value at the initial moment of the disturbance as a benchmark, the relative power angle offset trajectory of each unit relative to the initial moment of the disturbance is calculated through zero-bias correction, as shown in the following formula:

[0094] Δδ off,j(t)=Δδ j (t)-Δδ j (t0), t0≤t≤t end

[0095] Where, Δδ off,j (t) represents the relative power angle offset of the j-th generating unit at time t relative to the initial time t0 of the disturbance; Δδ j (t0) represents the relative power angle reference value of the j-th unit at the initial time t0 of the disturbance.

[0096] Step 3: Calculate the time of grouping

[0097] Step 31: Calculate the relative angle deviation range R at each time point according to the following formula. off (t):

[0098]

[0099] Where, Δδ off,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0100] Step 32: When the above range exceeds the preset threshold α for the first time, the time t1 is determined as the clustering time; otherwise, if the range of all times does not exceed the preset threshold α, proceed to step 33.

[0101] Step 33: Calculate the variance of the relative work angle offset at each time point using the following formula:

[0102]

[0103] in, Δδ represents the variance of the relative work angle offset at time t. off,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0104] Step 34: Determine the time t2 with the largest variance as the clustering time.

[0105] Step 4: Candidate Pattern Filtering

[0106] Step 41: At the time of grouping, sort the relative power angle offset values ​​of each unit.

[0107] We can sort them in descending order (largest to smallest) to obtain the sorted offset sequence: Δδ ord,1 ≥Δδord,2 ≥…≥Δδ ord,N ; where Δδ ord,i This represents the relative power angle offset value of the i-th unit after sorting;

[0108] Step 42: Record the original unit number mapping relationship corresponding to each relative power angle offset value in the sorted offset sequence, as shown in the following formula:

[0109] ID(i) = j,stΔδ ord,i =Δδ off,j (t c )

[0110] Where ID(i) represents the number of the i-th unit in the original data after sorting; Δδ off,j (t c ) indicates that the j-th unit is at the grouping time t c The relative power angle offset value relative to the initial time t0 of the disturbance;

[0111] Step 43: For the sorted offset sequence, calculate the difference (i.e., the "gap") between adjacent items, and normalize each gap using the maximum gap as the benchmark, as shown in the following formula:

[0112] g ord,i =Δδ ord,i -Δδ ord,i+1 ,1≤i <N

[0113]

[0114] Among them, g ord,i This represents the gap between the i-th unit and the (i+1)-th unit after sorting; g ord,max Indicates the maximum gap; This represents the per-unit gap between the i-th unit and the (i+1)-th unit after sorting.

[0115] Step 44: Determine the significance of each per-unit gap: if it is greater than the preset per-unit gap threshold β, it is a significant gap; for each significant gap corresponding to index k, divide the first k sorted units into the leading group and the remaining Nk units into the lagging group, and use this division result as a candidate pattern, as shown in the following formula:

[0116]

[0117] GS k ={ID(1),ID(2),…,ID(k)},k∈K

[0118] GA k ={ID(k+1),…,ID(N)},k∈K

[0119] Where K represents the set of indices corresponding to significant gaps; GS k GA represents the leading group of units based on a significant gap k partition; k This represents the set of lag group units partitioned based on significant gap k.

[0120] Step 5: Dominant Pattern Recognition

[0121] Step 51: Within the time period from the initial disturbance moment to the grouping moment, calculate the inertia center offset trajectories of the leading and lagging groups for each candidate mode, as shown in the following formula:

[0122] t0≤t≤t c And k∈K

[0123] t0≤t≤t c And k∈K

[0124] Where, Δδ off,S,k (y) represents the relative work angle offset of the leading group inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode.

[0125] Step 52: Calculate the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode. off,max,k As shown in the formula below:

[0126]

[0127] Step 53: Calculate the global maximum value D according to the following formula. off,max :

[0128]

[0129] Step 54: Place D off,max The corresponding kth * Candidate mode As the dominant stable mode.

[0130] The transient power angle stable pattern recognition system based on relative power angle offset trajectory of the present invention includes a relative power angle calculation module, an offset trajectory calculation module, a grouping time calculation module, a candidate pattern screening module, and a dominant pattern recognition module connected in sequence.

[0131] The relative power angle calculation module is used to calculate the relative power angle curve of each unit relative to the system inertia center.

[0132] Furthermore, the relative work angle calculation module includes an inertia center calculation unit and a relative work angle calculation unit.

[0133] The center of inertia calculation unit is used to calculate the power angle curve of the system's center of inertia according to the following formula:

[0134]

[0135] Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This refers to the end time of the measurement or simulation data.

