CN106327028A - Terminal energy consumption prediction method and device - Google Patents

Abstract

The invention provides a terminal energy consumption prediction method and device. The method comprises the steps of acquiring energy consumption data; utilizing the energy consumption data for prediction through an electric energy substitution prediction model to acquire a first energy consumption prediction value and a first total energy consumption prediction value; utilizing the energy consumption data for prediction through a grey combined wavelet neural network model and a nonlinear regression model to acquire a second energy consumption prediction value, a second total energy consumption prediction value, a third energy consumption prediction value and a third total energy consumption prediction value; determining a virtual prediction value of a target year through terminal energy increments in target historical years; constructing a cross entropy model based on all the prediction values and determining a target energy consumption prediction value and a target total energy consumption prediction value through the cross entropy model. Through the terminal energy consumption prediction method and device, accurate prediction of the terminal electric energy consumption and total energy consumption in the prediction year is achieved.

Description

A kind of terminal energy sources predicting method for consumption and device

Technical field

The present invention relates to energy technology field, particularly relate to a kind of terminal energy sources predicting method for consumption and device.

Background technology

Along with socio-economic development, the contradiction that fossil energy of electricity needs and terminal is short, for electricity Feature and the advantages such as energy is convenient, safe and clean, efficient, actively pushing forward " using electricity instead of coal, replacing oil by electricity, electricity come " from a distant place is core The electric energy replacement work of intracardiac appearance, puts forth effort to improve electric energy proportion in final energy consumption, reduces terminated to greatest extent to the greatest extent The burning and exhausting of the stone energy has become the problem that China is in the urgent need to address.

In objective terms, all of primary energy can be converted to electric energy, and electric energy is relative to coal, oil, natural gas etc. More convenient, safety that the energy has and the advantage of cleaning, and other shape such as mechanical energy, heat energy can be converted to more conveniently The energy of formula, and realize accurate control.These characteristics of electric energy make it be used widely in modern economy society, electrified One of important symbol having become as modernization.Along with energy prices comparison relation progressively tends to reasonable, oil, Gas Prices To rise steadily, the electric energy competitiveness in final energy consumption market can further enhance, and electric energy will be made to substitute project and have more Add significant economic benefit.And the prediction to following electric energy with terminal energy sources can provide data for electric energy replacement work with analyzing Theoretical foundation, to instruct the propelling of future work.

Summary of the invention

In view of this, the invention provides a kind of terminal energy sources predicting method for consumption and device, in order to realize future Power consumption and the prediction of energy resource consumption total amount, its technical scheme is as follows:

A kind of terminal energy sources predicting method for consumption, described method includes:

Obtain target energy-output ratio data；

Utilize the described target energy-output ratio data acquisition electric energy replacement amount forecast model power consumption to target year Amount and energy resource consumption total amount are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount the first predictive value；

Utilize described target energy-output ratio data to be respectively adopted gray model and combine wavelet-neural network model, non-linear Power consumption and the energy resource consumption total amount of described target year are predicted by regression model, it is thus achieved that power consumption second is pre- Measured value and energy resource consumption total amount the second predictive value, and, power consumption flow control three predictive value and the total flow control of energy resource consumption three are predicted Value；

The virtual prognostication value of target year is determined by the terminal energy sources increment in target histories year；

Based on described power consumption the first predictive value and energy resource consumption total amount the first predictive value, described power consumption flow control Two predictive values and energy resource consumption total amount the second predictive value, described power consumption flow control three predictive value and energy resource consumption total flow control three are pre- Measured value and described virtual prognostication value build cross entropy model, and determine that target power consumption is pre-by described cross entropy model Measured value and target energy resource consumption Prediction of Total value.

Wherein, described utilize described target energy-output ratio data acquisition electric energy replacement amount forecast model to target year Power consumption and energy resource consumption total amount are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount first are pre- Measured value, including:

Power consumption rate of increase is calculated by the power consumption in described target energy-output ratio data；

The power consumption in described target energy-output ratio data and terminal energy sources total amount consumed is utilized to calculate electric energy Replacement amount；

By described power consumption rate of increase and described electric energy replacement amount predict described target year power consumption and Energy resource consumption total amount, it is thus achieved that described power consumption the first predictive value and energy resource consumption total amount the first predictive value.

Wherein, described described target energy-output ratio data acquisition gray model is utilized to combine wavelet-neural network model pair Power consumption and the energy resource consumption total amount of described target year are predicted, it is thus achieved that described power consumption the second predictive value and Energy resource consumption total amount the second predictive value, including:

Described gray model is used to calculate the match value of described target energy-output ratio data；

Residual values is calculated with described match value based on described target energy-output ratio data；

Use described Wavelet-network model that described residual values is trained, it is thus achieved that training result；

Described power consumption the second predictive value and energy resource consumption is determined by described match value and described training result Total amount the second predictive value；

Described utilize the described target energy-output ratio data acquisition nonlinear regression model (NLRM) electric energy to described target year Consumption and energy resource consumption total amount are predicted, it is thus achieved that described power consumption flow control three predictive value and energy resource consumption total flow control three are pre- Measured value, including:

Described target energy-output ratio data are utilized to set up nonlinear regression model (NLRM)；

Described nonlinear regression model (NLRM) is converted into linear model；

Estimate the parameter of described linear model, and set up terminal energy sources consumption predictive equation by described parameter；

Described power consumption flow control three predictive value and energy resource consumption is determined by described terminal energy sources consumption predictive equation Total flow control three predictive value.

Wherein, described based on described power consumption the first predictive value and energy resource consumption total amount the first predictive value, described electricity Can consumption the second predictive value and energy resource consumption total amount the second predictive value, described power consumption flow control three predictive value and energy resource consumption Total flow control three predictive value and described virtual prognostication value build cross entropy model, and determine target electricity by described cross entropy model Energy consumption predictive value and target energy resource consumption Prediction of Total value, including:

Based on described power consumption the first predictive value, described power consumption the second predictive value and described power consumption 3rd predictive value and described virtual prognostication value build the cross entropy model of power consumption, and by the total flow control of described energy resource consumption One predictive value, described energy resource consumption total amount the second predictive value, described energy resource consumption total flow control three predictive value and described virtual pre- Measured value builds the cross entropy model of energy resource consumption total amount；

By the cross entropy model of described power consumption determine the weight coefficient of power consumption, probability density function with Energy resource consumption flow function, and by the cross entropy model of described energy resource consumption total amount determine energy resource consumption total amount weight coefficient, Probability density function and energy resource consumption flow function；

Described target is determined with energy resource consumption flow function by the weight coefficient of described power consumption, probability density function Power consumption predictive value, and by the weight coefficient of described energy resource consumption total amount, probability density function and energy-output ratio letter Number determines described target energy resource consumption Prediction of Total value.

