CN104881718A - Regional power business index constructing method based on multi-scale leading economic indicators - Google Patents

Abstract

The invention provides a regional power business index constructing method based on multi-scale leading economic indicators, comprising the steps of establishing a multi-scale economic index database, acquiring regional power consumption of a historical sample interval and monthly data of each economic indicator, screening leading economic indicators of regional power consumption, constructing a regional power consumption monthly prediction regression model, predicting regional power consumption of a target month, and calculating the regional power business index of the target month. According to the invention, a power business index used for predicting the trend of regional power consumption is constructed by extracting information related to international and domestic economic situation changes from multiple economic indicators and taking the time lag effect between the economic indicators and regional power consumption into consideration. The problem that an argument indicator needs to be selected manually and the optimal model is hard to determine is solved effectively, and a scientific basis is provided for planning and decision making of power grid companies.

Description

Based on the region electric power consumer confidence index construction method of multiple dimensioned economic leading indicators

Technical field

The present invention relates to electrical network electricity demand forecasting technical field, specifically based on the region electric power consumer confidence index construction method of multiple dimensioned economic leading indicators.

Background technology

Research finds, electricity consumption is very responsive to national economy variation, and therefore, the change of need for electricity can be predicted according to the change of macroeconomic operation situation.And reflect that the index of macroeconomic operation situation can be divided into leading indicators, coincident indicator and lagging indicator, wherein only have the prediction of leading indicators to need for electricity to have directive significance.Under the background complicated in economic situation, economic fluctuation frequency is more and more higher, amplitude is increasing, traditional electricity demand forecasting method shows the deficiency in predictive ability gradually, and the development of data mining technology and econometrics, then for electricity demand forecasting provides new thinking and countermeasure, be conducive to the ageing and degree of accuracy improving electricity demand forecasting.

Summary of the invention

The object of the present invention is to provide a kind of region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators, by setting up the multivariate regression model of region power consumption and multiple dimensioned economic leading indicators, further structure electric power consumer confidence index, to realize estimation range power consumption and the external function issuing power consumption variation tendency.

Technical scheme of the present invention is:

Based on a region electric power consumer confidence index construction method for multiple dimensioned economic leading indicators, comprise the following steps:

(1) from region, the whole nation and international three yardsticks choose some economic targets, set up multiple dimensioned economic target database;

(2) obtain the monthly data of the interval region power consumption of historical sample and each economic target, and seasonal adjustment is carried out to the monthly data of region power consumption;

(3) filter out the economic leading indicators of region power consumption, and determine the advanced issue of every economic leading indicators;

(4) based on the monthly data of the interval region power consumption of historical sample and every economic leading indicators, the monthly prediction regression model of region power consumption is built;

(5) according to the advanced issue of every economic leading indicators, the monthly data of the monthly corresponding issue of the advanced target of every economic leading indicators is obtained, then according to the monthly prediction regression model of region power consumption, the region power consumption that target of prediction is monthly;

(6) following formulae discovery is adopted to obtain the monthly region electric power consumer confidence index of target:

${\mathrm{EBI}}_t=0.5*\frac{Y_t-Y_{t-\mathrm{\Δ}t}}{Y_{t-\mathrm{\Δ}t}}+100$

Wherein, EBI _trepresent the region electric power consumer confidence index that target is monthly, Y _trepresent the region power consumption that target is monthly, Y _{t-Δ t}represent the region power consumption of advanced target monthly Δ t phase.

The described region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators, in described step (2), adopts seasonal adjustment multiplied model to carry out seasonal adjustment to the monthly data of region power consumption.

The described region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators, in described step (3), adopts K-L information Contents Method to filter out the economic leading indicators of region power consumption.

The described region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators, described step (4) comprising:

(41) based on akaike information criterion AIC and Bayesian Information amount criterion BIC, build respectively one with region power consumption be dependent variable, the economic leading indicators monthly prediction regression model of alternative area power consumption that is independent variable;

(42) two monthly prediction regression models of alternative area power consumption are tested, compare the error of fitting of two models, using model less for wherein error of fitting as the monthly prediction regression model of final region power consumption.

As shown from the above technical solution, the present invention is by extensively collecting the world, domestic and the multiple dimensioned economic target in region, incorporate developed country and domestic macroeconomy, industrial economy data, brisk market data etc., extract from some economic targets and change relevant information with international and domestic economic situation, and consider the time-lag effect between economic target and region power consumption, build the electric power consumer confidence index that can be used for estimation range power consumption tendency, efficiently solve independent variable index and need artificial selection, optimization model is difficult to the problem determined, for grid company programmed decision-making provides scientific basis.

Accompanying drawing explanation

Fig. 1 is method flow diagram of the present invention;

Correlation curve figure before and after the Analyzing Total Electricity Consumption seasonal adjustment of Tu2Shi Anhui Province;

Tu3Shi Anhui Province Analyzing Total Electricity Consumption monthly prediction Regression Model Simulator figure.

