CN112257900A - Structural equation-based power distribution network frame optimization method with distributed power supply - Google Patents

Structural equation-based power distribution network frame optimization method with distributed power supply Download PDF

Info

Publication number
CN112257900A
CN112257900A CN202011007808.4A CN202011007808A CN112257900A CN 112257900 A CN112257900 A CN 112257900A CN 202011007808 A CN202011007808 A CN 202011007808A CN 112257900 A CN112257900 A CN 112257900A
Authority
CN
China
Prior art keywords
model
variable
distribution network
power distribution
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011007808.4A
Other languages
Chinese (zh)
Inventor
齐晓光
秦梁栋
王颖
王宁
张倩茅
习朋
张丽洁
徐田丰
朱天瞳
田家辉
陈宇
张磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Original Assignee
Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd filed Critical Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Priority to CN202011007808.4A priority Critical patent/CN112257900A/en
Publication of CN112257900A publication Critical patent/CN112257900A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
    • Y02E40/70—Smart grids as climate change mitigation technology in the energy generation sector
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

The invention relates to a structural equation-based optimization method for a power distribution network frame containing a distributed power supply. The method comprises the steps of adopting a final observation variable selected by five first-order factors to establish a structural equation model for evaluating the optimization effect of the power distribution network frame; and (3) identifying the model: each parameter in the model can be identified, and the model is identifiable, otherwise, the model needs to be reconstructed until the model is identifiable; selecting an observation variable and collecting data; selecting observation variables which accord with factor analysis through model evaluation and collecting data; fitting the observation variable and the collected data to the structural equation model; if the fitting does not meet the conditions, the model is corrected and then the model is identified again; and evaluating the optimization effect of the power distribution network frame by adopting the fitted model. The invention simultaneously tests the relationship among the dominant variable, the latent variable, the interference or the error contained in the model, and further obtains the direct effect, the indirect effect or the total effect of the influence of the independent variable on the dependent variable.

