Parallel markov chain monte carlo resampling method for power system risk detection

Abstract

The application discloses a power system risk detection method and device based on parallel Markov chain Monte Carlo resampling, and the method comprises the following steps: defining a risk index, wherein the risk index comprises a power equipment operation state and a load power loss, and generating a sample variable according to the power equipment operation state and the load power loss; simulating through parallel threads according to a Markov chain Monte Carlo method, obtaining a first sample set and a second sample set based on the sample variable simulation, obtaining a classification distribution parameter based on the first sample set, and obtaining a normal distribution parameter based on the second sample set; obtaining an ideal probability density function according to the classification distribution parameter and the normal distribution parameter; and obtaining a risk index importance sampling expression based on the ideal probability density function. The application shortens the sample collection time and improves the overall calculation efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power system risk assessment, and in particular to a power system risk detection method and device based on parallel Markov chain Monte Carlo resampling. BACKGROUND

[0002] The existing power system risk assessment methods are mainly divided into analytical method and simulation method. The analytical method is generally based on state enumeration, ignoring low-probability events, and only analyzing high-probability events or system states. The simulation method, mainly the Monte Carlo method, collects possible combined states of the system by random sampling, and then analyzes and calculates each system state one by one. Although it can be not limited by the complexity of the system, the calculation efficiency is low. The methods to improve the calculation efficiency of the Monte Carlo method mainly introduce variance reduction techniques, such as auxiliary variable method, stratified sampling and importance sampling.

[0003] The analytical method is more dependent on simplified model assumptions, and the solution speed for complex system problems is slow and the accuracy is not high. The simulation method, mainly the Monte Carlo method, although it can not be limited by the complexity of the system, the calculation efficiency is low. The methods to improve the calculation efficiency of the Monte Carlo method mainly introduce variance reduction techniques, which generally each targets some special problems and has obvious acceleration effect, but there are different degrees of limitations in universality, and the performance of the effect is not very stable. SUMMARY

[0004] The present application aims to at least solve one of the technical problems in the related art to some extent.

[0005] To this end, the purpose of the present application is to shorten the time of collecting training samples of the importance sampling method in the power system risk assessment problem without sacrificing the accuracy, improve the overall calculation efficiency, and propose a power system risk detection method based on parallel Markov chain Monte Carlo resampling.

[0006] Another purpose of the present application is to propose a power system risk detection device based on parallel Markov chain Monte Carlo resampling.

[0007] In order to achieve the above purpose, the present application proposes a power system risk detection method based on parallel Markov chain Monte Carlo resampling, which comprises the following steps:

[0008] Defining a risk index, the risk index comprising power equipment operating state and load power loss, generating a sample variable according to the power equipment operating state and load power loss;

[0009] Simulation is performed using parallel threads based on the Markov chain Monte Carlo method. The simulation yields a first sample set and a second sample set based on the sample variables. The classification distribution parameters are obtained based on the first sample set, and the normal distribution parameters are obtained based on the second sample set.

[0010] The ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters obtained by the parallel thread;

[0011] The importance sampling expression of the risk indicator is obtained based on the ideal probability density function.

[0012] According to an embodiment of the present invention, a power system risk detection method based on parallel Markov chain Monte Carlo resampling is proposed. This method defines risk indicators, including the operating status of power equipment and load power loss, and generates sample variables based on these indicators. Simulation is performed using a Markov chain Monte Carlo method with parallel threads. A first sample set and a second sample set are obtained based on the sample variables. Classification distribution parameters are obtained based on the first sample set, and normal distribution parameters are obtained based on the second sample set. An ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters. Finally, a sampling expression for the importance of the risk indicators is obtained based on the ideal probability density function. This invention shortens the sample collection time and improves the overall computational efficiency.

