CN1753010A - Classification model construction and rolling derivation for energy source optimization management of iron and steel enterprise - Google Patents

Abstract

The present invention discloses a classification modeling and roll-solving method for optimizing dispatch of energy source of iron/steel enterprise. It is characterized by that said method includes the following steps: according to energy source equipment and pipe network characteristics in the iron/steel enterprise making classification and creating models; utilizing these models to obtain a nonlinear mathematical programming problem with constrained force; setting object function and selecting decision variable quantity to convert it into a large-scale linear programming problem, then adopting a high-efficiency primary dual interior point algorithm specially-designed for large-scale linear programming problem to make quick resolution.

Description

The classification model construction and the rolling method for solving that are used for iron and steel enterprise's energy source optimization scheduling

Technical field

The present invention relates to iron and steel enterprise's energy management field energy balance and dispatching method, relate in particular to the classification model construction and the rolling method for solving that are used for iron and steel enterprise's energy source optimization scheduling.

Background technology

Iron and steel enterprise is the big power consumer, and the rationality of enterprise energy dispatching method etc. is the key factor that influences iron and steel enterprise's energy consumption, and good energy scheduling method is the energy-conservation and important practical sense that reduced production costs to iron and steel enterprise.The mathematical model of setting up enterprise energy equipment is the prerequisite of energy source optimization scheduling, and the mathematical model that makes up the energy source optimization scheduling problem on model based is crucial.At present, the most of iron and steel enterprise of China adopts by the abnormal conditions of observation site energy source data and carries out the manual dispatching method, the dispatcher provides the energy scheduling scheme according to the personal experience, the quality of scheduling result and dispatcher's quality, responsibility, mood are closely bound up, the poor reliability of scheduling result.When the energy scheduling system scale increased, the dispatcher was because the restriction of self-condition often is difficult to provide suitable scheduling scheme.There are numerous technological processes and a large amount of energy devices in iron and steel enterprise, and traditional energy device modeling method is come modeling by technological process, will understand technological process in depth during modeling, and the quantity of model is very big, and has increased the difficulty and the complicacy of modeling.Existing energy scheduling is not all considered the effect of energy storage equipment in energy scheduling.

The present invention has analyzed the characteristics of iron and steel enterprise's equipment energy product consumption situation, and considered the effect of energy storage equipment in energy scheduling, equipment to whole enterprise is classified, individual equipment is being carried out on the basis of energy dynamic modeling, set up the secondary energy electric power of iron and steel enterprise, the medium-term and long-term Optimization Dispatching model of steam and coal gas utilizes the quick computing power of computing machine, obtains the reasonable optimizing scheduling strategy by finding the solution of Optimization Dispatching algorithm.

Thereby, the present invention is intended to set up a kind of simple relatively but effective energy device modeling method, produce the characteristics of consumption situation according to iron and steel enterprise's equipment energy, equipment, pipe network are classified, under this sorting technique, the equipment mathematics model tormulation relevant with energy scheduling simplified greatly, and can generally be applicable to the iron and steel enterprise that each are different.By algorithm being carried out unique processing, former belt restraining nonlinear mathematics planning problem is converted to a large scale linear programming problem, and adopt specially and find the solution for the original antithesis interior point method of high-level efficiency of large scale linear programming problem design, the reliability height has guaranteed the real-time of Optimization Dispatching problem solving.When finding the solution, consider that iron and steel enterprise is a huge complication system, the business equipment operation has abnormal conditions to take place unavoidably, for the error on the energy source optimization scheduling strategy that reduces to cause by accident at random, adopt a kind of event-driven rolling optimization scheduling strategy, handle the reliability of accident at random to improve scheduling strategy based on moving window.

Summary of the invention

The purpose of this invention is to provide a kind of classification model construction and rolling method for solving that is used for iron and steel enterprise's energy source optimization scheduling.