Embodiment

The present invention is further illustrated as specific embodiment below using the building process of Anhui Province's electric power consumer confidence index and practice.

As shown in Figure 1, a kind of region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators, comprises the following steps:

S1, from region, the whole nation and international three yardsticks choose some economic targets, set up multiple dimensioned economic target database:

From Anhui Province, the whole nation and international three yardsticks choose 143 economic targets altogether, and wherein, Anhui Province's economic target type comprises: the economic overall objective of (1) the whole province; (2) financial class; (3) investment type: gross fixed assets investment, Investment by Various Sectors complete volume; (4) real estate investment volume; (5) industrial added value; (6) consumption of resident: state revenue, state financial spending; (7) foreign trade; (8) industrial goods output etc.Whole nation pointer type comprises: (1) index number of operation of manufacturing industry; (2) nonmanufacturing industry index; (3) money supply; (4) international trade and international investment; (5) communications and transportation; (6) key industry consumer confidence index; (7) the branch trade added value etc. in Shanghai City, developed area.International indicator type comprises: (1) U.S., Japan and euro area foreign trade: total import value, total export; (2) U.S., Japan and euro area unemployment rate.

The monthly data of S2, the interval region power consumption of acquisition historical sample and each economic target, and seasonal adjustment is carried out to the monthly data of region power consumption:

With in January, 2008 ~ 2014 year July for historical sample is interval, obtain the monthly data of Anhui Province's Analyzing Total Electricity Consumption and 143 economic targets, and seasonal adjustment is carried out to the monthly data of Anhui Province's Analyzing Total Electricity Consumption, to deduct the impact of Seasonal.

The rising tendency of power consumption and seasonal trend are also deposited, both with additive or multiplication form form composite model.In order to eliminate the impact of seasonal factor on electricity demand forecasting model, the present invention utilizes seasonal adjustment multiplied model to carry out seasonal adjustment to the power consumption data collected, and as shown in Figure 2, is the correlation curve before and after the adjustment of Anhui Province's Analyzing Total Electricity Consumption.It should be noted that, the present invention adopts the data after seasonal adjustment when building electricity demand forecasting regression model, model prediction result is do not consider the value of seasonal factor, finally also needs predicted value to be multiplied by corresponding seasonal factor and just can finally be predicted the outcome.

S3, filter out the economic leading indicators of region power consumption, and determine the advanced issue of every economic leading indicators:

K-L information content analysis is carried out to Anhui Province's Analyzing Total Electricity Consumption and 143 economic targets, filter out the economic leading indicators of Anhui Province's Analyzing Total Electricity Consumption, result and parameter (note: " Anhui " after index name represents that this index is Anhui Province's economic target as shown in table 1 thereof, " Shanghai " represents that this index is Shanghai City economic target, and other are the whole nation or international economy index).

Table 1

K-L information Contents Method, in order to weigh the similarity degree of two probability distribution, namely measures the distance between them, and less i.e. both representatives are more close more similar.Its principle take reference index as theoretical distribution, and alternative index is sample distribution, constantly the alternative index of change and the reference index time difference, calculating K-L quantity of information.Time difference corresponding when K-L quantity of information is minimum is defined as the final time difference of alternative index.When alternative index and reference index completely the same time, K-L quantity of information equals 0; Alternative index and reference index more close, K-L quantity of information absolute value is less, more close to 0.

The present embodiment with Anhui Province's Analyzing Total Electricity Consumption for reference index, be alternative index with 143 economic targets, Calculation Basis index and each alternative index are from-12 rank to the K-L quantity of information size on+12 rank respectively, and select minimum K-L quantity of information and the time lag exponent number of correspondence thereof, thus obtain the best lag period of all alternative indexs relative to reference index.If the best lag period of certain alternative index is less than 0, then this alternative index is screened is economic leading indicators.

S4, monthly data based on the interval region power consumption of historical sample and every economic leading indicators, build the monthly prediction regression model of region power consumption:

Historical sample interval is divided into two parts, wherein the historical data in January, 2008 ~ 2014 year April is used for setting up the monthly prediction regression model of Anhui Province's Analyzing Total Electricity Consumption, and the monthly prediction regression model of Anhui Province's Analyzing Total Electricity Consumption that the historical data in May, 2014 ~ in July, 2014 is used for setting up is tested, assessed.

32 the economic leading indicators (as shown in table 1) filtered out are converted according to its advanced issue, then as independent variable, with Anhui Province's Analyzing Total Electricity Consumption for dependent variable, use akaike information criterion AIC (Akaike Information Criterion), Bayesian Information amount criterion BIC (Bayesian Information Criterion) two kinds screens principle and chooses optimum prediction regression model respectively, again according to the fit solution of two kinds of optimum prediction regression models, choose the model that error of fitting is less, through comparing, the optimum prediction regression model selected based on AIC criterion becomes the final monthly prediction regression model of Anhui Province's Analyzing Total Electricity Consumption, output model parameter and fitting result, as shown in table 2.Wherein, conspicuousness: in the level of 0.001 significantly, " * * " represents remarkable in the level of 0.01 in " * * * " expression, and " * " represents remarkable in the level of 0.05.

