Power distribution network fault early warning method and device considering operation state uncertainty

Abstract

The invention discloses a power distribution network fault early warning method considering operation state uncertainty, and the method comprises the following steps: calculating each operation parameter of a power distribution system based on the historical data and prediction data of distributed photovoltaic and load, and calculating the difference value of the operation parameters under the two conditions; determining a time-phased probability distribution model of the parameter difference, and generating an operation parameter set based on the probability distribution model; according to the calculated parameters, constructing a future operation parameter prediction model based on LSTM, superposing the operation parameter set to obtain a prediction result of the operation situation of the power distribution network, and performing fault early warning on the power distribution network according to the superposition result; according to the method, the future operation situation of the power distribution network can be sensed and pre-judged in advance, and a powerful guarantee is provided for safe and stable operation of a power system.

Description

Technical Field

[0001] The present invention belongs to the field of distribution network fault early warning methods, and in particular, relates to a distribution network fault early warning method and device taking into account the uncertainty of operating states. Background Art

[0002] With the acceleration of urbanization and the continuous growth of electricity demand, the stable operation of the distribution network, as a key component of the power system, is crucial to ensuring the smooth development of the social economy and the normal electricity consumption of residents. However, due to various reasons such as equipment aging, environmental factors, and load fluctuations, the distribution network is often faced with various failure risks. In order to effectively respond to these challenges, distribution network fault warning technology has emerged. Through real-time monitoring and data analysis of the operating status of the distribution network, potential fault hazards can be discovered in a timely manner, and accurate warning information can be provided to operation and maintenance personnel, thereby greatly reducing the probability of failure and reducing the impact of power outages on society and residents.

[0003] The rapid development and widespread application of new energy technologies, especially the access to renewable energy such as wind and solar energy, has led to profound changes in the structure and operating characteristics of distribution networks. New energy power generation is intermittent and uncertain, and the fluctuation of its output power will have a significant impact on the stable operation of the distribution network. In order to effectively respond to the challenges brought by new energy power generation, distribution network fault warning needs to be integrated with new energy prediction technology. Through accurate prediction of new energy power generation, the early warning system can more accurately assess the load and stability of the distribution network, timely detect potential fault risks, and provide scientific decision-making basis for operation and maintenance personnel, ensuring that the distribution network can still operate safely and stably under the high proportion of new energy access. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and to propose a distribution network fault early warning method taking into account the uncertainty of the operating state, so as to achieve early warning of distribution network faults and provide a basis for fault recovery of the distribution network.

[0005] The present invention adopts the following technical solutions:

[0006] A distribution network fault early warning method considering the uncertainty of operating state comprises the following steps:

[0007] Based on the historical and predicted data of distributed photovoltaic and load, calculate the operating parameters of the distribution system and the difference between the operating parameters in two cases;

[0008] Determine the time-period probability distribution model of parameter differences, and generate an operating parameter set based on the probability distribution model;

[0009] According to the calculated parameters, a future operating parameter prediction model based on LSTM is constructed, and the operating parameter set is superimposed to obtain the prediction results of the distribution network operation status, and fault warnings are given to the distribution network based on the superimposed results.

[0010] Furthermore, the various operating parameters of the power distribution system include the active power and reactive power of each branch and the voltage and phase difference of each node.

[0011] Furthermore, the time-division probability distribution model of the difference is obtained by a kernel density estimation method, and the time correlation of different time periods is constructed by using a covariance matrix.

[0012] Furthermore, when the operating parameter set is generated based on the probability distribution model, the large number of generated scenarios are clustered by the K-means algorithm to obtain the optimal possible k-type scenarios S i ,i=1,2,...,k.

[0013] Furthermore, according to the calculated parameters, a future operation parameter prediction model based on LSTM is constructed, and the operation parameter set is superimposed to obtain the prediction result of the distribution network operation situation. The method for fault warning of the distribution network according to the superimposed result is as follows:

[0014] Through the time series prediction method, the distribution network operation parameters are predicted to obtain the predicted value scenario Then, superimpose the scenario S with the running parameter set i Get the corrected scene

[0015] By judging Whether the active power of the distribution network branch exceeds the threshold value, a fault warning is issued for the distribution network.