[0013] Furthermore, the power system risk detection method based on parallel Markov chain Monte Carlo resampling according to the above embodiments of the present invention also includes:

[0014] Furthermore, the operating status of the power equipment includes the operating status of the generator set, transformer, and transmission line, and the load power loss includes the operating power of wind power, photovoltaic power, and the load with continuous characteristics.

[0015] Furthermore, the defined risk indicator function includes:

[0016] Let R represent the risk index of the power system, defined as follows:

[0017]

[0018] Where, x: For the real number field An m-dimensional discrete random variable, y: For the real number field an n-dimensional continuous random variable Let f(x,y) represent the effective value space of the risk indicator R for x and y. Let S(x,y) be the joint probability density of x and y. This represents the system loss metric function.

[0019] Further, the simulation according to the Markov Chain Monte Carlo method is performed by parallel threads, and the first sample set and the second sample set are obtained based on the simulation of the sample variables, comprising:

[0020] defining a proportional function and initializing a set of samples;

[0021] simulating the first sample variable based on the proportional function and the initialized samples to obtain the first sample set;

[0022] simulating the second sample variable based on the proportional function and the initialized samples to obtain the second sample set.

[0023] Further, the defining a proportional function and initializing a set of samples comprises:

[0024] defining

[0025]

[0026] wherein κ i (x,y): is a proportional function of an unknown probability density, and the unknown probability density function is denoted as is a positive integer, representing the number of threads of parallel simulation;

[0027] initializing a set of samples {x0,y0}, and the initialized samples need to satisfy κ i (x0,y0)≠0.

[0028] Further, the simulating the first sample variable based on the proportional function and the initialized samples to obtain the first sample set, and the simulating the second sample variable based on the proportional function and the initialized samples to obtain the second sample set, comprises:

[0029] adopting a categorical distribution as a general expression of the probability distribution model of the first sample variable, denoted as:

[0030]

[0031] wherein is a vector composed of all possible values of x k , and P k =[P 1,k ,...,P z,k ,...P t,k is a vector composed of the probabilities corresponding to each value;

[0032] Based on the general expression of the first sample variable probability distribution model, the conditional probability distribution of the first sample variable is derived from the proportion function, denoted as:

[0033]

[0034] wherein x \k,j =[x 1,j ,...,x k-1,j ,x k+1,j ,x m,j ] represents the jth known sample of other random variables except x k,i

[0035] Based on the conditional probability distribution of the first sample variable, the first sample set is generated according to the initialization sample, denoted as M represents a preset sampling number;

[0036] The probability density equal-probability transformation of the second sample variable is converted into an independent normal distribution, denoted as:

[0037] y k ~N(y;μ k ,σ k )

[0038] wherein, μ k and σ k represent the mean and standard deviation of the normal distribution, respectively, which are known parameters;

[0039] The risk indicator function f(x,y) is expanded, and the general expression of the first sample variable probability distribution model is substituted into the risk indicator function to obtain a transformed form of the risk indicator function:

[0040]

[0041] wherein S'(x,y) represents the function form of S(x,y) after the equal-probability transformation of f(x,y), represents the new value space of y after the equal-probability transformation,

[0042] Based on the transformed form of the risk indicator function, the proportion function is transformed into:

[0043]

[0044] and further obtains a simplified form of the proportion function:

[0045] ​

[0046] Based on the simplified form of the ratio function, the second sample set is generated according to the initialization sample, denoted as M represents a preset sampling number.

[0047] Further, the classification distribution parameter is obtained based on the first sample set, and the normal distribution parameter is obtained based on the second sample set, including:

[0048] Based on the first sample set, the classification distribution parameter is obtained by maximum likelihood estimation, denoted as

[0049] Based on the second sample set, the normal distribution parameter is obtained by maximum likelihood estimation, denoted as And

[0050] Further, the ideal probability density function is obtained according to the classification distribution parameter and the normal distribution parameter obtained by the parallel thread, including:

[0051] The ideal probability density function is defined, denoted as:

[0052]