Described terminal energy sources predicting method for consumption also includes:

Described target year is determined based on described target power consumption predictive value and target energy resource consumption Prediction of Total value The electric energy replacement amount of degree and electric energy potential index.

A kind of terminal energy sources consumption prediction means, described device includes:

Acquisition module, is used for obtaining target energy-output ratio data；

First prediction module, is used for utilizing described target energy-output ratio data acquisition electric energy replacement amount forecast model to mesh Power consumption and the energy resource consumption total amount in mark year are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption are total Flow control one predictive value；

Second prediction module, is used for utilizing described target energy-output ratio data to be respectively adopted gray model and combines small echo god Through network model, power consumption and the energy resource consumption total amount of described target year are predicted, it is thus achieved that power consumption second Predictive value and energy resource consumption total amount the second predictive value；

3rd prediction module, for utilizing nonlinear regression model (NLRM) to disappear power consumption and the energy of described target year Consumption total amount is predicted, it is thus achieved that power consumption flow control three predictive value and energy resource consumption total flow control three predictive value；

4th prediction module, for determining the virtual pre-of target year by the terminal energy sources increment in target histories year Measured value；

Target prediction value determines module, for based on described power consumption the first predictive value and energy resource consumption total amount first Predictive value, described power consumption the second predictive value and energy resource consumption total amount the second predictive value, described power consumption flow control three are pre- Measured value and energy resource consumption total flow control three predictive value and described virtual prognostication value build cross entropy model, and by described cross entropy Model determines target power consumption predictive value and target energy resource consumption Prediction of Total value.

Wherein, described first prediction module includes, including:

Power consumption Growth Rate Calculation submodule, for by the power consumption in described target energy-output ratio data Calculate power consumption rate of increase；

Electric energy substitute gauge operator module, for utilize the power consumption in described target energy-output ratio data and Terminal energy sources total amount consumed calculates electric energy replacement amount；

First prediction submodule, for predicting described target by described power consumption rate of increase and described electric energy replacement amount The power consumption in year and energy resource consumption total amount, it is thus achieved that described power consumption the first predictive value and energy resource consumption total amount first Predictive value.

Wherein, described second prediction module, including:

Match value calculating sub module, for using described gray model to calculate the matching of described target energy-output ratio data Value；

Residual values calculating sub module, for calculating residual error based on described target energy-output ratio data with described match value Value；

Training submodule, is used for using described Wavelet-network model to be trained described residual values, it is thus achieved that training result；

Second prediction submodule, for determining described power consumption flow control by described match value and described training result Two predictive values and energy resource consumption total amount the second predictive value；

Described 3rd prediction module, including:

Building Nonlinear Model submodule, is used for utilizing described target energy-output ratio data to set up nonlinear regression mould Type；

Model conversion submodule, for being converted into linear mould by described nonlinear regression model (NLRM)；

Predictive equation sets up submodule, for estimating the parameter of described linear model, and sets up terminal by described parameter Energy-output ratio predictive equation；

3rd prediction submodule, for determining described power consumption flow control by described terminal energy sources consumption predictive equation Three predictive values and energy resource consumption total flow control three predictive value.

Wherein, described target prediction value determines module, including:

Submodule set up by cross entropy model, for based on described power consumption the first predictive value, described power consumption Second predictive value and described power consumption flow control three predictive value and described virtual prognostication value build the cross entropy mould of power consumption Type, and total by described energy resource consumption total amount the first predictive value, described energy resource consumption total amount the second predictive value, described energy resource consumption Flow control three predictive value and described virtual prognostication value build the cross entropy model of energy resource consumption total amount；

First determines submodule, for being determined the weight of power consumption by the cross entropy model of described power consumption Coefficient, probability density function and energy resource consumption flow function, and determine energy resource consumption by the cross entropy model of energy resource consumption total amount The weight coefficient of total amount, probability density function and energy resource consumption flow function；

Second determines submodule, for being disappeared with the energy by weight coefficient, the probability density function of described power consumption Consumption function determines described target power consumption predictive value, and close by weight coefficient, the probability of described energy resource consumption total amount Degree function determines described target energy resource consumption Prediction of Total value with energy resource consumption flow function.

Described terminal energy sources consumption prediction means also includes:

Evaluation index determines module, for based on described target power consumption predictive value and target energy resource consumption total amount Predictive value determines electric energy replacement amount and the electric energy potential index of described target year.

Technique scheme has the advantages that

The terminal energy sources predicting method for consumption of present invention offer and device, available energy consumption data is respectively adopted Electric energy replacement amount forecast model, gray model combine wavelet-neural network model, nonlinear regression model (NLRM) end to target year End energy-output ratio is predicted, and predicts the outcome based on each and utilize cross entropy model to determine final predictive value, i.e. this Invention can realize the Accurate Prediction to target year terminal energy sources consumption (power consumption and energy resource consumption total amount), it was predicted that number According to providing data theory foundation for electric energy replacement work, to instruct the propelling of future work.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this Inventive embodiment, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to according to The accompanying drawing provided obtains other accompanying drawing.

One schematic flow sheet of the terminal energy sources predicting method for consumption that Fig. 1 provides for the embodiment of the present invention；

Another schematic flow sheet of the terminal energy sources predicting method for consumption that Fig. 2 provides for the embodiment of the present invention；

In the terminal energy sources predicting method for consumption that Fig. 3 provides for the embodiment of the present invention, utilize target energy-output ratio number According to the realization side using electric energy replacement amount forecast model that power consumption and the energy resource consumption total amount of target year are predicted The schematic flow sheet of formula；

In the terminal energy sources predicting method for consumption that Fig. 4 provides for the embodiment of the present invention, utilize target energy-output ratio number According to using gray model to combine wavelet-neural network model, power consumption and the energy resource consumption total amount of target year are carried out pre- The schematic flow sheet of the implementation surveyed；

In the terminal energy sources predicting method for consumption that Fig. 5 provides for the embodiment of the present invention, utilize target energy-output ratio number According to using nonlinear regression model (NLRM) to the implementation that power consumption and the energy resource consumption total amount of target year are predicted Schematic flow sheet；

In the terminal energy sources predicting method for consumption that Fig. 6 provides for the embodiment of the present invention, pre-based on power consumption first Measured value and energy resource consumption total amount the first predictive value, power consumption the second predictive value and energy resource consumption total amount the second predictive value, electricity Flow control three predictive value and energy resource consumption total flow control three predictive value can be consumed and virtual prognostication value builds cross entropy model, and pass through Cross entropy model determines the flow process of the implementation of target power consumption predictive value and target energy resource consumption Prediction of Total value Schematic diagram；

The structural representation of the terminal energy sources consumption prediction means that Fig. 7 provides for the embodiment of the present invention.