The classification model construction and the rolling method for solving basic step that are used for iron and steel enterprise's energy source optimization scheduling are as follows:

1) according to the characteristic of energy device in the iron and steel enterprise and pipe network, all devices is divided into following five kinds: energy stationary device, energy changeable type equipment, energy conversion type equipment, energy storage/buffer-type equipment, non-energy storage/buffer-type equipment;

2) adopt modelling by mechanism and data fitting algorithm that the said equipment is classified and set up mathematical model;

3) based on the model of classification model construction method foundation, obtain the nonlinear mathematics planning problem of a belt restraining, be converted into a large scale linear programming problem;

4) adopt a kind of event-driven rolling optimization calculative strategy based on moving window;

5) adopt former antithesis interior point method to carry out rapid solving.

Described energy stationary device: the generation of this kind equipment energy and consumption are definite value after production task is assigned, can not arbitrarily distribute.

Described energy changeable type equipment: this kind equipment is to finish the given production task can adopt different energy combinations, and its energy combination must be satisfied certain proportion relation constraint.

Described energy conversion type equipment; The output of this kind equipment energy and the consumption that produces other energy of these energy can be regulated.

Described energy storage/buffer device: this kind equipment itself does not participate in the product consumption of the energy and calculates, but can keep the balance in the pipe network, changes the balance mode of pipe network.

Described employing modelling by mechanism and data fitting algorithm are classified to the said equipment and are set up mathematical model: to energy stationary device, energy changeable type equipment, energy conversion type equipment, adopt the funtcional relationship between moving average filter algorithm, partial least square method match devices consume or the generation energy and the production task respectively, and can change the transformational relation between the energy mutually; Energy storage/buffer-type equipment is adopted the method for modelling by mechanism.

The described nonlinear mathematics planning problem that obtains a belt restraining, be converted into a large scale linear programming problem: the form of energy equation of constraint can be divided into changeable type energy device equation of constraint, the equation of constraint of energy conversion equipment, the balanced supply and demand of energy equation of pipe network; The equation of constraint of energy buffer type equipment does not satisfy successional requirement, presents nonlinear characteristic, by the setting of objective function and choosing of decision variable, former belt restraining nonlinear mathematics planning problem is converted to a large scale linear programming problem.

A kind of event-driven rolling optimization calculative strategy of described employing: during the incident burst based on moving window, determine device type and corresponding process constraint condition thereof, scheduling slot with temporary changes is a moving window, recomputates energy balance and Optimization Dispatching strategy.

The former antithesis interior point method of described employing carries out rapid solving: make iteration point near the feasible zone border time by the function of placing obstacles on the border of feasible zone, the target function value that provides increases rapidly, and in iterative process suitable control step-length, thereby make iteration point all the time in feasible zone inside, algorithm convergence is in the globally optimal solution of former problem.

The device class method that the present invention proposes has certain versatility for the iron and steel enterprise of all kinds and scale; The computing time of problem solving algorithm is insensitive to the scale of problem, and calculation times can not increase with the increase of problem scale, and good convergence and robustness are arranged, and the reliability height has guaranteed the real-time of Optimization Dispatching problem solving; Rolling optimization makes scheduling strategy have excellent adaptability to the abnormal conditions that equipment takes place, and improved the reliability that scheduling strategy is handled accident at random.

Embodiment

1. iron and steel enterprise's energy device category of model modeling

At iron and steel enterprise's equipment power consumption and the mode of production capacity and the not same-action that various device is risen in the energy source optimization scheduling process, the present invention proposes a kind of device class mode of whole iron and steel enterprise newly, specifically iron and steel enterprise's equipment is divided into following five classes: energy stationary device, energy changeable type equipment and energy conversion type equipment, energy storage/buffer-type equipment and non-energy storage/buffer-type equipment.This device class method is to classify according to the basis of equipment energy operating position and energy source optimization scheduling modeling, and therefore the iron and steel enterprise for all kinds and scale has certain versatility.

Classification declaration about equipment is as follows:

1) energy stationary device: in this kind equipment, after production task was assigned, the generation of the equipment energy and consumption were definite value, can not arbitrarily distribute, or can produce the delay of production task or the decline of product quality after arbitrarily distributing.Its mathematic(al) representation is

x(t)＝g( u(t))

Wherein x (t) represents energy stationary device energy resource consumption or generation; U (t) expression energy stationary device production task; The funtcional relationship of g (u (t)) expression energy stationary device energy resource consumption or generation and production task.

2) energy changeable type equipment: in this kind equipment, can adopt different energy sources combination, and its energy combination must satisfy certain proportion relation constraint, the constraint of energy use amount bound, and the energy constraint of regulating the speed for finishing the given production task.Its mathematic(al) representation is

c ₁x ₁(t)+c ₂x ₂(t)+...+c _nx _n(t)＝g( u(t))

C wherein _iThe coefficient of the consumption of expression energy changeable type equipment energy i; x _i(t) consumption of expression energy changeable type equipment energy i; U (t) expression energy changeable type device fabrication task; The funtcional relationship of g (u (t)) expression energy changeable type equipment energy resource consumption or generation and production task.

3) energy conversion type equipment: in this kind equipment, the output of certain energy (mainly comprising the generator that produces steam and electric power and boiler etc.) scalable, and the consumption that produces the various coal gas of these energy can be regulated.Its mathematic(al) representation is

y _j(t)＝h(x ₁(t)，x ₂(t)，…x _i(t)，…，y ₁(t)，y ₂(t)，…，y _i(t)，…)

X wherein _i(t) expression energy conversion type equipment produces the consumption that energy j needs energy i; y _i(t) expression energy conversion type equipment produces the generation that energy j produces energy i simultaneously; y _i(t) expression energy conversion type equipment produces the generation of energy j.

4) energy storage/buffer-type equipment: this kind equipment itself does not participate in the product consumption of the energy and calculates, but can keep the balance in the pipe network, changes the balance mode of pipe network.This kind equipment arrives on device security in limited time in its store/buffer ability, for guaranteeing that device security will produce the energy and diffuse its mathematic(al) representation and be:

$h\left(t_{i+1}\right)=\left\{\begin{array}{cc}{h}_{\max }& h\left(t_i\right)+k_1\mathrm{\&Delta;Q}\left(t_i\right)>h_{\max }\\ h\left(t_i\right)+k_1(Q_{\mathrm{in}}\left(t_i\right)-Q_{\mathrm{out}}\left(t_i\right))& h_{\min }\&le;h\left(t_i\right)+k_1\mathrm{\&Delta;Q}\left(t_i\right)\&le;h_{\max }\\ h_{\min }& h\left(t_i\right)+k_1\mathrm{\&Delta;Q}\left(t_i\right)<h_{\min }\end{array}\right.$

Or

$p\left(t_{i+1}\right)=\left\{\begin{array}{cc}{p}_{\max }& p\left(t_i\right)+k_2\mathrm{\&Delta;Q}\left(t_i\right)>p_{\max }\\ p\left(t_i\right)+k_2(Q_{\mathrm{in}}\left(t_i\right)-Q_{\mathrm{out}}\left(t_i\right))& p_{\min }\&le;h\left(t_i\right)+k_2\mathrm{\&Delta;Q}\left(t_i\right)\&le;p_{\max }\\ p_{\min }& p\left(t_i\right)+k_2\mathrm{\&Delta;Q}\left(t_i\right)<p_{\min }\end{array}\right.$

Q wherein _InThe energy influx of expression energy buffer equipment; Q _OutThe energy flow output of expression energy buffer equipment; k _iThe scale-up factor that the fluctuations in discharge of (i=1,2) expression energy buffer equipment and energy buffer equipment state change; Δ h represents the high variable quantity of energy buffer equipment (gas chamber); Δ p represents the pressure change amount of energy buffer equipment (steam pipe system); Δ Q (t _i)=Q _In(t _i)-Q _Out(t _i) energy fluctuations in discharge amount of expression energy buffer equipment; k ₁=T/S, S represent the floorage of gas chamber, and T represents the scheduling decision time interval; k ₂=RT ₁T/ (VM), V represents the volume of steam pipe system, and M represents the molal weight of steam, and R represents universal gas constant, 8.3143J/mol.K, T ₁The absolute temperature of steam in the expression steam pipe system.

5) non-energy storage/buffer-type equipment: in such energy device, do not have energy storage equipment or pipe network itself can not regard energy storage equipment as.Its mathematic(al) representation is:

Q _in-Q _out＝0

On the basis of above category of model, adopt modelling by mechanism and data fitting algorithm to set up mathematical model respectively.The variable that influences equipment capacity or power consumption,, be defined as the input variable of model as production task, product category etc.The output of model is defined as the energy consumption of equipment.Input variable is divided into conventional input and dummy variables input.Conventional input is generally continuous real number.The codomain of dummy variables input is some discrete points, such as the molten steel kind.To energy stationary device, energy changeable type equipment, energy conversion type equipment, historical data based on energy device, adopt the funtcional relationship between moving average filter algorithm, partial least square method match devices consume or the generation energy and the production task respectively, and can change the transformational relation between the energy mutually; To the method for energy storage/buffer-type equipment employing modelling by mechanism, according to the thermodynamics formula, utilize the parameter of energy buffer equipment, calculate the state variation situation of energy buffer equipment.

2. the structure of energy source optimization scheduling problem

When making up the energy source optimization scheduling problem, based on following condition precedent:

Condition A1: in each scheduling decision time interval T, but the energy sendout of energy controlling equipment and the production task amount of equipment are definite value, this also meets energy scheduling reality, by Fixed Time Interval the equipment energy is distributed and to call, can reduce equipment energy adjustment personnel's working strength, the equipment that also prevented is simultaneously adjusted the wearing and tearing that cause and problem such as wear out because of frequent.

Condition A2: energy changeable type equipment (as heating furnace, heat generator etc.) is finished production task by the linear combination of energy use amount, is nonlinear relationship between the linear combination of energy use amount and the production task.

Condition A3: be linear relationship between energy-output ratio and the generation in the energy conversion type equipment, in actual production, it also is the most economical reliable method of operation that power plant and boiler equal energy source conversion equipment are operated in linearity range.

Condition A4: the energy in every kind of energy pipe network can and be taken out the energy balance that reach pipe network by outsourcing.

On above condition precedent basis, the energy source optimization scheduling problem can be described as following mathematical form:

The Optimization Dispatching target: full enterprise finishes under the prerequisite of given production in assurance in whole scheduling slot, produces consumption by the energy of adjusting each production equipment, realizes at all energy pipe networks:

Sub-goal 1: maximization energy pipe network is taken out energy quantity;

Sub-goal 2: minimize energy pipe network outsourcing energy quantity;

Sub-goal 3: minimize the energy pipe network energy amount of diffusing; The wastage minimum of the energy, the maximization of economic benefit after the conversion in feasible the production.

The form of energy equation of constraint can be divided three classes: the changeable type energy produces consumption facility constraints equation, energy conversion facility constraints equation, balanced supply and demand of energy equation of constraint.Convenient for problem analysis, be defined as follows variable and numbering, establishing whole energy source optimization scheduling slot is [t ₀, t _f], the scheduling decision time interval is T, make have in the scheduling slot N scheduling time interval T, i.e. NT=t _f-t ₀, make t _i=[t ₀+ (i-1) T, t ₀+ iT], i=1 wherein ..., N.

Electric power, coal gas, steam pipe system is numbered k=1 ..., NK, NK energy pipe network altogether

Be respectively described below:

1) the changeable type energy produces consumption facility constraints equation

Make changeable type energy device (as heating furnace and soaking pit or the like) numbering s=1 ..., N _s

The changeable type energy produces consumption restricted model equation:

c _s1x _s1(t _i)+c _s2x _s2(t _i)+…＝g _s(u(t _i)) (1)

The a certain production task of changeable type finish to(for) needs produces the power consumption source device, and the needed energy will satisfy certain calorific value proportioning process requirement, the constraint of changeable type energy proportioning:

${\mathrm{\&lambda;}}_{s_{\min }\left(12\right)}\&le;x_{s1}\left(t_i\right):x_{s2}\left(t_i\right){\&le;\mathrm{\&lambda;}}_{s_{\max }\left(12\right)},{\mathrm{\&lambda;}}_{s_{\min }\left(13\right)}{\&le;x}_{s1}\left(t_i\right):x_{s3}\left(t_i\right)\&le;{\mathrm{\&lambda;}}_{s_{\max }\left(13\right)},\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;---\left(2\right)$

The constraint of changeable type energy use amount bound:

a _sk(min)≤x _sk(t _i)≤a _sk(max) (3)

The changeable type energy of the energy uses the constraint of regulating the speed:

|x _sk(t _i)-x _sk(t _i+1)|≤δ _sk (4)

2) energy conversion facility constraints equation

Make energy conversion equipment (as gas turbine, boiler etc.) be numbered l=1 ..., N _l

The energy of energy conversion equipment produces the model constrained equation of consumption:

y _lj(t)＝h(x _l1(t _i)，x _l2(t _i)，…x _li(t _i)，…，y _l1(t _i)，y _l2(t _i)，…，y _li(t _i)，…)?(5)

Energy conversion equipment also has the calorific value proportioning process constraint of energy-output ratio:

${\mathrm{\&lambda;}}_{l_{\min }\left(12\right)}\&le;x_{l1}\left(t_i\right):x_{l2}\left(t_i\right){\&le;\mathrm{\&lambda;}}_{l_{\max }\left(12\right)},{\mathrm{\&lambda;}}_{l_{\min }\left(13\right)}{\&le;x}_{l1}\left(t_i\right):x_{l3}\left(t_i\right)\&le;{\mathrm{\&lambda;}}_{l_{\max }\left(13\right)},\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;---\left(6\right)$

The bound constraint of energy use amount: a _{Lk (min)}≤ x _1k(t _i)≤a _{Lk (max)}(7)

The bound constraint of energy generation: b _{Lk (min)}≤ y _Lk(t _i)≤b _{Lk (max)}(8)

The energy constraint of regulating the speed: | x _Lk(t _o)-x _Lk(t _I+1) |≤δ _Lk, y _Lk(t _i)-y _Lk(t _I+1) |≤r _Lk(9)

1) balanced supply and demand of energy equation of constraint

Before energy balance equation was discussed, the various energy that at first will calculate energy stationary device produced the consumption situation.Produce the stationary device numbering j=1 of the energy ..., N _j, all produce the stationary device of the energy at t _iProduce total generation of the k kind energy in time period ${\stackrel{\&OverBar;}{y}}_k\left(t_i\right)=\underset{j=1}{\overset{N_j}{\mathrm{\&Sigma;}}}{g}_{\mathrm{jk}}^{\&prime;}\left({\stackrel{\&OverBar;}{u}}_j\left(t_i\right)\right)$

The stationary device numbering m=1 of consumes energy ..., N _Mk, the stationary device of all consumes energy is at t _iProduce the total flow of the k kind energy in time period ${\stackrel{\&OverBar;}{x}}_k\left(t_i\right)=\underset{m=1}{\overset{N_m}{\mathrm{\&Sigma;}}}{g}_{\mathrm{mk}}\left({\stackrel{\&OverBar;}{u}}_m\left(t_i\right)\right)$

Because the type difference of energy pipe network, the balanced supply and demand of energy equation of constraint of energy pipe network divides two classes:

I. non-energy storage/buffer-type pipe network I=1 such as electric power ... N _I, its energy balance equation is as follows

E _pI(t _i)-E _cI(t _i)-E _sI(t _i)+E _bI(t _i)＝0 (10)

E wherein _PIEnergy generation for pipe network I $E_{\mathrm{pI}}={\stackrel{\&OverBar;}{y}}_I+\underset{s=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{y}_{\mathrm{sI}}+\underset{l=1}{\overset{N_l}{\mathrm{\&Sigma;}}}{y}_{\mathrm{lI}},$ E _CIEnergy-output ratio for pipe network I $E_{\mathrm{cI}}={\stackrel{\&OverBar;}{x}}_I+\underset{s=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{x}_{\mathrm{sI}}+\underset{l=1}{\overset{N_l}{\mathrm{\&Sigma;}}}{x}_{\mathrm{lI}},$ E _SIBe the take-away amount of pipe network I, E _SIOutsourcing amount for pipe network I

II. steam, coal gas equal energy source store/buffer type pipe network J=1 ... N _J, its energy balance equation is as follows

V _J(t _i+1)＝V _J(t _i)+[E _pJ(t _i)-E _cJ(t _i)-E _sJ(t _i)+E _bJ(t _i)-E _wJ(t _i)]K _JT (11)

V _J(min)≤V _J(t _i)≤V _J(max) (12)

V wherein _J(t _i) be gas chamber position or steam pipe system force value, $E_{\mathrm{pI}}={\stackrel{\&OverBar;}{y}}_I+\underset{s=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{y}_{\mathrm{sI}}+\underset{l=1}{\overset{N_l}{\mathrm{\&Sigma;}}}{y}_{\mathrm{lI}},$ $E_{\mathrm{cJ}}={\stackrel{\&OverBar;}{x}}_J+\underset{s=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{x}_{\mathrm{sJ}}+\underset{l=1}{\overset{N_l}{\mathrm{\&Sigma;}}}{x}_{\mathrm{lJ}},$ K _JFor the energy is adjusted coefficient, K when pipe network is gaspipe network _J=k ₁, k when pipe network is steam pipe system _J=k ₂, E _SJBe the take-away amount of pipe network, E _SJBe the outsourcing amount of pipe network, E _WJBe the amount of diffusing of pipe network, because the energy storage pipe network just has the amount of diffusing, E in limited time reaching on its storage capacity _WJAlso should satisfy following constraint condition

$\left\{\begin{array}{ccc}{E}_{\mathrm{wJ}}\left(t_i\right)=0,& \mathrm{if}& V_J\left(t_i\right)<V_{J\left(\max \right)}\\ E_{\mathrm{wJ}}\left(t_i\right)\&GreaterEqual;0,& \mathrm{if}& V_J\left(t_i\right)\&GreaterEqual;V_{J\left(\max \right)}\end{array}\right.;---\left(13\right)$

The energy of energy storage type pipe network constraint: the V that regulates the speed _J(t _I+1)-V _J(t _i)≤δ V _{J (mas)}(14)

But changeable type energy device that relates in above-mentioned equation of constraint and energy conversion equipment all are controlling equipments, be that the energy that relates in the equipment running process produces consumption scalable within the specific limits, but therefore the energy of controlling equipment produce the decision variable that consumption becomes optimization problem.And energy stationary device is a non-scheduling equipment, and it is unadjustable that promptly the energy that relates in operational process of equipment produces consumption, but can influence the balanced supply and demand of energy equation of constraint.

3. the conversion of energy source optimization scheduling model

Because the existence of energy storage/buffer-type equipment makes the balanced supply and demand of energy equation of constraint not satisfy continuity and slickness requirement.Therefore after setting up the energy source optimization scheduling model according to above objective function and equation of constraint, what obtain is a belt restraining, non-smooth nonlinear mathematics planning problem.But the existing algorithm of finding the solution this type of nonlinear optimal problem needs constantly search in row space, and the time of finding the solution is long, and precision is low, can't satisfy the real-time requirement of energy scheduling.

The present invention is by the setting of objective function and choosing of decision variable, the energy amount of diffusing of energy storage/buffer-type equipment and the nonlinear relationship of its storage capacity have been handled, former belt restraining nonlinear mathematics planning problem is converted to a large scale linear programming problem, thereby can finds the solution by high efficiency linear programming algorithm effectively.

One of sub-goal in the energy source optimization scheduling problem that the present invention relates to is that the energy amount of diffusing that will realize energy pipe network minimizes.If directly with the energy amount of diffusing as decision variable target approach function, the value that can run into the energy amount of diffusing in then finding the solution becomes non-this difficult problem of smooth nonlinear relationship with the state of energy storage/buffer-type equipment.Maximize this form if in objective function, the energy amount of diffusing is minimized the state that replaces with energy storage/buffer-type equipment, in conjunction with the bound constraint of this type of equipment state value, the energy amount of diffusing of energy storage in the former problem/buffer-type equipment can satisfy in solution procedure automatically with the nonlinear relationship of its store/buffer ability.Therefore this nonlinear relationship can be rejected from the balanced supply and demand of energy equation of constraint, thus former belt restraining, non-smooth nonlinear mathematics planning problem is converted into following linear programming problem:

Its constraint condition is that the foregoing changeable type energy produces consumption facility constraints equation (1)～(4), energy conversion facility constraints equation (5)～(9), energy pipe network equilibrium of supply and demand equation of constraint (10)～(14).

wherein _SkBe the cost coefficient that k energy pipe network taken out quantity of energy, E _SkBe that k energy pipe network taken out quantity of energy; _BkBe the cost coefficient of k energy pipe network outsourcing quantity of energy, E _BkBe k energy pipe network outsourcing quantity of energy; _MkBe the cost coefficient of k energy pipe network storage quantity of energy, V _MkIt is the amount that k energy pipe network represented the energy storage ability.

On the other hand, because the problem of the present invention's research is a multi-objective optimization question, the purchase volume of the energy and take-away amount cost coefficient in objective function can be set according to real price information in the energy pipe network.And the cost coefficient of pipe network storage capacity can not get zero, is met with the amount of diffusing of assurance pipe network and the nonlinear relationship of pipe network storage capacity.In order not influence the value of preceding two variablees, the cost coefficient of pipe network storage capacity can be got very little value, thereby has avoided and the interacting of preceding two sub-goals.

4. the rolling method for solving of energy source optimization scheduling problem

1) utilize original antithesis interior point method to find the solution the energy source optimization scheduling problem

By the conversion method of above energy source optimization scheduling model as can be known, at condition precedent A1, A2, A3, under the A4, the objective function of energy source optimization scheduling problem is a linear function, and decision variable is that all energy dispatched that energy changeable type equipment and energy conversion type equipment relate to produce consumption, and the quantity of state of energy storage/buffer-type equipment, constraint condition useable linear equation and inequality that these decision variables satisfy are described.Consider that the energy a large amount of in the iron and steel enterprise produces consumption equipment scheduling decision variable and scheduling slot is handled by the discretize of time granularity, so the energy source optimization scheduling problem after transforming becomes the sparse linear planning problem of extensive a, multiple constraint, the quantity of decision variable and constraint condition all reaches 10 ³～10 ⁴The order of magnitude.

In order to find the solution this type of energy source optimization scheduling problem fast and efficiently, adopted original antithesis interior point method among the present invention at the sparse linear planning problem of extensive, multiple constraint.Former antithesis interior point method is from the inside of feasible zone trend optimum solution, rather than with simplicial method optimizing on the feasible zone border vertices like that.Interior point method requires iterative process to carry out in feasible zone inside all the time, its basic thought is that initial point is taken at feasible zone inside, and make iteration point the time by the function of placing obstacles on the border of feasible zone near the feasible zone border, the target function value that provides increases rapidly, and in iterative process, suitably control step-length, thereby make iteration point all the time in feasible zone inside.Along with the minimizing of obstruction factor, the effect of barrier function will reduce gradually, and algorithm will converge on the minimax solution of former problem.

Using the advantage that former antithesis interior point method finds the solution linear programming problem is that computing time is insensitive to the scale of problem, and calculation times can not increase with the increase of problem scale, and good convergence and robustness are arranged.

2) dispatch solution strategies based on the event-driven rolling optimization of moving window

Because iron and steel enterprise is a huge complication system, the business equipment operation has abnormal conditions to take place unavoidably, as the maintenance and the unexpected damping down of blast furnace, power plant, the damage and the scheduled maintenance of boiler equal energy source conversion equipment, the unexpected change of production task and device parameter etc.Scheduled overhaul for equipment can be anticipated its constraint condition in the energy scheduling algorithm, some at random accident then be unpredictable before scheduling.For the error on the energy source optimization scheduling strategy that reduces to be caused by accident at random, the present invention adopts a kind of event-driven rolling optimization scheduling strategy based on moving window, handles the reliability of accident at random to improve scheduling strategy.Here, the implication of incident is the unexpected damage of equipment, and price volalility is bought and taken out to the energy, the change of production task change and the constraint of equipment energy process conditions.If there is certain incident to take place in the production run, algorithm will carry out the energy source optimization scheduling problem by moving window again and find the solution.