Description

Structural equation-based power distribution network frame optimization method with distributed power supply
Technical Field
The invention belongs to the field of power distribution networks, and particularly relates to a structural equation-based optimization method for a power distribution network frame containing a distributed power supply.
Background
The method for optimizing the operation mode of the power distribution network is used for researching the existing operation mode of the power distribution network (Shanghai transportation university, 2009), and an operation mode optimization comprehensive evaluation system is constructed by combining an analytic hierarchy process and considering equipment operation, economy and safety indexes. The optimization theory is introduced into the power distribution network operation mode evaluation system according to the theoretical basis and the practical application (electrical application, 2013(S1): 430-. A scientific decision method is adopted in distributed power robust planning (Yanshan university, 2018) of active power distribution network neutralization network reconstruction, an operation mode decision model of multi-objective coordination optimization is established, multi-objectives are subjected to weighting processing, a power distribution network topological structure is determined by combining an improved evolutionary planning algorithm, and an optimal operation mode is determined. Research and application (the university of northeast electric power, 2015(02):48-52.) of the comprehensive quantitative evaluation system for operation of the distribution network analyzes and researches evaluation indexes of the real-time operation state of the distribution network based on SCADA/EMS on the basis of conventional evaluation indexes, and establishes an evaluation system for the real-time operation state. A comprehensive evaluation system (power grid technology, 2012(8):95-99.) of the urban power distribution network analyzes and researches main factors of power distribution network construction, constructs a comprehensive evaluation system for transverse comparison of the power distribution network, and establishes an index scoring standard by adopting a Delphi method, thereby quantitatively evaluating the running state of the power distribution network. A gridding medium and low voltage intelligent power distribution network evaluation index system and method (power grid technology, 2016,40(01): 249-plus 255) are used for constructing a refined power distribution network evaluation index system from the perspective of different interest correlators, meanwhile, a least square method is adopted for determining index subjective weight, an entropy weight method is adopted for determining objective weight, the subjective objectivity of the evaluation index is considered, a weak link in a power distribution network can be obtained by using the index evaluation method, and the index evaluation method has certain guiding significance for refined management of the power distribution network. The power distribution network evaluation index system research (power construction, 2013(02):18-21.) carries out differentiation classification on indexes according to the attribute characteristics of the indexes, and provides a practical power distribution network grading method. A comprehensive evaluation index system and an evaluation method of the power distribution network (Guangdong power, 2013,26(011):20-25.) comprehensively use methods such as an analytic hierarchy process, a Delphi method and fuzzy evaluation to construct a comprehensive operation evaluation model of the power distribution network, and take a local power distribution network as an example to carry out quantitative evaluation on an actual operation state.
The existing power distribution network optimization scheme evaluation method is mostly aimed at a general power distribution network, power distribution network optimization scheme evaluation aiming at the consideration of distributed power supply grid connection is lacked, the influence of randomness, volatility and instability of new energy on the power distribution network optimization scheme is difficult to consider, evaluation indexes are not comprehensive enough, and pertinence is lacked.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a power distribution network frame optimization method containing a distributed power supply based on a structural equation.
The present invention is achieved in such a way that,
a power distribution network frame optimization method based on a structural equation and comprising a distributed power supply comprises the following steps:
step 1: adopting the final observation variables selected by the five first-order factors to establish a structural equation model for evaluating the optimization effect of the power distribution network frame;
step 2: and (3) identifying the model: each parameter in the model can be identified, and the model is identifiable, otherwise, the model needs to be reconstructed until the model is identifiable;
and step 3: selecting an observation variable and collecting data;
and 4, step 4: selecting observation variables which accord with factor analysis through model evaluation and collecting data;
and 5: fitting the observation variable and the collected data in the step 4 to the structural equation model;
step 6: if the fitting does not meet the conditions, performing model correction and returning to the step 2;
and 7: and evaluating the optimization effect of the power distribution network frame by adopting the fitted model.
Furthermore, the model in the step 1 is determined by equations (1) and (2), wherein the optimization result of the power distribution grid frame is an exogenous latent variable xi, the reliability, the safety, the adaptability, the economic benefit and the environmental benefit are endogenous latent variables eta, and 13 indexes are observation variables yijEpsilon is the measured residual error of the observed variable which is not completely explained by the latent variable, zeta is the estimation error of the endogenous latent variable which can not be completely explained, LambdayFor the regression matrix of the observed variable interpreted by the endogenous latent variable, the factor loads of η to y, Γ is the regression matrix of the endogenous latent variable interpreted by the exogenous latent variable, the regression matrix of η to ξ:
y＝Λyη+ε (1)
Figure BDA0002696560650000031
further, the selection of the observation variables in step 3 includes:
defining a research variable, and collecting sample data;
and preprocessing, reliability analysis and validity analysis are carried out on the collected data.
Further, the pre-processing comprises: the method comprises the steps of processing missing values, unifying data types, normalizing non-normal distribution variables, analyzing reliability and analyzing validity, wherein:
processing missing values: filling the missing value by adopting a multivariate calculation method in PRELIS in LISREL 8.8;
data type reconciliation: converting each type of index into a maximum index, and carrying out non-dimensionalization treatment on the index;
reliability analysis: performing Cronbach's alpha test on the passing sample data by using SPSS 19.0, wherein the minimum requirement of the reliability is at least up to more than 0.5;
and (3) effectiveness analysis: firstly, KMO test and Bartlett sphere test are carried out to judge whether the observed variable is suitable for factor analysis, and the factor analysis is adopted for validity test.
Further, performing data fitting on the model by using the processed sample data through Liserel8.80 to obtain the fitted model.
Compared with the prior art, the invention has the beneficial effects that:
(1) the method considers the particularity of the grid-connected power supply and evaluates the optimization effect of the power distribution network frame containing the distributed power supply in a targeted manner.
(2) The method integrates the statistical technology of 'factor analysis' and 'regression analysis of linear model' in the traditional multivariate statistical analysis, and simultaneously tests the relation among dominant variables, latent variables, interference or errors contained in the model, thereby obtaining the direct effect, indirect effect or total effect of the influence of the independent variables on the dependent variables and increasing the scientificity of the evaluation result.
(3) The method evaluates the optimization effect of the power distribution network frame from five dimensions of reliability, safety, adaptability, economic benefit and social benefit, and the evaluation result is more comprehensive.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a structural equation model structure diagram;
fig. 3 is a plan for optimizing the grid structure of a certain county in fig. 3.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1 and fig. 2, a method for optimizing a grid structure of a power distribution network including a distributed power source based on a structural equation includes the following steps:
step 1: adopting the final observation variables selected by the five first-order factors to establish a structural equation model for evaluating the optimization effect of the power distribution network frame;
step 2: and (3) identifying the model: each parameter in the model can be identified, and the model is identifiable, otherwise, the model needs to be reconstructed until the model is identifiable;
and step 3: selecting an observation variable and collecting data;
and 4, step 4: selecting observation variables which accord with factor analysis through model evaluation and collecting data;
and 5: fitting the observation variable and the collected data in the step 4 to the structural equation model;
step 6: if the fitting does not meet the conditions, performing model correction and returning to the step 2;
and 7: and evaluating the optimization effect of the power distribution network frame by adopting the fitted model.
And in the step 1, a structural equation model for evaluating the optimization effect of the power distribution network frame is established according to the final observation variables selected by the five first-order factors. The second order factor optimization effect is a potential concept that is higher or broader than the five first order factors.
A second order validation factor model can be determined by equations (1) and (2). In the formula, the optimization result of the power distribution network frame is an exogenous latent variable (xi), the reliability, the safety, the adaptability, the economic benefit and the environmental benefit are endogenous latent variables (eta), and 13 indexesMarked as observed variable (y)ij) Epsilon is the measurement residual error of the observed variable which is not completely explained by the latent variable, and zeta is the estimation error of the endogenous latent variable which can not be completely explained. LambdayIs the regression matrix (factor load of eta to y) for the interpretation of the observed variable by the endogenous latent variable, and Γ is the regression matrix (regression matrix of eta to ξ) for the interpretation of the endogenous latent variable by the exogenous latent variable.
y＝Λyη+ε (1)
Figure BDA0002696560650000051
Step 2, identifying the model: each parameter in the model can be identified, and the model is identifiable, otherwise, the model needs to be reconstructed until the model is identifiable;
the identification problem of a model is related to whether parameters in the model can be identified. The identification of parameters can be divided into identifiable and non-identifiable. Recognizable again includes just recognition and over recognition. An unknown parameter is said to be identifiable when it can be expressed by a representative function of one or more elements in the covariance matrix of the observed variable, the former representing proper identification and the latter representing over-identification. If the parameters in the model are all identified exactly, the model is the exactly identified model; if the model contains over-identified parameters, the model is an over-identified model; if the model contains unidentifiable parameters, the model is an unidentifiable model. The unrecognizable model has a relationship with the construction of the model and has no relationship with the sample, and the unrecognizable model can be caused by few degrees of freedom or interaction among variables. A check in parameter identification may be provided by LISREL8.80 during parameter estimation, and if a problem is found, the program alerts the user to modify the model to eliminate the problem.
And 3, selecting an observation variable and collecting data:
the method comprises the following steps: defining a research variable, and collecting sample data;
and preprocessing, reliability analysis and validity analysis are carried out on the collected data.
In this embodiment: and analyzing the importance degree of the index on the total target evaluation by a five-level Lekter scale method on the collected sample data. The judgment standard is as follows: generally, 1 is important, 2 is slightly important, 3 is relatively important, 4 is very important, and 5 is very important.
In order to ensure the accuracy of the SEM analysis result, the sample volume should preferably be at least 100, preferably more than 200. Whereas questionnaires are costly. The sample capacity needs to be determined within the range allowed by the cost, and the accuracy of the statistical result is improved to the maximum extent. Considering that the design of questionnaires and the selection of samples in the previous stage are strict and ensure high data quality, the sample capacity of this embodiment is preferably about 200.
Before the sample data is fitted with the model, early data analysis is required, including data preprocessing, reliability analysis and validity analysis.
(1) Pre-processing of data
The preprocessing of data mainly includes processing of missing values, data type uniformization, normalization of non-normal distribution variables, and the like. The sample data is preprocessed by the PRELIS subroutine in LISREL 8.8.
1) Processing missing values
Missing values refer to data missing due to mechanical or human causes. The PRELIS subroutine in LISREL8.8 deals with missing values in two ways, one is deleting samples containing missing values and one is filling in missing values. Since the data required by the questionnaire is difficult to obtain, in order to ensure the accuracy of the statistical result, the missing value is filled by using a multivariate calculation method in PRELIS in LISREL 8.8.
2) Data type reconciliation
The sample data used in this example was measured using a 5-grade litters scale. The index system type mainly includes a very large index, a very small index, an intermediate index and an interval index. Before the optimization effect of the power distribution network frame is evaluated, indexes need to be subjected to consistency processing, otherwise, the evaluation result cannot be qualitatively judged whether the value is getting larger and better or getting smaller and better. Generally, each type of index is converted into a maximum index. Furthermore, it is necessary to make the indicators dimensionless to avoid the inequitability of the indicators due to the difference of the measurement units.
3) Normalization of non-normally distributed variables
When fitting the data to the model, the maximum likelihood estimation method is adopted, which requires that the sample data must be in multivariate normal distribution. Therefore, it is very critical to normalize the data. The measurement unit of the sample data of the research has no practical significance, and the sample data can be normalized by using the normal value of the sample data.
(2) Reliability analysis
Reliability (Reliability) is the degree of consistency of the results obtained when the same measurement is repeated on the same subject using the same method. Currently, there are the following methods for reliability analysis: a double confidence measure method, a duplicate confidence measure, a semi-confidence measure, and an alpha confidence coefficient method. The alpha reliability coefficient method is the most common reliability analysis method at present, has the advantages of other reliability analysis methods, and is suitable for reliability analysis of attitudes or opinion questionnaires. For the above reasons, the reliability analysis will be performed by using the α reliability coefficient method.
In this example, SPSS 19.0 was used to perform a Cronbach's α test on data collected by questionnaires, and the test results are shown in Table 2. The reliability of the whole questionnaire is 0.850, and the reliability of reliability, safety, adaptability, economic benefit and environmental benefit is about 0.6. Generally, the overall reliability is better to be more than 0.8, and Cuieford (1965) research shows that the reliability is low when Cronbach's alpha is less than 0.35, the reliability is medium-intensity when Cronbach's alpha is more than or equal to 0.35 and less than 0.7, and the reliability is high when Cronbach's alpha is more than or equal to 0.7; nunnally (1978) also considers that Cronbach's α ≧ 0.7 is within a very trusted range, with confidence minimum requirements of at least 0.5 or more. Therefore, the overall reliability belongs to high reliability, and the other five parts belong to medium reliability, which all meet the reliability requirement.
TABLE 1 Cronbach's alpha confidence test results
Figure BDA0002696560650000071
(3) Efficacy analysis
Validity (Validity), which means the degree to which a measuring tool or means can accurately measure the object to be measured, is mainly classified into content Validity, criterion Validity, and structure Validity. Thus, the more the sample data matches the content that is desired to be examined, the higher the validity and vice versa. The content validity mainly considers whether the sample data is representative or not, and the criterion validity is to examine the effectiveness degree of predicting the behavior of an individual under a certain situation, which are related to the design of the internal logic of the data and the selection of the sample. The data acquisition design is obtained by researching and modifying relevant experts, teachers and researchers for many times on the basis of reading a large amount of domestic and foreign documents and relevant theories, and the content effectiveness and the criterion effectiveness can be considered to reach the standard. The structural validity refers to the degree of correspondence between a certain structure represented by the measurement result and the measured value. Many scholars consider that model fitting coefficients and standardized factor loads obtained by the analysis of the verification factors are the most suitable indexes for detecting the structural validity at present. With a good model fit, the lowest criterion for the normalization factor load to meet the validity requirement is 0.45, with greater than 0.71 indicating a higher validity. This example will first perform the KMO (Kaiser-Meyer-Olkin) test and the Bartlett sphere test to determine if the variables are suitable for factoring, and on a suitable basis, to perform the validity test using factoring.