[0016] Furthermore, when considering the time correlation of different time periods, the entire operating parameter difference can be regarded as a random vector Z = (Z1, Z1, ..., Z T ),T=24, assuming that the random vector Z obeys the multivariate standard normal distribution Z~N(0,∑), and the covariance matrix is:

[0017]

[0018] Among them, σ n,m = cov(Z n ,Z m ) is the random variable Z n and Z m The covariance between , where n, m = 1, 2, ..., T; ∑ is a positive definite matrix and its diagonal elements are all 1;

[0019] Random variable Z n and Zm The correlation coefficient between them is:

[0020]

[0021] In the formula, σ n and σ m are random variables Z n and Z m Variance; Since Z follows a multivariate standard normal distribution, σ n =σ m =1, so ρ n,m = cov(Z n ,Z m )=σ n,m , that is, the random variable Z n and Z m The covariance and correlation of Z are equivalent, so we only need to calculate the covariance matrix ∑ of Z to get the correlation between the random variables.

[0022] The distribution network fault early warning device considering the uncertainty of operation status includes:

[0023] An operating parameter difference acquisition module is used to calculate various operating parameters of the distribution system based on the historical data and predicted data of distributed photovoltaic and load, and calculate the difference of operating parameters under two conditions;

[0024] An operation parameter set acquisition module is used to determine a time period probability distribution model of parameter differences and generate an operation parameter set based on the probability distribution model;

[0025] The distribution network fault warning module is used to build a future operation parameter prediction model based on LSTM according to the calculated parameters, and superimpose the operation parameter set to obtain the prediction result of the distribution network operation situation, and issue a fault warning for the distribution network according to the superimposed result.

[0026] A non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the distribution network fault early warning method considering the uncertainty of operating status as described above.

[0027] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the distribution network fault early warning method taking into account the uncertainty of the operating state as described above is implemented.

[0028] The beneficial technical effects of the present invention are:

[0029] 1. The present invention can capture the changing rules of the system's operating parameters in different time periods by using a time-period probability distribution model that statistically calculates the difference in operating parameters. The operating parameter set generated based on this probability distribution model provides rich input data for the prediction model, thereby enhancing the accuracy and robustness of the model's prediction of future situations.

[0030] 2. Use the long short-term memory neural network (LSTM) to build a future operating parameter prediction model, and superimpose the operating parameter set to achieve accurate prediction of the distribution network operation status. The LSTM model can capture the long-term dependencies in time series data and is suitable for processing systems such as distribution systems with complex dynamic characteristics. Through this step, the future operation status of the distribution network can be perceived and predicted in advance, providing a strong guarantee for the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A block diagram of a fault warning method provided by an embodiment of the present invention;

[0032] Figure 2 A schematic diagram of a neural unit of an LSTM provided in an embodiment of the present invention;

[0033] Figure 3 Schematic diagram of the LSTM prediction model structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following is a further clear and complete description of a method and device for clearing a power capacity market taking into account flexibility provided by the present invention in conjunction with the accompanying drawings:

[0035] Example 1

[0036] A distribution network fault early warning method considering the uncertainty of operating state comprises the following steps:

[0037] Based on the historical data and predicted data of distributed photovoltaic and load, various operating parameters of the distribution system are calculated, and the difference between the operating parameters under two conditions is calculated; the various operating parameters of the distribution system include the active power and reactive power of each branch and the voltage and phase difference of each node;

[0038] Determine the time-period probability distribution model of parameter differences, and generate an operating parameter set based on the probability distribution model;

[0039] According to the calculated parameters, a future operating parameter prediction model based on LSTM is constructed, and the operating parameter set is superimposed to obtain the prediction results of the distribution network operation status, and fault warnings are given to the distribution network based on the superimposed results.

[0040] Based on the historical data and forecast data of distributed photovoltaic and load, the operating parameters of the distribution system are calculated, and the method for calculating the difference between the operating parameters in the two cases is as follows:

[0041] 101: Collect the historical actual and predicted data of distributed photovoltaic and load, and use formula (1) (2) to calculate the operating parameters of the distribution: the active power and reactive power of each branch, the voltage and phase difference of each node; the distribution network parameter results calculated using the actual value and the predicted value are respectively expressed as vectors Y and express;

[0042] 102: The historical actual and predicted data and the admittance parameters of the distribution network are introduced, and the power flow model of the distribution network is solved by the Newton-Raphson method, that is, equations (1) and (2), to obtain the active power and reactive power of each branch, the voltage and phase difference of each node as described in 101.