[0053] The classification distribution and the normal distribution are substituted into the ideal probability density function to obtain a transformed form of the ideal probability density function:

[0054]

[0055] Wherein,

[0056] Further, the risk indicator importance sampling expression is obtained based on the ideal probability density function, including:

[0057] M samples are collected according to the transformed form of the ideal probability density function by using the Monte Carlo method, and a set IS

[0058] According to the set An expression for importance sampling estimation of the transformed form of the risk indicator function is generated:

[0059]

[0060] To achieve the above purpose, another aspect of the present application provides a power system risk detection device based on parallel Markov chain Monte Carlo resampling, including:

[0061] ​A definition module is used to define risk indicators, which include the operating status of power equipment and load power loss, and to generate sample variables based on the operating status of power equipment and load power loss.

[0062] The parameter acquisition module is used to perform simulation using the Markov chain Monte Carlo method through parallel threads, obtain a first sample set and a second sample set based on the sample variables, obtain classification distribution parameters based on the first sample set, and obtain normal distribution parameters based on the second sample set.

[0063] The parameter substitution module is used to obtain the ideal probability density function based on the classification distribution parameters and the normal distribution parameters obtained by the parallel thread;

[0064] An expression construction module is used to obtain a sampling expression for the importance of risk indicators based on the ideal probability density function.

[0065] This invention discloses a power system risk detection device based on parallel Markov chain Monte Carlo resampling. By defining risk indicators, including the operating status of power equipment and load power loss, sample variables are generated based on these indicators. Simulation is performed using the Markov chain Monte Carlo method through parallel threads. A first sample set and a second sample set are obtained based on the sample variables. Classification distribution parameters are obtained from the first sample set, and normal distribution parameters are obtained from the second sample set. An ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters. Finally, a sampling expression for the importance of the risk indicators is obtained based on the ideal probability density function. This invention shortens the sample collection time and improves overall computational efficiency.

[0066] The beneficial effects of this invention are:

[0067] This invention allocates the same training sample requirements to parallel computing resources, shortening the time for collecting training samples in power system risk assessment problems without sacrificing accuracy, and improving overall computational efficiency.

[0068] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0069] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0070] Figure 1 This is a flowchart of a power system risk detection method based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention;

[0071] Figure 2 This is a flowchart illustrating the power system risk detection method based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention.

[0072] Figure 3(a) is a schematic diagram comparing the computation time of parallel computation of the EENS of the power generation system according to an embodiment of the present invention;

[0073] Figure 3(b) is a schematic diagram comparing the evaluation values ​​of the parallel calculation of the EENS of the power generation system according to an embodiment of the present invention;

[0074] Figure 4 This is a schematic diagram of a power system risk detection device based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention. Detailed Implementation

[0075] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0076] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0077] The following describes, with reference to the accompanying drawings, a power system risk detection method and apparatus based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention. First, the power system risk detection method based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention will be described with reference to the accompanying drawings.

[0078] Figure 1 This is a flowchart of a power system risk detection method based on parallel Markov chain Monte Carlo resampling, according to an embodiment of the present invention.

[0079] like Figure 1 As shown, the power system risk detection method based on parallel Markov chain Monte Carlo resampling includes the following steps:

[0080] Step S1: Define risk indicators, which include the operating status of power equipment and load power loss. Generate sample variables based on the operating status of power equipment and load power loss.

[0081] Specifically, risk indicators are defined, including:

[0082] Let x: For the real number field An m-dimensional discrete random variable, y: For the real number field An n-dimensional continuous random variable. In general problems of power system risk assessment, x can represent the operating status of components such as generating units, transformers, and transmission lines, such as x i =1 indicates that the unit is operating normally, x i =0 represents the unit being taken out of operation. y can represent the continuous operating power of wind power, photovoltaic power, and loads. Assuming the risk index of the power system is represented by R, it can be uniformly defined using equation (1).

