New energy and load sample sampling method

Abstract

The invention relates to a new energy and load sample sampling method, which comprises the following steps of: firstly, constructing a Beta distribution model and a nonparametric kernel density estimation model of new energy, and drawing a green density curve; then, inputting network structure parameters, probability model information of new energy and load, and mutually independent standard normal random samples, so as to determine variables, variable numbers and correlation coefficients of correlation between the mutually independent standard normal random samples, and setting a sampling scale; then, on the basis of a median Latin hypercube sampling technology and a Cholesky decomposition technology, the mutually independent standard normal random samples are converted into related standard normal random samples, and finally, on the basis of an equal probability conversion principle, the standard normal random samples are obtained. Related standard normal random samples are converted into new energy and load random samples which obey specified distribution and have given correlation coefficients; the method has the advantages of good use effect, rapidness and high efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of power sampling, and in particular relates to a sampling method for new energy and load samples. Background Art

[0002] With the increase in wind and solar penetration rates, the uncertainty of wind and solar output and the peak trend of electricity load have become increasingly prominent, causing a huge impact on the safe and stable operation of the power system; at the same time, the significant climate correlation between wind and solar output and electricity load has led to increased climate sensitivity and vulnerability of the power system; therefore, it is of great significance to build a new power system scenario model with new energy as the main body; at present, domestic and foreign scholars have conducted extensive research on scenario generation methods that consider source and load uncertainty and correlation. In the specific research process, probability analysis usually requires the generation of random samples of variables such as new energy power output power and load; however, for new energy power and load that are correlated, their random samples cannot be generated directly through sampling; therefore, it is necessary to develop a sampling method for new energy and load samples that is effective, fast and efficient. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a method for sampling new energy and load samples that is effective, fast and efficient.

[0004] The object of the present invention is achieved as follows: A method for sampling new energy and load samples comprises the following steps: Step 1: Construct the Beta distribution model and non-parametric kernel density estimation model of new energy, and draw the dark green density curve; Step 2: Input network structure parameters, probability model information of new energy and load, and independent standard normal random samples. Then, determine the variables and number of variables and correlation coefficients that exist between the independent standard normal random samples, and set the sampling scale. Step 3: Based on the median Latin hypercube sampling technique, independent standard normal random variables are sampled to generate standard normal random samples. Then, the standard normal random samples determine the standard normal random vector and the standard normal random sample matrix. Step 4: Convert the correlation coefficients between independent standard normal random sample variables into correlation coefficients between standard normal variables. Then, calculate the covariance matrix of the standard normal random vector from the correlation coefficients. Then, perform Cholesky decomposition on the covariance matrix of the standard normal random vector to obtain a lower triangular matrix. Step 5: Based on the lower triangular matrix obtained in step 4 and the standard normal random sample matrix generated by sampling in step 3, and based on the principle of equal probability conversion, convert the random vectors related to new energy and load that follow the non-standard normal distribution into related standard normal random vectors. Then, the related standard normal random vectors are further converted into independent standard normal random vectors. Step 6: By transforming each element in the relevant standard normal random vector in step 5, a random sample of new energy and load that obeys the specified distribution and has a given correlation coefficient can be obtained.

[0005] Furthermore, the probability density curve in step 1 can be drawn by statistically analyzing the power flow results such as node voltage, line transmission power and line loss rate obtained from each power flow calculation, and using non-parametric kernel density estimation theory to obtain the probability distribution of the above indicators.

[0006] Furthermore, the Latin hypercube sampling technique in step 3 is a generalization of the Latin square in multi-dimensions. Each hyperplane perpendicular to the axis contains at most one sample, thereby ensuring the randomness of the sample.

[0007] The beneficial effects of the present invention are as follows: the present invention first constructs a Beta distribution model and a non-parametric kernel density estimation model for new energy, and draws a dark green density curve; then, inputs network structure parameters, probability model information of new energy and load, and mutually independent standard normal random samples, thereby determining the variables and number of variables and correlation coefficients existing between the mutually independent standard normal random samples, and setting the sampling scale; then, based on the median Latin hypercube sampling technology and the Cholesky decomposition technology, the mutually independent standard normal random samples are converted into correlated standard normal random samples; finally, based on the equal probability conversion principle, the correlated standard normal random samples are converted into new energy and load random samples that obey a specified distribution and have a given correlation coefficient; in this process, the Latin hypercube sampling technology utilizes the generalization of the Latin square in multiple dimensions, wherein each hyperplane perpendicular to the axis contains at most one sample, thereby ensuring the randomness of the sample; the probability density curve can be obtained by statistically analyzing the flow results such as node voltage, line transmission power and line loss rate obtained from each flow calculation, and using the non-parametric kernel density estimation theory to obtain the probability distribution of the above indicators, and then draws it; in general, the present invention has the advantages of good use effect, fastness and high efficiency. DETAILED DESCRIPTION

[0008] The present invention will be further described below.