Detailed description of the invention

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Describe, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments wholely.Based on Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under not making creative work premise Embodiment, broadly falls into the scope of protection of the invention.

Embodiments provide a kind of terminal energy sources predicting method for consumption, refer to Fig. 1, it is shown that the method Schematic flow sheet, may include that

Step S101: obtain target energy-output ratio data.

Step S102: utilize the target energy-output ratio data acquisition electric energy replacement amount forecast model electric energy to target year Consumption and energy resource consumption total amount are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount first are predicted Value.

Step S103: utilize target energy-output ratio data be respectively adopted gray model combine wavelet-neural network model, Power consumption and the energy resource consumption total amount of target year are predicted by nonlinear regression model (NLRM), it is thus achieved that power consumption second Predictive value and energy resource consumption total amount the second predictive value, and, power consumption flow control three predictive value and energy resource consumption total flow control three are pre- Measured value.

Step S104: determined the virtual prognostication value of target year by the terminal energy sources increment in target histories year.

Step S105: based on power consumption the first predictive value and energy resource consumption total amount the first predictive value, power consumption Second predictive value and energy resource consumption total amount the second predictive value, power consumption flow control three predictive value and the total flow control of energy resource consumption three are predicted Value and virtual prognostication value build cross entropy model, and determine target power consumption predictive value and mesh by cross entropy model Mark energy resource consumption Prediction of Total value.

The terminal energy sources predicting method for consumption that the embodiment of the present invention provides, available energy consumption data is respectively adopted Electric energy replacement amount forecast model, gray model combine wavelet-neural network model, nonlinear regression model (NLRM) end to target year End energy-output ratio is predicted, and predicts the outcome based on each and utilize cross entropy model to determine final predictive value, i.e. this Inventive embodiments can realize the Accurate Prediction to target year terminal energy sources consumption (power consumption and energy resource consumption total amount), Prediction data can provide data theory foundation for electric energy replacement work, to instruct the propelling of future work.

Refer to Fig. 2, it is shown that another flow process of the terminal energy sources predicting method for consumption that the embodiment of the present invention provides is shown Being intended to, the method may include that

Step S201: obtain target energy-output ratio data.

Step S202: utilize the target energy-output ratio data acquisition electric energy replacement amount forecast model electric energy to target year Consumption and energy resource consumption total amount are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount first are predicted Value.

Step S203: utilize target energy-output ratio data be respectively adopted gray model combine wavelet-neural network model, Power consumption and the energy resource consumption total amount of target year are predicted by nonlinear regression model (NLRM), it is thus achieved that power consumption second Predictive value and energy resource consumption total amount the second predictive value, and, power consumption flow control three predictive value and energy resource consumption total flow control three are pre- Measured value.

Step S204: determined the virtual prognostication value of target year by the terminal energy sources increment in target histories year.

Step S205: based on power consumption the first predictive value and energy resource consumption total amount the first predictive value, power consumption Second predictive value and energy resource consumption total amount the second predictive value, power consumption flow control three predictive value and the total flow control of energy resource consumption three are predicted Value and virtual prognostication value build cross entropy model, and determine target power consumption predictive value and mesh by cross entropy model Mark energy resource consumption Prediction of Total value.

Step S206: determine target year based on target power consumption predictive value and target energy resource consumption Prediction of Total value The electric energy replacement amount of degree and electric energy potential index.

The terminal energy sources predicting method for consumption that the embodiment of the present invention provides, available energy consumption data is respectively adopted Electric energy replacement amount forecast model, gray model combine wavelet-neural network model, nonlinear regression model (NLRM) end to target year End energy-output ratio is predicted, and predicts the outcome based on each and utilize cross entropy model to determine final predictive value, also may be used Electric energy replacement amount and the electric energy potential index of forecast year is determined based on final predictive value.The i.e. embodiment of the present invention not only can be real The now Accurate Prediction to target year terminal energy sources consumption (power consumption and energy resource consumption total amount), is also based on predicting number Carrying out quantitatively evaluating according to electric energy is substituted potentiality, these data can provide data theory foundation for electric energy replacement work, to refer to Lead the propelling of future work.

In the terminal energy sources predicting method for consumption that any of the above-described embodiment provides, electric energy replacement amount forecast model supposes Forecast year power consumption, on the premise of consistent with rate of increase the previous year, keep the electric energy replacement amount of the previous year, refers to Fig. 3, it is shown that utilize the target energy-output ratio data acquisition electric energy replacement amount forecast model power consumption to target year And energy resource consumption total amount is predicted, it is thus achieved that power consumption the first predictive value and the realization of energy resource consumption total amount the first predictive value The schematic flow sheet of mode, may include that

Step S301: calculate power consumption rate of increase by the power consumption in target energy-output ratio data.

Concrete, the power consumption rate of increase k of t_e,tCalculated by following formula:

$k_{e,t}=\frac{Y_{e,t}-Y_{e,t-1}}{Y_{e,t}}\×100\%$

Wherein, Y_e,tIt it is t power consumption.

Step S302: utilize the power consumption in target energy-output ratio data and terminal energy sources total amount consumed to calculate Electric energy replacement amount.

It is ε that the present invention defines electric energy replacement amount_e,t, its represent electric energy account for terminal energy sources consume proportion constant on the premise of, phase The electric energy amount of increasing consumption of relatively the previous year.Concrete, electric energy replacement amount is ε_e,tAvailable following formula (1) calculating electric energy replacement amount:

${\mathrm{\ϵ}}_{e,t}=Y_{e,t}-\frac{Y_{e,t-1}}{Y_{t-1}}\·Y_t---\left(1\right)$

Wherein, Y_tIt is t terminal energy resource consumption total amount, Y_t-1It is t-1 terminal energy resource consumption total amount, Y_e,tIt is t The terminal power consumption in year, Y_e,t-1It it is the terminal power consumption of t-1.