TABLE 2 KMO test and Bartlett's sphericity test
Figure BDA0002696560650000081
The results of KMO and Bartlett sphere tests on the power distribution network optimization effect evaluation population are shown in table 2. The table shows that the overall KMO sample measurement value is 0.856 and the Bartlett sphere test results are 0.000. The lowest standard value of KMO is generally considered to be 0.5, and a closer value to 1 indicates that it is more suitable for factorial analysis. The Bartlett sphere test is relatively large and the corresponding concomitant probability value is less than a given significance level (e.g., 0.001), which may be considered suitable for factorial analysis. Therefore, it can be seen that the measurement value of the KMO sample of the data population is greater than 0.5, which is more suitable for the factor analysis, the result of Bartlett sphere test is 0.000 and less than 0.001, the zero hypothesis that the correlation coefficient matrix is the identity matrix is rejected, and the factor analysis is also supported, so that the sample data collected in the data survey meets the requirement of the factor analysis.
Evaluating and modifying a model includes the following three aspects:
1) and (6) parameter checking. It is generally considered that, when the degree of freedom is large and the significance level is 0.05, the t value corresponding to the free parameter should be greater than 1.96, otherwise, the free parameter is considered to be not significant and should be removed from the model.
2) The fitting index was examined. The statistic used to evaluate the degree of fit of the model to the data, the fitting index (goodness of fit index), will pass mainly the chi-square test (χ) in this study2The model is evaluated by determining the degree of fitting of the model using indices such as/df, P value), model fitting index (GFI, AGFI, NFI, NNFI, etc.), substitution index (NCP, CFI, RMSEA, AIC, etc.), residual analysis (RMR, SRMR, etc.).
3) The full normalization factor load and correction index (MI) of each variable are examined to determine the model correction scheme. It is generally believed that the factor load for the tables compiled by the social scientific research will not be too high, and that a fully normalized factor load for each observed variable will only be greater than 0.45. This may be a controversy limited by the nature of the measurement (e.g., the separation of the attitude measurements is too broad to focus, the structure is too ambiguous, etc.), the influence of external interference and measurement errors, or even the nature of the structure being formative or reactive. If the normalization factor load of the individual observation variables is less than 0.45, the decision of whether to delete or modify the dependency reconstruction model needs to be made by combining the relevant theory, economic significance and correction coefficient provided by software.
Model fitting is a relatively complex, progressive process. In the research process, the theoretical basis and the model are combined fully, and the fitting degree of the whole model is tried to be improved continuously. Therefore, the part firstly carries out parameter estimation inspection and model fitting degree analysis, and then carries out model correction by combining the economic significance of each variable on the basis of checking the standardized factor load and the correction index of each variable.
In this embodiment, the processed sample data, including 270 samples and 17 observation variables, is substituted into the initial verification factor analysis model, and the feasibility of the model is examined, so as to further correct the model. And performing data fitting on the model by using Liserel8.80 to obtain a model M1.
Table 3 is the raw estimates (non-normalized values), standard error and statistical significance. Wherein the significance test is performed as a t-test. If the number of samples is 270, the absolute value of t is considered significant if it exceeds 1.96. As shown in Table 4, the non-normalized regression parameter for first order factor market stability for the second order factor optimization was 0.89, the standard error was 0.12, and the t value was 7.60(t value is the original estimator divided by the standard error). Since the value of t is greater than 1.96, a significance level is reached, indicating that this parameter is statistically significant. Other parameters also pass significance tests and have statistical significance.
TABLE 3 Model1 parameter estimation results
Figure BDA0002696560650000101
TABLE 4 Model1 fitting results
Figure BDA0002696560650000102
The former analysis data is the result of parameter estimation, and can be used to determine the statistical significance of individual parameters. The overall effect of these parameter estimates should reflect the statistical significance of the Model, and can be evaluated by the fitting coefficients, and the fitting result of the initial Model1 is shown in table 4. As can be seen from the table, the degree of freedom of the entire model is 114, the chi-squared value is 268.77, and the P value is 0.000, indicating that there is a significant difference between the assumed model and the observed value. NFI and NNFI are the normal and denormal fit indices, respectively, that reflect the degree of difference between the hypothetical model and an independent model without any co-variation assumption between the observed variables. It is generally greater than 0.90, indicating a good Model fit, while Model1 has an NFI of 0.900, not greater than 0.9. GFI and AGFI are both hypothetical models that can account for the proportion of observed data, while AGFI takes complexity into account. Both indices need to be greater than 0.9 to be considered as having a desirable degree of fit, whereas Model1 has a GFI of 0.89 and an AGFI of 0.86, which does not reach the desired value. The RMR is the overall residual of the normalized hypothesis Model, which fits less than 0.05, whereas the RMR of Model1 is 0.067, which does not reach the ideal state. In summary, the Model1 cannot be said to have the best fitting degree, and there is still room for correction.
Model correction
