Wind power plant protection action sensitive boundary analysis method and device based on probability graph representation and storage medium

Abstract

The invention belongs to the technical field of power system safety operation and control, and provides a wind power plant protection action sensitive boundary analysis method and device based on probability graph representation and a storage medium, and the method comprises the steps: building a wind power plant voltage protection model with a confidence interval; establishing a wind power plant protection sensitive boundary probability graph analysis model; establishing a wind power plant voltage protection linearization model; and calculating a protection action result of each wind power plant under the cascading failure background to form a novel power system cascading failure simulation failure chain. According to the method, the influence of the random change of the wind speed in a novel power system on the unexpected protection action of the wind power plant is considered, a solution thought is provided for quantification of the sensitive boundary of the protection action of the wind power plant, and the actual simulation condition can be represented more accurately and comprehensively.

Description

Technical Field

[0001] The present invention belongs to the technical field of safe operation and control of power systems, and particularly relates to a method, device, and storage medium for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation. Background Art

[0002] In the current research on the cascading fault protection characteristics considering new energy, on the one hand, it focuses on considering the power flow changes after the transient process of each stage of the fault disappears, and less considers the transient process and system stability problems after the fault occurs. On the other hand, it focuses on analyzing the impact of the uncertainty of new energy output from the perspective of line overload, and does not pay attention to the action risk of the voltage protection of new energy units, and the impact of the randomness of output after new energy access on the action characteristics of the voltage protection of new energy units. However, in the evolution process of cascading faults in the new energy grid-connected system, the random changes of external environmental factors such as wind speed will lead to the uncertainty and volatility of wind farm output. Coupled with the promoting effect of the hidden faults of relay protection devices, the voltage and frequency of wind farms participating in the large power grid cascading fault process show high uncertainty characteristics, and problems such as frequency and voltage are intertwined during the fault transient evolution process, increasing the risk of unexpected actions of wind farm protection, and thus may induce problems such as large-scale disconnection of wind farms; the existing unit protection action model ignores the impact of wind farm output randomness on wind farm outage and ignores the actual unexpected actions of wind farm protection in actual simulation.

[0003] Therefore, it is necessary to quantitatively analyze the impact of changes in external environmental factors, etc. on the sensitive boundary of wind farm protection actions in different stages of faults. There is a lack of an effective and feasible method for quantitatively analyzing the sensitive boundary of wind farm protection actions in the existing technology. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, device, and storage medium for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation in view of the deficiencies of the existing technology, providing a solution idea for quantifying the sensitive boundary of wind farm protection actions under the background of random changes in wind speed, being more accurate, more comprehensive, and more capable of representing the actual simulation situation.

[0005] According to one aspect of the specification of the present invention, a method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation is provided, including:

[0006] Step 1, establishing a wind farm voltage protection model with a confidence interval;

[0007] Step 2, establishing a probabilistic graph analysis model for the sensitive boundary of wind farm protection, completing parameter learning of the Bayesian network and inferring to obtain the confidence interval of the sensitive boundary of wind farm protection actions;

[0008] Step 3, establishing a linearized model of wind farm voltage protection;

[0009] Step 4: Based on the linearized model of wind farm voltage protection established in Step 3, calculate the protection action results of each wind farm under the background of cascading faults, and form a simulation fault chain of cascading faults in the new power system.

[0010] As a further technical solution, based on the technical requirements for high and low voltage protection of wind farms in the current national standard, referring to the modeling process of the action-induced outage probability of traditional generator protection, a wind farm voltage protection model with a confidence interval considering wind speed changes, measurement errors, and the influence of hidden faults is established.

[0011] As a further technical solution, in Step 2, establish a probability map analysis model for the sensitive boundary of wind farm protection, including:

[0012] Bayesian network modeling of wind farm nodes: Construct a parent node u representing its on-grid state WT , a parent node e representing the prediction error of its active power output affected by random wind speed changes WT , a child node p representing its active power output WT , the wind farm child node p WT The connections between the aforementioned u WT , e WT and the parent nodes are respectively expressed as conditional probabilities pr{u WT}, pr{e WT}, and the overall is expressed as pr{p WT |u WT , e WT}. Generally speaking, the marginal probability of the DBN network of wind farm nodes is:

[0013] pr{p WT} = Σpr{p WT |u WT}pr{p WT |e WT};

[0014] Bayesian network modeling of synchronous machine nodes: A parent node u representing its on-grid state SG , a child node p representing its active power output SG , the child node p SG The connection between and the parent node u SG is expressed as conditional probabilities pr{u SG}, pr{p SG |u SG}. Generally speaking, the marginal probability of the DBN network of synchronous machine nodes is:

[0015] pr{p SG} = Σpr{p SG |u SG}pr{uSG};

[0016] Bayesian network modeling of line nodes: The parent node u representing its in-network state L , the child node p representing its active power flow L , the child node p L and u L The connection between the child node and the parent node is expressed as conditional probabilities pr{u L}}, pr{p L |u L}}. Overall, the marginal probability of the DBN network of the line node is:

[0017] pr{p L}} = ∑pr{p L |u L}}pr{u L}};

[0018] Bayesian network modeling of load nodes: The parent node u representing its in-network state D , the child node p representing its active power output D , the child node p D and u D The connection between the child node and the parent node is expressed as conditional probabilities pr{u D}}, pr{p D |u D}}. Overall, the marginal probability of the DBN network of the load node is:

[0019] pr{p D}} = Σpr{p D |u D}}pr{p D |e D}}.