Step S303: predicted power consumption and the energy of target year by power consumption rate of increase and electric energy replacement amount Total amount consumed, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount the first predictive value.

Concrete, t+1 terminal power consumptionCalculated by following formula (2):

${\hat{Y}}_{e,t+1}=(1+k_{e,t})\·Y_{e,t}+{\mathrm{\ϵ}}_{e,t}---\left(2\right)$

Terminal power consumption the first predictive value Y obtained by electric energy replacement amount forecast model_{1_e,t}And terminal energy sources Total amount consumed the first predictive value Y_{1_t}。

In the terminal energy sources predicting method for consumption that any of the above-described embodiment provides, Lycoperdon polymorphum Vitt is used to combine Wavelet Neural Network Terminal power consumption and terminal energy sources total amount consumed are entered by network built-up pattern, two kinds of Forecasting Methodologies of nonlinear regression model (NLRM) respectively Row prediction.Relatively big to the predictive value relatively actual value error of energy-output ratio in view of gray model, the present invention uses gray model In conjunction with the built-up pattern of wavelet-neural network model, the prediction residual value drawn gray model by wavelet neural network is carried out Revise, to improve the accuracy of prediction.Refer to Fig. 4, it is shown that the terminal energy sources consumption prediction side that above-described embodiment provides In method, utilize target energy-output ratio data acquisition gray model to combine wavelet-neural network model and the electric energy of target year is disappeared Consumption and energy resource consumption total amount are predicted, it is thus achieved that power consumption the second predictive value and energy resource consumption total amount the second predictive value The schematic flow sheet of implementation, including:

Step S401: use gray model to calculate the match value of target energy-output ratio data.

Assume that target energy-output ratio data sequence is y⁽⁰⁾T (), then calculate matching by Lycoperdon polymorphum Vitt GM (1,1) model Value

Step S402: calculate residual values with match value based on target energy-output ratio data.

By target energy-output ratio data sequence y⁽⁰⁾(t) and match valueDiffer from, it is thus achieved that the residual sequence of moment t, It is denoted as e (t), it may be assumed that

$e\left(t\right)=y^{\left(0\right)}\left(t\right)-{\hat{y}}^{\left(0\right)}\left(t\right)---\left(3\right)$

Step S403: use wavelet-neural network model that residual values is trained, it is thus achieved that training result.

Wavelet-neural network model is used to be trained obtaining training result to residual sequence e (t)

Step S404: determine power consumption the second predictive value and energy resource consumption total amount by match value and training result Second predictive value.

Concrete, by the match value using Lycoperdon polymorphum Vitt GM (1,1) model to calculateWith employing wavelet neural network Model is trained the training result obtained to residual sequenceDetermine predictive value

${\hat{Y}}^{\left(0\right)}\left(t\right)={\hat{y}}^{\left(0\right)}\left(t\right)+\hat{e}\left(t\right)---\left(4\right)$

Combine wavelet-neural network model by gray model and obtain terminal power consumption the second predictive value Y_{2_e,t}And Terminal energy sources total amount consumed the second predictive value Y_{2_t}。

It is not linear regression simply owing to terminal energy sources consume growth trend, and presents non-linear growth feature, Therefore, by nonlinear regression model (NLRM), terminal energy sources consumption can be predicted.Refer to Fig. 5, it is shown that any of the above-described reality Execute in the terminal energy sources predicting method for consumption that example provides, utilize target energy-output ratio data acquisition nonlinear regression model (NLRM) pair Power consumption and the energy resource consumption total amount of target year are predicted, it is thus achieved that power consumption flow control three predictive value and energy resource consumption The schematic flow sheet of the implementation of total flow control three predictive value, the method may include that

Step S501: utilize target energy-output ratio data to set up nonlinear regression model (NLRM).

If end objectives energy-output ratio data sequence is Y_t(t=1,2 ..., T), set up Nonlinear regression equation as follows:

Y_t=a+bt+ct²+u_t, a, b, c ＞ 0 (5)

Wherein a, b, c are model parameter, u_tFor stochastic error.

Step S502: nonlinear regression model (NLRM) is converted into linear model.

Make t₁=t, t₂=t², thus obtain linear model:

Y=a+bt₁+ct₂+u (6)

Step S503: estimate the parameter of linear model, and set up terminal energy sources consumption predictive equation by parameter.

Utilize method of least square that model parameter is estimated, obtain estimates of parametersAnd then determine terminal Energy-output ratio predictive equation is:

${\hat{Y}}_t=\hat{a}+\hat{b}t+\hat{c}{t}^2---\left(7\right)$

Step S504: determine power consumption flow control three predictive value and energy resource consumption by terminal energy sources consumption predictive equation Total flow control three predictive value.

Terminal power consumption flow control three predictive value is designated as Y_{3_e,t}, terminal energy sources total amount consumed is designated as Y_{3_t}。

In the above-described embodiments, the virtual prognostication of target year is determined by the terminal energy sources increment in target histories year Value.Concrete, according to the situation of change of the influence factor such as GDP, population Yu terminal energy sources consumption, history year is divided into A, B, C, D Four classifications, such as: economic quickly growth, the time that electrified level significantly improves can be divided into A class；Economic steady-state growth, The time that electrified horizontal speed stabilizing improves can be divided into B class；Economic development is stable, and electrified level can without the time significantly improved It is divided into C class；Economical occur gliding, and electrified level stagnation or even the time declined can be divided into D class, wherein, each classification Middle terminal each energy resource consumption accounting and growth pattern all similar.

Assume that X apoplexy due to endogenous wind contains M_XIn individual history year, obtain M_XIndividual terminal energy sources increment { k_{X_m}(X=A, B, C, D) (m= 1,2 ..., M_X)。

Select annual classification X close with forecast year economic development, GDP and population rate of rise, obtain category M_X Individual terminal energy sources increment { k_{X_m}(m=1,2 ..., M_X), Y_{X_m, t}Virtual prognostication value as this forecast year:

Y_{X_m, t}=(1+k_{X_m})·Y_t-1, (m=1,2 ..., M_X) (8)

Refer to Fig. 6, it is shown that in the terminal energy sources predicting method for consumption that any of the above-described embodiment provides, based on electric energy Consumption the first predictive value and energy resource consumption total amount the first predictive value, power consumption the second predictive value and the total flow control of energy resource consumption Two predictive values, power consumption flow control three predictive value and energy resource consumption total flow control three predictive value and virtual prognostication value build cross entropy Model, and determine target power consumption predictive value and target energy resource consumption Prediction of Total value by cross entropy model, including:

Step S601: based on power consumption the first predictive value, power consumption the second predictive value and power consumption flow control Three predictive values and virtual prognostication value build the cross entropy model of power consumption, and by energy resource consumption total amount the first predictive value, Energy resource consumption total amount the second predictive value, energy resource consumption total flow control three predictive value and virtual prognostication value build energy resource consumption total amount Cross entropy model.

