CN102646216B - Energy demand prediction method based on S-shaped model - Google Patents

Abstract

An energy demand prediction method based on an S-shaped model belongs to the technical field of national, regional or industrial energy demand prediction. The method solves the problems that other prediction methods lack theoretical support, result deviation is large and reliability is low, an energy demand prediction model equation with universality and taking the human-average GDP as an independent variable is constructed according to an S-shaped physical model between the human-average energy consumption and the human-average GDP by applying mathematical methods such as a hyperbolic tangent function, and the like, relevant parameters are determined by analyzing historical data of different countries or regions, a specific equation of a prediction object is established, and the specific equation is substituted into the human-average GDP in a given prediction period to obtain a corresponding prediction result. The method has the advantages that the basic law of energy consumption is taken as guidance, the universal and targeted energy demand prediction method is established, the reliability and the confidence coefficient are high, and the method is practical, convenient and easy to realize and popularize.

Description

Based on the Energy Demand Forecast method of serpentine model

Technical field:

A kind of quantitative Energy Demand Forecast method of system of the present invention, be consume based on GDP per capita and per capita energy between serpentine physical model, use the forecasting techniques that the mathematical methods such as hyperbolic tangent function build, for medium-term and long-term Energy Demand Forecast, directly apply to the fields such as the energy is reconnoitred, develops, produces, utilizes, transported, trade.

Background technology:

Current Energy Demand Forecast method totally can be divided into qualitative forecasting and the large class of quantitative forecast two.

Qualitative forecasting judges to provide future source of energy demand according to subjective experience usually, and most is representational Delphi method and analogy method two kinds.Delphi method is judged by expert estimation assignment and determines future source of energy demand; Analogy rule, by the developed country or region analogy similar with development pattern, extracts its identical developing stage energy consumption indicators, and draws the energy demand of forecasting object on this basis.Qualitative forecasting has stronger subjectivity, lacks preciseness and science, and its accuracy and reliability are difficult to ensure.

Quantitative forecast is according to predicting that the approach difference chosen can be divided into several main method such as department's predicted method, Energy consumption intensity method, elastic coefficient method, inputoutput analysis method, energy total amount mathematical simulation method.(1) department's predicted method is on the basis of mutual relationship between the different department's energy-consuming historical data of system summary and sector development, by the judgement to all departments' future developing trend, and prediction department and even national energy demand.Current international main energy sources mechanism this method of many employings.Its major defect shows that the data related in one aspect to is numerous and jumbled, very easily produces statistically deviation; Its two, the judgement of department's future development is often flowed in the deduction from historical data in the past, lacks the assurance to department's energy-consuming basic law, thus produce prediction deviation; In addition, the energy kind related to from terminal department to primary energy transfer process and department numerous, conversion efficiency varies, and too increases and produces the risk of prediction deviation.Showing the history predictive result inspection that IEA and EIA uses this method to draw, there is more than 10% error mostly to the prediction of typical developing countries and regions in it; (2) Energy consumption intensity method is the historical variations according to Energy consumption intensity index, deduces and judges future trends, provides time span of forecast consumption intensity index value, and then draws corresponding energy demand.The deduction of the change of this method to following intensity index many foundations subjective judgement or historical trend, lacks the assurance to its inherent law, often occurs very large deviation in prediction.(3) elastic coefficient method and intensity method similar, by determining time span of forecast elasticity coefficient value indirect predictions future source of energy demand.Because the predicted value of elasticity coefficient index is for thinking setting, have stronger subjectivity and be difficult to precise value, and its small change all can causes the larger variation of total demand, therefore the accuracy of its predict is difficult to ensure.(4) inputoutput analysis method is the econometrics method of input and output relation of interdependence between various piece in research economic system (national economy, regional economy, sectional, company or business economic unit).The prerequisite using the method is that whole system is comparatively stable, and the relation in simultaneity factor between each unit is changeless, and the prerequisite of this method determines it and there is systematic uncertainty for medium-term and long-term demand forecast.(5) energy demand total amount mathematical simulation method refers to uses mathematical method simulation energy-consuming historical track, draw model equation, and predict tomorrow requirement with this, main method has gray prediction, regretional analysis etc., these methods are used for the energy and the socio-economic development system of the complexity of continuous evolution, are often limited to limited historical data, from time series or cause-effect relationship, be difficult to the inner link comprehensively, correctly between reflection things, predict the outcome and often there is very large error.

The common defects that above-mentioned these methods exist lacks the assurance to yardstick quantitative relationship long between energy demand and economic development, and only use digital simulation, deduction or analogy future in the past, prediction lacks scientific and accuracy, and result error is large.Improve the accuracy of medium-term and long-term Energy Demand Forecast, objectivity all has very important significance to the development that science formulates national planning and future source of energy industry and relevant enterprise.