[0083]

[0084] in, Let f(x,y) represent the effective value space of the risk indicator R for x and y. Let S(x,y) be the joint probability density of x and y. This represents the system loss metric function. The importance sampling method (re-sampling) is used to estimate R, aiming to find a probability density g that is as close as possible to the ideal probability density defined in equation (2). idea (x).

[0085]

[0086] Step S2 involves performing simulations using the Markov chain Monte Carlo method in parallel threads. Based on the sample variables, a first sample set and a second sample set are obtained. The classification distribution parameters are obtained from the first sample set, and the normal distribution parameters are obtained from the second sample set. Specifically, the parallel simulation process includes:

[0087] definition

[0088]

[0089] Among them, κ i (x,y): The proportional function represents a certain unknown probability density, and we first denote this unknown probability density as... κ is a positive integer representing the number of threads in the parallel simulation. These parallel threads can be multiple threads on a single computer or multiple threads across multiple computers. The Markov chain Monte Carlo method is used to analyze κ. i We sample from (x, y), assuming the preset number of samples is M, and denote the resulting sample set as... By applying maximum likelihood estimation, the estimate made in the embodiments of the present invention can be obtained. For the case where x and y are independent, the following are embodiments of the present invention. The detailed principles and steps of the simulation method.

[0090] First, because x and y are independent, It can be broken down into and For clarity, vector x i,j and y i,j Expand separately, i.e., x i,j =[x 1,i,j ,...,x m,i,j ] and y i,j =[y 1,i,j ,...,y n,i,j First, for The simulation principle is explained. According to the definition of equation (3), due to the κ defined by each parallel thread i (x, y) are all the same. Unless otherwise stated, the subscript "i" is omitted in the following content. Since each element of x is a discrete random variable, a categorical distribution can be used as the general expression for its probability distribution model, denoted as .

[0091]

[0092] in, Represents x k P is a vector consisting of all possible values. k =[P 1,k ,...,P z,k ,...P t,k The vector [R] is composed of the probabilities corresponding to each value. According to the definition of the risk index R, and P k All conditions are known. Therefore, based on equation (4), the x required by the Markov chain Monte Carlo method can be derived from equation (3). k Conditional probability distribution, denoted as Where, x \k,j =[x 1,j ,...,x k-1,j ,x k+1,j ,x m,j ] represents the division by x k,i The j-th known sample of other random variables besides the one mentioned above. The detailed expression is as follows:

[0093]

[0094] in, Equation (5) is the given sample {x} j ,yj When x is generated, j+1 The conditional probability density. Next, let's look at... The simulation principle will be explained. First, considering that any continuous probability density can be transformed into an independent normal distribution through equal probability transformation, such as... Where, μ k and σ k and represent the mean and standard deviation of the normal distribution, respectively, and are known parameters. Based on this, f(x,y) in equation (1) can be expanded, and equation (4) can be substituted into equation (1). Then, an equal probability transformation is applied to the probability distribution of the continuous variable y, ultimately yielding...

[0095]

[0096] Where S′(x,y) represents the functional form of f(x,y) after S(x,y) has been modified accordingly after undergoing an equal probability transformation. This represents the new space of values ​​for y after an equal probability transformation. Based on the transformation of equation (6), equation (3) will be changed accordingly to

[0097]

[0098] The conditional probability distribution of y can be obtained according to equation (7), taking into account... Sampling can be performed according to equation (5) and is independent of equation (7), therefore equation (7) can be further simplified to

[0099]

[0100] Equation (8) is The proportional function of the probability density it follows can be generated using the Metropolis–Hastings algorithm. In this embodiment of the invention, the required proposal density is defined as a normal distribution N(y k,j |y k,j-1 0.08y k,j-1 To generate from equations (5) and (8) According to the sampling principle of Markov chain Monte Carlo, a set of samples {x0, y0} needs to be initialized first before all subsequent samples can be generated iteratively, provided that κ is satisfied. i (x0, y0) ≠ 0, so any initialization value can be used. Based on Assuming that it follows a categorical distribution, the parameters of the categorical distribution are obtained through maximum likelihood estimation. based on Assuming it follows a normal distribution, the parameters of the normal distribution are obtained through maximum likelihood estimation. and

[0101] The above The simulation and parameter estimation principles are applicable to every Furthermore, they are independent of each other and can be obtained through parallel computing.