Step S602: determine the weight coefficient of power consumption, probability density by the cross entropy model of power consumption Function and energy resource consumption flow function, and the weight system of energy resource consumption total amount is determined by the cross entropy model of energy resource consumption total amount Number, probability density function and energy resource consumption flow function.

It should be noted that build the cross entropy model of power consumption true by the cross entropy model of power consumption Determine process and the structure energy resource consumption total amount of the weight coefficient of power consumption, probability density function and energy resource consumption flow function Cross entropy model, and cross entropy model based on energy resource consumption total amount determines the weight coefficient of energy resource consumption total amount, probability density Function is essentially identical with the process of energy resource consumption flow function, the present embodiment with build energy resource consumption total amount cross entropy model, and Cross entropy model based on energy resource consumption total amount determines the weight coefficient of energy resource consumption total amount, probability density function and energy resource consumption It is described in detail as a example by the process of flow function:

(1) definition terminal energy sources total amount consumed probability density function:

Assume that the terminal energy sources total amount consumed of t meets normal distribution, f_nX () is corresponding with the n-th forecast model The terminal energy sources total amount consumed probability density function in t year, it meets normal distribution:

$f_n\left(x\right)=\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}}{e}^{-{(x-{\mathrm{\μ}}_n)}^2/\left(2{{\mathrm{\σ}}_n}^2\right)}---\left(9\right)$

Wherein, μ_nFor meansigma methods, σ_nFor variance, n=1,2 ..., 3.

For the terminal energy sources total amount consumed predictive value Y obtained by the n-th forecast model_{n_t}Can predict as with n-th The terminal energy sources total amount consumed mean of a probability distribution μ that model is corresponding_nEven, μ_n=Y_{n_t}.Virtual prognostication based on target year Value can calculate sample variance σ of the terminal energy sources total amount consumed probability distribution corresponding with the n-th forecast model_{n t}:

${\mathrm{\σ}}_{n\_t}=\frac{1}{M_X}\underset{m=1}{\overset{M_X}{\mathrm{\Σ}}}{(Y_{X\_m,t}-Y_{n\_t})}^2---\left(10\right)$

If f (x) combines wavelet neural network built-up pattern and nonlinear regression model (NLRM) by cross entropy pair for using Lycoperdon polymorphum Vitt Electric energy replacement amount forecast model revised terminal energy sources total amount consumed probability distributing density function, meets:

$\underset{n=1}{\overset{N}{\Σ}}{\mathrm{\ω}}_n=1---\left(11\right)$

$f\left(x\right)=\underset{n=1}{\overset{N}{\Σ}}{\mathrm{\ω}}_n{f}_n\left(x\right)---\left(12\right)$

Wherein N=3, ω_nWeight coefficient for the probability density function corresponding with the n-th forecast model.

(2) set up and support vector, determine cross entropy object function:

Set up and support vector S:S=[S₁, S₂..., S_N]

S_n=D [f (x) | | f_n(x)]=∫ f (x) ln [f (x)/f_n(x)]dx

S.t. ∫ f (x) y (x) dx=E [y (x)] (13)

∫ f (x) dx=1

Wherein, y (x) is terminal energy sources total amount consumed functions, according to the t year terminal energy sources required by this probability density function It is mathematic expectaion during x that total amount consumed is y (x) value.

$A_n=S_n/\underset{n=1}{\overset{N}{\Σ}}S---\left(14\right)$

${\mathrm{\ω}}_n=\frac{1/A_n}{\underset{n=1}{\overset{N}{\Σ}}1/A_n}=\frac{1}{1+\underset{n=1,i\≠n}{\overset{N}{\Σ}}{A}_n/A_i}---\left(15\right)$

Set up the object function of minimum cross entropy optimization problem:

(3) minimum cross entropy optimization problem is solved:

Object function F is about weight coefficient ω_nFunction, use its optimization problem of DFP Algorithm for Solving, specifically wrap Include:

First, object function F is done 2 times and approximates Taylor expansions:

$F\left(\mathrm{\ω}\right)=F\left({\mathrm{\ω}}_0\right)+{(\mathrm{\ω}-{\mathrm{\ω}}_0)}^T\frac{\∂L}{\∂\mathrm{\ω}}+{(\mathrm{\ω}-{\mathrm{\ω}}_0)}^T/2A(\mathrm{\ω}-{\mathrm{\ω}}_0)---\left(17\right)$

In formula, ω is weight coefficient matrix, ω₀For initial weight coefficient matrix, A is extra large gloomy matrix, is made up of 2 rank local derviations Symmetrical matrix.

Secondly, above formula is differentiated, obtains the gradient vector of ω:

$G\left(\mathrm{\ω}\right)=\frac{\∂F\left(\mathrm{\ω}\right)}{\∂\mathrm{\ω}}+\frac{1}{2}(A+A^T)(\mathrm{\ω}-{\mathrm{\ω}}_0)---\left(18\right)$

Set up each searching method D_iRelation with corresponding gradient vector G:

D_i=-H_iG_i (19)

H_iIt is a symmetric positive definite matrix, initial matrix H₀For unit matrix

H_i+1=H_i+B_i+C_i (20)

$B_i=\frac{{\mathrm{\σ}}_i{{\mathrm{\σ}}_i}^T}{{{\mathrm{\σ}}_i}^T{z}_i}---\left(21\right)$

$C_i=\frac{-\left(H_i{z}_i\right){\left(H_i{z}_i\right)}^T}{{z_i}^T{H}_i{z}_i}---\left(22\right)$

σ_i=ω_i+1-ω_i (23)

z_i=G_i+1-G_i (24)

Above-mentioned steps is iterated, until searching minimum；

Above-mentioned steps gives the cross entropy model setting up terminal energy sources total amount consumed, and determines end by cross entropy model The mistake of weight coefficient matrix ω, probability density function f (x) and terminal energy sources total amount consumed function y (x) of end energy resource consumption total amount Journey, and for terminal power consumption, use same method to set up cross entropy model, and determined by cross entropy model The weight coefficient matrix ω of terminal power consumption_e, probability density function f_e(x) and terminal energy sources total amount consumed function y_e(x)。

Step S603: determine mesh with energy resource consumption flow function by the weight coefficient of power consumption, probability density function Mark power consumption predictive value, and by the weight coefficient of energy resource consumption total amount, probability density function and energy resource consumption flow function Determine target energy resource consumption Prediction of Total value.

