Load cluster regulation and control domain parameter online identification method and system

Abstract

The invention discloses a load cluster regulation and control domain parameter online identification method, which comprises the following steps: S1, for a load cluster system, obtaining N0 groups of multi-element load physical parameters according to the value range of the multi-element load physical parameters, and generating parameters of each multi-element load in the whole system for each group of multi-element load physical parameters; s2, calculating parameters of the aggregation model through independent simulation to obtain N0 output quantities; s3, acquiring an HDMR model of the load cluster system through N0 groups of input and output samples; s4, obtaining a correlation coefficient to obtain global sensitivity of the multi-element load physical parameters so as to identify important physical parameters; and S5, verifying the HDMR model and storing correlation coefficients. The method can accurately and efficiently identify, understand and provide the characteristics of the load cluster so as to adapt to load changes and improve the regulation and control capability of the system.

Description

Technical Field

[0001] The present invention relates to the field of cluster load regulation, and in particular to an online identification method and system for load cluster regulation domain parameters. Background Art

[0002] At the system level, relevant dispatching departments need to coordinate and optimize different types of resources to ensure safe and stable operation of the power system while improving economic benefits. Therefore, the accuracy of the external characteristic parameters of the load cluster directly affects the system's dispatch results. However, for load clusters, taking the aggregation of multiple load resources as an example, due to factors such as inaccurate power measurement, random user behavior, and user information protection, the parameters of multiple loads are often difficult to obtain. Due to differences in factors such as user preferences and equipment differences, multiple loads have obvious heterogeneity. Establishing a heterogeneous aggregation model for multiple loads without considering actual scenarios makes it difficult to accurately obtain their aggregated power and response potential. At the same time, traditional load model parameters based on equivalent parameter models mostly use reference values from literature and do not consider the time-varying characteristics of the equivalent parameters, resulting in poor model adaptability. Summary of the Invention

[0003] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a method and system for online identification of load cluster control domain parameters, which can accurately and efficiently identify, understand and provide the characteristics of the load cluster to adapt to load changes and improve the system control capability.

[0004] A technical solution to achieve the above object is: an online identification method for load cluster control domain parameters, comprising the following steps:

[0005] S1, for the load cluster system, according to the value range of the multi-load physical parameters, obtain N0 groups of multi-load physical parameters, and generate the parameters of each multi-load in the entire system for each group of multi-load physical parameters;

[0006] S2, calculate the parameters of the aggregation model through independent simulation to obtain N0 output quantities;

[0007] S3, obtains the HDMR model of the load cluster system through the N0 group of input and output samples;

[0008] S4, obtain the correlation coefficient to obtain the global sensitivity of the physical parameters of the multivariate load to identify the important physical parameters;

[0009] S5, verify the HDMR model and store the correlation coefficient.

[0010] Furthermore, in S1, the Quasi Monte Carlo method is applied to generate N0 groups of input samples. The input of the multivariate system can be expressed as:

[0011]

[0012] (s=1,2,…,N0)

[0013] Among them, R_μ (s) is the average resistance of the sth group, C a _μ (s) is the average capacitance of the sth group, P_μ (s) is the average power of the sth group, T set _μ (s) is the average set temperature of the sth group, δ_μ (s) is the average adjustment rate of the sth group, η_μ (s) is the average efficiency of the sth group, R_χ (s) is the resistance variance of the sth group, C a _χ (s) is the capacitance variance of the sth group, P_χ (s) is the power variance of the sth group, T set _χ (s) is the set temperature variance of the sth group, δ_χ (s) is the variance of the regulation efficiency of the sth group, η_χ (s) is the efficiency variance of the sth group, T o (s) is the initial temperature of the sth group.

[0014] Furthermore, in S2, for each input sample, 10,000 multivariate load parameters are generated according to the parameter distribution type, and then the values of the aggregate model parameters are obtained based on the independent simulation calculation results of the 10,000 multivariate loads, and N0 groups of input samples obtain N0 output values.

[0015] Furthermore, the parameter distribution type is specifically divided into normal distribution, lognormal distribution or uniform distribution.

[0016] Furthermore, in S3, after obtaining N0 pairs of input and output values, each pair of values is normalized before being input into the HDMR model, that is:

[0017]

[0018] Where x i ,x * i Represent the actual value and standard value of the input variable respectively; x min ,x max Represents the minimum and maximum values of the input respectively.

[0019] Furthermore, in S3, the specific method of the HDMR model is:

[0020] Assume that the system input is x=(x1,x2,…,x n ) T , the output is f(x), which is equal to the sum of the independent effects of the input variables and the coupling effects between the input variables. Therefore, f(x) can be expressed as:

[0021]

[0022] Where nf is the number of input variables; f0 is the zero-order component function, which is a constant; f i (x i ) is a first-order component function, representing x i Independent effect on output; f ij (x i ,x j ) is a second-order component function, representing x i and x j The effect of coupling between them on the output; 1,2,…,nf (x) is the nf-order component function, which is the influence of the coupling between all input variables on the output.

[0023] Furthermore, in S4, the correlation coefficients α and β are calculated for the multivariate HDMR model established using N0 for the input and output values. The linear coefficient α1 of the first-order component function of the input is calculated according to the following formula:

[0024]

[0025] in,

[0026] The quadratic coefficient α2 of the first-order component function of the input quantity is calculated according to the following formula:

[0027]

[0028] in,

[0029] By analogy, higher-order coefficients of the first-order component function of the input quantity are calculated;

[0030] β represents the effect of the coupling between different input quantities on the output, and is the first-order coefficient of the second-order component function of the input quantity. 11 ,Right now:

[0031]

[0032] By analogy, the higher-order coefficients of the second-order component function of the input quantity are calculated;

[0033] The global sensitivity of the independent and coupled inputs is obtained by the obtained values of α and β.

[0034] Furthermore, after verifying the HDMR model and storing the correlation coefficients, an HDMR model database of spatial aggregation model parameters at different time scales is established.

[0035] Furthermore, when multiple loads participate in demand response, the parameters of the multiple load aggregation model will be in a constantly changing state. By obtaining the physical parameters of the multiple loads after regulation, the physical parameter values of the multiple loads are updated online, and the parameters of the multiple load aggregation model are quickly calculated, the parameters of the multiple load aggregation model can be updated in real time.

[0036] A system for the above-mentioned load cluster control domain parameter online identification method includes an input module, an independent simulation calculation module, an HDMR model module, a global sensitivity calculation module, an HDMR model verification module and an HDMR model database;

[0037] The input module applies the Quasi Monte Carlo method to generate N0 groups of input samples;

[0038] The independent simulation calculation module obtains N0 output values through input samples;

[0039] The HDMR model module obtains the HDMR model of the load cluster system through the N0 group of input and output samples;

[0040] The global sensitivity calculation module obtains the correlation coefficient to obtain the global sensitivity of the physical parameters of the multivariate load to identify the important physical parameters;

[0041] The HDMR model database verifies the HDMR model and stores the correlation coefficients, and establishes an HDMR model database of spatial aggregation model parameters at different time scales.

[0042] The advantages of the present invention are: the present invention analyzes the uncertainty of the physical parameters of multiple loads, uses high-dimensional model expression technology (HDMR) to identify the external characteristic parameters of flexible load clusters at different time scales, and uses high-dimensional space mapping technology to perform online identification of the control domain of flexible load clusters. It can provide more characteristic information, improve data separability, process large-scale data, and discover hidden associations, more accurately and efficiently identify and understand the characteristics of load clusters, and provide more reliable support for the management of flexible loads. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a method flow for online identification of load cluster control domain parameters according to the present invention;

[0044] Figure 2 Schematic diagram of the architecture of an online identification system for load cluster control domain parameters of the present invention. DETAILED DESCRIPTION

[0045] In order to better understand the technical solution of the present invention, the following is a detailed description through specific embodiments:

[0046] A method for online identification of load cluster control domain parameters of the present invention includes the following steps:

[0047] S1, for the load cluster system, according to the value range of the multiple load physical parameters, obtain N0 groups of multiple load physical parameters, and generate parameters of each multiple load in the entire system for each group of multiple load physical parameters.

[0048] Specifically, in order to obtain faster convergence speed, the Quasi Monte Carlo method is applied to sample the input to generate N0 sets of input samples. Then the input of the multivariate system can be expressed as:

[0049]

[0050] (s=1,2,…,N0)

[0051] Among them, R_μ (s) is the average resistance of the sth group, C a _μ (s) is the average capacitance of the sth group, P_μ (s) is the average power of the sth group, T set _μ (s) is the average set temperature of the sth group, δ_μ (s) is the average adjustment rate of the sth group, η_μ (s) is the average efficiency of the sth group, R_χ (s) is the resistance variance of the sth group, C a _χ (s) is the capacitance variance of the sth group, P_χ (s) is the power variance of the sth group, T set _χ (s) is the set temperature variance of the sth group, δ_χ (s) is the variance of the regulation efficiency of the sth group, η_χ (s) is the efficiency variance of the sth group, is the initial temperature of the sth group.

[0052] S2, calculate the parameters of the aggregation model through independent simulation to obtain N0 output quantities.

[0053] Specifically, for each input sample, 10,000 multivariate load parameters are generated according to a specific parameter distribution type. The parameter distribution type can be normal, lognormal, or uniform. The values of the aggregate model parameters are then calculated based on the independent simulation results of the 10,000 multivariate loads. N0 sets of input samples will produce N0 output values. Taking the normal distribution as an example, it can be expressed as:

[0054]

[0055] Where C (s) Represents the value of the sth output quantity.

[0056] Then, each input quantity is normalized, that is:

[0057]

[0058] Where x i ,x * i Represent the actual value and standard value of the input variable respectively; x min ,x max Represents the minimum and maximum values of the input respectively.

[0059] S3, obtain the HDMR model of the load cluster system through the N0 group of input and output samples.

[0060] HDMR is used to describe the mapping relationship between multiple inputs and outputs of a system. HDMR has demonstrated significant effectiveness in sensitivity analysis, correlation modeling, and reliability analysis, and has been applied in fields such as bioengineering, communications technology, and information storage. Therefore, this section uses HDMR to analyze the impact of uncertainty in the physical parameters of multivariate loads on aggregate model parameters and identify the relevant aggregate parameters.

[0061] Assume that the system input is x=(x1,x2,…,x n ) T , the output is f(x). In general, the output is equal to the sum of the independent effects of the input variables and the coupling effects between the input variables. Therefore, f(x) can be expressed as:

[0062]

[0063] Where nf is the number of input variables; f0 is the zero-order component function, which is a constant; f i (x i ) is a first-order component function, representing x i Independent effect on output; f ij (x i ,x j) is a second-order component function, representing x i and x j The effect of coupling between them on the output; 1,2,…,nf (x) is the nf-order component function, which is the influence of the coupling between all input variables on the output.

[0064] S4, obtain the correlation coefficient to obtain the global sensitivity of the multivariate load physical parameters to identify important physical parameters.

[0065] Specifically, based on GUI-HDMR, the multivariate HDMR model is established using the input and output values of N0 to calculate the correlation coefficients α, β and the global sensitivity of the multivariate load physical parameters. For example, the first-order coefficient of the first-order component function of the input, α1, is calculated according to the following formula:

[0066]

[0067] in,

[0068] The quadratic coefficient α2 of the first-order component function of the input quantity is calculated according to the following formula:

[0069]

[0070] in,

[0071] In this way, higher-order coefficients of the first-order component function of the input quantity are calculated.

[0072] β characterizes the effect of coupling between different input quantities on the output. The first-order coefficient of the second-order component function of the input quantity, β 11 , there are 78 values in total, namely:

[0073]

[0074] Similarly, higher-order coefficients of the second-order component function of the input can be obtained.

[0075] Once the values of α and β are known, the global sensitivity of the input variables to both independent and coupled effects can be calculated. Multivariate physical parameters with greater global sensitivity have a greater influence on the aggregate model parameters.

[0076] S5, verify the HDMR model and store the correlation coefficients, and store important multivariate physical parameters and their corresponding global sensitivities and coefficients.

[0077] After validating the HDMR model and storing the correlation coefficients, an HDMR model database of spatial aggregation model parameters at different time scales was established.

[0078] When multiple loads participate in demand response, the parameters of the multiple load aggregation model are in a constantly changing state due to factors such as the constantly changing outdoor temperature and the randomness of user behavior. This will bring certain difficulties to the optimization and coordination of multiple loads and other resources at the system level. Therefore, it is necessary to update the parameters of the multiple load aggregation model in real time. By identifying important physical parameters of multiple loads through HDMR and establishing a mapping relationship between aggregation model parameters and physical parameters, the real-time online identification of multiple load aggregation model parameters can be achieved by monitoring key physical parameters of multiple loads. The specific process is as follows: Figure 1 As shown (taking air conditioning load control as an example).

[0079] The present invention also includes an online identification system for load cluster control domain parameters, including an input module, an independent simulation calculation module, an HDMR model module, a global sensitivity calculation module, an HDMR model verification module and an HDMR model database.

[0080] The input module applies the Quasi Monte Carlo method to generate N0 groups of input samples;

[0081] The independent simulation calculation module obtains the value of N0 output quantities through input samples;

[0082] The HDMR model module obtains the HDMR model of the load cluster system through the N0 group of input and output samples;

[0083] The global sensitivity calculation module obtains the correlation coefficient to obtain the global sensitivity of the multi-element load physical parameters to identify important physical parameters;

[0084] The HDMR model database verifies the HDMR model and stores the correlation coefficients, and establishes an HDMR model database of spatial aggregation model parameters at different time scales.

[0085] Those skilled in the art should recognize that the above embodiments are merely intended to illustrate the present invention and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.

Claims

1. A method for online identification of load cluster control domain parameters, characterized in that: The steps include: S1, for the load cluster system, according to the value range of the multi-load physical parameters, obtain N0 groups of multi-load physical parameters, and generate the parameters of each multi-load in the entire system for each group of multi-load physical parameters; S2, calculate the parameters of the aggregation model through independent simulation to obtain N0 output quantities; S3, obtains the HDMR model of the load cluster system through the N0 group of input and output samples; S4, obtain the correlation coefficient to obtain the global sensitivity of the physical parameters of the multivariate load to identify the important physical parameters; S5, verify the HDMR model and store the correlation coefficient.

2. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: In S1, the Quasi Monte Carlo method is applied to generate N0 groups of input samples. The input of the multivariate system can be expressed as: x (s) =[R_μ (s) ,C a _m (s) ,P_μ (s) ,T set _m (s) ,d_m (s) ,h_m (s) , R_x (s) ,C a _x (s) ,P_x (s) ,T set _x (s) ,d_x (s) ,h_x (s) ,T o (s) ] T (s=1,2,…,N0) Among them, R_μ (s) is the average resistance of the sth group, C a _μ (s) is the average capacitance of the sth group, P_μ (s) is the average power of the sth group, T set _μ (s) is the average set temperature of the sth group, δ_μ (s) is the average adjustment rate of the sth group, η_μ (s) is the average efficiency of the sth group, R_χ (s) is the resistance variance of the sth group, C a _χ (s) is the capacitance variance of the sth group, P_χ (s) is the power variance of the sth group, T set _χ (s) is the set temperature variance of the sth group, δ_χ (s) is the variance of the regulation efficiency of the sth group, η_χ (s) is the efficiency variance of the sth group, T o (s) is the initial temperature of the sth group.

3. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: In S2, for each input sample, 10,000 multivariate load parameters are generated according to the parameter distribution type, and then the values of the aggregate model parameters are obtained based on the independent simulation calculation results of the 10,000 multivariate loads. N0 groups of input samples obtain N0 output values.

4. The method for online identification of load cluster control domain parameters according to claim 3 is characterized in that: Parametric distribution types are classified as normal, lognormal, or uniform.

5. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: In S3, after obtaining the values of N0 pairs of input and output quantities, each pair of values is standardized before entering the HDMR model, that is: Where x i ,x * i Represent the actual value and standard value of the input variable respectively; x min ,x max Represents the minimum and maximum values of the input respectively.

6. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: In S3, the specific method of the HDMR model is: Assume that the system input is x=(x1,x2,…,x n ) T , the output is f(x), which is equal to the sum of the independent effects of the input variables and the coupling effects between the input variables. Therefore, f(x) can be expressed as: Where nf is the number of input variables; f0 is the zero-order component function, which is a constant; f i (x i ) is a first-order component function, representing x i Independent effect on output; f ij (x i ,x j ) is a second-order component function, representing x i and x j The effect of coupling between them on the output; 1,2,…,nf (x) is the nf-order component function, which is the influence of the coupling between all input variables on the output.

7. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: In S4, the correlation coefficients α and β are calculated for the multivariate HDMR model established using N0 for the input and output values. The linear coefficient α1 of the first-order component function of the input is calculated as follows: in, The quadratic coefficient α2 of the first-order component function of the input quantity is calculated according to the following formula: in, By analogy, higher-order coefficients of the first-order component function of the input quantity are calculated; β represents the effect of the coupling between different input quantities on the output, and is the first-order coefficient of the second-order component function of the input quantity. 11 ,Right now: By analogy, the higher-order coefficients of the second-order component function of the input quantity are calculated; The global sensitivity of the independent and coupled inputs is obtained by the obtained values of α and β.

8. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: After validating the HDMR model and storing the correlation coefficients, an HDMR model database of spatial aggregation model parameters at different time scales was established.

9. The method for online identification of load cluster control domain parameters according to claim 1, characterized in that: When multiple loads participate in demand response, the parameters of the multiple load aggregation model will be in a constantly changing state. By obtaining the physical parameters of the multiple loads after regulation, the physical parameter values of the multiple loads are updated online, and the parameters of the multiple load aggregation model are quickly calculated, the parameters of the multiple load aggregation model can be updated in real time.

10. A system for implementing the method for online identification of load cluster control domain parameters according to any one of claims 1 to 9, characterized in that: It includes input module, independent simulation calculation module, HDMR model module, global sensitivity calculation module, HDMR model verification module and HDMR model database; The input module applies the Quasi Monte Carlo method to generate N0 groups of input samples; The independent simulation calculation module obtains the value of N0 output quantities through input samples; The HDMR model module obtains the HDMR model of the load cluster system through the N0 group of input and output samples; The global sensitivity calculation module obtains the correlation coefficient to obtain the global sensitivity of the multi-element load physical parameters to identify important physical parameters; The HDMR model database verifies the HDMR model and stores the correlation coefficients, and establishes an HDMR model database of spatial aggregation model parameters at different time scales.

Images