[0102] Step S3: Obtain the ideal probability density function based on the classification distribution parameters and normal distribution parameters obtained by the parallel thread.

[0103] Specifically, based on the results obtained in step S2 and The g produced by the embodiment of the present invention is constructed using equation (9). idea The (x,y) estimate is denoted as...

[0104]

[0105] Specifically, the categorical distribution of x and the normal distribution of y obtained from parallel computation are substituted into (9), that is,

[0106]

[0107] in, Step S4: Obtain the sampling expression for the importance of risk indicators based on the ideal probability density function.

[0108] Specifically, the Monte Carlo method is used from M collected in the middle IS The set consists of samples. Finally, the final expression for importance sampling estimation of R is:

[0109]

[0110] Figure 2 This is a flowchart illustrating the power system risk detection method based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention, describing the steps outlined above. As an example, the example analysis system and initialization instructions of this embodiment are as follows:

[0111] Based on the publicly available IEEE-RTS79 standard test system, the beneficial effects of the embodiments of the present invention are explained below in conjunction with the reliability assessment of power generation systems. First, the Expected Energy Shortage (EENS) is selected as the risk indicator to be assessed. The random operating states of all generators are described using a Bernoulli distribution, where x... k =1 represents that unit k is operating normally, with a state probability of p. k ;xk =0 represents a fault in unit k, with a state probability of 1-p. k The capacity of unit k is represented by the symbol U. k This indicates that the load demand is assumed to follow a normal distribution, such that for the j-th load, y j ~N(μ) j ,σ j ), where parameter μ j Equal to the load peak in the original data file, σ j =5%μ j When initializing x0, set the status of all units to "0" and y0 to the average of all loads.

[0112] When the risk index in equation (1) is determined to be EENS, for the EENS assessment problem of power generation systems,

[0113] Furthermore, the specific expression of formula (5) is as follows:

[0114]

[0115] in,

[0116] The specific expression of formula (8) is as follows:

[0117]

[0118] To illustrate the effects and advantages of the embodiments of the present invention, the algorithm was developed using Julia (ver 1.7.2), and the hardware platform was an Intel(R) Core(TM) i9-10885H with 32GB RAM, and the operating system was Windows 10. M was determined by equation (11) with a variance coefficient not exceeding 0.01. IS The basis for the load size, using the ordinary Monte Carlo method and the EENS evaluation results of this invention, is shown in Figure 3. In the figure, the load size is expressed in per-unit values, with a baseline value of 100MW and N. par =1 represents the evaluation result of a standard Monte Carlo simulation. As can be clearly seen from Figure 3(a), in this embodiment of the invention, N... par When N=3, the overall computation time is minimized, only 1 / 6 of that of the ordinary Monte Carlo method. Furthermore, judging from the EENS evaluation values ​​in Figure 3(b), if the ordinary Monte Carlo method is used as the baseline, the embodiments of the present invention achieve the best N values ​​across various N values. par Under these values, the maximum relative error of the given evaluation value is only 0.11%.

[0119] Through the above steps, risk indicators are defined, including the operating status of power equipment and load power loss. Sample variables are generated based on the operating status of power equipment and load power loss. Simulation is performed using a Markov chain Monte Carlo method with parallel threads. Based on the sample variables, a first sample set and a second sample set are obtained. Classification distribution parameters are obtained based on the first sample set, and normal distribution parameters are obtained based on the second sample set. An ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters. Finally, a sampling expression for the importance of the risk indicators is obtained based on the ideal probability density function. This invention shortens the sample collection time and improves the overall computational efficiency.