By above-mentioned steps weight coefficient matrix ω and ω_eWith probability density function f (x) and f_eX () is calculated consumption letter Number y (x) and y_e(x), thus obtain marking energy resource consumption Prediction of Total valueWith target power consumption predictive value

${\hat{Y}}_t=E\[y\left(x\right)\]---\left(25\right)$

${\hat{Y}}_{e,t}=E\[y_e\left(x\right)\]---\left(26\right)$

After determining target power consumption predictive value and target energy resource consumption Prediction of Total value, can be based on target electricity Energy consumption predictive value and target energy resource consumption Prediction of Total value determine electric energy replacement amount and the electric energy potential index of target year.

Concrete, electric energy potential index uses Triangle Module to merge operator, and power consumption rate of increase and electric energy are accounted for terminal energy Source consumes proportion and merges, and expression is:

${\mathrm{\ϵ}}_{e,t}=\frac{{\hat{Y}}_{e,t}}{{\hat{Y}}_t}---\left(27\right)$

$k_{e,t}=\frac{{\hat{Y}}_{e,t}-{\hat{Y}}_{e,t-1}}{{\hat{Y}}_{e,t-1}}---\left(28\right)$

${\mathrm{\γ}}_t=\frac{{\mathrm{\ϵ}}_{e,t}\·k_{e,t}}{1-{\mathrm{\ϵ}}_{e,t}-k_{e,t}+2{\mathrm{\ϵ}}_{e,t}\·k_{e,t}}---\left(29\right)$

Wherein, ε_e,tIt is that t electric energy accounts for terminal energy sources consumption proportion,It is t power consumption predictive value,It is T terminal energy sources wastage in bulk or weight predictive value, k_e,tIt is t power consumption rate of increase, γ_tIt it is the electric energy potential index of t.

Corresponding with said method, the embodiment of the present invention additionally provides a kind of terminal energy sources consumption prediction means, please join Read Fig. 7, it is shown that the structural representation of this device, may include that acquisition module the 701, first prediction module 702, second is predicted Module the 703, the 3rd prediction module the 704, the 4th prediction module 705 and target prediction value determine module 706.Wherein:

Acquisition module 701, is used for obtaining target energy-output ratio data.

First prediction module 702, is used for utilizing target energy-output ratio data acquisition electric energy replacement amount forecast model to mesh Power consumption and the energy resource consumption total amount in mark year are predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption are total Flow control one predictive value.

Second prediction module 703, is used for utilizing target energy-output ratio data to be respectively adopted gray model and combines small echo god Through network model, power consumption and the energy resource consumption total amount of target year are predicted, it is thus achieved that power consumption second is predicted Value and energy resource consumption total amount the second predictive value.

3rd prediction module 704, for utilizing nonlinear regression model (NLRM) to disappear power consumption and the energy of target year Consumption total amount is predicted, it is thus achieved that power consumption flow control three predictive value and energy resource consumption total flow control three predictive value.

4th prediction module 705, for determining the void of target year by the terminal energy sources increment in target histories year Intend predictive value.

Target prediction value determines module 706, for based on power consumption the first predictive value and energy resource consumption total amount first Predictive value, power consumption the second predictive value and energy resource consumption total amount the second predictive value, power consumption flow control three predictive value and energy Source total amount consumed the 3rd predictive value and virtual prognostication value build cross entropy model, and determine target electric energy by cross entropy model Consumption predictive value and target energy resource consumption Prediction of Total value.

The terminal energy sources consumption prediction means that the embodiment of the present invention provides, available energy consumption data is respectively adopted Electric energy replacement amount forecast model, gray model combine wavelet-neural network model, nonlinear regression model (NLRM) end to target year End energy-output ratio is predicted, and predicts the outcome based on each and utilize cross entropy model to determine final predictive value, i.e. this Inventive embodiments can realize the Accurate Prediction to target year terminal energy sources consumption (power consumption and energy resource consumption total amount), Prediction data can provide data theory foundation for electric energy replacement work, to instruct the propelling of future work.

In the terminal energy sources consumption prediction means that above-described embodiment provides, the first prediction module includes: power consumption Growth Rate Calculation submodule, electric energy substitute gauge operator module and the first prediction submodule.

Power consumption Growth Rate Calculation submodule, for calculating by the power consumption in target energy-output ratio data Power consumption rate of increase.

Electric energy substitutes gauge operator module, for utilizing the power consumption in target energy-output ratio data and terminal Energy resource consumption total amount calculates electric energy replacement amount.

By power consumption rate of increase and electric energy replacement amount, first prediction submodule, for predicting that the electric energy of target year disappears Consumption and energy resource consumption total amount, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount the first predictive value.

In the terminal energy sources consumption prediction means that above-described embodiment provides, the second prediction module includes: match value meter Operator module, residual values calculating sub module, training submodule and the second prediction submodule.

Match value calculating sub module, for using gray model to calculate the match value of target energy-output ratio data.

Residual values calculating sub module, for calculating residual values based on target energy-output ratio data with match value.

Training submodule, is used for using Wavelet-network model to be trained described residual values, it is thus achieved that training result.

Second prediction submodule, for determining power consumption the second predictive value and energy by match value and training result Source total amount consumed the second predictive value.

In the terminal energy sources consumption prediction means that above-described embodiment provides, the 3rd prediction module includes: nonlinear model Submodule set up by type, model conversion submodule, predictive equation set up submodule and the 3rd prediction submodule.Wherein:

Building Nonlinear Model submodule, is used for utilizing target energy-output ratio data to set up nonlinear regression model (NLRM).

Model conversion submodule, for being converted into linear mould by nonlinear regression model (NLRM).

Predictive equation sets up submodule, for estimating the parameter of linear model, and sets up terminal energy sources consumption by parameter Amount predictive equation.