[0043]

[0044]

[0045] Where P G,i,t and Q G,i,t P is the active power and reactive power injected into bus i at time t; PV,i,t , Q PV,i,t P is the active power and reactive power injected by the photovoltaic power plant at node i at time t; L,i,t and Q L,i,t are the active power and

[0046]

[0047] 103: Calculate the difference between the actual value and the predicted value of the distribution network parameter:

[0048]

[0049] The method for determining the time-period probability distribution model of parameter differences and generating an operating parameter set based on the probability distribution model is as follows:

[0050] 201: In actual situations, the parameters of random variables are unknown, which requires non-parametric estimation. Kernel density estimation is a type of non-parametric estimation. It is a method that does not require any prior knowledge and studies the distribution characteristics of data completely from data samples.

[0051] The kernel density estimation method does not rely on the pre-set assumptions about the potential distribution form of the data, but directly explores its distribution law based on the intrinsic characteristics of the data sample. Therefore, this method has attracted much attention in the field of statistics. Specifically, kernel density estimation applies a smooth "kernel" function to adapt the observed data samples, and then simulates the probability distribution curve followed by the data. Its core is to use kernel functions such as Gaussian functions, take each data point and its bandwidth as the parameters of the kernel function, generate N kernel functions, and linearly combine these kernel functions to construct an estimation function of the kernel density.

[0052] Assume x1,x2,...,x n For n sample points of independent and identical distribution F, such as active power and reactive power of a branch, let its probability density function be f, and the kernel density estimate is as follows:

[0053]

[0054] Where h is the bandwidth, n is the number of samples, is the kernel function.

[0055] Taking the branch active power difference as an example, the main steps of establishing its time-division probability distribution model are as follows:

[0056] Step 221: Divide the daily historical data of branch active power difference into 24 hours, and exclude several hours in which the historical data are all zero;

[0057] Step 222: Calculate the probability of the active power difference being 0 in each time period;

[0058] Step 2023: Calculate the probability distribution of the data after removing 0 using the kernel density estimation method described above;

[0059] Step 2024: Combine the two probability distribution functions of step 2022 and step 2023 to obtain the segmented probability distribution model at that moment.

[0060] 203: The above steps construct the probability distribution function of the active power difference in different time periods. In the process of forming the scene, the time correlation of different time periods needs to be considered. The present invention uses the covariance matrix to construct its correlation. The entire active power difference can be regarded as a random vector Z = (Z1, Z1, ..., Z T ),T=24, assuming that the random vector Z obeys the multivariate standard normal distribution Z~N(0,∑), and the covariance matrix is:

[0061]

[0062] Among them, σ n,m = cov(Z n ,Zm ) is the random variable Z n and Z m The covariance between , where n,m=1,2,...,T. ∑ is a positive definite matrix and its diagonal elements are all 1.

[0063] Random variable Z n and Z m The correlation coefficient between them is:

[0064]

[0065] In the formula, σ n and σ m are random variables Z n and Z m Since Z follows a multivariate standard normal distribution, σ n =σ m =1, so ρ n,m = cov(Z n ,Z m )=σ n,m , that is, the random variable Z n and Z m The covariance and correlation of Z are equivalent, so we only need to calculate the covariance matrix ∑ of Z to get the correlation between the random variables.

[0066] 204: By considering the time correlation, the difference generalization density function of the historical data parameters of the distribution network is sampled, so as to obtain the curve of the difference of the distribution network operation parameters in different time periods. In the present invention, each curve is called a scene.

[0067] 205: Cluster the large number of generated scenarios using the K-means algorithm to obtain the most likely k-type scenarios S i , i=1,2,...,k, the K-means algorithm does not belong to the content protected by the present invention and is a mature content, so the present invention will not describe it in detail.