[0020] As a further technical solution, model the grid-connected nodes of the wind farm and establish a DBN protection node model related to voltage

[0021] , including: the child node V representing the voltage of the wind farm Bi , the marginal probability of its DBN network is:

[0022]

[0023] Representing the leaf nodes LOV min,Bi , RIV max,Bi indicating whether the voltage of the wind farm exceeds the limit compared with the low-voltage protection setting value V Bi and the high-voltage protection setting value V Bi , the marginal probability of its DBN network is:

[0024]

[0025] Leaf node indicating whether the high and low voltage protection of the wind farm operates The edge probability of its DBN network is as follows:

[0026]

[0027] As a further technical solution, in step 2, completing the parameter learning of the Bayesian network and inferring the confidence interval of the sensitivity boundary of the wind farm protection action includes:

[0028] 1) Parameter learning of the Bayesian model of the wind farm protection sensitivity boundary

[0029] Estimate the probability distribution of each node in the probability map model of the wind farm protection action sensitivity boundary based on the maximum likelihood parameter estimation method to complete the parameter learning of the Bayesian network;

[0030] 2) Inference calculation of the Bayesian model of the wind farm protection sensitivity boundary

[0031] Input the state data of each component at different stages of fault evolution for inference, and calculate the marginal distribution p(x f-act ) of the defined wind farm node protection action probability; Based on the mean mu and variance Sigma of the inference calculation result p(x f-act ), calculate the confidence interval of the wind farm protection action sensitivity boundary according to "mu ± z*(Sigma^0.5)", where the parameter z value is related to the significance level of the confidence interval calculation, and the calculation result is the sensitive boundary error range ±ε of the wind farm protection trigger value U set . U .

[0032] As a further technical solution, the inference calculation of the Bayesian model of the wind farm protection sensitivity boundary further includes:

[0033] Based on the central limit theorem, approximate the (0,1) discrete marginal distribution indicating whether the high and low voltage protection of the wind farm operates in the Bayesian model of the wind farm protection sensitivity boundary as a normal Gaussian distribution.

[0034] As a further technical solution, the establishment of the linearized model of the wind farm voltage protection in step 3 includes:

[0035] Use the broken line model to approximate the wind farm voltage protection model after quantization of the protection sensitivity boundary ±ε U to establish a linearized probability model of the high and low voltage protection action of the wind farm leading to outage.

[0036] According to one aspect of the specification of the present invention, there is provided an analysis device for the sensitivity boundary of the wind farm protection action represented by a probability graph, including:

[0037] The first main module is used to establish a wind farm voltage protection model with a confidence interval;

[0038] The second main module is used to establish a probability map analysis model for the sensitive boundary of wind farm protection, complete the parameter learning of the Bayesian network, and infer the confidence interval of the sensitive boundary of the wind farm protection action;

[0039] The third main module is used to establish a linearized model of wind farm voltage protection;

[0040] The fourth main module is used to calculate the protection action results of each wind farm under the background of cascading faults based on the established linearized model of wind farm voltage protection, and form a simulation fault chain of cascading faults in a new power system.

[0041] According to one aspect of the specification of the present invention, there is provided an analysis device for the sensitive boundary of wind farm protection action based on probabilistic graph representation, including a memory and a processor. The memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the method for analyzing the sensitive boundary of wind farm protection action based on probabilistic graph representation.

[0042] According to one aspect of the specification of the present invention, there is provided a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the steps of the method for analyzing the sensitive boundary of wind farm protection action based on probabilistic graph representation.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0044] The calculation results of the present invention can include different fault connection branches of protection action and non-action, the calculation results are more accurate, the simulation is more comprehensive, the influence of random wind speed changes on the risk of wind farm disconnection is considered, and it can better represent the actual simulation situation, providing a solution idea for the quantification of the sensitive boundary of wind farm protection action under the background of random wind speed changes. Description of the Drawings

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0046] Figure 1 It is a flowchart of the method for analyzing the sensitive boundary of wind farm protection action based on probabilistic graph representation in Embodiment 1 of the present invention.

[0047] Figure 2Schematic diagram of the probability map analysis model for the voltage protection sensitive boundary in Wind Farm in Embodiment 1 of the present invention.

[0048] Figure 3 Schematic diagram of the linearized model for the voltage protection in Wind Farm in Embodiment 1 of the present invention. Detailed implementation manners

[0049] The terms "comprising" and "having" and any variations thereof in the description and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0050] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention. In addition, the technical features in each embodiment or individual embodiment provided by the present invention can be arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the order of steps and / or the mode of structural composition, but must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0051] As shown in the attached Figures 1 to 3 figure, the embodiments of the present invention disclose a method for analyzing the sensitive boundary of wind farm protection actions represented by a probability graph, including the following steps:

[0052] Step 1, establish a wind farm voltage protection model with a confidence interval.