After the first correction (removal of the observed variable y22), the model fit results are shown in table 5. The outcome p <0.05 of the chi-square test is still significant, and NFI, GFI, AGFI, RFI and RMR still do not meet the adaptation criteria. The model needs further modification, and the modification results are shown in table 6 below:
TABLE 5 Model2 fitting results
Figure BDA0002696560650000111
Figure BDA0002696560650000121
TABLE 6 Model3 fitting results
Figure BDA0002696560650000122
Fitting exponent χ of Model32/df<2、RMSEA<0.08、RMR<0.05, NFI, NNFI, CFI, IFI, GFI, etc. exceed the adaptation standard of 0.90, which indicates that the Model3 is much improved over the Model2, and the goodness of fit to the data is already good. But its chi-square check P value is still<0.05, the significance level is reached, the virtual null hypothesis (the virtual null hypothesis is that the theoretical matrix and the observation matrix have no difference) is not established, and the fitting degree of the model is not good. Since chi-squared distribution is susceptible to the degree of freedom and the number of samples, the greater the degree of freedom or sample, the greater the chi-squared value. That is, as the degree of freedom or sample is larger, the number of parameters to be estimated is larger, the more factors affecting a hypothetical model, and the higher the possibility of poor fitting of the hypothetical model. Therefore, when examining the SEM model using the chi-square distribution, the fitting degree examination of the hypothesis model is affected by the technical characteristics of the number of parameters and the number of samples. And the chi-square degree of freedom also takes the influence of the degree of freedom into consideration on the basis of the chi-square value. The chi-square degree of freedom of the Model3 is 1.681, which is less than 2, and shows that the Model3 has ideal fitting degree if the complexity of the Model is considered.
The output result of the Liserel shows that the t values corresponding to all the free parameters of the Model3 are all larger than 1.96, and the parameter test requirements are met, as shown in Table 7. The normalized path coefficients of the five first-order factors of reliability, safety, adaptability, economic benefit and social benefit are respectively 0.90, 0.96, 0.75, 0.88 and 0.73 (shown in tables 5-12), and are all between 0.7 and 1, which indicates that the second-order factor model has enough polymerization efficiency, and 5 first-order factors can be used as the measurement indexes of the optimization effect of the second-order factor. Furthermore, the residual scatter plots from the Model3 fitting showed that the normalized residuals were all distributed around the 45 degree diagonal and very close together, indicating that the Model3 fit to the data was acceptable.
TABLE 7 Model3 parameter estimation results
Figure BDA0002696560650000131
For the analysis of the higher-order factors, the most important coefficient is the path coefficient of the higher-order factors, and the parameter reflects the explanatory power of the higher-order factors on the initial-order factors and represents the relative importance of the first-order factors on the second-order factors. The results show that: the five first-order factors of reliability, safety, adaptability, economic benefit and environmental benefit have normalized path coefficients of 0.90, 0.96, 0.75, 0.88 and 0.73 respectively at the second-order factor of the "optimization effect".
And evaluating the optimization effect of the power distribution network frame by using the obtained final model of the second-order factor. The report adopts a correlation weight method to determine the weight of the index system. The correlation weight method is a method of determining a weight using a correlation between variables. The method is to measure and calculate the correlation coefficient between variables through a large amount of sample data, and determine the weight according to the relative importance degree. The normalization factor loading in the structural equation model fit results substantially reflects the correlation coefficient between each observed variable and the corresponding latent variable. Accordingly, the weights corresponding to the indexes can be obtained by normalizing the normalized factor load.
The normalized formula is:
Figure BDA0002696560650000141
where ρ isijIs a first order factor etaiThe same method is applied to the weight determination of the first-order factor.
The specific measurement method for the second-order factor power distribution network frame optimization effect comprises the following steps:
Figure BDA0002696560650000142
in which ξ1Represents the optimization effect of the network frame of the power distribution network, betaiRepresenting a first order factor etaiWeight of (p) ()ijIs a first order factor etaiWeight of the jth measurement index of (1), yijRepresenting a first order factorηiThe index value of the jth measurement index of (1). H represents the number of first-order factors, and K represents the weight distribution condition of the power distribution network frame optimization effect evaluation index system of the corresponding index number corresponding to the first-order factors, as shown in Table 8.
Table 8 weight coefficient table of power distribution network frame optimization effect evaluation model variables
Figure BDA0002696560650000143
Figure BDA0002696560650000151
Taking the power distribution network optimization scheme of a power distribution network in a certain county as an example, the method is applied to the scheme evaluation, and the optimization scheme is shown in fig. 3.
Based on the county power distribution network frame optimization result, five variable values of reliability, safety, adaptability, economic benefit and social benefit are subjected to unified variable processing, the five variable values are respectively 93.25, 94.48, 90.16, 99.01 and 91.11 and are substituted into a power distribution network frame optimization effect comprehensive evaluation model, and the comprehensive optimization effect score is 93.830252. Based on the structural equation evaluation model principle, the county power distribution network frame optimization result has implementability in five dimensions of reliability, safety, adaptability, economic benefit and social benefit.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (5)

