Distribution network distributed photovoltaic bearing capacity estimation method based on hidden Markov model

Abstract

The invention discloses a distribution network distributed photovoltaic bearing capacity estimation method based on a hidden Markov model, and the method comprises the steps: firstly building a power distribution network power flow simulation model based on OpenDSS, and determining the safe and stable operation constraint conditions of a target power distribution network; secondly, considering the uncertainty of a distributed photovoltaic access position and access capacity, establishing a hidden Markov model by analyzing the hidden Markov characteristics of the distributed photovoltaic access distribution network, and constructing a load minimum power consumption curve and a distributed photovoltaic maximum power generation curve for researching the extreme condition of the distributed photovoltaic access power grid; and finally, randomly generating scene sets of different distributed photovoltaic access conditions according to a hidden Markov model, performing time sequence analogue simulation under the extreme condition that the distributed photovoltaic output is maximum and the load is minimum, and constructing a target function based on the power distribution network safety and stability constraint condition so as to traverse the generated scene sets. And calculating the distributed photovoltaic bearing capacity of the target power distribution network. According to the method, the problems that the distributed photovoltaic bearing capacity estimation calculation amount of the distribution network is large and the scene applicability is limited due to the fact that the distributed photovoltaic access position and the access capacity are uncertain are solved.

Description

Technical Field

[0001] The present invention relates to the field of energy and power technology, and in particular to a method for estimating the carrying capacity of distributed photovoltaic power generation in a distribution network based on a hidden Markov model. Background Art

[0002] According to statistics from the National Energy Administration of China, in the first three quarters of 2024, the country's distributed photovoltaic power generation will be newly connected to the grid by 85.22 million kilowatts. By the end of September 2024, the national photovoltaic power generation installed capacity has reached 340 million kilowatts. Distributed photovoltaic power generation is a non-dispatched distributed energy resource, and it has uncertainty in access location and access capacity. Therefore, the large-scale access of distributed photovoltaic power generation has a significant impact on the voltage characteristics, equipment overload characteristics, and system power balance of the distribution network. In addition, due to the natural uncertainty of solar radiation, the volatility of distributed photovoltaic power generation, and factors such as temperature and time affect the changes in load demand, resulting in large differences in the power injected into the distribution network. Therefore, in order to prevent the large-scale access of distributed photovoltaic power generation from having a negative impact on the distribution network, scientific evaluation of the carrying capacity of distributed photovoltaic power generation in the distribution network is of great significance to promote its reasonable grid connection.

[0003] Due to the uncertainty of the access location and access capacity of distributed photovoltaics, it is increasingly difficult to estimate the carrying capacity of distribution networks. Based on an accurate and representative distributed photovoltaic access model, the analytical method has high accuracy in estimating the carrying capacity of distribution networks for distributed photovoltaics, but it is difficult to deal with the uncertainty of large and complex distribution networks and distributed photovoltaic power generation, and its representative level in the actual distributed photovoltaic access distribution network is limited. Therefore, the method of random simulation distribution network for estimating the carrying capacity of distributed photovoltaics has attracted the attention of relevant researchers. Among them, the method combining Latin hypercube sampling and Cholesky decomposition performs Cholesky decomposition on the covariance matrix describing the correlation between the access location and access capacity of distributed photovoltaics, and combines Latin hypercube sampling to simulate and calculate the carrying capacity of distribution networks for distributed photovoltaics. It has high calculation accuracy, but the implementation is more complex and requires high accuracy of the covariance matrix. In addition, researchers also use the cumulative method to obtain the statistical characteristics of variables and use the correlation between variables to achieve rapid estimation of the carrying capacity of distribution networks for distributed photovoltaics. However, this method has a high dependence on model assumptions and also needs to ensure the accuracy of distributed photovoltaic access data.

[0004] To address the above problems, the Monte Carlo method is proposed to estimate the carrying capacity of the distribution network for distributed photovoltaics. The Monte Carlo simulation randomly selects distributed photovoltaic access according to the Gaussian distribution of each distributed photovoltaic access to the distribution network, and makes multiple attempts according to the Gaussian distribution to achieve the transition from the current distributed photovoltaic access to the next distributed photovoltaic access. With the emergence of distributed photovoltaic access such as distributed photovoltaic integrated batteries, when using Monte Carlo simulation to estimate the carrying capacity of the distribution network for distributed photovoltaics, additional calculations are required on the basis of considering the distributed photovoltaic access location and access capacity as random variables, which cannot reduce the calculation cost. At the same time, Monte Carlo estimates the carrying capacity of the distribution network for distributed photovoltaics by randomly generating a large number of distributed photovoltaic access scenarios, lacking an understanding of the dynamic behavior of the system, that is, when considering the multiple uncertainties of distributed photovoltaic access to the target distribution network, Monte Carlo simulation can generate a variety of distributed photovoltaic access schemes through simulation, but lacks accurate modeling of internal states, and can only obtain the distribution network's estimation of the carrying capacity of distributed photovoltaics, and it is difficult to understand the impact of the differences in the access location and access capacity of distributed photovoltaics on the carrying capacity estimation process. Monte Carlo simulation is essentially an approximate method to estimate the carrying capacity of distributed photovoltaic power generation in distribution networks. How to ensure that the access of distributed photovoltaic power generation to distribution networks will not violate the constraints of safe and stable operation under any circumstances is also an urgent problem to be solved. Summary of the invention