[0053] In a specific implementation manner, the voltage protection model of the new energy power station with a confidence interval in Step 1 is as follows:

[0054] Table 1 Parameter settings for high and low voltage protection in wind farms

[0055]

[0056] Based on the technical requirements for high and low voltage protection of wind farms in the current national standard shown in Table 1, referring to the modeling process of the action-induced outage probability of traditional generator protection, a wind farm voltage protection model with a confidence interval considering the influence of wind speed changes, measurement errors, hidden faults, etc. is established.

[0057] Assume that due to the influence of wind speed changes, measurement errors, and hidden faults, there is a triggering voltage value U in the entire system set with an error of ±ε U , and it follows a truncated normal distribution with a mean of U set0 and a standard deviation of σ U , and the range is [U set0 (1 - ε U ), U set0 (1 + ε U )], and its probability density function is:

[0058]

[0059] where Φ is the standard normal distribution function, and a U represents the total probability of the original normal distribution within the truncation interval.

[0060] Let U re be the operating voltage of the wind farm. Denote event A = {voltage protection operates to disconnect the equipment}, event A L = {under - voltage protection operates to disconnect the equipment}, event A L0 = {under - voltage < 0.2 protection operates to disconnect the equipment}, event A L1 = {under - voltage ≤ 0.9 protection operates to disconnect the equipment}; event A H = {over - voltage protection operates to disconnect the equipment}, event A H0 = {over - voltage > 1.1 protection operates to disconnect the equipment}, event A H1 = {over - voltage > 1.2 protection operates to disconnect the equipment}, event A H2 = {over - voltage > 1.25 protection operates to disconnect the equipment}, event A H3 = {over - voltage > 1.3 protection operates to disconnect the equipment}.

[0061] Denote the under - voltage protection event B L1 = {U ≤ U Lset0}, event B L2 = {U > U Lset0}; the 0.2 fixed - value sensitive boundary event B L01 = {U ≤ U Lset0}, event B L02 = {U > U Lset0}; the 0.9 fixed - value sensitive boundary event B L11 = {U ≤ U Lset1}, event B L12 = {U > U Lset1}. The over - voltage protection event B H1 = {U ≤ U Hset0}, event B H2 = {U > UHset0}; 1.1 Fixed - value sensitive boundary event B H01 ={U≤U Hset0}, event B H02 ={U > U Hset0}; 1.2 Fixed - value sensitive boundary event B H11 ={U≤U Hset1}, event B H12 ={U > U Hset1}; 1.25 Fixed - value sensitive boundary event B H21 ={U≤U Hset2}, event B H22 ={U > U Hset2}; 1.3 Fixed - value sensitive boundary event B H31 ={U≤U Hset3}, event B H32 ={U > U Hset3}.

[0062] Let P(A) be the probability P that the fan stops due to the voltage protection action reU . P(A L |B L1 ) is the probability P of the correct operation of the low - voltage protection Luz , P(A L |B L2 ) is the probability P of the malfunction of the low - voltage protection LUw ; P(A L0 |B L01 ) is the probability P of the correct operation of the low - voltage protection < 0.2 Luz0 , P(A L0 |B L02 ) is the probability P of the malfunction of the low - voltage protection < 0.2 LUw0 ; P(A L1 |B L11 ) is the probability P of the correct operation of the low - voltage protection < 0.9 Luz1 , P(A L1 |B L12 ) is the probability P of the malfunction of the low - voltage protection < 0.9 LUw1 .

[0063] P(A H |B H1 ) is the probability P of the malfunction of the high - voltage protection Huw , P(A H |B H2 ) is the probability P of the correct operation of the high - voltage protection Huz ; P(A H0 |B H01 ) is the probability P of the malfunction of the high - voltage protection > 1.1 HUw0 , P(A H0 |B H02) Probability P of correct operation of the high-voltage protection > 1.1 Huz0 ; P(A H1 |B H11 ) Probability P of incorrect operation of the high-voltage protection > 1.2 Huw1 , P(A H1 |B H12 ) Probability P of correct operation of the high-voltage protection > 1.2 Huz1 ; P(A H2 |B H21 ) Probability P of incorrect operation of the high-voltage protection > 1.25 Huw2 , P(A H2 |B H22 ) Probability P of correct operation of the high-voltage protection > 1.25 Huz2 ; P(A H3 |B H31 ) Probability P of incorrect operation of the high-voltage protection > 1.3 Huw3 , P(A H3 |B H32 ) Probability P of correct operation of the high-voltage protection > 1.3 Huz3 .