[0136] The relative power angle calculation unit is used to calculate the relative power angle curve of each unit according to the following formula:

[0137] Δδ j (t)=δ j (t)-δ c (t), t0≤t≤t end

[0138] Where, Δδ j (t) represents the relative power angle of the j-th generator unit at time t.

[0139] The offset trajectory calculation module is used to generate a relative power angle offset trajectory through zero offset correction.

[0140] Furthermore, the offset trajectory calculation module performs the following steps during zero-offset correction:

[0141] Using the relative power angle value at the initial moment of the disturbance as a benchmark, the relative power angle offset trajectory of each unit relative to the initial moment of the disturbance is calculated through zero-bias correction, as shown in the following formula:

[0142] Δδ off,j (t)=Δδ j (t)-Δδ j (t0), t0≤t≤t end

[0143] Where, Δδ off,j (t) represents the relative power angle offset of the j-th generating unit at time t relative to the initial time t0 of the disturbance; Δδ j (t0) represents the relative power angle reference value of the j-th unit at the initial time t0 of the disturbance.

[0144] The grouping time calculation module is used to calculate and determine the grouping time based on the relative power angle offset range or variance at each time.

[0145] Furthermore, the clustering time calculation module includes a calculation unit and a clustering time determination unit.

[0146] The calculation unit is used to calculate the range and variance of the relative power angle offset at each time point.

[0147] The formula for calculating the relative power angle deviation range at each time point is as follows:

[0148]

[0149] Wherein, the R off (t) represents the range of relative work angle deviation at each time point; Δδ off,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0150] The formula for calculating the variance of the relative power angle offset at each time point is as follows:

[0151]

[0152] in, Δδ represents the variance of the relative work angle offset at time t. off,j (t) represents the relative power angle offset of the j-th generator at time t relative to the initial time t0 of the disturbance; N is the total number of synchronous machines and virtual synchronous machines in the system; t end This refers to the end time of the measurement or simulation data.

[0153] The grouping time determination unit is used to monitor the relative power angle offset range calculated by the calculation unit.

[0154] R off (t), when the R off (t) When the range of all times exceeds the preset threshold α for the first time, the time t1 is determined as the time of grouping; otherwise, if the range of all times does not exceed the preset threshold α, the maximum value of the relative power angle offset variance calculated by the calculation unit is detected, and the time t2 is determined as the time of grouping.

[0155] The candidate pattern filtering module is used to divide the leading group and the lagging group based on the per-unit gap between two adjacent units in the sorted sequence of relative power angle offset values ​​of each unit and its preset threshold, and obtain several candidate patterns.

[0156] Furthermore, the candidate pattern filtering module includes a sorting unit and a boundary division unit.

[0157] The sorting unit can be a descending sorting unit, used to arrange the relative power angle offset values ​​of each unit at the time of grouping in descending order, to obtain the sequence: Δδ ord,1 ≥Δδ ord,2 ≥…≥Δδ ord,N And map it to the original unit number, i.e., ID(i) = j,stΔδ ord,i =Δδ off,j (t c ); where Δδ ord,i Δδ represents the relative power angle offset of the i-th unit after sorting, where ID(i) represents the number of the i-th unit in the original data after sorting; off,j (t c ) indicates that the j-th unit is at the grouping time t c The relative power angle offset value relative to the initial time t0 of the disturbance.

[0158] The boundary division unit is used to calculate the gap between two adjacent items in the sorted offset sequence, and to normalize each gap based on the maximum gap, and to regard those exceeding a preset threshold as significant gaps; the leading group and the lagging group are divided by the significant gaps to obtain several candidate patterns.

[0159] The formula for calculating the gap between the two adjacent items is:

[0160] g ord,i =Δδ ord,i -Δδ ord,i+1 ,1≤i <N

[0161] The formula for calculating the per-unit standardization is:

[0162]

[0163] Among them, g ord,i This represents the gap between the i-th unit and the (i+1)-th unit after sorting; g ord,max Indicates the maximum gap; This represents the per-unit gap between the i-th unit and the (i+1)-th unit after sorting.

[0164] The division of the leading group and the lagging group includes:

[0165] For each significant gap corresponding to index k, the first k units after sorting are divided into the leading group, and the remaining Nk units are divided into the lagging group. This division result is used as a candidate pattern, as shown in the following formula:

[0166]

[0167] GS k={ID(1),ID(2),…,ID(k)},k∈K

[0168] GA k ={ID(k+1),…,ID(N)},k∈K

[0169] Where K represents the set of indices corresponding to significant gaps; GS k GA represents the leading group of units based on a significant gap k partition; k This represents the set of lag group units partitioned based on significant gap k.