By terminal energy sources consumption predictive equation, 3rd prediction submodule, for determining that described power consumption flow control three is pre- Measured value and energy resource consumption total flow control three predictive value.

In the terminal energy sources consumption prediction means that above-described embodiment provides, target prediction value determines module, including: hand over Fork entropy model is set up submodule, first is determined that submodule and second determines submodule.

Submodule set up by cross entropy model, for predicting based on power consumption the first predictive value, power consumption second Value and power consumption flow control three predictive value and described virtual prognostication value build the cross entropy model of power consumption, and pass through the energy Total amount consumed the first predictive value, energy resource consumption total amount the second predictive value, energy resource consumption total flow control three predictive value and virtual prognostication Value builds the cross entropy model of energy resource consumption total amount.

First determines submodule, for being determined the weight system of power consumption by the cross entropy model of power consumption Number, probability density function and energy resource consumption flow function, and determine that energy resource consumption is total by the cross entropy model of energy resource consumption total amount The weight coefficient of amount, probability density function and energy resource consumption flow function.

Second determines submodule, for by weight coefficient, probability density function and the energy-output ratio of power consumption Function determines target power consumption predictive value, and by the weight coefficient of energy resource consumption total amount, probability density function and the energy Consumption function determines target energy resource consumption Prediction of Total value.

The terminal energy sources consumption prediction means that above-described embodiment provides can also include: evaluation index determines module.

Evaluation index determines module, for based on target power consumption predictive value and target energy resource consumption Prediction of Total Value determines electric energy replacement amount and the electric energy potential index of target year.

In this specification, each embodiment uses the mode gone forward one by one to describe, and what each embodiment stressed is and other The difference of embodiment, between each embodiment, identical similar portion sees mutually.

In several embodiments provided herein, it should be understood that disclosed method, device and equipment, permissible Realize by another way.Such as, device embodiment described above is only schematically, such as, and described unit Dividing, be only a kind of logic function and divide, actual can have other dividing mode, the most multiple unit or assembly when realizing Can in conjunction with or be desirably integrated into another system, or some features can be ignored, or does not performs.Another point, shown or The coupling each other discussed or direct-coupling or communication connection can be by between some communication interfaces, device or unit Connect coupling or communication connection, can be electrical, machinery or other form.

The described unit illustrated as separating component can be or may not be physically separate, shows as unit The parts shown can be or may not be physical location, i.e. may be located at a place, or can also be distributed to multiple On NE.Some or all of unit therein can be selected according to the actual needs to realize the mesh of the present embodiment scheme 's.It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it is also possible to be each Unit is individually physically present, it is also possible to two or more unit are integrated in a unit.

If described function is using the form realization of SFU software functional unit and as independent production marketing or use, permissible It is stored in a computer read/write memory medium.Based on such understanding, technical scheme is the most in other words The part contributing prior art or the part of this technical scheme can embody with the form of software product, this meter Calculation machine software product is stored in a storage medium, including some instructions with so that a computer equipment (can be individual People's computer, server, or the network equipment etc.) perform all or part of step of method described in each embodiment of the present invention. And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory (ROM, Read-Only Memory), random access memory are deposited The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic disc or CD.

Described above to the disclosed embodiments, makes professional and technical personnel in the field be capable of or uses the present invention. Multiple amendment to these embodiments will be apparent from for those skilled in the art, as defined herein General Principle can realize without departing from the spirit or scope of the present invention in other embodiments.Therefore, the present invention It is not intended to be limited to the embodiments shown herein, and is to fit to and principles disclosed herein and features of novelty phase one The widest scope caused.

Claims (10)

1. a terminal energy sources predicting method for consumption, it is characterised in that described method includes:

Obtain target energy-output ratio data；

Utilize described target energy-output ratio data acquisition electric energy replacement amount forecast model to the power consumption of target year and Energy resource consumption total amount is predicted, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount the first predictive value；

Utilize described target energy-output ratio data to be respectively adopted gray model and combine wavelet-neural network model, nonlinear regression Power consumption and the energy resource consumption total amount of described target year are predicted by model, it is thus achieved that power consumption the second predictive value And energy resource consumption total amount the second predictive value, and, power consumption flow control three predictive value and energy resource consumption total flow control three predictive value；

The virtual prognostication value of target year is determined by the terminal energy sources increment in target histories year；

Pre-based on described power consumption the first predictive value and energy resource consumption total amount the first predictive value, described power consumption second Measured value and energy resource consumption total amount the second predictive value, described power consumption flow control three predictive value and energy resource consumption total flow control three predictive value And described virtual prognostication value builds cross entropy model, and determine target power consumption predictive value by described cross entropy model And target energy resource consumption Prediction of Total value.

Terminal energy sources predicting method for consumption the most according to claim 1, it is characterised in that described utilize described target energy Source consumption data uses electric energy replacement amount forecast model to carry out pre-to power consumption and the energy resource consumption total amount of target year Survey, it is thus achieved that power consumption the first predictive value and energy resource consumption total amount the first predictive value, including:

Power consumption rate of increase is calculated by the power consumption in described target energy-output ratio data；

Utilize the power consumption in described target energy-output ratio data and terminal energy sources total amount consumed to calculate electric energy to substitute Amount；

Power consumption and the energy of described target year is predicted by described power consumption rate of increase and described electric energy replacement amount Total amount consumed, it is thus achieved that described power consumption the first predictive value and energy resource consumption total amount the first predictive value.

Terminal energy sources predicting method for consumption the most according to claim 1, it is characterised in that described utilize described target energy Source consumption data uses gray model to combine wavelet-neural network model to the power consumption of described target year and the energy Total amount consumed is predicted, it is thus achieved that described power consumption the second predictive value and energy resource consumption total amount the second predictive value, including:

Described gray model is used to calculate the match value of described target energy-output ratio data；

Residual values is calculated with described match value based on described target energy-output ratio data；

Use described Wavelet-network model that described residual values is trained, it is thus achieved that training result；

Described power consumption the second predictive value and energy resource consumption total amount is determined by described match value and described training result Second predictive value；

Described utilize the described target energy-output ratio data acquisition nonlinear regression model (NLRM) power consumption to described target year Amount and energy resource consumption total amount are predicted, it is thus achieved that described power consumption flow control three predictive value and the total flow control of energy resource consumption three are predicted Value, including:

Described target energy-output ratio data are utilized to set up nonlinear regression model (NLRM)；

Described nonlinear regression model (NLRM) is converted into linear model；

Estimate the parameter of described linear model, and set up terminal energy sources consumption predictive equation by described parameter；

Described power consumption flow control three predictive value and energy resource consumption total amount is determined by described terminal energy sources consumption predictive equation 3rd predictive value.