[0005] In order to solve the above technical problems, the present invention discloses a method for estimating the carrying capacity of distributed photovoltaic power generation in a distribution network based on a hidden Markov model. First, a simulation model of a target distribution network is established based on OpenDSS, taking into account safety constraints such as the maximum deviation of node voltage, the maximum reverse power of the distribution transformer, and the maximum current carrying capacity of the line. Then, a hidden Markov model is established for the uncertainty of the access location and capacity of distributed photovoltaic power generation, and a minimum load power consumption curve and a maximum photovoltaic power generation curve are constructed at the same time. On this basis, a scene set of different access locations and capacities of distributed photovoltaic power generation is randomly generated through a hidden Markov model, and a time series simulation is performed in extreme cases to calculate the carrying capacity of the distribution network for distributed photovoltaic power generation.

[0006] The method for estimating the carrying capacity of distributed photovoltaic power distribution networks based on a hidden Markov model described in the present invention adopts a technical solution comprising the following steps:

[0007] S1. According to the target distribution system topology, equipment characteristics and source-load historical data, establish a distribution network power flow simulation model based on OpenDSS, and establish the target distribution network safe and stable operation constraints including the maximum node voltage deviation, the maximum reverse power of the distribution transformer and the maximum current carrying capacity of the line;

[0008] S2. Considering the uncertainty of the access location and access capacity of distributed photovoltaics, a hidden Markov model is established based on the hidden Markov characteristics of distributed photovoltaic access to the distribution network. At the same time, the minimum load power consumption curve and the maximum distributed photovoltaic power generation curve are constructed based on all load power consumption curves and photovoltaic power generation curves within a year;

[0009] S3. Based on the hidden Markov model, a scenario set considering different access locations and access capacities of distributed photovoltaics is randomly generated. A time series simulation is performed under the extreme condition where the distributed photovoltaic output is maximum and the load is minimum. Based on the generated scenario set traversed based on the safety and stability constraints of the distribution network, the carrying capacity of the target distribution network for distributed photovoltaics is calculated.

[0010] Furthermore, in step S1, the target distribution network is first determined, and the target distribution network is configured according to the data provided by IEEE PEC, taking into full account the topological structure and equipment characteristics of the distribution network, the electricity consumption habits and demand curves of different types of users, and the overall power quality of the system, and establishing a distribution network simulation model based on OpenDSS. Based on the actual operation experience of the distribution network, three safe and stable operation constraints including the maximum deviation of node voltage, the maximum reverse power of the distribution transformer, and the maximum current carrying capacity of the line are considered when evaluating the photovoltaic absorption capacity of the distribution network model.

[0011] Furthermore, in step S1, the safe and stable operation constraints including the maximum deviation of node voltage, the maximum reverse power of distribution transformer and the maximum current carrying capacity of line considered when evaluating the photovoltaic absorption capacity of distribution network model are as follows:

[0012] a. Node voltage deviation constraint:

[0013]

[0014] Where U b is the node voltage reference value, U i is the voltage at node i, U i,max , U i,min are the upper and lower limits of the node voltage respectively, and ε is the node voltage deviation rate limit.

[0015] b. Reverse power constraints of distribution transformers:

[0016]

[0017] In the formula, S L is the reverse power of the substation, S r,max is the maximum reverse power limit allowed by the substation, and γ is the reverse power over-limit threshold of the substation.

[0018] c. Line current constraints:

[0019] 0≤I L≤I L,max (4)

[0020] In the formula, I L is the actual current carrying value of line L, I L,max is the current carrying limit of line L.

[0021] Furthermore, step S2 is specifically as follows:

[0022] S2-1. Clarify the access locations of distributed photovoltaics, including residential users, commercial users and other users, and determine the access capacity of distributed photovoltaics based on the historical data of the target distribution network;

[0023] S2-2. Considering the uncertainty of the access location and access capacity of distributed photovoltaics, according to the hidden Markov characteristics of distributed photovoltaic access to the distribution network, the access location of distributed photovoltaics is defined as a hidden state. Based on this, a transition probability matrix describing the mutual conversion between hidden states is constructed. According to the historical data of the distribution network, the access capacity of distributed photovoltaics can be observed and determined as an observable state. The emission probability matrix from the hidden state to the output observable state is constructed, and the initial state matrix is ​​constructed at the same time. Then, the first distributed photovoltaic access location is determined, and a hidden Markov model is constructed from the five characteristics of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix.

[0024] S2-3. Preprocess the load power consumption and distributed power generation data, construct the load minimum power consumption curve based on all annual load power consumption curves, and construct the distributed photovoltaic maximum power generation curve based on all annual distributed photovoltaic power generation curves.

[0025] Furthermore, in step S2, the hidden Markov model constructed based on the hidden state, observable state, transition probability matrix, emission probability matrix and initial probability matrix is ​​as follows:

[0026] a. The hidden Markov model is recorded as a five-tuple σ consisting of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix:

[0027] σ=(H,O,C,E,V) (5)

[0028] Where H is the set of all hidden states of the distributed photovoltaic access location, O is the set of all observable states of the distributed photovoltaic access capacity, C is the transition probability matrix describing the probability of mutual conversion between states, E is the emission probability matrix describing the transition from the hidden state to the output observable state, and V is the initial probability matrix describing the initial state in the hidden Markov model.