Summary of the invention:

The present invention is directed to the deficiencies in the prior art, according to the serpentine physical model between the per capita energy's consumption built and GDP per capita, use the mathematical methods such as hyperbolic tangent function, define brand-new Quantitative Forecasting Technology, fundamentally solve the problem in the past predicting that shortage theory support, result error are large, with a low credibility.

The present invention specifically comprises the following steps and content:

Set up the serpentine physical model between per capita energy's consumption and GDP per capita

First, serpentine correlationship between per capita energy's consumption and GDP per capita is established;

Numerous developed countries such as system summary English, U.S., day are over 150 years, from agricultural society-industrial society-postindustrialization society's energy-consuming course, disclose per capita energy's consumption and present complete period serpentine variation relation with GDP per capita, namely agricultural society per capita energy consumption is in low and slow rising tendency, the stage of industrialization presents a rapidly rising trend, afterwards along with the transformation of economic structure, wealth of society accumulating level improve constantly and be gradually improved with infrastructure, per capita energy's demand culminates successively, is tending towards thereafter declining.

Secondly, determine per capita energy and consume serpentine three key points--takeoff point, turning point, zero growth rate point quantitative target;

The takeoff point of energy-consuming is the starting point that demand enters the high growth phase; Turning point place demand speedup then starts to slow down; Zero growth rate point is then the summit of energy demand.Three key points have relatively-stationary GDP per capita numerical value, takeoff point place GDP per capita concentrates on 2500-3000 dollar (Gai Kai, lower same), the corresponding GDP per capita 10000-12000 dollar of turning point, zero growth rate point place GDP per capita then concentrates between 20000-22000 dollar;

3rd, serpentine curve is divided into four sections by three key points, corresponding to the energy demand trend of different stages of development.

Curve is divided into and slowly increases district, increases district, speedup slowing-down area and zero growth rate district/negative growth four intervals fast by three key points, each region has the energy demand growth pattern relatively determined, thus forms the non-linear forecast of growth ultimate principle of serpentine.

Build Energy Demand Forecast equation

Per capita energy's consumption and the S shape track of GDP per capita from a Relative steady-state to another Relative steady-state, can adopt trend analysis, provide the trend relation equation of per capita energy consumption and GDP per capita.

First provide per capita energy and consume E and GDP per capita ^gfit equation:

$\underset{k=1}{\overset{j}{\Σ}}{a}_k\frac{d^{k}E}{{\mathrm{dG}}^k}=f(E,G)---\left(1\right)$

Here a _kfor undetermined constant, the polynomial expression that j=2, f (E, G) they are E and G or periodic function.

Then based on the analysis to data with existing, providing the simple concrete equation form of above formula is:

$\frac{dE}{dG}+{\mathrm{\σ}}_1{(E-E_i)}^2+{\mathrm{\σ}}_2=0---\left(2\right)$

Here σ ₁, σ ₂for undetermined constant, σ ₁σ ₂<0; And curve is at turning point P _i(G _i, E _i) there is equation at place:

$\frac{{\mathrm{\&Delta;}}^{2}E}{{\mathrm{\&Delta;G}}^2}=\frac{\mathrm{\&Delta;}}{\mathrm{\&Delta;}G}\frac{\mathrm{\&Delta;}E}{\mathrm{\&Delta;}G}=0---\left(3\right)$

The solution of equation (2) formula is hyperbolic tangent function

E-E _i＝Atanh(α(G-G _i))(4)

Wherein A is the amplitude of hyperbolic tangent function, and unit is identical with E, for undetermined constant.

Equation (4) formula describe curve smooth before takeoff point and behind summit time situation.When curve is non-flat forms, equation (4) formula then becomes:

$E-E_i=A\frac{\exp \left({\mathrm{\&alpha;}}_1\right(G-G_i\left)\right)-\exp (-{\mathrm{\&alpha;}}_3(G-G_i))}{2\cosh \left({\mathrm{\&alpha;}}_2\right(G-G_i))}---\left(5\right)$

Here α ₁, α ₂, α ₃for exponential constant, unit and G ^-1identical

According to the historical data determination correlation parameter of country variant, thus obtain the concrete predictive equation in each country (or area).

Setting prediction duration and relevant GDP per capita

According to country variant economic development planning or rising tendency, determine time span of forecast GDP per capita, as independent variable.

Use predictive equation to draw to predict the outcome accordingly

The GDP per capita of pre-timing points is substituted into predictive equation, draws corresponding Energy Demand Forecast result.

The invention has the advantages that with energy-consuming basic law for instruct, fundamentally solve the problem that the ubiquitous prediction deviation of Energy Demand Forecast is large in the past, improve predicting reliability and degree of confidence.Establish one and have universality and Energy Demand Forecast method targetedly concurrently, Forecasting Methodology practical convenient, be easy to realize and promote.