[0064] 1) When the generator terminal voltage of the fan crosses the low-voltage limit value U < U Lset0 *(1 - ε U ), the probability that the fan voltage protection action causes shutdown is taken as:

[0065] P(A) = P Luz + P Huw - P Luz P Huw (3)

[0066] 2) When U Lset0 *(1 - ε U ) < U < U Lset0 *(1 + ε U ), the probability that the fan voltage protection action causes shutdown is taken as:

[0067]

[0068] 3) When U Lset0 *(1 + ε U ) < U < U Lset1 *(1 - ε U ), the probability that the fan voltage protection action causes shutdown is taken as:

[0069] P(A) = P Luz1 + P Huw - P Luz1 P Huw + (1 - P Luz1 - P Huw + PLuz1 P Huw )P Luw0 (5)

[0070] 4) When U Lset1 *(1 - ε U ) < U < U Lset1 *(1 + ε U )), the probability of the fan voltage protection action causing shutdown is taken as:

[0071]

[0072] 5) When the fan terminal voltage is within the normal range, i.e., U Lset1 *(1 + ε U ) < U < U Hset0 *(1 - ε U ), the influence of voltage change on the generator shutdown probability is very small, and the probability of the fan voltage protection action causing shutdown is taken as:

[0073] P(A) = P Luw +P Huw -P Luw P Huw (7)

[0074] 6) When U Hset0 *(1 - ε U ) < U < U Hset0 *(1 + ε U )), the probability of the fan voltage protection action causing shutdown is taken as:

[0075]

[0076] 7) When U Hset0 *(1 + ε U ) < U < U Hset1 *(1 - ε U )), the probability of the fan voltage protection action causing shutdown is taken as:

[0077] P(A) = P Luw +P Huw1-3 -P Luw P Huw1-3 +(1 - P Luw -P Huw1-3 +P Luw P Huw1-3 )P Huz0 (9)

[0078] 8) When U Hset1 *(1 - ε U ) < U < U Hset1 *(1 + ε U) When the probability of the fan voltage protection action resulting in shutdown is taken as:

[0079]

[0080] 9) When U Hset1 *(1 + ε U ) < U < U Hset2 *(1 - ε U ) When the probability of the fan voltage protection action resulting in shutdown is taken as:

[0081] P(A) = P Luw +P Huw0,2-3 -P Luw P Huw0,2-3 +(1 - P Luw -P Huw0,2-3 +P Luw P Huw0,2-3 )P Huz1 (11)

[0082] 10) When U Hset2 *(1 - ε U ) < U < U Hset2 *(1 + ε U ) When the probability of the fan voltage protection action resulting in shutdown is taken as:

[0083]

[0084] 11) When U Hset2 *(1 + ε U ) < U < U Hset3 *(1 - ε U ) When the probability of the fan voltage protection action resulting in shutdown is taken as:

[0085] P(A) = P Luw +P Huw0-1,3 -P Luw P Huw0-1,3 +(1 - P Luw -P Huw0-1,3 +P Luw P Huw0-1,3 )P Huz2 (13)

[0086] 12) When U Hset3 *(1 - ε U ) < U < U Hset3 *(1 + ε U ) When the probability of the fan voltage protection action resulting in shutdown is taken as:

[0087]

[0088] 13) When the fan terminal voltage exceeds the high - voltage limit value, i.e., U > UHset3 *(1 + ε U ) When the probability of the fan voltage protection action causing shutdown is taken as:

[0089] P(A) = P Luw + P Huw0-2 - P Luw P Huw0-2 +(1 - P Luw - P Huw0-2 + P Luw P Huw0-2 )P Huz3 (15)

[0090] Then, enter Step 2. Establish a probability map analysis model for the protection sensitive boundary of the wind farm; complete the parameter learning of the Bayesian network and infer to obtain the confidence interval of the protection action sensitivity boundary of the wind farm.

[0091] In the specific implementation manner, the probability map analysis model of the wind farm voltage protection sensitive boundary described in Step 2 is specifically defined as:

[0092] 1) The Bayesian network modeling of the wind farm node includes: the parent node u representing its on-network state WT , u WT = 1 represents in-network operation, u WT = 0 represents that the wind farm is disconnected from the network due to factors such as hidden faults, and is modeled as a DBN(0,1) discrete node; the parent node e representing the active power prediction error affected by the random wind speed change WT , following the N(μ, σ WT ) normal distribution, and is modeled as a DBN continuous node:

[0093]

[0094] c is the scale parameter, k is the shape parameter, which determines the basic shape of the normal distribution curve, and Γ is the gamma function.

[0095] The child node p representing its active power output WT , is modeled as a DBN continuous node, and the wind farm child node p WT and the aforementioned u WT , e WT The connections between the parent nodes are respectively represented as conditional probabilities pr{u WT}}, pr{e WT}}, and the overall is represented as pr{p WT |u WT , e WT}}.

[0096] Overall, the marginal probability of the DBN network of the wind farm node is:

[0097] pr{pWT} = ∑ pr{p WT |u WT} pr{p WT |e WT} (17)

[0098] 2) The Bayesian network modeling of the synchronous machine node includes: the parent node u representing its in-network state SG , u SG = 1 indicates in-network operation, and u SG = 0 indicates that the synchronous machine is disconnected from the network due to factors such as hidden faults, and it is modeled as a DBN(0,1) discrete node; the child node p representing its active power output SG , which is modeled as a DBN continuous node. The connection between the child node p SG and the u SG parent node is represented as the conditional probabilities pr{u SG}, pr{p SG |u SG}.