[0068] According to the calculated parameters, a future operation parameter prediction model based on LSTM is constructed, and the operation parameter set is superimposed to obtain the prediction result of the distribution network operation situation. The method for fault warning of the distribution network according to the superposition result is as follows:

[0069] 301:The LSTM recurrent module has a unique structure, which is completely different from the structure of a single unit. Each LSTM module contains four components, which cooperate with each other in a highly specialized way.

[0070] In the LSTM structure, the first component is called the forget gate, as shown in the figure Figure 2This component uses the sigmoid function to decide which information to discard based on the output information at the previous moment and the input value at the current moment. Its mathematical expression is:

[0071] f t =σ(W f [h t-1 ,x t ]+b f ) (6)

[0072] The second component is the input gate. Figure 2 As shown in ② in the figure, the function of this department is to determine which information needs to be updated. The output value of the input gate is selected and filtered by the sigmoid and tanh functions respectively. The expression of the information that needs to be updated is as follows.

[0073] After the σ function:

[0074] i t =σ(W i [h t-1 ,x t ]+b i ) (7)

[0075] After the tanh function:

[0076]

[0077] In the figure, it is marked Figure 2 The third part is the information update link. This step updates the cell state by multiplying the forget gate with the information update value of the previous moment and adding the result to the product of the two values ​​of the output gate. The specific update formula is:

[0078] C t =f t C t-1 +i t C t (9)

[0079] The fourth component is the output gate, whose output value depends on the results of all the previous steps. This process first uses the sigmoid function to filter out some of the information in the input gate; then, the tanh function is used to filter and filter the updated information; finally, the two filtered information are multiplied to obtain the cell output value at the current moment. The mathematical expression of this process is as follows.

[0080] After the σ function:

[0081] o t =σ(W o [h t-1 ,x t]+b o ) (10)

[0082] h t =o t tanh(C t ) (11)

[0083] In the above formulas, x t is the input value at time t, h t-1 is the output value at time t-1, C t Update value for cell state, W [] is the weight value, is the bias value.

[0084] 302: The above is the model of the neural unit of LSTM. In terms of the overall structure, it adopts the form of input, hidden and output layers, such as Figure 3 As shown, taking the active power of the distribution network branch as an example, the details are as follows:

[0085] Input layer: predict the actual value of the active power of the distribution network branch M times before the time t;

[0086] Output layer: predict the value of active power of distribution network branch N time after prediction time t;

[0087] Hidden layer: contains L neural units;

[0088] 303: Use the time series prediction method to predict the active power of the distribution network branch and obtain the predicted value scenario Then, superimpose the scenario S with the running parameter set i Get the corrected scene

[0089] By judging Whether the active power of the distribution network branch exceeds the threshold value, a fault warning is issued for the distribution network.

[0090] Example 2

[0091] This embodiment provides a distribution network fault early warning device taking into account the uncertainty of the operating state, including:

[0092] An operating parameter difference acquisition module is used to calculate various operating parameters of the distribution system based on the historical data and predicted data of distributed photovoltaic and load, and calculate the difference of operating parameters under two conditions;

[0093] An operation parameter set acquisition module is used to determine a time period probability distribution model of parameter differences and generate an operation parameter set based on the probability distribution model;

[0094] The distribution network fault warning module is used to build a future operation parameter prediction model based on LSTM according to the calculated parameters, and superimpose the operation parameter set to obtain the prediction result of the distribution network operation situation, and issue a fault warning for the distribution network according to the superimposed result.

[0095] A non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the distribution network fault early warning method considering the uncertainty of operating status as described above.

[0096] Furthermore, the present invention adopts the following technical solutions:

[0097] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the distribution network fault early warning method taking into account the uncertainty of the operating state as described above is implemented.

[0098] Through the description of the above embodiments, it can be clearly understood by those skilled in the art that the facilities of the present invention can be implemented by means of software plus the necessary general hardware platform. The embodiments of the present invention can be implemented using an existing processor, or by a dedicated processor used for this purpose or other purposes for an appropriate system, or by a hard-wired system. The embodiments of the present invention also include a non-transitory computer-readable storage medium, which includes a machine-readable medium for carrying or having a machine-executable instruction or data structure stored thereon; such a machine-readable medium can be any available medium that can be accessed by a general or special-purpose computer or other machine with a processor. For example, such a machine-readable medium can include RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the required program code in the form of machine-executable instructions or data structures, and can be accessed by a general or special-purpose computer or other machine with a processor. When information is transmitted or provided to a machine via a network or other communication connection (hard wiring, or wireless, or a combination of hard wiring and wireless), the connection is also considered a machine-readable medium.