Accompanying drawing illustrates:

Fig. 1 per capita energy consumption and GDP per capita graph of a relation

Numerous developed country per capita energy consumption such as the U.S., Britain, France shows with the track of GDP per capita change, along with the growth of various countries' GDP per capita, per capita energy's consumption change in serpentine, namely from rise to quick growth at a low speed, thereafter speedup reduce and culminate (figure a); Country variant because of development model, energy-consuming custom difference serpentine can be divided into high, medium and low three classes (figure b)

Fig. 2 per capita energy consumes serpentine three key point schematic diagram

Per capita energy consumes serpentine and there are three key points, i.e. takeoff point, turning point and zero growth rate point, GDP per capita value corresponding to three key points is relatively fixing, takeoff point place GDP per capita is 2500-3000 dollar, turning point corresponds to 10000-12000 dollar, and the GDP per capita that zero growth rate point is corresponding concentrates on 20000-22000 dollar.Curve is four districts by three key points, and each region has the energy demand growth pattern relatively determined;

Fig. 3 is based on the Energy Demand Forecast method flow diagram of serpentine model

First this method builds serpentine physical model, thereafter Energy Demand Forecast model equation is set up based on this, by compiling the national historical data of prediction to the parameter assignment of model equation, set up prediction state energy demand equation, GDP per capita value in given prediction duration and time span of forecast, substitutes into equation and draws and predict the outcome.

The following 20 years Energy Demand Forecast result figure of Fig. 4 China

Embodiment:

For understanding technical scheme of the present invention better, below in conjunction with instantiation, embodiments of the present invention are described further.For the following 20 years Energy Demand Forecasts of China, concrete Forecasting Methodology following (Fig. 3):

1) China's per capita energy's consumption in nearly 60 years and GDP per capita data are extracted

2) determine equation (5) correlation parameter, it is as follows for obtaining Chinese energy demand equation:

E _m＝0.68+0.30tanh(0.0025(G _m-670))+0.20tanh(0.0007(G _m-2500))+0.30tanh(0.0008(G _m-5800))(6)

3) according to China's economic history rising tendency and following economic development planning in 20 years, GDP per capita in given time span of forecast

4) substitute into GDP per capita in time span of forecast, draw corresponding Energy Demand Forecast result (Fig. 4)

Claims (1)

1., based on the Energy Demand Forecast method of serpentine model, it is characterized in that:

Based on the serpentine physical model between per capita energy's consumption and GDP per capita, take GDP per capita as independent variable, use tanh mathematical method, energy-consuming math equation is set up in conjunction with country variant or regional historical data, thus the accurate quantitative analysis prediction realized country, region or the medium-term and long-term demand of the industry energy, concrete steps are as follows:

1) serpentine physical model between per capita energy's consumption and GDP per capita is set up;

2) utilization tanh mathematical method structure take GDP per capita as the Energy Demand Forecast equation of independent variable;

3) setting prediction duration and relevant GDP per capita value;

4) Energy Demand Forecast result is tried to achieve by predictive equation;

Wherein, step 2) in, utilization tanh mathematical method structure take GDP per capita as the Energy Demand Forecast equation of independent variable, and its implementation is as follows:

Per capita energy's consumption and the S shape track of GDP per capita from a Relative steady-state to another Relative steady-state, can adopt trend analysis, provide the trend relation equation of per capita energy consumption and GDP per capita:

First the fit equation that per capita energy consumes E and GDP per capita G is provided:

$\underset{k=1}{\overset{j}{\&Sigma;}}{a}_k\frac{d^{k}E}{{\mathrm{dG}}^k}=f(E,G)---\left(1\right)$

Wherein, a _kfor undetermined constant, the polynomial expression that j=2, f (E, G) they are E and G or periodic function;

Then based on the analysis to data with existing, providing the simple concrete equation form of above formula is:

$\frac{dE}{dG}+{\mathrm{\&sigma;}}_1{(E-E_i)}^2+{\mathrm{\&sigma;}}_2=0---\left(2\right)$

Wherein, σ ₁, σ ₂for undetermined constant, σ ₁σ ₂< 0; And curve is at turning point P _i(G _i, E _i) there is equation at place:

$\frac{{\mathrm{\&Delta;}}^{2}E}{{\mathrm{\&Delta;G}}^2}=\frac{\mathrm{\&Delta;}}{\mathrm{\&Delta;}G}\frac{\mathrm{\&Delta;}E}{\mathrm{\&Delta;}G}=0---\left(3\right)$

The solution of equation (2) formula is hyperbolic tangent function

E-E _i＝Atanh(α(G-G _i))(4)

Wherein A is the amplitude of hyperbolic tangent function, and unit is identical with E, for undetermined constant;

Equation (4) formula describe curve smooth before takeoff point and behind summit time situation, when curve is non-flat forms, equation (4) formula then becomes:

$E-E_i=A\frac{\exp \left({\mathrm{\&alpha;}}_1\right(G-G_i))-\exp (-{\mathrm{\&alpha;}}_3(G-G_i))}{2\cosh \left({\mathrm{\&alpha;}}_2\right(G-G_i))}---\left(5\right)$

Wherein, α ₁, α ₂, α ₃for exponential constant, unit and G ^-1identical.