[0099] Overall, the marginal probability of the DBN network of the synchronous machine node is:

[0100] pr{p SG} = ∑ pr{p SG |u SG} pr{u SG} (18)

[0101] 3) The Bayesian network modeling of the line node includes: the parent node u representing its in-network state L , u L = 1 indicates in-network operation, and u L = 0 indicates that the line is disconnected from the network due to factors such as hidden faults, and it is modeled as a DBN(0,1) discrete node; the child node p representing its active power flow L , which is modeled as a DBN continuous node. The connection between the child node p L and the u L parent node is represented as the conditional probabilities pr{u L}, pr{p L |u L}. Overall, the marginal probability of the DBN network of the line node is:

[0102] pr{p L} = ∑ pr{p L |u L} pr{u L} (19)

[0103] 4) The Bayesian network modeling of the load node includes: the parent node u representing its in-network state D , u D= 1 indicates in-network operation, u D = 0 indicates load disconnection from the network, modeled as a DBN(0,1) discrete node; the child node p representing its active power output D , modeled as a DBN continuous node, child node p D and u D The relationship between the parent nodes is expressed as conditional probabilities pr{u D}}、pr{p D |u D}}. Overall, the marginal probability of the DBN network of the load node is:

[0104] pr{p D}} = ∑pr{p D |u D}}pr{p D |e D}(20)

[0105] 5) Further, model the grid-connected nodes of the wind farm and establish a DBN protection node model related to voltage. It includes: the child node V representing the wind farm voltage Bi , and its marginal probability of the DBN network is:

[0106]

[0107] The leaf nodes LOV min,Bi , RIV max,Bi indicating whether the wind farm voltage exceeds the limit compared with the low-voltage protection setting value V Bi and the high-voltage protection setting value V Bi , and their marginal probabilities of the DBN network are:

[0108]

[0109] The leaf nodes indicating whether the high and low voltage protections of the wind farm act

[0110]

[0111] In addition, in step 2, the parameter learning of the Bayesian network is completed and the confidence interval of the sensitivity boundary of the wind farm protection action is inferred, which is specifically defined as:

[0112] 1) Parameter learning of the Bayesian model of the wind farm protection sensitive boundary

[0113] Estimate the probability distribution of each node in the probability map model of the wind farm protection action sensitive boundary based on the maximum likelihood parameter estimation method, and complete the parameter learning of the Bayesian network, which is expressed as follows:

[0114]

[0115] Among them, D represents a new power system transient stability time-scale historical simulation data sample set containing the high and low voltage protection action information of the wind farm, θ is the parameter vector of each node in the Bayesian model of the wind farm protection sensitive boundary, and p(D|θ) is the probability function. It represents the parameter θ that maximizes p(D|θ).

[0116] 2) Inference calculation of the Bayesian model of the wind farm protection sensitive boundary

[0117] Input the state data of each component at different stages of fault evolution for inference, and calculate the marginal distribution p(x f-act ) of the wind farm node protection action probability defined in formula (23). Here, the Bayesian method adopted is improved. Based on the central limit theorem, the (0,1) discrete marginal distribution representing whether the high and low voltage protection of the wind farm acts or not in the Bayesian model of the wind farm protection sensitive boundary is approximated as a normal Gaussian distribution. Further, based on the mean mu and variance Sigma of the inference calculation result p(x f-act ), calculate the confidence interval of the wind farm protection action sensitivity boundary according to "mu ± z*(Sigma^0.5)". The value of the parameter z is related to the significance level of the confidence interval calculation. For a 95% confidence level, z takes 1.96, and the calculation result is the sensitive boundary error range ±ε set of the wind farm protection trigger value U U .

[0118] Then, enter step 3. Establish a linearized model of the wind farm voltage protection.

[0119] In a specific implementation manner, the specific definition of the wind farm voltage protection linearized model in step 3 is as follows:

[0120] To make the model convenient for engineering applications, a piecewise model is used to approximate the wind farm voltage protection model after quantifying the protection sensitive boundary ±ε U , and a linearized model of the high and low voltage protection action-induced outage probability of the wind farm is established. The specific results are as follows:

[0121] 1) When the grid connection point voltage of the wind farm crosses the low voltage limit value U < U g,min0 , the probability of the wind farm voltage protection action causing outage is taken as:

[0122] P(A) = P Luz +P Huw -P Luz P Huw (25)

[0123] U g,min0 is the low voltage U Lset0 of the wind farm.(0.2 p.u.) lower limit value, taken as the setting value U of the low-voltage protection Lset0 times of 1 - ε U , that is, U Lset0 *(1 - ε U )

[0124] 2) When , the probability of the wind farm voltage protection action resulting in outage is taken as:

[0125]

[0126] is the lower limit of the normal value of the low voltage U of the wind farm, taken as 1 + ε Lset0 times of the specified value U of the low-voltage protection Lset0 , that is, U U *(1 + ε Lset0 *(1 + ε U )