Terminal energy sources predicting method for consumption the most according to claim 1, it is characterised in that described disappear based on described electric energy Consumption the first predictive value and energy resource consumption total amount the first predictive value, described power consumption the second predictive value and energy resource consumption total amount Second predictive value, described power consumption flow control three predictive value and energy resource consumption total flow control three predictive value and described virtual prognostication value Build cross entropy model, and determine that target power consumption predictive value and target energy resource consumption are total by described cross entropy model Amount predictive value, including:

Based on described power consumption the first predictive value, described power consumption the second predictive value and described power consumption flow control three Predictive value and described virtual prognostication value build the cross entropy model of power consumption, and pre-by described energy resource consumption total amount first Measured value, described energy resource consumption total amount the second predictive value, described energy resource consumption total flow control three predictive value and described virtual prognostication value Build the cross entropy model of energy resource consumption total amount；

The weight coefficient of power consumption, probability density function and the energy is determined by the cross entropy model of described power consumption Consumption function, and determine the weight coefficient of energy resource consumption total amount, probability by the cross entropy model of described energy resource consumption total amount Density function and energy resource consumption flow function；

Described target electric energy is determined with energy resource consumption flow function by the weight coefficient of described power consumption, probability density function Consumption predictive value, and true with energy resource consumption flow function by weight coefficient, the probability density function of described energy resource consumption total amount Fixed described target energy resource consumption Prediction of Total value.

5. according to the terminal energy sources predicting method for consumption described in any one in claim 1-4, it is characterised in that described side Method also includes:

Described target year is determined based on described target power consumption predictive value and target energy resource consumption Prediction of Total value Electric energy replacement amount and electric energy potential index.

6. a terminal energy sources consumption prediction means, it is characterised in that described device includes:

Acquisition module, is used for obtaining target energy-output ratio data；

First prediction module, is used for utilizing described target energy-output ratio data acquisition electric energy replacement amount forecast model to target year Power consumption and the energy resource consumption total amount of degree are predicted, it is thus achieved that power consumption the first predictive value and the total flow control of energy resource consumption One predictive value；

Second prediction module, is used for utilizing described target energy-output ratio data to be respectively adopted gray model and combines Wavelet Neural Network Power consumption and the energy resource consumption total amount of described target year are predicted by network model, it is thus achieved that power consumption second is predicted Value and energy resource consumption total amount the second predictive value；

3rd prediction module, for utilizing nonlinear regression model (NLRM) total to power consumption and the energy resource consumption of described target year Amount is predicted, it is thus achieved that power consumption flow control three predictive value and energy resource consumption total flow control three predictive value；

4th prediction module, for determining the virtual prognostication of target year by the terminal energy sources increment in target histories year Value；

Target prediction value determines module, for predicting based on described power consumption the first predictive value and energy resource consumption total amount first Value, described power consumption the second predictive value and energy resource consumption total amount the second predictive value, described power consumption flow control three predictive value And energy resource consumption total flow control three predictive value and described virtual prognostication value build cross entropy model, and by described cross entropy model Determine target power consumption predictive value and target energy resource consumption Prediction of Total value.

Terminal energy sources consumption prediction means the most according to claim 6, it is characterised in that described first prediction module bag Include, including:

Power consumption Growth Rate Calculation submodule, for calculating by the power consumption in described target energy-output ratio data Power consumption rate of increase；

Electric energy substitutes gauge operator module, for utilizing the power consumption in described target energy-output ratio data and terminal Energy resource consumption total amount calculates electric energy replacement amount；

First prediction submodule, for predicting described target year by described power consumption rate of increase and described electric energy replacement amount Power consumption and energy resource consumption total amount, it is thus achieved that described power consumption the first predictive value and energy resource consumption total amount first are predicted Value.

Terminal energy sources consumption prediction means the most according to claim 6, it is characterised in that described second prediction module, Including:

Match value calculating sub module, for using described gray model to calculate the match value of described target energy-output ratio data；

Residual values calculating sub module, for calculating residual values based on described target energy-output ratio data with described match value；

Training submodule, is used for using described Wavelet-network model to be trained described residual values, it is thus achieved that training result；

By described match value and described training result, second prediction submodule, for determining that described power consumption second is pre- Measured value and energy resource consumption total amount the second predictive value；

Described 3rd prediction module, including:

Building Nonlinear Model submodule, is used for utilizing described target energy-output ratio data to set up nonlinear regression model (NLRM)；

Model conversion submodule, for being converted into linear mould by described nonlinear regression model (NLRM)；

Predictive equation sets up submodule, for estimating the parameter of described linear model, and sets up terminal energy sources by described parameter Consumption predictive equation；

By described terminal energy sources consumption predictive equation, 3rd prediction submodule, for determining that described power consumption flow control three is pre- Measured value and energy resource consumption total flow control three predictive value.

Terminal energy sources consumption prediction means the most according to claim 6, it is characterised in that described target prediction value determines Module, including:

Submodule set up by cross entropy model, for based on described power consumption the first predictive value, described power consumption second Predictive value and described power consumption flow control three predictive value and described virtual prognostication value build the cross entropy model of power consumption, and By described energy resource consumption total amount the first predictive value, described energy resource consumption total amount the second predictive value, the total flow control of described energy resource consumption Three predictive values and described virtual prognostication value build the cross entropy model of energy resource consumption total amount；

First determines submodule, for being determined the weight system of power consumption by the cross entropy model of described power consumption Number, probability density function and energy resource consumption flow function, and determine that energy resource consumption is total by the cross entropy model of energy resource consumption total amount The weight coefficient of amount, probability density function and energy resource consumption flow function；

Second determines submodule, for by weight coefficient, probability density function and the energy-output ratio of described power consumption Function determines described target power consumption predictive value, and by the weight coefficient of described energy resource consumption total amount, probability density letter Number and energy resource consumption flow function determine described target energy resource consumption Prediction of Total value.

10. according to the terminal energy sources consumption prediction means described in any one in claim 6-9, it is characterised in that described Device also includes:

Evaluation index determines module, for based on described target power consumption predictive value and target energy resource consumption Prediction of Total Value determines electric energy replacement amount and the electric energy potential index of described target year.