The key step of dispatching solution strategies based on the event-driven rolling optimization of moving window is as follows:

Step 1 is determined scheduling slot moving window length according to the scheduling needs;

Step 2 is read the state value of energy storage/buffer-type equipment from the on-line measurement instrument of production run, as the initial value of current scrolling Optimization Dispatching problem solving, and buys and takes out price and reset cost coefficient relevant in the objective function by the current energy;

Step 3 makes up the energy source optimization scheduling model in the current scheduling period moving window length, and is converted into sparse linear planning problem extensive, multiple constraint, uses former antithesis interior point method and calculates corresponding scheduling decision;

Step 4 is carried out the energy scheduling decision-making in the current scheduling period moving window length; In the process of implementation, judge whether that new incident takes place,, do not turn to step 5 if having if turn to step 1;

Step 5 judges whether to arrive the whole scheduling termination time, turns to step 4 if not, if turn to step 6;

Step 6 Optimization Dispatching stops.

This event-driven rolling optimization method based on moving window makes scheduling strategy have excellent adaptability to the abnormal conditions that equipment takes place.

Claims (9)

1. one kind is applied to classification model construction and the rolling method for solving that iron and steel enterprise's energy source optimization is dispatched, and it is characterized in that following method step:

1) according to the characteristic of energy device in the iron and steel enterprise and pipe network, all devices is divided into following five kinds: energy stationary device, energy changeable type equipment, energy conversion type equipment, energy storage/buffer-type equipment, non-energy storage/buffer-type equipment;

2) adopt modelling by mechanism and data fitting algorithm that the said equipment is classified and set up mathematical model;

3) based on the model of classification model construction method foundation, obtain the nonlinear mathematics planning problem of a belt restraining, be converted into a large scale linear programming problem;

4) adopt a kind of event-driven rolling optimization calculative strategy based on moving window;

5) adopt former antithesis interior point method to carry out rapid solving.

2. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, described energy stationary device: the generation of this kind equipment energy and consumption are definite value after production task is assigned, can not arbitrarily distribute.

3. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, described energy changeable type equipment: this kind equipment is to finish the given production task can adopt different energy combinations, and its energy combination must be satisfied certain proportion relation constraint.

4. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1 is characterized in that described energy conversion type equipment; The output of this kind equipment energy and the consumption that produces other energy of these energy can be regulated.

5. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, described energy storage/buffer device: this kind equipment itself does not participate in the product consumption of the energy and calculates, but can keep the balance in the pipe network, change the balance mode of pipe network.

6. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, described employing modelling by mechanism and data fitting algorithm are classified to the said equipment and are set up mathematical model: to energy stationary device, energy changeable type equipment, energy conversion type equipment, adopt the funtcional relationship between moving average filter algorithm, partial least square method match devices consume or the generation energy and the production task respectively, and can change the transformational relation between the energy mutually; Energy storage/buffer-type equipment is adopted the method for modelling by mechanism.

7. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, the described nonlinear mathematics planning problem that obtains a belt restraining, be converted into a large scale linear programming problem: the form of energy equation of constraint can be divided into changeable type energy device equation of constraint, the equation of constraint of energy conversion equipment, the balanced supply and demand of energy equation of pipe network; The equation of constraint of energy buffer type equipment does not satisfy successional requirement, presents nonlinear characteristic, by the setting of objective function and choosing of decision variable, former belt restraining nonlinear mathematics planning problem is converted to a large scale linear programming problem.

8. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, a kind of event-driven rolling optimization calculative strategy of described employing: during the incident burst based on moving window, determine device type and corresponding process constraint condition thereof, scheduling slot with temporary changes is a moving window, recomputates energy balance and Optimization Dispatching strategy.

9. a kind of classification model construction and rolling method for solving that is applied to iron and steel enterprise's energy source optimization scheduling according to claim 1, it is characterized in that, the former antithesis interior point method of described employing carries out rapid solving: make iteration point near the feasible zone border time by the function of placing obstacles on the border of feasible zone, the target function value that provides increases rapidly, and in iterative process, control step-length, thereby make iteration point all the time in feasible zone inside, algorithm convergence is in the globally optimal solution of former problem.