[0170] The dominant mode recognition module is used to calculate the global maximum value of the inertia center offset trajectory under each candidate mode, and determine the corresponding candidate mode as the dominant stable mode.

[0171] Furthermore, the dominant pattern recognition module performs the following process when determining the dominant stable pattern:

[0172] Within the time period from the initial disturbance to the grouping time, the inertia center offset trajectories of each candidate mode leading and lagging groups are calculated according to the following formula:

[0173] t0≤t≤t c And k∈K

[0174] t0≤t≤t c And k∈K

[0175] Where, Δδ off,S,k (t) represents the relative work angle offset of the leading group's inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode;

[0176] Based on the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode... off,max,k Calculate the global maximum value D off,max As shown in the formula below:

[0177]

[0178] D off,max The corresponding kth * Candidate mode As the dominant stable mode.

Claims

1. A method for identifying transient power angle stability patterns based on relative power angle offset trajectories, characterized in that, The method includes: (1) Calculate the relative power angle curves of each generator unit in the power system with respect to the system inertia center, wherein the generator units include synchronous machines and virtual synchronous machines; (2) Using the relative power angle value at the initial moment of the disturbance as a reference, the relative power angle offset trajectory of each unit is calculated through zero bias correction; (3) Calculate the relative power angle offset range at each time point, and determine the time when the range first exceeds the preset threshold as the grouping time point; if the range at all times does not exceed the preset threshold, calculate the relative power angle offset variance at each time point, and determine the time with the largest variance as the grouping time point. (4) Sort the relative power angle offset values ​​of each unit at the grouping time, calculate the adjacent gaps of the sorted offset sequence, and normalize them based on the maximum gap; regard all the per-unit values ​​that exceed the preset threshold as significant gaps, divide the leading group and lagging group based on the significant gaps, and obtain several candidate modes. (5) During the period from the initial time of the disturbance to the time of grouping, calculate the maximum difference between the inertia center offset trajectories of the leading group and the lagging group of each candidate mode, and determine the corresponding candidate mode as the dominant stable mode.

2. The transient power angle stability pattern recognition method based on relative power angle offset trajectory according to claim 1, characterized in that, The physical rotor angle of the synchronous machine is considered its power angle, and the virtual rotor angle in the controller of the virtual synchronous machine is considered its power angle.

3. The transient power angle stability pattern recognition method based on relative power angle offset trajectory according to claim 1, characterized in that, Step (1) includes: (11) Based on the power angle curves of each synchronous machine and the virtual synchronous machine in the power system after the disturbance, calculate the power angle curve of the system's center of inertia, as shown in the following formula: Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This is the end time of the measurement or simulation data; (12) Calculate the relative power angle curves of each unit, as shown in the following formula: Dd j (t)=δ j (t)-d c (t),t0≤t≤t end Where, Δδ j (t) represents the relative power angle of the j-th generator unit at time t.

4. The transient power angle stability pattern recognition method based on relative power angle offset trajectory according to claim 1, characterized in that: The formula for calculating the relative work angle offset trajectory in step (2) is as follows: Dd off,j (t)=Δδ j (t)-Δδ j (t0),t0≤t≤t end Where, Δδ off,j (t) represents the relative power angle offset of the j-th generating unit at time t relative to the initial time t0 of the disturbance; Δδ j (t) represents the relative power angle of the j-th generator unit at time t; Δδ j (t0) represents the relative power angle reference value of the j-th unit at the initial time t0 of the disturbance.

5. The transient power angle stability pattern recognition method based on relative power angle offset trajectory according to claim 1, characterized in that, Step (4) includes: (41) At the time of grouping, the relative power angle offset values ​​of each unit are arranged in descending order to obtain the sorted offset sequence: Δδ ord,1 ≥Δδ ord,2 ≥…≥Δδ ord,N And record the corresponding original unit number mapping relationship: ID(i)=j,stΔδ ord,i =Δδ off,j (t c ); where Δδ ord,i The relative power angle offset of the i-th unit after sorting is represented by ID(i); ID(i) represents the number of the i-th unit in the original data after sorting; Δδ off,j (t c ) indicates that the j-th unit is at the grouping time t c The relative power angle offset value relative to the initial time t0 of the disturbance; (42) Calculate the gap between adjacent items in the sorted offset sequence, and normalize each gap based on the maximum gap, as shown in the following formula: g ord,i =Dd ord,i -Dd ord,i+1 ,1≤i <N Among them, g ord,i This represents the gap between the i-th unit and the (i+1)-th unit after sorting; g ord,max Indicates the maximum gap; This represents the per-unit gap between the i-th unit and the (i+1)-th unit after sorting. (43) Judge each per-unit gap: if it is greater than the preset per-unit gap threshold β, it is a significant gap; For each significant gap corresponding to index k, the first k sorted units are divided into a leading group, and the remaining Nk units are divided into a lagging group. This division result is used as a candidate pattern, as shown in the following formula. GS k ={ID(1),ID(2),…,ID(k)},k∈K GA k ={ID(k)+1),…,ID(N)},k∈K Where K represents the set of indices corresponding to significant gaps; GS k GA represents the leading group of units based on a significant gap k partition; k This represents the set of lag group units partitioned based on significant gap k.