1. A power distribution network frame optimization method based on a structural equation and comprising a distributed power supply is characterized by comprising the following steps:
step 1: adopting the final observation variables selected by the five first-order factors to establish a structural equation model for evaluating the optimization effect of the power distribution network frame;
step 2: and (3) identifying the model: each parameter in the model can be identified, and the model is identifiable, otherwise, the model needs to be reconstructed until the model is identifiable;
and step 3: selecting an observation variable and collecting data;
and 4, step 4: selecting observation variables which accord with factor analysis through model evaluation and collecting data;
and 5: fitting the observation variable and the collected data in the step 4 to the structural equation model;
step 6: if the fitting does not meet the conditions, performing model correction and returning to the step 2;
and 7: and evaluating the optimization effect of the power distribution network frame by adopting the fitted model.
2. The method of claim 1 wherein the model in step 1 is determined by equations (1) and (2) wherein the grid optimization results in an exogenous latent variable ξ, the reliability, safety, adaptability, economic and environmental benefits are endogenous latent variables η, and the 13 indices are observed variables yijEpsilon is the measured residual error of the observed variable which is not completely explained by the latent variable, zeta is the estimation error of the endogenous latent variable which can not be completely explained, LambdayFor the regression matrix of the observed variable interpreted by the endogenous latent variable, the factor loads of η to y, Γ is the regression matrix of the endogenous latent variable interpreted by the exogenous latent variable, the regression matrix of η to ξ:
y＝Λyη+ε (1)
Figure FDA0002696560640000011
3. the method of claim 1, wherein step 3 observing the selection of variables comprises:
defining a research variable, and collecting sample data;
and preprocessing, reliability analysis and validity analysis are carried out on the collected data.
4. The method of claim 3, wherein the pre-processing comprises: the method comprises the steps of processing missing values, unifying data types, normalizing non-normal distribution variables, analyzing reliability and analyzing validity, wherein:
processing missing values: filling the missing value by adopting a multivariate calculation method in PRELIS in LISREL 8.8;
data type reconciliation: converting each type of index into a maximum index, and carrying out non-dimensionalization treatment on the index;
reliability analysis: performing Cronbach's alpha test on the passing sample data by using SPSS 19.0, wherein the minimum requirement of the reliability is at least up to more than 0.5;
and (3) effectiveness analysis: firstly, KMO test and Bartlett sphere test are carried out to judge whether the observed variable is suitable for factor analysis, and the factor analysis is adopted for validity test.
5. The method of claim 1, wherein the processed sample data is subjected to data fitting on the model by using Liserel8.80 to obtain a fitted model.
CN202011007808.4A 2020-09-23 2020-09-23 Structural equation-based power distribution network frame optimization method with distributed power supply Pending CN112257900A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011007808.4A CN112257900A (en) 2020-09-23 2020-09-23 Structural equation-based power distribution network frame optimization method with distributed power supply