[0029] b. Clarify the user types of the distributed photovoltaic access location, including residential users, commercial users, and other users including educational users and military users, and construct all hidden state sets H based on the user types of the distributed photovoltaic access location, as shown in formula (6):

[0030] H={h1,h2,...,h m}={h r ,h c ,h o} (6)

[0031] Where, H = {h1,h2,...,h m} represents that as time moves forward, the access location of distributed photovoltaics will transfer from one hidden state to another hidden state, m is the total number of hidden states, h r is the hidden state of the access residential user, h c is the hidden state of the access commercial user, h o To access the hidden state of other users.

[0032] c. According to formula (7), the probability of the hidden state of the distributed photovoltaic access location is selected from the hidden state set H, and the transition probability matrix C representing the mutual transition process between hidden states is generated. At the same time, the generated transition probability matrix must ensure that the total probability of transferring from the current hidden state to the next hidden state is 1, as follows:

[0033] Q={q1,q2,...,q t} (7)

[0034] In the formula, Q is a random variable sequence, q t is the state of the random sequence at time t.

[0035] c ij =P(q t+1 =h j |q t =h i ) (8)

[0036] In the formula, c ij is the transition probability, q t Corresponding to the i-th hidden state h at time t i Information, q t+1 Corresponding to the jth hidden state h at time t+1 j information.

[0037]

[0038] Wherein, the dimension of the transition probability matrix C is m′m, and the total number of hidden states m is 3 in the case of equation (6).

[0039] d. Clearly define the distributed photovoltaic access capacity as the observable state of the hidden Markov model, and define the observable state set O according to the size of the distributed photovoltaic access capacity. The specific expression is as follows:

[0040] O={o1,o2,...,o n} (10)

[0041] Where n is the total number of observable states, and o is the access capacity of distributed photovoltaics.

[0042]

[0043] In the formula, O H is the set of observable states of all hidden states under the corresponding distributed photovoltaic access capacity, and A is the set of historical data of all distributed photovoltaic access capacity.

[0044] e. Apply the transition probability matrix C to the hidden state h i , get the next hidden state h j , update q t , the probability of selecting the next distributed photovoltaic access capacity from the observable state set is defined as the emission probability when the distributed photovoltaic access position has been selected from the hidden state set, so as to construct the emission probability matrix E. At the same time, the generated emission probability matrix needs to ensure that for each hidden state in the hidden state set, the total probability of the distributed photovoltaic access capacity from the observable state set is 1. The specific expression is as follows:

[0045] e jz =P(b t =o z |q t =h j ) (12)

[0046] In the formula, z is the total number of observable states, e jz is the emission probability, b t Corresponding to the zth observable state o at time t z information.

[0047]

[0048] Wherein, the dimension of the emission probability matrix E is m′n, and the total number of observable states n is 5 in the case of equation (11).

[0049] f. Define the probability of selecting a hidden state from the hidden state set as the first distributed photovoltaic access location as the initial probability, and then construct the initial probability matrix to ensure that the total probability of all distributed photovoltaic access locations as the first access to the distribution network is 1. The specific expression is as follows:

[0050] vi =P(q t =h i ) (15)

[0051] In the formula, v i is the initial probability of selecting the i-th hidden state.

[0052] V=[v1,v2,...,v m ] (16)

[0053]

[0054] In the formula, the dimension of the initial probability matrix V is 1×m, and m is 3 in combination with formula (6).

[0055] Furthermore, in step S2, a load minimum power consumption curve is generated based on the annual load power consumption curve, as follows:

[0056] x min (t g )=min{x1(t),x2(t),x3(t),...,x l (t)} (18)

[0057] In the formula, the load power consumption curve is recorded once a day with a time resolution of 30 minutes, x min (t g ) is the minimum load power consumption curve in the g-th time interval, x1(t), x l (t) are the load power consumption curves for the gth time interval on the first and last day of a year, respectively.

[0058] X min (t) = {x min (t1),x min (t2),x min (t g ),...,x min (t d )} (19)

[0059] Where, X min (t) is the load minimum power consumption curve, x min (t1), x min (t d ) are the minimum load power consumption curves in the first and last time intervals of each day respectively.

[0060] Furthermore, in step S2, a distributed photovoltaic maximum power generation curve is generated based on the annual distributed photovoltaic power generation curve, as follows:

[0061] y max (t g)=max{y1(t),y2(t),y3(t),...,y l (t)} (20)

[0062] In the formula, distributed photovoltaic power generation data is recorded once a day with a time resolution of 1 minute, y max (t g ) is the maximum power generation curve of distributed photovoltaic in the gth time interval, y1(t), y l (t) are the distributed photovoltaic power generation curves in the gth time interval on the first and last day of the year, respectively.

[0063] Y max (t) = {y max (t1),y max (t2),y max (t g ),...,y max (t d )} (twenty one)

[0064] Where Y max (t) is the annual distributed photovoltaic maximum power generation curve, y max (t1), x max (t d ) are the maximum power generation curves of distributed photovoltaics in the first and last time intervals, respectively.