[0120] To achieve the above embodiments, such as Figure 4 As shown, this embodiment also provides a power system risk detection device 10 based on parallel Markov chain Monte Carlo resampling. The device 10 includes: a definition module 100, a parameter acquisition module 200, a parameter substitution module 300, and an expression construction module 400.

[0121] Module 100 is used to define risk indicators, which include the operating status of power equipment and load power loss. Sample variables are generated based on the operating status of power equipment and load power loss.

[0122] The parameter acquisition module 200 is used to perform simulations using the Markov chain Monte Carlo method through parallel threads, obtain a first sample set and a second sample set based on the sample variables, obtain classification distribution parameters based on the first sample set, and obtain normal distribution parameters based on the second sample set.

[0123] The parameter input module 300 is used to obtain the ideal probability density function based on the classification distribution parameters and normal distribution parameters obtained by the parallel thread.

[0124] Expression construction module 400 is used to obtain the sampling expression for the importance of risk indicators based on the ideal probability density function.

[0125] The power system risk detection device based on parallel Markov chain Monte Carlo resampling according to an embodiment of the present invention defines risk indicators, including the operating status of power equipment and load power loss, and generates sample variables based on the operating status of power equipment and load power loss. Simulation is performed using a Markov chain Monte Carlo method through parallel threads, and a first sample set and a second sample set are obtained based on the sample variables. Classification distribution parameters are obtained based on the first sample set, and normal distribution parameters are obtained based on the second sample set. An ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters. Finally, a sampling expression for the importance of risk indicators is obtained based on the ideal probability density function. This invention shortens the sample collection time and improves the overall computational efficiency.

[0126] It should be noted that the foregoing explanation of the power system risk detection method based on parallel Markov chain Monte Carlo resampling also applies to the power system risk detection device based on parallel Markov chain Monte Carlo resampling in this embodiment, and will not be repeated here.

[0127] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0128] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0129] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A power system risk detection method based on parallel Markov chain Monte Carlo resampling, characterized in that, include: Define risk indicators, which include the operating status of power equipment and load power loss, and generate sample variables based on the operating status of power equipment and load power loss. Simulation is performed using parallel threads based on the Markov chain Monte Carlo method. The simulation yields a first sample set and a second sample set based on the sample variables. The classification distribution parameters are obtained based on the first sample set, and the normal distribution parameters are obtained based on the second sample set. The ideal probability density function is obtained based on the classification distribution parameters and the normal distribution parameters obtained by the parallel thread; The sampling expression for the importance of risk indicators is obtained based on the ideal probability density function. The defined risk indicator function includes: use R Risk indicators for power systems are defined as follows: in, For the real number field within m 3D discrete random variables, For the real number field Inside n 3D continuous random variable, This indicates the definition of risk indicators. R of and Effective value space, for and The joint probability density, y represents the system loss metric function, x represents the operating status of the generator, transformer and transmission line, and y represents the continuous operating power of wind power, photovoltaic power and load. The simulation, performed using the Markov chain Monte Carlo method and parallel threads, obtains a first sample set and a second sample set based on the sample variables, including: Define a scaling function and initialize a set of samples; The first sample set is obtained by simulating the first sample variable based on the scaling function and the initial sample; The second sample set is obtained by simulating the second sample variables based on the aforementioned proportional function and the initial samples. The definition of the scaling function and the initialization of a set of samples include: definition ， in, The proportional function representing the unknown probability density is denoted as . , A positive integer representing the number of threads in the parallel simulation; Initialize a set of samples The initial sample must meet the following requirements: .