[0009] Example: A method for sampling new energy and load samples includes the following steps: Step 1: Construct a Beta distribution model and a nonparametric kernel density estimation model for new energy, and draw a dark green density curve. The probability density curve is drawn by statistically analyzing the power flow results such as node voltage, line transmission power, and line loss rate obtained from each power flow calculation, and using the nonparametric kernel density estimation theory to obtain the probability distribution of the above indicators. Step 2: Input network structure parameters, probability model information of new energy and load, and independent standard normal random samples. Then, determine the variables and number of variables and correlation coefficients that exist between the independent standard normal random samples, and set the sampling scale. Step 3: Based on the median Latin hypercube sampling technique, independent standard normal random variables are sampled to generate standard normal random samples. Then, the standard normal random vector and standard normal random sample matrix are determined through the standard normal random samples. The Latin hypercube sampling technique is a generalization of the Latin square in multidimensional space. Each hyperplane perpendicular to the axis contains at most one sample, thus ensuring the randomness of the sample. Step 4: Convert the correlation coefficients between independent standard normal random sample variables into correlation coefficients between standard normal variables. Then, calculate the covariance matrix of the standard normal random vector from the correlation coefficients. Then, perform Cholesky decomposition on the covariance matrix of the standard normal random vector to obtain a lower triangular matrix. Step 5: Based on the lower triangular matrix obtained in step 4 and the standard normal random sample matrix generated by sampling in step 3, and based on the principle of equal probability conversion, convert the random vectors related to new energy and load that follow the non-standard normal distribution into related standard normal random vectors. Then, the related standard normal random vectors are further converted into independent standard normal random vectors. Step 6: By transforming each element in the relevant standard normal random vector in step 5, a random sample of new energy and load that obeys the specified distribution and has a given correlation coefficient can be obtained.

[0010] When the present invention is used, first, a Beta distribution model and a non-parametric kernel density estimation model of new energy are constructed, and a dark green density curve is drawn; then, network structure parameters, probability model information of new energy and load, and mutually independent standard normal random samples are input, thereby determining the variables and the number of variables and correlation coefficients existing between the mutually independent standard normal random samples, and setting the sampling scale; then, based on the median Latin hypercube sampling technology, the mutually independent standard normal random variables are sampled to generate standard normal random samples, and then, the standard normal random vector and the standard normal random sample matrix are determined through the standard normal random samples; Then, the correlation coefficients between the independent standard normal random sample variables are converted into the correlation coefficients between the standard normal variables in turn, and then the covariance matrix of the standard normal random vector is calculated by the correlation coefficient, and finally the covariance matrix of the standard normal random vector is subjected to Cholesky decomposition to obtain the lower triangular matrix; finally, according to the lower triangular matrix and the standard normal random sample matrix, and based on the principle of equal probability conversion, the related random vectors such as new energy and load that obey the non-standard normal distribution are converted into related standard normal random vectors, and then the related standard normal random vectors are further converted into independent standard normal random vectors. After completing the above operations, each element in the relevant standard normal random vector is converted in this way to obtain a random sample of new energy and load that obeys the specified distribution and has a given correlation coefficient; in this process, the Latin hypercube sampling technology is a promotion of the Latin square in multiple dimensions, in which each hyperplane perpendicular to the axis contains at most one sample, thereby ensuring the randomness of the sample, and the probability density curve can be obtained by statistically calculating the node voltage, line transmission power, line loss rate and other flow results of each flow calculation, and using the non-parametric kernel density estimation theory to obtain the probability distribution of the above indicators and draw it; in general, the present invention has the advantages of good use effect and high efficiency.

[0011] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A sampling method for new energy and load samples, characterized in that: The following steps are included: Step 1: Construct the Beta distribution model and non-parametric kernel density estimation model of new energy, and draw the dark green density curve; Step 2: Input network structure parameters, probability model information of new energy and load, and independent standard normal random samples. Then, determine the variables and number of variables and correlation coefficients that exist between the independent standard normal random samples, and set the sampling scale. Step 3: Based on the median Latin hypercube sampling technique, independent standard normal random variables are sampled to generate standard normal random samples. Then, the standard normal random vector and standard normal random sample matrix are determined through the standard normal random samples. Step 4: Convert the correlation coefficients between independent standard normal random sample variables into correlation coefficients between standard normal variables. Then, calculate the covariance matrix of the standard normal random vector from the correlation coefficients. Then, perform Cholesky decomposition on the covariance matrix of the standard normal random vector to obtain a lower triangular matrix. Step 5: Based on the lower triangular matrix obtained in step 4 and the standard normal random sample matrix generated by sampling in step 3, and based on the principle of equal probability conversion, convert the random vectors related to new energy and load that follow the non-standard normal distribution into related standard normal random vectors. Then, the related standard normal random vectors are further converted into independent standard normal random vectors. Step 6: By transforming each element in the relevant standard normal random vector in step 5, a random sample of new energy and load that obeys the specified distribution and has a given correlation coefficient can be obtained.

2. The method for sampling new energy and load samples according to claim 1, wherein: The probability density curve in step 1 can be drawn by statistically analyzing the power flow results such as node voltage, line transmission power and line loss rate obtained from each power flow calculation, and using the non-parametric kernel density estimation theory to obtain the probability distribution of the above indicators.

3. The method for sampling new energy and load samples according to claim 1, wherein: The Latin hypercube sampling technique in step 3 is a generalization of the Latin square in multi-dimensions. Each hyperplane perpendicular to the axis contains at most one sample, thereby ensuring the randomness of the sample.