[0065] Furthermore, the step S3 is specifically as follows:

[0066] S3-1, randomly generate a scenario set that fully considers the uncertainty of the distributed photovoltaic access location and access capacity according to the hidden Markov model in step S2;

[0067] S3-2. According to the different access capacity and access location of distributed photovoltaic in the corresponding scenario, combined with the minimum load power consumption curve and the maximum power generation curve of distributed photovoltaic, a time series simulation is performed;

[0068] S3-3. Construct a safe operation constraint penalty function based on the safety and stability constraints, and then generate an objective function that satisfies the safety and stability constraints. Determine whether the corresponding scenario violates the safety and stability operation conditions based on whether the objective function is 0. If it is 0, the scenario does not violate the safety and stability operation conditions, and extract the next scenario. If it is not 0, adjust the access position and access capacity of the distributed photovoltaics in the scenario accordingly until the safety constraints are met, and then extract the next scenario, traverse the generated scenario set, and finally calculate the distribution network's carrying capacity for distributed photovoltaics.

[0069] Furthermore, in step S3, a safe operation constraint penalty function is constructed from the safe and stable constraint conditions, and then an objective function satisfying the safe and stable operation constraint is generated according to the general form of the constraint conditions, as follows:

[0070] a. Introduced safe operation constraint penalty function l w (k) As shown in formula (22):

[0071]

[0072] Where K is the scenario set considering the uncertainty of the access location and access capacity of distributed photovoltaics, l1(k), l2(k), l3(k), l4(k), l5(k) are the penalty functions of the node voltage upper limit constraint, voltage lower limit constraint, distribution transformer reverse power constraint, line current carrying capacity upper limit constraint and current carrying capacity lower limit constraint in the corresponding scenario k, respectively. w is the subscript of the safe operation penalty function, and its set is recorded as W = {1, 2, 3, 4, 5}.

[0073] b. The general form of the safe and stable operation constraints is as follows:

[0074] l w (x)≤0,w∈W (23)

[0075] c. The specific expression of the objective function F(k) that meets the safe and stable operation constraints is as follows:

[0076]

[0077] In the formula, λ w is the penalty coefficient, λ1, λ2, λ3, λ4, and λ5 are the penalty coefficients of penalty functions l1(k), l2(k), l3(k), l4(k), and l5(k), respectively.

[0078] A computer-readable storage medium having computer instructions stored thereon, characterized in that: when the computer instructions are executed by a processor, a method for estimating the distributed photovoltaic carrying capacity of a distribution network based on a hidden Markov model is implemented.

[0079] In view of the problem that the distribution network is difficult to estimate the carrying capacity of distributed photovoltaic power generation, the present invention proposes a method for estimating the carrying capacity of distributed photovoltaic power generation in the distribution network based on the hidden Markov model. According to the hidden Markov characteristics of the distributed photovoltaic access to the distribution network, the access location and access capacity of the distributed photovoltaic power generation are determined as hidden states and observable states, respectively. The transition probability matrix and the emission probability matrix are used to realize the transition between hidden states and the transition from hidden states to observable states, respectively, providing an in-depth understanding of the transition between states, and facilitating the analysis of the carrying capacity estimation of distributed photovoltaic power generation under different circumstances. At the same time, the load minimum power consumption curve and the photovoltaic maximum power generation curve are used to characterize extreme scenarios to ensure that the access of distributed photovoltaic power generation to the power grid will not violate the safe and stable operation constraints in any scenario. On this basis, a safe operation constraint penalty function is introduced, and the objective function is constructed using the distributed photovoltaic safe and stable operation constraints, so as to more intuitively judge the access situation of distributed photovoltaic power generation in various scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 Schematic diagram of a method for estimating the carrying capacity of distributed photovoltaic power generation in distribution networks based on a hidden Markov model. DETAILED DESCRIPTION

[0081] In order to make the contents of the present invention more clearly understood, the following is based on specific embodiments in combination with the accompanying drawings: The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and cannot be used to limit the scope of protection of the present application.

[0082] Example: Figure 1 A schematic diagram of a method for estimating the carrying capacity of distributed photovoltaic power distribution networks based on a hidden Markov model is provided as an example of the present invention; Figure 1 As shown, the method includes:

[0083] In step S1, it is necessary to establish a distribution network power flow simulation model based on OpenDSS according to the target distribution network topology, equipment characteristics and source and load historical data, and establish the safe and stable operation constraints of the target distribution network, including the maximum deviation of node voltage, the maximum reverse power of distribution transformers and the maximum current carrying capacity of lines.

[0084] First, the target distribution network is determined, and the target distribution network is configured according to the data provided by IEEE PEC. The topology and equipment characteristics of the distribution network, the electricity consumption habits and demand curves of different types of users, and the overall power quality of the system are fully considered, and a distribution network simulation model based on OpenDSS is established. Based on the actual operation experience of the distribution network, three safe and stable operation constraints, including the maximum deviation of node voltage, the maximum reverse power of the distribution transformer, and the maximum current carrying capacity of the line, are considered when evaluating the photovoltaic absorption capacity of the distribution network model.

[0085] The safe and stable operation constraints considered when evaluating the photovoltaic absorption capacity of the distribution network model include the maximum deviation of node voltage, the maximum reverse power of the distribution transformer, and the maximum current carrying capacity of the line, as follows:

[0086] a. Node voltage deviation constraint:

[0087]

[0088] Where U b is the node voltage reference value, U i is the voltage at node i, U i,max , U i,min are the upper and lower limits of the node voltage respectively, and ε is the node voltage deviation rate limit.