[0099] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A distribution network fault early warning method considering the uncertainty of operating status, characterized in that: The method comprises the following steps: Based on the historical and predicted data of distributed photovoltaic and load, calculate the operating parameters of the distribution system and the difference between the operating parameters in two cases; Determine the time-period probability distribution model of parameter differences, and generate an operating parameter set based on the probability distribution model; According to the calculated parameters, a future operating parameter prediction model based on LSTM is constructed, and the operating parameter set is superimposed to obtain the prediction results of the distribution network operation status, and fault warnings are given to the distribution network based on the superimposed results.

2. The distribution network fault early warning method considering the uncertainty of the operating state according to claim 1 is characterized in that: The operating parameters of the power distribution system include the active power and reactive power of each branch and the voltage and phase difference of each node.

3. The distribution network fault early warning method considering the uncertainty of the operating state according to claim 1 is characterized in that: The time-division probability distribution model of the difference is obtained by a kernel density estimation method, and the time correlation of different time periods is constructed by using a covariance matrix.

4. The distribution network fault early warning method considering the uncertainty of the operating state according to claim 1 is characterized in that: When the operating parameter set is generated based on the probability distribution model, the large number of generated scenarios are clustered by the K-means algorithm to obtain the optimal possible k-type scenarios S i ,i=1,2,...,k.

5. The distribution network fault early warning method considering the uncertainty of the operating state according to claim 1 is characterized in that: According to the calculated parameters, a future operation parameter prediction model based on LSTM is constructed, and the operation parameter set is superimposed to obtain the prediction result of the distribution network operation situation. The method for fault warning of the distribution network according to the superposition result is as follows: Through the time series prediction method, the distribution network operation parameters are predicted to obtain the predicted value scenario Then, superimpose the scenario S with the running parameter set i Get the corrected scene By judging whether S exceeds the threshold value of the active power of the distribution network branch, a fault warning is issued for the distribution network.

6. The distribution network fault early warning method considering the uncertainty of the operating state according to claim 1 is characterized in that: When considering the time correlation of different time periods, the entire operating parameter difference can be regarded as a random vector Z = (Z1, Z1, ..., Z T ),T=24, assuming that the random vector Z obeys the multivariate standard normal distribution Z~N(0,∑), and the covariance matrix is: Among them, σ n,m = cov(Z n ,Z m ) is the random variable Z n and Z m The covariance between , where n, m = 1, 2, ..., T; ∑ is a positive definite matrix and its diagonal elements are all 1; Random variable Z n and Z m The correlation coefficient between them is: In the formula, σ n and σ m are random variables Z n and Z m Variance; Since Z follows a multivariate standard normal distribution, σ n =σ m =1, so ρ n,m = cov(Z n ,Z m )=σ n,m , that is, the random variable Z n and Z m The covariance and correlation of Z are equivalent, so we only need to calculate the covariance matrix ∑ of Z to get the correlation between the random variables.

7. A distribution network fault warning device considering the uncertainty of the operating state, characterized in that: include: An operating parameter difference acquisition module is used to calculate various operating parameters of the distribution system based on the historical data and predicted data of distributed photovoltaic and load, and calculate the difference of operating parameters under two conditions; An operation parameter set acquisition module is used to determine a time period probability distribution model of parameter differences and generate an operation parameter set based on the probability distribution model; The distribution network fault warning module is used to build a future operation parameter prediction model based on LSTM according to the calculated parameters, and superimpose the operation parameter set to obtain the prediction result of the distribution network operation situation, and issue a fault warning for the distribution network according to the superimposed result.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the distribution network fault early warning method taking into account the uncertainty of the operating state as described in any one of claims 1 to 6 is implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the distribution network fault early warning method considering the uncertainty of the operating state as described in any one of claims 1 to 6 is implemented.

Images