2. The method according to claim 1, characterized in that, The process of simulating the first sample set based on the scaling function and the initial samples to obtain the first sample set, and simulating the second sample set based on the scaling function and the initial samples to obtain the second sample set, includes: The general expression for the probability distribution model using the categorical distribution as the first sample variable is denoted as: , in, represent All t A vector consisting of possible values. It is a vector consisting of the probabilities corresponding to each value; Based on the general expression of the probability distribution model of the first sample variable, the conditional probability distribution of the first sample variable is derived from the proportional function, denoted as: in, Representative except Other random variables besides j A known sample, , , , ; Based on the conditional probability distribution of the first sample variables, the first sample set is generated according to the initial samples, denoted as... M represents the preset number of samples; The probability density function of the second sample variable is transformed into an independent normal distribution with equal probability, denoted as: in, , and ... In the risk indicator function Expanding on this, and substituting the general expression of the probability distribution model of the first sample variable into the risk indicator function, we obtain the transformed form of the risk indicator function: in, express After undergoing an equal probability transformation The function form after making the corresponding changes, Represents the result of equal probability transformation The new value space, ; Based on the transformed form of the risk indicator function, the proportional function is transformed as follows: ， ， And further, the simplified form of the proportional function is obtained: ， ； Based on the simplified form of the proportional function, the second sample set is generated according to the initial samples, denoted as... M represents the preset number of samples.

3. The method according to claim 2, characterized in that, The step of obtaining classification distribution parameters based on the first sample set and normal distribution parameters based on the second sample set includes: Based on the first sample set, the classification distribution parameters are obtained through maximum likelihood estimation, denoted as . ; Based on the second sample set, the normal distribution parameters are obtained through maximum likelihood estimation, denoted as . and , .

4. The method according to claim 3, characterized in that, The step of obtaining the ideal probability density function based on the classification distribution parameters and the normal distribution parameters obtained by the parallel thread includes: Define the ideal probability density function, denoted as: ； Substituting the classification distribution and the normal distribution into the ideal probability density function, we obtain the transformed form of the ideal probability density function: in, , , .

5. The method according to claim 4, characterized in that, The method for obtaining the risk indicator importance sampling expression based on the ideal probability density function includes: The Monte Carlo method is used to collect data based on the transformed form of the ideal probability density function. Given samples, construct a set. ; According to the set Generate an expression for importance sampling estimation of the transformed form of the risk indicator function: 。 6. A power system risk detection device based on parallel Markov chain Monte Carlo resampling, characterized in that, include: A definition module is used to define risk indicators, which include the operating status of power equipment and load power loss, and to generate sample variables based on the operating status of power equipment and load power loss. The parameter acquisition module is used to perform simulation using the Markov chain Monte Carlo method through parallel threads, obtain a first sample set and a second sample set based on the sample variables, obtain classification distribution parameters based on the first sample set, and obtain normal distribution parameters based on the second sample set. The parameter substitution module is used to obtain the ideal probability density function based on the classification distribution parameters and the normal distribution parameters obtained by the parallel thread; The expression construction module is used to obtain the risk indicator importance sampling expression based on the ideal probability density function; The defined risk indicator function includes: use R Risk indicators for power systems are defined as follows: in, For the real number field within m 3D discrete random variables, For the real number field Inside n 3D continuous random variable, This indicates the definition of risk indicators. R of and Effective value space, for and The joint probability density, y represents the system loss metric function, x represents the operating status of the generator, transformer and transmission line, and y represents the continuous operating power of wind power, photovoltaic power and load. The simulation, performed using the Markov chain Monte Carlo method and parallel threads, obtains a first sample set and a second sample set based on the sample variables, including: Define a scaling function and initialize a set of samples; The first sample set is obtained by simulating the first sample variable based on the scaling function and the initial sample; The second sample set is obtained by simulating the second sample variables based on the aforementioned proportional function and the initial samples. The definition of the scaling function and the initialization of a set of samples include: definition ， in, The proportional function representing the unknown probability density is denoted as . , A positive integer representing the number of threads in the parallel simulation; Initialize a set of samples The initial sample must meet the following requirements: .

Images