[0127] 3) When , the probability of the wind farm voltage protection action resulting in outage is taken as:

[0128] P(A) = P Luz1 + P Huw - P Luz1 P Huw +(1 - P Luz1 - P Huw + P Luz1 P Huw )P Luw0 (27)

[0129] U g,min1 is the lower limit of the low voltage U of the wind farm, taken as 1 - ε Lset1 times of the setting value U of the low-voltage protection Lset1 , that is, U U *(1 - ε Lset1 *(1 - ε U )

[0130] 4) When , the probability of the wind farm voltage protection action resulting in outage is taken as:

[0131]

[0132] is the lower limit of the normal value of the low voltage U of the wind farm, taken as 1 + ε Lset1 times of the specified value U of the low-voltage protection Lset1 , that is, U U *(1 + ε Lset1 *(1 + ε U )

[0133] 5) When When , the probability of the wind farm voltage protection causing shutdown is:

[0134] P(A)=P Luw +P Huw -P Luw P Huw (29)

[0135] The high voltage U Hset0 The upper limit of the normal value is taken as the high voltage protection setting value U Hset0 1-ε U times, that is, U Hset0 *(1-ε U ).

[0136] 6) When When , the probability of the wind farm voltage protection causing shutdown is:

[0137]

[0138] U g,max0 The high voltage U Hset0 The upper limit value is taken as the high voltage protection setting value U Hset0 1+ε U times, that is, U Hset0 *(1+ε U ).

[0139] 7) When When , the probability of the wind farm voltage protection causing shutdown is:

[0140] P(A)=P Luw +P Huw1-3 -P Luw P Huw1-3 +(1-P Luw -P Huw1-3 +P Luw P Huw1-3 ) Huz0 (31)

[0141] The high voltage U Hset1 The upper limit of the normal value is taken as the high voltage protection setting value U Hset1 1-ε U times, that is, U Hset1 *(1-ε U ).

[0142] 8) When When , the probability of the wind farm voltage protection causing shutdown is:

[0143]

[0144] U g,max1 is the upper limit value of the high voltage U of the wind farm, and is taken as 1+ε times the setting value of the high voltage protection U Hset1 , that is, U Hset1 *(1+ε U ). Hset1 *(1+ε U )

[0145] 9) When , the probability of the wind farm voltage protection action causing outage is taken as:

[0146] P(A) = P Luw +P Huw0,2-3 -P Luw P Huw0,2-3 +(1 - P Luw -P Huw0,2-3 +P Luw P Huw0,2-3 )P Huz1 (33)

[0147] is the upper limit of the normal value of the high voltage U of the wind farm, and is taken as 1 - ε times the setting value of the high voltage protection U Hset2 , that is, U Hset2 *(1 - ε U ). Hset2 *(1 - ε U )

[0148] 10) When , the probability of the wind farm voltage protection action causing outage is taken as:

[0149]

[0150] U g,max2 is the upper limit value of the high voltage U of the wind farm, and is taken as 1+ε times the setting value of the high voltage protection U Hset2 , that is, U Hset2 *(1+ε U ). Hset2 *(1+ε U )

[0151] 11) When , the probability of the wind farm voltage protection action causing outage is taken as:

[0152] P(A) = P Luw +P Huw0-1,3 -P Luw P Huw0-1,3 +(1 - P Luw -P Huw0-1,3 +P Luw P Huw0-1,3 )P Huz2 (35)

[0153] For the high voltage U of the wind farm Hset3 The upper limit of the normal value is taken as the setting value U of the over-voltage protection Hset3 times of 1-ε U That is, U Hset3 *(1 - ε U ).

[0154] 12) When , the probability of the wind farm voltage protection action causing outage is taken as:

[0155]

[0156] U g,max3 For the upper limit value U of the high voltage of the wind farm Hset3 is taken as 1+ε Hset3 times of the setting value U of the over-voltage protection U That is, U Hset3 *(1 + ε U ).

[0157] 13) When U > U g,max3 , the probability of the wind farm voltage protection action causing outage is taken as:

[0158] P(A) = P Luw +P Huz -P Luw P Huz (37)

[0159] Step 4. Based on the wind farm voltage protection linearization model established in Step 3, calculate the protection action results of each wind farm under the background of cascading faults, and form a new power system cascading fault simulation fault chain.

[0160] In order to verify the reliability and effectiveness of this method, under the background of random wind speed variation, this method is compared with the deterministic protection setting value and the conventional outage probability model without considering the confidence interval through cascading fault simulation. The model proposed by this method can take into account the influence of uncertainty factors such as wind speed on the sensitive boundary of wind farm protection, and the calculation results are more accurate, more comprehensive, and more able to represent the actual simulation situation.

[0161] The implementation basis of each embodiment of the present invention is achieved through programmed processing by a device with processor functions. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this actual situation, on the basis of the above embodiments, the embodiments of the present invention provide an analysis device for the sensitive boundary of wind farm protection action based on probabilistic graph representation, and this device is used to execute an analysis method for the sensitive boundary of wind farm protection action based on probabilistic graph representation in the above method embodiments.