[0089] b. Reverse power constraints of distribution transformers:

[0090]

[0091] In the formula, S L is the reverse power of the substation, S r,max is the maximum reverse power limit allowed by the substation, and γ is the reverse power over-limit threshold of the substation.

[0092] c. Line current constraints:

[0093] 0≤I L ≤I L,max (4)

[0094] In the formula, I L is the actual current carrying value of line L, I L,max is the current carrying limit of line L.

[0095] Step S2 is specifically as follows:

[0096] First, the access location of distributed photovoltaics is clarified, including residential users, commercial users and other users, and the size of distributed photovoltaic access capacity is determined based on the historical data of the target distribution network.

[0097] Secondly, the uncertainty of the access location and access capacity of distributed photovoltaics is considered. According to the hidden Markov characteristics of distributed photovoltaic access to the distribution network, the access location of distributed photovoltaics is defined as a hidden state. Based on this, a transition probability matrix describing the mutual conversion between hidden states is constructed. According to the historical data of the distribution network, the access capacity of distributed photovoltaics can be observed and determined as an observable state. The emission probability matrix from the hidden state to the output observable state is constructed, and the initial state matrix is ​​constructed at the same time. Then, the first distributed photovoltaic access location is determined, and a hidden Markov model is constructed from the five characteristics of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix.

[0098] Finally, the load power consumption and distributed power generation data are preprocessed, and the load minimum power consumption curve is constructed based on all annual load power consumption curves. At the same time, the distributed photovoltaic maximum power generation curve is constructed based on all annual distributed photovoltaic power generation curves.

[0099] The hidden Markov model constructed in step S2 based on the hidden state, observable state, transition probability matrix, emission probability matrix and initial probability matrix is ​​as follows:

[0100] a. The hidden Markov model is recorded as a five-tuple σ consisting of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix:

[0101] σ=(H,O,C,E,V) (5)

[0102] Where H is the set of all hidden states of the distributed photovoltaic access location, O is the set of all observable states of the distributed photovoltaic access capacity, C is the transition probability matrix describing the probability of mutual conversion between states, E is the emission probability matrix describing the transition from the hidden state to the output observable state, and V is the initial probability matrix describing the initial state in the hidden Markov model.

[0103] b. Clarify the user types of the distributed photovoltaic access location, including residential users, commercial users, and other users including educational users and military users, and construct all hidden state sets H based on the user types of the distributed photovoltaic access location, as shown in formula (6):

[0104] H={h1,h2,...,h m}={h r ,h c ,h o} (6)

[0105] Where, H = {h1,h2,...,h m} represents that as time moves forward, the access location of distributed photovoltaics will transfer from one hidden state to another hidden state, m is the total number of hidden states, h r is the hidden state of the access residential user, h c is the hidden state of the access commercial user, h o To access the hidden state of other users.

[0106] c. According to formula (7), the probability of the hidden state of the distributed photovoltaic access location is selected from the hidden state set H, and the transition probability matrix C representing the mutual transition process between hidden states is generated. At the same time, the generated transition probability matrix must ensure that the total probability of transferring from the current hidden state to the next hidden state is 1, as follows:

[0107] Q={q1,q2,...,q t} (7)

[0108] In the formula, Q is a random variable sequence, q t is the state of the random sequence at time t.

[0109] c ij =P(q t+1 =h j |q t =h i ) (8)

[0110] In the formula, c ij is the transition probability, q t Corresponding to the i-th hidden state h at time t i Information, q t+1 Corresponding to the jth hidden state h at time t+1 j information.

[0111]

[0112] Wherein, the dimension of the transition probability matrix C is m′m, and the total number of hidden states m is 3 in the case of equation (6).

[0113] d. Clearly define the distributed photovoltaic access capacity as the observable state of the hidden Markov model, and define the observable state set O according to the size of the distributed photovoltaic access capacity. The specific expression is as follows:

[0114] O={o1,o2,...,o n} (10)

[0115] Where n is the total number of observable states, and o is the access capacity of distributed photovoltaics.

[0116]

[0117] In the formula, O H is the set of observable states of all hidden states under the corresponding distributed photovoltaic access capacity, and A is the set of historical data of all distributed photovoltaic access capacity.

[0118] e. Apply the transition probability matrix C to the hidden state h i , get the next hidden state h j , update q t , the probability of selecting the next distributed photovoltaic access capacity from the observable state set is defined as the emission probability when the distributed photovoltaic access position has been selected from the hidden state set, so as to construct the emission probability matrix E. At the same time, the generated emission probability matrix needs to ensure that for each hidden state in the hidden state set, the total probability of the distributed photovoltaic access capacity from the observable state set is 1. The specific expression is as follows:

[0119] ejz =P(b t =o z |q t =h j ) (12)

[0120] In the formula, z is the total number of observable states, e jz is the emission probability, b t Corresponding to the zth observable state o at time t z information.

[0121]

[0122] Wherein, the dimension of the emission probability matrix E is m′n, and the total number of observable states n is 5 in the case of equation (11).

[0123] f. Define the probability of selecting a hidden state from the hidden state set as the first distributed photovoltaic access location as the initial probability, and then construct the initial probability matrix to ensure that the total probability of all distributed photovoltaic access locations as the first access to the distribution network is 1. The specific expression is as follows:

[0124] v i =P(q t =h i ) (15)

[0125] In the formula, v i is the initial probability of selecting the i-th hidden state.

[0126] V=[v1,v2,...,v m ] (16)

[0127]

[0128] In the formula, the dimension of the initial probability matrix V is 1×m, and m is 3 in combination with formula (6).

[0129] In step S2, the minimum load power consumption curve is generated based on the annual load power consumption curve, as follows:

[0130] x min (t g )=min{x1(t),x2(t),x3(t),...,x l (t)} (18)

[0131] In the formula, the load power consumption curve is recorded once a day with a time resolution of 30 minutes, x min (t g ) is the minimum load power consumption curve in the g-th time interval, x1(t), x l(t) are the load power consumption curves for the gth time interval on the first and last day of a year, respectively.

[0132] X min (t) = {x min (t1),x min (t2),x min (t g ),...,x min (t d )} (19)

[0133] Where, X min (t) is the load minimum power consumption curve, x min (t1), x min (t d ) are the minimum load power consumption curves in the first and last time intervals of each day respectively.

[0134] In step S2, a distributed photovoltaic maximum power generation curve is generated based on the annual distributed photovoltaic power generation curve, as follows:

[0135] y max (t g )=max{y1(t),y2(t),y3(t),...,y l (t)} (20)

[0136] In the formula, distributed photovoltaic power generation data is recorded once a day with a time resolution of 1 minute, y max (t g ) is the maximum power generation curve of distributed photovoltaic in the gth time interval, y1(t), y l (t) are the distributed photovoltaic power generation curves in the gth time interval on the first and last day of the year, respectively.

[0137] Y max (t) = {y max (t1),y max (t2),y max (t g ),...,y max (t d )} (twenty one)

[0138] Where Y max (t) is the annual distributed photovoltaic maximum power generation curve, y max (t1), x max (t d ) are the maximum power generation curves of distributed photovoltaics in the first and last time intervals, respectively.

[0139] In step S3, random simulation is performed according to the hidden Markov model to obtain the estimation of distributed photovoltaic carrying capacity, which is specifically:

[0140] First, according to the hidden Markov model in step S2, a scenario set that fully considers the uncertainty of the distributed photovoltaic access location and access capacity is randomly generated.

[0141] Secondly, according to the different access capacity and access location of distributed photovoltaics in the corresponding scenarios, combined with the minimum load power consumption curve and the maximum power generation curve of distributed photovoltaics, a timing simulation is carried out.

[0142] Finally, a safe operation constraint penalty function is constructed according to the safety and stability constraints, and then an objective function that satisfies the safety and stability constraints is generated. Depending on whether the objective function is 0, it is judged whether the corresponding scenario violates the safe and stable operation conditions. If it is 0, the scenario does not violate the safe and stable operation conditions, and the next scenario is extracted. If it is not 0, the access position and access capacity of the distributed photovoltaics in the scenario are adjusted accordingly until the safety constraints are met, and then the next scenario is extracted, the generated scenario set is traversed, and finally the carrying capacity of the distribution network for distributed photovoltaics is calculated.

[0143] In step S3, a safe operation constraint penalty function is constructed from the safe and stable constraint conditions, and then an objective function satisfying the safe and stable operation constraint is generated according to the general form of the constraint conditions, which is specifically as follows:

[0144] a. Introduced safe operation constraint penalty function l w (k) As shown in formula (22):

[0145]

[0146] Where K is the scenario set considering the uncertainty of the access location and access capacity of distributed photovoltaics, l1(k), l2(k), l3(k), l4(k), l5(k) are the penalty functions of the node voltage upper limit constraint, voltage lower limit constraint, distribution transformer reverse power constraint, line current carrying capacity upper limit constraint and current carrying capacity lower limit constraint in the corresponding scenario k, respectively. w is the subscript of the safe operation penalty function, and its set is recorded as W = {1, 2, 3, 4, 5}.

[0147] b. The general form of the safe and stable operation constraints is as follows:

[0148] l w (x)≤0,w∈W (23)

[0149] c. The specific expression of the objective function F(k) that meets the safe and stable operation constraints is as follows:

[0150]

[0151] In the formula, λ w is the penalty coefficient, λ1, λ2, λ3, λ4, and λ5 are the penalty coefficients of penalty functions l1(k), l2(k), l3(k), l4(k), and l5(k), respectively.

[0152] It should be noted that the above embodiments are not intended to limit the protection scope of the present invention, and equivalent changes or substitutions made on the basis of the above technical solutions all fall within the protection scope of the claims of the present invention.

Claims

1. A method for estimating the carrying capacity of distributed photovoltaic power generation in a distribution network based on a hidden Markov model, characterized in that: The following steps are involved: S1. According to the target distribution network topology, equipment characteristics and source-load historical data, establish a distribution network power flow simulation model based on OpenDSS, and establish the target distribution network safe and stable operation constraints including the maximum node voltage deviation, the maximum reverse power of the distribution transformer and the maximum current carrying capacity of the line; S2. Considering the uncertainty of the access location and access capacity of distributed photovoltaics, a hidden Markov model is established based on the hidden Markov characteristics of distributed photovoltaic access to the distribution network. At the same time, the minimum load power consumption curve and the maximum distributed photovoltaic power generation curve are constructed based on all load power consumption curves and photovoltaic power generation curves within a year; S3. Based on the hidden Markov model, a scenario set considering different access locations and access capacities of distributed photovoltaics is randomly generated. A time series simulation is performed under the extreme condition where the distributed photovoltaic output is maximum and the load is minimum. Based on the generated scenario set traversed based on the safety and stability constraints of the distribution network, the carrying capacity of the target distribution network for distributed photovoltaics is calculated.