Table 2

S5, advanced issue according to every economic leading indicators, obtain the monthly data of the monthly corresponding issue of the advanced target of every economic leading indicators, then according to the monthly prediction regression model of region power consumption, the region power consumption that target of prediction is monthly;

The fitting result of the monthly prediction regression model of Anhui Province's Analyzing Total Electricity Consumption as shown in Figure 3.Model is tested and Fitting Analysis, result show, the regression criterion of model meets white noise characteristics, analog result through seasonal adjustment reduction after match value and actual value variation tendency be consistent, matching average error rate is 1.8%.Fitting result is better, and this model can be used for the Anhui Province's Analyzing Total Electricity Consumption predicted monthly.

S6, following formulae discovery is adopted to obtain the monthly region electric power consumer confidence index of target:

${\mathrm{EBI}}_t=0.5*\frac{Y_t-Y_{t-\mathrm{\Δ}t}}{Y_{t-\mathrm{\Δ}t}}+100$

Wherein, EBI _trepresent the region electric power consumer confidence index that target is monthly, Y _trepresent the region power consumption that target is monthly, Y _{t-Δ t}represent the region power consumption of advanced target monthly Δ t phase.

For Δ t=2, calculate the electric power consumer confidence index in April, 2013 ~ 2014, Anhui Province year March, as shown in table 3.

Month	Amplification	Consumer confidence index	Month	Amplification	Consumer confidence index
						In May, 2013	3％	101	In November, 2013	-9％	95
In June, 2013	7％	103	In Dec, 2013	11％	105
						In July, 2013	22％	111	In January, 2014	14％	107
In August, 2013	29％	114	In February, 2014	-16％	92
						In September, 2013	-9％	95	In March, 2014	-2％	99
In October, 2013	-22％	89	In April, 2014	7％	103

Table 3

The present invention is based on the impact that region power consumption is subject to international and domestic economic situation to a certain extent, and there is certain time-lag effect, thus build region electric power consumer confidence index by economic leading indicators estimation range power consumption.Region electric power consumer confidence index is except except the turning point of estimation range electricity consumption cyclic swing, can reflecting the amplitude of electricity consumption cyclic swing in some sense, have ageing and degree of accuracy preferably.

The above embodiment is only be described the preferred embodiment of the present invention; not scope of the present invention is limited; under not departing from the present invention and designing the prerequisite of spirit; the various distortion that those of ordinary skill in the art make technical scheme of the present invention and improvement, all should fall in protection domain that claims of the present invention determine.

Claims (4)

1., based on a region electric power consumer confidence index construction method for multiple dimensioned economic leading indicators, it is characterized in that, comprise the following steps:

(1) from region, the whole nation and international three yardsticks choose some economic targets, set up multiple dimensioned economic target database;

(2) obtain the monthly data of the interval region power consumption of historical sample and each economic target, and seasonal adjustment is carried out to the monthly data of region power consumption;

(3) filter out the economic leading indicators of region power consumption, and determine the advanced issue of every economic leading indicators;

(4) based on the monthly data of the interval region power consumption of historical sample and every economic leading indicators, the monthly prediction regression model of region power consumption is built;

(5) according to the advanced issue of every economic leading indicators, the monthly data of the monthly corresponding issue of the advanced target of every economic leading indicators is obtained, then according to the monthly prediction regression model of region power consumption, the region power consumption that target of prediction is monthly;

(6) following formulae discovery is adopted to obtain the monthly region electric power consumer confidence index of target:

${\mathrm{EBI}}_t=0.5*\frac{Y_t-Y_{t-\mathrm{\Δ}t}}{Y_{t-\mathrm{\Δ}t}}+100$

Wherein, EBI _trepresent the region electric power consumer confidence index that target is monthly, Y _trepresent the region power consumption that target is monthly, Y _{t-Δ t}represent the region power consumption of advanced target monthly Δ t phase.

2. the region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators according to claim 1, it is characterized in that: in described step (2), adopt seasonal adjustment multiplied model to carry out seasonal adjustment to the monthly data of region power consumption.

3. the region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators according to claim 1, is characterized in that: in described step (3), adopts K-L information Contents Method to filter out the economic leading indicators of region power consumption.

4. the region electric power consumer confidence index construction method based on multiple dimensioned economic leading indicators according to claim 1, it is characterized in that, described step (4) comprising:

(41) based on akaike information criterion AIC and Bayesian Information amount criterion BIC, build respectively one with region power consumption be dependent variable, the economic leading indicators monthly prediction regression model of alternative area power consumption that is independent variable;

(42) two monthly prediction regression models of alternative area power consumption are tested, compare the error of fitting of two models, using model less for wherein error of fitting as the monthly prediction regression model of final region power consumption.