[0162] The device includes: a first main module for establishing a wind farm voltage protection model with a confidence interval; a second main module for establishing a probability map analysis model of the wind farm protection sensitive boundary, completing the parameter learning of the Bayesian network and inferring the confidence interval of the wind farm protection action sensitivity boundary; a third main module for establishing a linearized model of the wind farm voltage protection; and a fourth main module for calculating the protection action results of each wind farm under the background of cascading faults based on the established linearized model of the wind farm voltage protection, and forming a simulation fault chain of cascading faults in the new power system.

[0163] An analysis device for the sensitive boundary of wind farm protection actions based on probabilistic graph representation provided by an embodiment of the present invention, aiming at the need to quantitatively analyze the influence of changes in external environmental factors, etc. on the sensitive boundary of wind farm protection actions in different stages of faults, adopts the foregoing several modules. By establishing a probability map analysis model of the wind farm protection sensitive boundary, the parameter learning of the Bayesian network is completed and the confidence interval of the wind farm protection action sensitivity boundary is inferred, providing a solution idea for the quantification of the sensitive boundary of wind farm protection actions under the background of random wind speed changes, being more accurate, more comprehensive, and more capable of characterizing the actual simulation situation.

[0164] It should be noted that the device embodiment provided by the present invention, in addition to being used to implement the method in the above method embodiment, is also used to implement the methods in other method embodiments provided by the present invention. The difference is only in setting corresponding functional modules, and its principle is basically the same as that of the above device embodiment provided by the present invention. As long as those skilled in the art, based on the above device embodiment, refer to the specific technical solutions in other method embodiments, obtain corresponding technical means by combining technical features, and the technical solutions constituted by these technical means, and on the premise of ensuring the practicality of the technical solutions, improve the device in the above device embodiment to obtain corresponding device type embodiments for implementing the methods in other method type embodiments.

[0165] Based on the same inventive concept as the foregoing embodiment, an embodiment of the present invention also provides an analysis device for the sensitive boundary of wind farm protection actions based on probabilistic graph representation, including a memory and a processor. The memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation.

[0166] In an embodiment of the present invention, the memory may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), etc., or may also be a volatile memory, such as a random-access memory (RAM). The memory is any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory in the embodiment of the present invention may also be a circuit or any other device capable of implementing a storage function, for storing program instructions and / or data.

[0167] In an embodiment of the present invention, the processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present invention can be directly embodied as being executed and completed by a hardware processor, or can be executed and completed by a combination of hardware and software modules in the processor.

[0168] Based on the same inventive concept as the foregoing embodiments, an embodiment of the present invention further provides a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the method steps of the wind farm protection action sensitivity boundary analysis method based on a probabilistic graph representation as follows:

[0169] Step 1, establish a wind farm voltage protection model with a confidence interval;

[0170] Step 2, establish a probabilistic graph analysis model for the wind farm protection sensitivity boundary, complete the parameter learning of the Bayesian network, and infer the confidence interval of the wind farm protection action sensitivity boundary;

[0171] Step 3, establish a linearized model of the wind farm voltage protection;

[0172] Step 4, based on the linearized model of the wind farm voltage protection established in Step 3, calculate the protection action results of each wind farm under the background of cascading faults, and form a new power system cascading fault simulation fault chain.

[0173] In summary, the embodiments of the present invention provide a method, device, and storage medium for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation, including: establishing a wind farm voltage protection model with a confidence interval; establishing a probabilistic graph analysis model for the sensitive boundary of wind farm protection; establishing a linearized model of wind farm voltage protection; calculating the protection action results of each wind farm under the background of cascading faults to form a cascading fault simulation fault chain for a new power system. The present invention takes into account the influence of random wind speed changes in the new power system on the unexpected actions of wind farm protection, provides a solution for quantifying the sensitive boundary of wind farm protection actions, and is more accurate, comprehensive, and capable of representing the actual simulation situation.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features. However, such modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation, characterized in that, Including: Step 1: Establish a wind farm voltage protection model with confidence intervals; Step 2: Establish a probability graph analysis model for the sensitive boundary of wind farm protection, complete the parameter learning of the Bayesian network, and infer the confidence interval of the sensitive boundary of wind farm protection action; Step 3: Establish a linearized model of wind farm voltage protection; Step 4: Based on the linearized model of wind farm voltage protection established in Step 3, calculate the protection action results of each wind farm under the background of cascading faults, and form a simulation fault chain of cascading faults in the new power system.

2. The method for analyzing the sensitive boundary of the protection action of a wind farm based on the probabilistic graph representation according to claim 1, wherein Based on the technical requirements for high and low voltage protection of wind farms in the current national standard, referring to the modeling process of the outage probability caused by the action of traditional generator protection, establish a wind farm voltage protection model with confidence intervals considering the influence of wind speed changes, measurement errors, and hidden faults.

3. The method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation according to claim 1, characterized in that, The establishment of the probability graph analysis model for the sensitive boundary of wind farm protection in Step 2 includes: Bayesian network modeling of wind farm nodes: constructing the parent node u representing its on-grid state WT , the parent node e representing the prediction error of the active power output affected by the random wind speed change WT , the child node p representing its active power output WT , the wind farm child node p WT and the aforementioned u WT , e WT The connections between the parent nodes are respectively represented as the conditional probabilities pr{u WT}, pr{e WT}, and the overall is represented as pr{p WT |u WT , e WT}. Overall, the marginal probability of the DBN network of the wind farm nodes is as follows: pr{p WT} = ∑ pr{p WT | u WT} pr{p WT e WT}; Bayesian network modeling of the synchronous machine node: the parent node u representing its on-network state SG , the child node p representing its active power output SG , the child node p SG and u SG The connection between the child node p and the parent node u is represented as the conditional probabilities pr{u SG}, pr{p SG |u SG}. Overall, the marginal probability of the DBN network of the synchronous machine node is: pr{p SG} = ∑ pr{p SG | u SG} pr{u SG}; Bayesian network modeling of line nodes: the parent node u representing its in-network state L , the child node p representing its active power flow L , the child node p L and u L The relationship between the child node p and the parent node u is expressed as conditional probabilities pr{u L}, pr{p L |u L}. Overall, the marginal probability of the DBN network of the line node is as follows: pr{p L} = ∑pr{p L | u L} pr{u L}; Bayesian network modeling of load nodes: parent node u representing its network status D , indicating its active output child node p D , child node p D with u D The relationship between parent nodes is expressed as the conditional probability pr{u D }、pr{p D |u D }, overall, the DBN network edge probability of the load node is: pr{p D} = ∑ pr{p D | u D} pr{p D e D}.

4. The method for analyzing the sensitive boundary of the protection action of a wind farm based on a probabilistic graph representation according to claim 3, wherein Model the grid connection node of the wind farm and establish a voltage-related DBN protection node model, including: a sub-node V representing the voltage of the wind farm Bi , and its DBN network edge probability is: Indicating leaf nodes LOV min,Bi and high-voltage protection setting value V max,Bi respectively, to compare whether the wind farm voltage exceeds the limit, and RIV Bi , Bi The edge probability of its DBN network is as follows: The leaf nodes TGLOVE BWT and TGRIVE BWT , the edge probabilities of its DBN network are as follows:

5. The method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation according to claim 3, wherein The completion of the parameter learning of the Bayesian network and the inference of the confidence interval of the sensitive boundary of wind farm protection action in Step 2 includes: 1) Parameter learning of the Bayesian model for the sensitive boundary of wind farm protection Estimate the probability distribution of each node in the probability graph model of the sensitive boundary of wind farm protection action based on the maximum likelihood parameter estimation method, and complete the parameter learning of the Bayesian network; 2) Inference calculation of the Bayesian model for the sensitive boundary of wind farm protection Infer by inputting the state data of each component at different stages of input fault evolution, and calculate the marginal distribution p(x f-act ) of the defined wind farm node protection action probability; Based on the mean mu and variance Sigma of the inference calculation result p(x f-act ), calculate the confidence interval of the sensitivity boundary of the wind farm protection action according to "mu±z*(Sigma^0.5)". The value of the parameter z is related to the significance level of the confidence interval calculation, and the calculation result is the sensitivity boundary error range ±ε set of the wind farm protection trigger value U U .

6. The method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation according to claim 5, wherein The inference calculation of the Bayesian model for the sensitive boundary of wind farm protection also includes: Based on the central limit theorem, approximate the (0,1) discrete marginal distribution representing whether the high and low voltage protection of the wind farm acts in the Bayesian model of the sensitive boundary of wind farm protection as a normal Gaussian distribution.

7. The method for analyzing the sensitive boundary of wind farm protection actions based on probabilistic graph representation according to claim 1, wherein The establishment of the linearized model of wind farm voltage protection described in Step 3 includes: Adopt a broken-line model to approximate the voltage protection model of the wind farm after quantization of the protection sensitive boundary ±ε U and establish a linearized probability model of the high and low voltage protection actions of the wind farm leading to outage.

8. An analysis device for the sensitive boundary of wind farm protection actions based on probabilistic graph representation, characterized in that, Including: The first main module is used to establish a wind farm voltage protection model with confidence intervals; The second main module is used to establish a probability graph analysis model for the sensitive boundary of wind farm protection, complete the parameter learning of the Bayesian network, and infer the confidence interval of the sensitive boundary of wind farm protection action; The third main module is used to establish a linearized model of wind farm voltage protection; The fourth main module is used to calculate the protection action results of each wind farm under the background of cascading faults based on the established linearized model of wind farm voltage protection, and form a simulation fault chain of cascading faults in the new power system.

9. An analysis device for the sensitive boundary of wind farm protection actions based on probabilistic graph representation, characterized in that, Including a memory and a processor, the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the method for analyzing the sensitive boundary of wind farm protection action represented by a probability graph according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the steps of the method for analyzing the sensitive boundary of wind farm protection action represented by a probability graph according to any one of claims 1 to 7.

Images