6. The transient power angle stability pattern recognition method based on relative power angle offset trajectory according to claim 1, characterized in that, Step (5) includes: (51) From the initial disturbance time t0 to the grouping time t c Within the defined time period, the inertia center offset trajectories of the leading and lagging groups of each candidate mode are calculated using the following formula: t0≤t≤t c And k∈K t0≤t≤t c And k∈K Where, Δδ off,S,k (t) represents the relative work angle offset of the leading group's inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode; K represents the index set corresponding to the significant gap; (52) Based on the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode, the maximum difference D is calculated. off,max,k Calculate the global maximum value D off,max As shown in the formula below: (53) D off,max The corresponding kth * Candidate mode As the dominant stable mode.

7. A transient power angle stability pattern recognition system based on relative power angle offset trajectory, characterized in that, The system includes: The relative power angle calculation module is used to calculate the relative power angle curve of each unit relative to the system's center of inertia. The offset trajectory calculation module is used to generate a relative power angle offset trajectory through zero offset correction; The grouping time calculation module is used to calculate and determine the grouping time based on the relative power angle offset range or variance at each time point. The candidate pattern filtering module is used to divide the leading group and the lagging group based on the per-unit gap between two adjacent units in the sorted sequence of relative power angle offset values ​​of each unit and its preset threshold, and obtain several candidate patterns. The dominant mode recognition module is used to calculate the global maximum value of the inertia center offset trajectory under each candidate mode, and determine the corresponding candidate mode as the dominant stable mode.

8. The transient power angle stability pattern recognition system based on relative power angle offset trajectory according to claim 7, characterized in that, The relative work angle calculation module includes: The center of inertia calculation unit is used to calculate the power angle curve of the system's center of inertia according to the following formula: Where, δ c (t) represents the power angle of the system's center of inertia at time t; N is the total number of synchronous machines and virtual synchronous machines in the system; M j δ represents the moment of inertia of the j-th unit; j (t) represents the power angle of the j-th generator at time t; t0 is the initial time of the disturbance; t end This is the end time of the measurement or simulation data; The relative power angle calculation unit is used to calculate the relative power angle curve of each unit according to the following formula: Dd j (t)=δ j (t)-d c (t),t0≤t≤t end Where, Δδ j (t) represents the relative power angle of the j-th generator unit at time t.

9. The transient power angle stability pattern recognition system based on relative power angle offset trajectory according to claim 7, characterized in that, The candidate pattern filtering module includes: The sorting unit is used to sort the relative power angle offset values ​​of each unit at the time of grouping. The boundary division unit is used to calculate the gap between two adjacent items in the sorted offset sequence, and to normalize each gap based on the maximum gap, and to regard those exceeding a preset threshold as significant gaps; the leading group and the lagging group are divided by the significant gaps to obtain several candidate patterns.

10. The transient power angle stability pattern recognition system based on relative power angle offset trajectory according to claim 7, characterized in that, The dominant pattern recognition module performs the following process when determining the dominant stable pattern: Within the time period from the initial disturbance to the grouping time, the inertia center offset trajectories of each candidate mode leading and lagging groups are calculated according to the following formula: t0≤t≤t c And k∈K t0≤t≤t c And k∈K Where, Δδ off,S,k (t) represents the relative work angle offset of the leading group's inertia center at time t under the k-th candidate mode; Δδ off,A,k (t) represents the relative work angle offset of the hysteresis group inertia center at time t under the k-th candidate mode; K represents the index set corresponding to the significant gap; Based on the maximum difference D between the inertia center offset trajectories of the leading and lagging groups of each candidate mode... off,max,k Calculate the global maximum value D off,max As shown in the formula below: D off,max The corresponding kth * Candidate mode As the dominant stable mode.

Images