2. According to the method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model in claim 1, it is characterized in that: Specifically, step S1 includes first determining the target distribution network, configuring the target distribution network according to the data provided by IEEE PEC, fully considering the topological structure and equipment characteristics of the distribution network, the electricity consumption habits and demand curves of different types of users, and the overall power quality of the system, establishing a distribution network simulation model based on OpenDSS, and considering three safe and stable operation constraints including the maximum deviation of node voltage, the maximum reverse power of distribution transformers, and the maximum current carrying capacity of lines when evaluating the photovoltaic absorption capacity of the distribution network model based on the actual operation experience of the distribution network.

3. According to claim 2, a method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model, characterized in that: When evaluating the photovoltaic absorption capacity of the distribution network model, the safe and stable operation constraints considered include the maximum deviation of node voltage, the maximum reverse power of the distribution transformer, and the maximum current carrying capacity of the line, as follows: a. Node voltage deviation constraint: Where U b is the node voltage reference value, U i is the voltage at node i, U i,max , U i,min are the upper and lower limits of the node voltage respectively, ε is the node voltage deviation rate limit, b. Reverse power constraints of distribution transformers: In the formula, S L is the reverse power of the substation, S r,max is the maximum reverse power limit allowed by the substation, γ is the reverse power over-limit threshold of the substation, c. Line current constraints: 0≤I L ≤I L,max (4) In the formula, I L is the actual current carrying value of line L, I L,max It is the maximum current carrying limit of line L.

4. According to the method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model in claim 1, it is characterized in that: The step S2 is specifically as follows: S2-1. Clarify the access locations of distributed photovoltaics, including residential users, commercial users and other users, and determine the access capacity of distributed photovoltaics based on the historical data of the target distribution network; S2-2. Considering the uncertainty of the access location and access capacity of distributed photovoltaics, according to the hidden Markov characteristics of distributed photovoltaic access to the distribution network, the access location of distributed photovoltaics is defined as a hidden state. Based on this, a transition probability matrix describing the mutual conversion between hidden states is constructed. According to the historical data of the distribution network, the access capacity of distributed photovoltaics can be observed and determined as an observable state. The emission probability matrix from the hidden state to the output observable state is constructed, and the initial state matrix is ​​constructed at the same time. Then, the first distributed photovoltaic access location is determined, and a hidden Markov model is constructed from the five characteristics of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix. S2-3. Preprocess the load power consumption and distributed power generation data, construct the load minimum power consumption curve based on all annual load power consumption curves, and construct the distributed photovoltaic maximum power generation curve based on all annual distributed photovoltaic power generation curves.

5. The method for evaluating the distributed photovoltaic absorption capacity of a distribution network based on a hidden Markov model according to claim 4 is characterized in that: The hidden Markov model constructed based on the hidden state, observable state, transition probability matrix, emission probability matrix and initial probability matrix is ​​as follows: a. The hidden Markov model is recorded as a five-tuple σ consisting of hidden state, observable state, transition probability matrix, emission probability matrix and initial state matrix: σ=(H,O,C,E,V) (5) In the formula, H is the set of all hidden states of the distributed photovoltaic access location, O is the set of all observable states of the distributed photovoltaic access capacity, C is the transition probability matrix describing the probability of mutual conversion between states, E is the emission probability matrix describing the transition from the hidden state to the output observable state, V is the initial probability matrix describing the initial state in the hidden Markov model, b. Clarify the user types of the distributed photovoltaic access location, including residential users, commercial users and other users, and construct all hidden state sets H based on the user types of the distributed photovoltaic access location, as shown in formula (6): H={h1,h2,...,h m }={h r ,h c ,h o } (6) Where, H = {h1,h2,...,h m } represents that as time moves forward, the access location of distributed photovoltaics will transfer from one hidden state to another hidden state, m is the total number of hidden states, h r is the hidden state of the access residential user, h c is the hidden state of the access commercial user, h o To access other users’ hidden states, c. According to formula (7), the probability of the hidden state of the distributed photovoltaic access location is selected from the hidden state set H, and the transition probability matrix C representing the mutual transition process between hidden states is generated. At the same time, the generated transition probability matrix must ensure that the total probability of transferring from the current hidden state to the next hidden state is 1, as follows: Q={q1,q2,...,q t } (7) In the formula, Q is a random variable sequence, q t is the state of the random sequence at time t, c ij =P(q t+1 =h j |q t =h i ) (8) In the formula, c ij is the transition probability, q t Corresponding to the i-th hidden state h at time t i Information, q t+1 Corresponding to the jth hidden state h at time t+1 j Information, In the formula, the dimension of the transition probability matrix C is m′m, and the total number of hidden states m is 3 in the case of formula (6). d. Clearly define the distributed photovoltaic access capacity as the observable state of the hidden Markov model, and define the observable state set O according to the size of the distributed photovoltaic access capacity. The specific expression is as follows: O={o1,o2,...,o n } (10) Where n is the total number of observable states, o is the access capacity of distributed photovoltaics, In the formula, O H is the set of observable states of all hidden states under the corresponding distributed photovoltaic access capacity, A is the set of historical data of all distributed photovoltaic access capacity, e. Apply the transition probability matrix C to the hidden state h i , get the next hidden state h j , update q t , the probability of selecting the next distributed photovoltaic access capacity from the observable state set is defined as the emission probability when the distributed photovoltaic access position has been selected from the hidden state set, so as to construct the emission probability matrix E. At the same time, the generated emission probability matrix needs to ensure that for each hidden state in the hidden state set, the total probability of the distributed photovoltaic access capacity from the observable state set is 1. The specific expression is as follows: e jz =P(b t =o z |q t =h j ) (12) In the formula, z is the total number of observable states, e jz is the emission probability, b t Corresponding to the zth observable state o at time t z Information, In the formula, the dimension of the emission probability matrix E is m′n, and the total number of observable states n is 5 in the case of formula (11). f. Define the probability of selecting a hidden state from the hidden state set as the first distributed photovoltaic access location as the initial probability, and then construct the initial probability matrix to ensure that the total probability of all distributed photovoltaic access locations as the first access to the distribution network is 1. The specific expression is as follows: v i =P(q t =h i ) (15) In the formula, v i is the initial probability of selecting the i-th hidden state, V=[v1,v2,...,v m ] (16) In the formula, the dimension of the initial probability matrix V is 1×m, and m is 3 in combination with formula (6).

6. The method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model according to claim 4, characterized in that: The generation of the load minimum power consumption curve based on the annual load power consumption curve is as follows: x min (t g )=min{x1(t),x2(t),x3(t),...,x l (t)} (18) In the formula, the load power consumption curve is recorded once a day with a time resolution of 30 minutes, x min (t g ) is the minimum load power consumption curve in the g-th time interval, x1(t), x l (t) are the load power consumption curves for the first and last days of the year in the g-th time interval, X min (t)={x min (t1),x min (t2),x min (t g ),...,x min (t d )} (19) In the formula, X min (t) is the load minimum power consumption curve, x min (t1), x min (t d ) are the minimum power consumption curves of the load in the first and last time intervals of each day respectively.

7. The method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model according to claim 4, characterized in that: The distributed photovoltaic maximum power generation curve is generated based on the annual distributed photovoltaic power generation curve, as follows: y max (t g )=max{y1(t),y2(t),y3(t),...,y l (t)} (20) In the formula, distributed photovoltaic power generation data is recorded once a day with a time resolution of 1 minute, y max (t g ) is the maximum power generation curve of distributed photovoltaic in the gth time interval, y1(t), y l (t) are the distributed photovoltaic power generation curves in the gth time interval on the first and last day of the year, AND max (t)={and max (t1),and max (t2),and max (t g ),...,and max (t d )} (21) Where Y max (t) is the annual distributed photovoltaic maximum power generation curve, y max (t1), x max (t d ) are the maximum power generation curves of distributed photovoltaics in the first and last time intervals, respectively.

8. The method for estimating the carrying capacity of distributed photovoltaic power distribution network based on hidden Markov model according to claim 1, characterized in that: The step S3 is specifically as follows: S3-1, randomly generate a scenario set that fully considers the uncertainty of the distributed photovoltaic access location and access capacity according to the hidden Markov model in step S2; S3-2. According to the different access capacity and access location of distributed photovoltaic in the corresponding scenario, combined with the minimum load power consumption curve and the maximum power generation curve of distributed photovoltaic, a time series simulation is performed; S3-3. Construct a safe operation constraint penalty function based on the safety and stability constraints, and then generate an objective function that satisfies the safety and stability constraints. Determine whether the corresponding scenario violates the safety and stability operation conditions based on whether the objective function is 0. If it is 0, the scenario does not violate the safety and stability operation conditions, and extract the next scenario. If it is not 0, adjust the access position and access capacity of the distributed photovoltaics in the scenario accordingly until the safety constraints are met, and then extract the next scenario, traverse the generated scenario set, and finally calculate the distribution network's carrying capacity for distributed photovoltaics.

9. A method for estimating the carrying capacity of distributed photovoltaic power distribution networks based on a hidden Markov model according to claim 8, characterized in that: The safe operation constraint penalty function is constructed from the safe and stable constraint conditions, and then the objective function that satisfies the safe and stable operation constraint is generated according to the general form of the constraint conditions, as follows: a. Introduced safe operation constraint penalty function l w (k) As shown in formula (22): Where K is the scenario set considering the uncertainty of the access location and access capacity of distributed photovoltaics, l1(k), l2(k), l3(k), l4(k), l5(k) are the penalty functions of the node voltage upper limit constraint, voltage lower limit constraint, distribution transformer reverse power constraint, line current carrying capacity upper limit constraint and current carrying capacity lower limit constraint in the corresponding scenario k, w is the subscript of the safe operation penalty function, and its set is denoted as W = {1,2,3,4,5}. b. The general form of the safe and stable operation constraints is as follows: l w (k)≤0 (w∈W,k∈K) (23) c. The specific expression of the objective function F(k) that meets the safe and stable operation constraints is as follows: In the formula, λ w is the penalty coefficient, λ1, λ2, λ3, λ4, and λ5 are the penalty coefficients of penalty functions l1(k), l2(k), l3(k), l4(k), and l5(k), respectively.

10. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instruction is executed by the processor, a method for estimating the carrying capacity of distributed photovoltaic power distribution networks based on a hidden Markov model as described in any one of claims 1 to 9 is implemented.

Images