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011007808.4A CN112257900A (en) 2020-09-23 2020-09-23 Structural equation-based power distribution network frame optimization method with distributed power supply

Publications (1)

Publication Number Publication Date
CN112257900A true CN112257900A (en) 2021-01-22

Family

ID=74232967

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011007808.4A Pending CN112257900A (en) 2020-09-23 2020-09-23 Structural equation-based power distribution network frame optimization method with distributed power supply

Country Status (1)

Country Link
CN (1) CN112257900A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113256446A (en) * 2021-05-31 2021-08-13 华南理工大学 Power distribution network robust optimization scheduling method considering data loss and achieving information physical fusion
WO2023236472A1 (en) * 2022-06-08 2023-12-14 东南大学 Built-environment quality measurement method, and device and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FI20116321A (en) * 2011-12-23 2013-06-24 Aalto Korkeakoulusaeaetioe Method for Detecting Faults in an Industrial Process or Part thereof or for Controlling and / or Optimizing an Industrial Process or Part thereof
CN107180160A (en) * 2017-06-21 2017-09-19 西南交通大学 Public bicycles consumer loyalty degree based on SEM models determines method
CN108537419A (en) * 2018-03-22 2018-09-14 安徽建筑大学 The method of coal mine safety management risk

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FI20116321A (en) * 2011-12-23 2013-06-24 Aalto Korkeakoulusaeaetioe Method for Detecting Faults in an Industrial Process or Part thereof or for Controlling and / or Optimizing an Industrial Process or Part thereof
CN107180160A (en) * 2017-06-21 2017-09-19 西南交通大学 Public bicycles consumer loyalty degree based on SEM models determines method
CN108537419A (en) * 2018-03-22 2018-09-14 安徽建筑大学 The method of coal mine safety management risk

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113256446A (en) * 2021-05-31 2021-08-13 华南理工大学 Power distribution network robust optimization scheduling method considering data loss and achieving information physical fusion
WO2023236472A1 (en) * 2022-06-08 2023-12-14 东南大学 Built-environment quality measurement method, and device and storage medium

Similar Documents

Publication Publication Date Title
CN111950918B (en) Market risk assessment method based on power transaction data
CN116150897A (en) Machine tool spindle performance evaluation method and system based on digital twin
CN112257900A (en) Structural equation-based power distribution network frame optimization method with distributed power supply
CN113866642A (en) Lithium ion battery fault diagnosis method based on gradient lifting tree
CN107832927A (en) 10kV circuits line based on gray relative analysis method becomes relation evaluation method
CN113671394A (en) Lithium ion battery expected life prediction method and system
CN113641959B (en) Temperature trend prediction method for high-voltage cable connector
CN113627735A (en) Early warning method and system for safety risk of engineering construction project
CN110705859A (en) PCA-self-organizing neural network-based method for evaluating running state of medium and low voltage distribution network
CN116933010A (en) Load rate analysis and evaluation method and system based on multi-source data fusion and deep learning
CN114881374A (en) Multi-element heterogeneous energy consumption data fusion method and system for building energy consumption prediction
CN113869732A (en) Scientific and technological project industrialization evaluation method based on factor analysis method
CN117195505A (en) Evaluation method and system for informatization evaluation calibration model of electric energy meter
JP2024015998A (en) Automatic capacity calibration method for distribution transformers with voltage level below 315kV
CN107644285A (en) The screening of power sales profitability evaluation index and Weight Determination and system
Guo et al. Information fusion and XGBoost algorithm used for bearing remaining useful life prediction
CN114358160A (en) Data anomaly detection method in power system
CN114139408A (en) Power transformer health state assessment method
Yun et al. Line loss calculation of low-voltage districts based on improved K-Means
CN112116139A (en) Power demand prediction method and system
CN116500456B (en) Ternary lithium battery capacity detection method based on machine learning
CN117808327A (en) Power quality comprehensive evaluation method and system based on combined weighting association analysis
CN116226767B (en) Automatic diagnosis method for experimental data of power system
CN117114468A (en) Power supply reliability assessment method, device, equipment and medium for power distribution system
Fan Synthetic Evaluation of Power Quality Based on Prospect Theory and Membership Grade

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination