CN104298191A - Heat prediction management based energy consumption control method in intelligent building - Google Patents

Abstract

The invention provides a heat prediction management based energy consumption control method in an intelligent building, which comprises the steps of 1, establishing a heat model of the building; 2, establishing a thermodynamic model in a state space; 3, carrying out a schedulability test and judging whether a heat load of the building has schedulability or not under given energy consumption load budget; and 4, operating a model predictive control (MPC) based heat control strategy to solve a corresponding control output variable. According to the invention, an HVAC (heating, ventilating and air conditioning) system in the building is adjusted, and the peak power is enabled to be reduced under various constraints such that the room temperature is small in variation, a task processing constraint is met, and the like. In addition, the system can carry out schedulability analysis very well, and carries out good judgment and correction on whether the current energy consumption budget can meet requirements or not.

Description

Based on the consumption control method of heat prediction management in intelligent building

Technical field

The present invention relates to and consume management, particularly, relate to the consumption control method based on heat prediction management in intelligent building.

Background technology

These years recently, the energy resource consumption in buildings increases quickly.The correlative study display of 2009, in American Architecture thing, energy resource consumption accounts for U.S.'s total energy consumption 40% nearly.Recently, energy information portion (the Energy Information Administration) prediction of the U.S., between 2012 to the year two thousand thirty, the total energy consumption in buildings will reach 4.74QBtu.Energy resource consumption in buildings has salient feature, i.e. electricity consumption peak demand is more obvious.For ensureing the power supply service quality to user, during building construction, need to make it meet the ability of user power utilization peak value, this will improve the construction cost of buildings.On the other hand, for reducing the operation risk of electrical network, electrical network is often applied with comparatively severe price punishment for peak load, namely improves electrical network peak value electricity price.Therefore, mild buildings electricity consumption curve, for electricity consumption user and supplier, is all useful.

Under construction, consumption a large portion of energy is from HVAC (Heating, Ventilating and Air Conditioning, heating ventilation and artificial atmosphere) system in building.Meanwhile, in large building, generally also need to relate to complicated HVAC system.Therefore, for the HVAC system in building, carry out electric energy management, make it obtain electricity consumption curve comparatively stably, user can be made to obtain larger benefit.HVAC system is mainly used in eliminating the heat that in buildings, energy dissipation applications produces.Therefore, we use MPC (Model Predictive Control, Model Predictive Control) method to consider thermal load prediction management, and comprehensive energy consumption control method, the electricity consumption load peak problem in buildings HVAC system carries out modeling control.

Current, less for the combine mode carrying out controlling of the dissipative system unified model Forecasting Methodology in buildings and consumption control method.Existing method, concentrates under a multi-level structure, carries out energy dissipation applications operation and to control and optimum energy consumption manages.This structure comprises a multilayer dispatching system, and bottom carries out online load dispatch strategy based on Mixed integer linear programming.This strategy for minimizing total operation cost, and has carried out the restriction of energy consumption ability.Mutual by demand response manager processes between load system and electrical network.Can but, this strategy study schedulability problem quantitatively, namely under restriction energy consumption, make HVAC system regulate buildings temperature to reach specific range of temperatures.

Summary of the invention

For defect of the prior art, the object of this invention is to provide the consumption control method based on heat prediction management in a kind of intelligent building.

According to the consumption control method based on heat prediction management in a kind of intelligent building provided by the invention, comprise the steps:

Step 1: set up heat model in buildings;

Step 2: set up the thermodynamical model under state space;

Step 3: carry out schedulability test, judges under fixed load budget budget, whether the thermal load of this buildings has schedulability;

Step 4: run the heat control strategy based on Model Predictive Control MPC, minimizes formula (10) and solves, solve and control output variable accordingly;

Wherein, described step 4 is specific as follows:

For the objective function that formula (11) represents, meeting under the constraint in formula (10), adopting the mode solving the integer least square optimization problem of standard to solve:

Wherein, represent temperature reference vector, represent F zone heater reference power input rate matrix, represent budget reference vector, W _ck () represents perturbation vector, L represents well heater number, and N represents prediction length, represent status predication vector, represent state model parameter, represent state model parameter, represent the ON/OFF state of i-th well heater in a jth region, U _pthe new system input variable used after (k) expression (10) distortion;

Implication for formula in formula (10) is as follows:

${\stackrel{\‾}{Y}}_r\left(k\right)={\left(\begin{array}{cccc}{Y}_r(k+1|k)& Y_r(k+2|k)& ...& Y_r(k+M|k)\end{array}\right)}^T.$

$\stackrel{\‾}{X}\left(k\right)=\left(\begin{array}{c}\mathrm{\ΦX}\left(k\right)\\ {\mathrm{\Φ}}^{2}X\left(k\right)\\ .\\ .\\ .\\ {\mathrm{\Φ}}^{M}X\left(k\right)\end{array}\right)$

$U_p\left(k\right)=\left(\begin{array}{c}U\left(k\right|k)\\ U(k+1|k)\\ .\\ .\\ .\\ U(k+N-1|k)\end{array}\right),$

$W_c\left(k\right)={\left(\begin{array}{ccccccc}{T}_a^1& {\mathrm{\Φ}}_s^1& O& ...& T_a^F& {\mathrm{\Φ}}_s^F& O\end{array}\right)}^T,$

$\stackrel{\‾}{\mathrm{\Ψ}}=\mathrm{diag}\left(\begin{array}{ccc}\mathrm{\Ψ}& ...& \mathrm{\Ψ}\end{array}\right),$

Wherein, M represents prediction length, Y _r(k+M|k) status predication of M sampling interval after a kth time is represented; X (k+M|k) represents the status predication of M sampling interval after a kth time, and M represents prediction length, and N represents prediction length, and U (k+N-1|k) represents the status predication of N-1 sampling interval after a kth time, represent the external air temperature in F region, X (k) represents k moment system state, R represents two interregional equivalent resistances, Ψ represents F zone heater power input rate matrix, G represents separate manufacturing firms model parameter, H represents separate manufacturing firms model parameter, and Φ represents separate manufacturing firms model parameter represent the total power budget number in buildings, represent the power input rate vector of F zone heater;

${\min }_{U_p\left(k\right)}{||\begin{array}{c}{S}_Q\stackrel{\‾}{G}{U}_p\left(k\right)-S_Q{\stackrel{\‾}{Y}}_r\left(k\right)\\ S_R{U}_p\left(k\right)\end{array}||}^2---\left(11\right)$

Wherein, $S_Q^T{S}_Q=Q,S_R^T{S}_R=R,$ S _qrepresent On Square-Rooting Matrices, represent S _qtransposed matrix, S _rrepresent On Square-Rooting Matrices, represent S _rtransposed matrix, U _pthe new system input variable used after (k) expression (10) distortion; Q and R is used to the compensation matrix to time error and power input respectively.

Preferably, described heat model, specific as follows:

For the jth region having n well heater, by law of conservation of energy, obtaining heat balance equation is

$\begin{array}{c}\frac{{\mathrm{dT}}_j}{\mathrm{dt}}=\frac{1}{R_j^a{C}_j}(T_a^j-T_j)+\frac{1}{C_j}\underset{i=1,i\≠j}{\overset{F}{\mathrm{\Σ}}}\frac{1}{R_{\mathrm{ij}}^r}(T_i-T_j)\\ +\frac{1}{C_j}{A}_j^w{\mathrm{\Φ}}_s^j+\frac{1}{C_j}\underset{n=1}{\overset{N_j}{\mathrm{\Σ}}}{\mathrm{\Φ}}_j^n+\frac{1}{C_j}{\mathrm{\σ}}_j{\mathrm{\ω}}_j,\end{array}---\left(1\right)$

T represents the time, and F represents areal, N _jrepresent well heater number in the j of region, σ _jrepresent that dimension receives noise variance, T _jrepresent a jth intra-zone temperature, represent that a jth region is with outside thermal resistance, C _jrepresent the thermal capacitance in a jth region, represent a jth region exterior temperature, represent the thermal resistance in i-th region and a jth region, T _irepresent i-th intra-zone temperature, represent solar radiation quantity Φ _seffective window areas that can enter, represent the energy variation that solar radiation produces, represent the power of the n-th well heater in a jth region, ω _jexpression standard dimension receives noise;

In the system model in a jth region, system state is indoor air temperature, and system input is heater power; Disturbance factor comprises three aspects, and ambient temperature, solar radiation heat, standard dimension receives noise.

Preferably, described thermodynamical model, specific as follows:

X(k)＝ΦX(k-1)+GU(k-1)+HW(k-1)，

Y(k)＝X(k)， (4)

X (k) represents k moment system state, Φ represents separate manufacturing firms model parameter, X (k-1) represents k-1 moment system state, G represents separate manufacturing firms model parameter, etching system input quantity when U (k-1) represents k-1, H represents separate manufacturing firms model parameter, and W (k-1) represents k-1 moment system disturbance amount, and Y (k) represents k moment Systematical control output quantity;

Wherein, kth time sampling interval T _slower system state, discrete-time system model matrix is expressed as $\mathrm{\Φ}=e^{\mathrm{ATs}},G={\∫}_0^{\mathrm{Ts}}{e}^{\mathrm{As}}\mathrm{Bds},$ With $H={\∫}_0^{\mathrm{Ts}}{e}^{\mathrm{As}}\mathrm{Dds}.$

Wherein, T _srepresent the time interval, e represents constant e, and s represents integration item, and A represents state-space model parameter, and B represents state-space model parameter, and D represents state-space model parameter, and G represents separate manufacturing firms model parameter, and H represents separate manufacturing firms model parameter.

Preferably, described step 3 comprises the steps:

Step 3.1: by schedulability testing algorithm, obtains the minimum load budget budget of demand fulfillment _min;

Step 3.2: the load budget budget < budget obtained if current _min, then perform step 3.3, otherwise, perform step 3.4;

Step 3.3: request increases energy consumption load budget, obtains a new load budget budget ', re-executes step 3.1, or loosen caloric restriction, obtains a new minimum load budget ' _min, re-execute step 3.2;

Step 3.4: perform step 4.

Preferably, described schedulability testing algorithm, is specially:

First, the design of schedulability problem is become following optimization problem:

$\underset{\mathrm{\&eta;}}{\min }\underset{j=1}{\overset{F}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{N^j}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j^i{P}_j^i,$

s.t.-CA ^-1(DW+Bη)∈int(Safe). (7)

Wherein, s.t. represent be tied in;

There is η=[η in definition _j] ∈ [0,1] ^f, make ${\mathrm{\&Sigma;}}_{j=1}^F{\mathrm{\&Sigma;}}_{i=1}^{N^j}{\mathrm{\&eta;}}_j^i{P}_j^i\&le;\mathrm{budget}$

η _jrepresent the number in [0, the 1] interval existed, [] ^frepresent F dimensional vector, F represents F-schedulability, N ^jrepresent a jth zone heater number, represent the η value of the existence of i-th well heater in a jth region, represent power input rate;

A represents state-space model parameter, and B represents state-space model parameter, and D represents state-space model parameter, and W represents system disturbance, and C represents state-space model parameter;

S set afe is defined as

$[{Y_l}^1,Y_u^1]\&times;[{Y_l}^2,Y_u^2]\&times;...\&times;[{Y_l}^F,Y_u^F].$

Wherein, Y _l ^frepresent the lower bound vector of F zone heater, represent the upper bound vector of F zone heater;

Minimum load budget budget _minobtained by this optimization problem; Budget _minshould be the load being not less than target function value to count in advance budget, arrange

Restriction in optimization problem is linear, is represented as:

Y _l+CA ^-1DW＜-CA ^-1BPη＜Y _u+CA ^-1DW.⑻

Wherein, Y _lrepresent the lower bound controlling output variable, Y _urepresent the upper bound controlling output variable, P represents power input vector;

Wherein

Y _l＝[Y _l ¹，Y _l ²，…，Y _l ^F] ^T，

$Y_u={[Y_u^1,Y_u^2,...,Y_u^F]}^T.$

This optimization problem passes through linear programming for solution.

Compared with prior art, the present invention has following beneficial effect:

Based on the consumption control method of heat prediction management in a kind of intelligent building of major design of the present invention, HVAC system in buildings is regulated, make it while the less variation of room temperature, meeting each item constraints such as task process constraint, reduce peak power.Whether further, system can carry out Schedulability Analysis well, can meet the demands, judge well and revise for current energy consumption budget.

By this inventive point, the effect that we can obtain is:

1, when making each regional temperature of buildings remain near certain certain value, temperature variation is steady.

2, effectively can do load balancing to heat load, the load peak of heat load is reduced.

3, well the schedulability of thermal load in buildings is analyzed, judge well and revise.

Accompanying drawing explanation

By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:

Fig. 1 is room layout.

Fig. 2 is ambient temperature and intensity of solar radiation disturbance.

Fig. 3 is well heater 1-4 power input under on/off control algolithm.

Fig. 4 is well heater 5-8 power input under on/off control algolithm.

Fig. 5 is well heater 1-4 power input under MPC control algolithm.

Fig. 6 is well heater 5-8 power input under MPC control algolithm.

Fig. 7 is that MPC controls lower room 1-4 range of temperature.

Fig. 8 is that MPC controls lower room 5-8 range of temperature.

Fig. 9 is the power peak comparing ON/OFF and MPC control.

Figure 10 is system architecture.

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.

In view of this, we devise the consumption control method based on heat prediction management in intelligent building.For the power dispatching problem in therrmodynamic system, define the optimization problem of a belt restraining.Use MPC technology, heat is predicted.Corresponding energy peak problem, becomes a constraint in optimization problem, is solved.And test solution is carried out to schedulability problem.

In the method, our basic structure as shown in Figure 10.

In the structurized system shown in Figure 10, mainly comprise several level: (1) PAC (Predictive Admission Controller, predictability admission controller) (2) LB (Load Balance, load balancing) (3) D/R (Demand Response, demand response) manager and load predicter.

PAC part is in the bottom, for interacting with physical equipment, carries out real-time power access control.At top layer, D/R manager is the inlet point of system, as the interface that same electrical network is mutual.LF layer supplies information to D/R manager and LB layer based on the load of prediction.LB layer, based on the load information predicted and current pricing information, provides energy consumption budget to PAC layer.

In the present invention, main concern PAC layer, for the operation entering and control electricity consumption application of management energy.PAC layer is submitted in all electricity consumption requests.PAC layer, based on total loading condition, determines that this request accepts or refusal.PAC layer mainly comprises two modules: schedulability test and heat control.Schedulability test module is used for when performing checking whether total energy consumption budget is observed.Heat control module, based on the temperature range of Current Temperatures, total energy consumption budget, setting, each well heater (refrigerator) in management buildings HVAC system.

Afterwards, modeling process is carried out to problem.Parameter list is as follows:

1. first, set up heat model in buildings

Between floors, heating and refrigeration system for controlling between buildings zones of different (such as, not chummery) or buildings zones of different with the energy transferring of external environment.Simplify processes, only considers heating system (refrigeration system situation is similar) in the present invention.The target that heating system controls guarantees that in room, each regional temperature maintains comfortable scope.Thermokinetics in each region, all affected by following three factors: outdoor energy (such as, solar radiation is to external surface of buildings or the heat interchange with external environment), indoor energy (such as, the heat that well heater or miscellaneous equipment produce, the activity etc. of people), interregional energy (such as, heat passes to another room from a room by wall).

For the jth region having n well heater, by law of conservation of energy, we can obtain heat balance equation and are

$\begin{array}{c}\frac{{\mathrm{dT}}_j}{\mathrm{dt}}=\frac{1}{R_j^a{C}_j}(T_a^j-T_j)+\frac{1}{C_j}\underset{i=1,i\&NotEqual;j}{\overset{F}{\mathrm{\&Sigma;}}}\frac{1}{R_{\mathrm{ij}}^r}(T_i-T_j)\\ +\frac{1}{C_j}{A}_j^w{\mathrm{\&Phi;}}_s^j+\frac{1}{C_j}\underset{n=1}{\overset{N_j}{\mathrm{\&Sigma;}}}{\mathrm{\&Phi;}}_j^n+\frac{1}{C_j}{\mathrm{\&sigma;}}_j{\mathrm{\&omega;}}_j,\end{array}---\left(1\right)$

T represents the time, and F represents areal, N _jrepresent well heater number in the j of region, σ _jrepresent that dimension receives noise variance;

In the system model in a jth region, system state is indoor air temperature, and system input is heater power.Disturbance factor comprises three aspects, and ambient temperature, solar radiation heat, standard dimension receives noise.Correlation parameter implication refers to parameter list.

The control inputs of well heater follows different operational circumstances be expressed as wherein represent well heater OFF (pass), represent well heater ON (opening).With represent the power of respective heater, then actual heater power for

${\mathrm{\&Phi;}}_j^n=P_j^n{u}_j^n.---\left(2\right)$

Therefore, well heater can be controlled by ON-OFF.Wherein, under ON state, the power of well heater is constant, constant.

2. the thermodynamical model under state space

Setting for the number of regions in buildings, and there is N in each jth region _jindividual well heater.Therefore, always have individual well heater.With X=[T ₁, T ₂..., T _f] ^trepresent state vector, use represent N _jthe control inputs vector of individual well heater, with U=[U ¹, U ²..., U ^f] ^trepresent Systematical control input vector.With represent the disturbance in a jth region.System disturbance vector representation is W=[W ¹, W ²..., W ^f] ^t.

State-space model is expressed as

$\stackrel{\&CenterDot;}{X}=\mathrm{AX}+\mathrm{BU}+\mathrm{DW},$

Y＝X， ⑶

Wherein, represent system state derivative, X represents system state vector, and A represents state-space model parameter, and B represents state-space model parameter, and D represents state-space model parameter, and W represents system disturbance, and Y represents Systematical control output vector, and U represents system input vector;

System matrix is defined as:

B＝diag(B ₁ B ₂ … B _F)，

D＝diag(D ₁ D ₂ … D _F)

A ^frepresent variable parameter name in A matrix, defined by following formula, C _frepresent F the corresponding thermal capacitance in region, B _frepresent variable parameter name in B matrix, defined by following formula, D _frepresent variable parameter name in D matrix, defined by following formula;

For j=1,2 ..., F

$A^j=-\frac{1}{R_j^a{C}_j}-\frac{1}{C_j}\underset{i=1,i\&NotEqual;j}{\overset{F}{\mathrm{\&Sigma;}}}\frac{1}{R_{\mathrm{ij}}^T},$

$B_j=\left(\begin{array}{cccc}\frac{P_j^1}{C_j}& \frac{P_j^2}{C_j}& ...& \frac{P_j^{N_j}}{C_j}\end{array}\right),$

$D_j=\left(\begin{array}{ccc}\frac{1}{R_j^a{C}_j}& \frac{A_j^w}{C_j}& \frac{{\mathrm{\&sigma;}}_j^T}{C_j}\end{array}\right).$

Wherein, represent power input rate, represent that dimension receives noise variance, R represents two interregional equivalent resistances;

In order to simplified design and analysis, we convert continuous time model to discrete time model.

X(k)＝ΦX(k-1)+GU(k-1)+HW(k-1)，

Y(k)＝X(k)， (4)

X (k) represents k moment system state, Φ represents separate manufacturing firms model parameter, X (k-1) represents k-1 moment system state, G represents separate manufacturing firms model parameter, etching system input quantity when U (k-1) represents k-1, H represents separate manufacturing firms model parameter, and W (k-1) represents k-1 moment system disturbance amount, and Y (k) represents k moment Systematical control output quantity;

Wherein, kth time sampling interval T _slower system state, discrete-time system model matrix is expressed as $\mathrm{\&Phi;}=e^{A{T}_8},G={\&Integral;}_0^{T_8}{e}^{\mathrm{As}}\mathrm{Bds},$ With $H={\&Integral;}_0^{T_8}{e}^{\mathrm{As}}\mathrm{Dds}.$

Wherein, T _srepresent the time interval, e represents constant e, and s represents integration item.

3. schedulability test

The target controlled makes the indoor air temperature Y of buildings zones of different remain on a comfortable interval.The bound in interval is expressed as Y _land Y _u

Y _l≤Y≤Y _u. ⑸

Control system submits to the energy consumption load determined by total energy consumption load budget and limits, and can be expressed as:

${\mathrm{\&Psi;}}^j=[P_j^1,P_j^2,...,P_j^{N_j}],$ be total energy consumption budget in buildings heating, Ψ represents F zone heater power input rate matrix, Ψ ^frepresent the power input rate vector of F zone heater, represent the real vector of 1*L, L represents total well heater number, Ψ ^jrepresent the power input rate vector of a jth zone heater, represent power input rate, N _jrepresent that there is N in a jth region _jindividual well heater;

In addition, the input of well heater must be ON/OFF state:

$U={\left[u_j^i\right]}^L,\&ForAll;u_j^i\&Element;\left\{\mathrm{0,1}\right\}.---\left(6\right)$

Wherein, U represents well heater ON/OFF state vector, represent the ON/OFF state of i-th well heater in a jth region, L represents total well heater number;

Therefore, schedulability problem is, while automatically carrying out thermal management, meets constraint (5) and (6).

We use k-schedulability to solve problem, are namely schedulable when guarantee k well heater.The control inputs U of system is considered and controls the ON/OFF state of single well heater.If when U inputs, can meet and be adjusted between comfort zone by all regional temperatures, and meet energy consumption budget constraints, so we claim this to be input as " safe ".Further, we are at define system " Safe ": life period τ >=0, and make Y (t) the ∈ Safe as t >=τ, wherein, t represents t, and Y (t) represents system output quantity.We claim system to be that k is schedulable, when meeting: there is sacurity dispatching U, make wherein, ‖ ‖ ₁represent 1-norm, U (t) represents the control inputs of system, represent the ON/OFF state of i-th well heater in a jth region, k represents schedulability tolerance k.

Further, we define, when there is η=[η _i] ∈ [0,1] ^f, make and meet-CA ^-1(D+B η) ∈ int (Safe).Then system is obviously also that k is schedulable.

Wherein, int (Safe) represents set of integers Safe;

Further, when there is η=[η _i] ∈ [0,1] ^f, make and-CA ^-1(D+B η) ∈ int (Safe), system is that budget is schedulable.Wherein, budget is load budget, represent the η value of the existence of i-th well heater in a jth region, represent power input rate.

Based on this, we design the schedulability testing algorithm in the present invention.

First, the design of schedulability problem is become following optimization problem.

$\underset{\mathrm{\&eta;}}{\min }\underset{j=1}{\overset{F}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{N^j}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j^i{P}_j^i,$

s.t.-CA ^-1(DW+Bη)∈int(Safe). (7)

Wherein, s.t. represent be tied in;

S set afe is defined as

$[{Y_l}^1,Y_u^1]\&times;[{Y_l}^2,Y_u^2]\&times;...\&times;[{Y_l}^F,Y_u^F].$

Wherein, Y _l ^frepresent the lower bound vector of F zone heater, represent the upper bound vector of F zone heater;

Minimum peak load budget budget _mincan be obtained by this optimization problem.Budget _minshould be the load being not less than target function value to count in advance, can arrange

Being restricted in optimization problem is linear, can be represented as:

Y _l+CA ^-1DW＜-CA ^-1BPη＜Y _u+CA ^-1DW.⑻

Wherein, Y _lrepresent the lower bound controlling output variable, Y _urepresent the upper bound controlling output variable;

Wherein

Y _l＝[Y _l ¹，Y _l ²，…，Y _l ^F] ^T，

$Y_u={[Y_u^1,Y_u^2,...,Y_u^F]}^T.$

Therefore, this optimization problem can pass through linear programming for solution.

Therefore, schedulability problem method of testing of the present invention is, within each time cycle, perform following algorithm:

1) solve the optimization problem in formula (7), obtain the minimum load budget budget of demand fulfillment _min

2) if the current load budget budget < budget obtained _min, then step 3 is performed), otherwise, perform step 4)

3) system increases energy consumption budget to upper strata (LB layer) request, obtains a new budget ', re-executes step 1), or loosen caloric restriction, obtain a new budget ' _min, re-execute step 2

4) system cloud gray model is based on the heat control strategy of MPC.

4. based on the heat control strategy of MPC

In MPC controlling calculation, controller controls input U (k), is used for minimizing following objective function J (K):

$\begin{array}{c}J\left(k\right)=\underset{s=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left|\right|Y(k+s|k)-Y_r(k+s|k)\left|\right|}_Q^2\\ +\underset{s=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{\left|\right|U(k+s|k)\left|\right|}_R^2,\end{array}---\left(9\right)$

M and N represents prediction length scope respectively, and M is the prediction length scope controlling output vector, and N is the prediction length scope of perturbation vector, and Y (k+s|k) is the status predication of s sampling interval after the kth time, Y _r(k|+s|k) be the status predication of s sampling interval after the kth time, the status predication of s sampling interval after U (k+s|k) the kth time.Q and R is used to the compensation matrix to time error and power input respectively.

This cost function J (K) is mainly and minimizes temperature tracking error and power demand.Even if while power demand is less, make the temperature variation of different time compartment less.

For the system model in formula (3), be again changed to following form,

$X_p\left(k\right)=\stackrel{\&OverBar;}{X}\left(k\right)+\stackrel{\&OverBar;}{G}{U}_p\left(k\right)+\stackrel{\&OverBar;}{H}{W}_c\left(k\right),$

Y _p(k)＝X _p(k)，

Wherein, represent status predication vector, represent state model parameter, represent state model parameter;

Therefore, in this strategy, need, in each time interval, to solve minimum value to this cost function, and meet corresponding constraint condition.Optimization problem arranges as follows:

Wherein, represent temperature reference vector, represent F zone heater reference power input rate matrix, represent budget reference vector, W _crepresent perturbation vector, represent and export reference vector lower bound, represent and export the reference vector upper bound, L represents well heater number, and N represents prediction length;

Corresponding matrix implication is as follows:

${\stackrel{\&OverBar;}{Y}}_r\left(k\right)={\left(\begin{array}{cccc}{Y}_r(k+1|k)& Y_r(k+2|k)& ...& Y_r(k+M|k)\end{array}\right)}^T.$

Wherein, M represents prediction length, Y _r(k+M|k) status predication of M sampling interval after a kth time is represented;

$X_p\left(k\right)=\left(\begin{array}{c}X(k+1|k)\\ X(k+2|k)\\ .\\ .\\ .\\ X(k+M|k)\end{array}\right),\stackrel{\&OverBar;}{X}\left(k\right)=\left(\begin{array}{c}\mathrm{\&Phi;X}\left(k\right)\\ {\mathrm{\&Phi;}}^{2}X\left(k\right)\\ .\\ .\\ .\\ {\mathrm{\&Phi;}}^{M}X\left(k\right)\end{array}\right),$

$U_p\left(k\right)=\left(\begin{array}{c}U\left(k\right|k)\\ U(k+1|k)\\ .\\ .\\ .\\ U(k+N-1|k)\end{array}\right),$

$W_c\left(k\right)={\left(\begin{array}{ccccccc}{T}_a^1& {\mathrm{\&Phi;}}_s^1& O& ...& T_a^F& {\mathrm{\&Phi;}}_s^F& O\end{array}\right)}^T,$

$\stackrel{\&OverBar;}{\mathrm{\&Psi;}}=\mathrm{diag}\left(\begin{array}{ccc}\mathrm{\&Psi;}& ...& \mathrm{\&Psi;}\end{array}\right),$

${\stackrel{\&OverBar;}{Y}}_u={\left(\begin{array}{ccc}{Y}_u^T& ...& Y_u^T\end{array}\right)}^T,{\stackrel{\&OverBar;}{Y}}_l={\left(\begin{array}{ccc}{Y_l}^T& ...& {Y_l}^T\end{array}\right)}^T,$

$\stackrel{\&OverBar;}{Q}=\mathrm{diag}\left(\begin{array}{ccc}Q& ...& Q\end{array}\right),\stackrel{\&OverBar;}{R}=\mathrm{diag}\left(\begin{array}{ccc}R& ...& R\end{array}\right).$

Wherein, X (k+M|k) represents the status predication of M sampling interval after a kth time, and M represents prediction length, and N represents prediction length, and U (k+N-1|k) represents the status predication of N-1 sampling interval after a kth time, represent the external air temperature in F region;

represent the matrix of compensating timing error matrix, represent the matrix of the compensation matrix of power input, represent the matrix exporting reference vector lower bound, represent the matrix exporting the reference vector upper bound, represent the total power budget number in buildings, represent separate manufacturing firms model parameter matrix;

represent and export reference vector lower bound, represent and export the reference vector upper bound, Q and R is used to the compensation matrix to time error and power input respectively;

Objective function can be deformed into:

${\min }_{U_p\left(k\right)}{||\begin{array}{c}{S}_Q\stackrel{\&OverBar;}{G}{U}_p\left(k\right)-S_Q{\stackrel{\&OverBar;}{Y}}_r\left(k\right)\\ S_R{U}_p\left(k\right)\end{array}||}^2---\left(11\right)$

Wherein, S _q, S _rrepresent $S_Q^T{S}_Q=Q,S_R^T{S}_R=R;$

Under this objective function, meet the corresponding constraint in formula (10), this problem is deformed into the integer least square optimization problem of standard, and YALMIP storehouse in Matlab can be used to solve.

Based on the consumption control method of heat prediction management in described intelligent building, specifically comprise the steps:

Step 1: set up heat model in buildings;

Step 2: set up the thermodynamical model under state space;

Step 3: carry out schedulability test, judges under fixed load budget budget, whether the thermal load of this buildings has schedulability;

Step 4: run the heat control strategy based on MPC (Model Predictive Control, Model Predictive Control), minimizes formula (10) and solves, solve and control output variable accordingly.

For making object of the present invention, technical scheme and a little clearly, below in conjunction with accompanying drawing and specific embodiment, progressive detailed description in detail is done to the present invention.

In this embodiment, we have selected one 120m ²single story building, each room is equipped with one or two well heater.Shown in its room layout's accompanying drawing 1.Partial heat parameter is shown in Fig. 2 and following allocation list:

Heat parameter configuration table

Thermal resistance non-zero variable set up

Heater power consumes

For MPC controller, arrange Q (1)=...=Q (M)=diag (10,10,10,10,10,10,10,10), M=5, R (1)=...=R (N)=diag (1,1,1,1,1,1,1,1), N=5, sampling interval is 2 minutes, i.e. T _s=120s.Arranging room temperature is between 22-24 DEG C,

By the method that the present invention introduces, with not using MPC PREDICTIVE CONTROL, only use ON-OFF mode to compare, in each room, power input results is shown in Fig. 3-Fig. 6.Its result gathered for Fig. 9, can find, the peak load under MPC control method is little compared with the tactful peak load of common ON-OFF.Energy consumption budget can be met preferably, show that the schedulability of system is better.Meanwhile, accompanying drawing 8-Fig. 9 shows, and MPC controls lower well heater, can well by each room temperature control between 22-24 degree Celsius that sets, and the amplitude that temperature changes at any time and changes is less.

Above specific embodiments of the invention are described.It is to be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or amendment within the scope of the claims, and this does not affect flesh and blood of the present invention.

Claims (5)

1. in intelligent building based on a consumption control method for heat prediction management, it is characterized in that, comprise the steps:

Step 1: set up heat model in buildings;

Step 2: set up the thermodynamical model under state space;

Step 3: carry out schedulability test, judges under fixed load budget budget, whether the thermal load of this buildings has schedulability;

Step 4: run the heat control strategy based on Model Predictive Control MPC, minimizes formula (10) and solves, solve and control output variable accordingly;

Wherein, described step 4 is specific as follows:

For the objective function that formula (11) represents, meeting under the constraint in formula (10), adopting the mode solving the integer least square optimization problem of standard to solve:

Wherein, represent temperature reference vector, represent F zone heater reference power input rate matrix, represent budget reference vector, W _ck () represents perturbation vector, L represents well heater number, and N represents prediction length, represent status predication vector, represent state model parameter, represent state model parameter, represent the ON/OFF state of i-th well heater in a jth region, U _pthe new system input variable used after (k) expression (10) distortion;

Implication for formula in formula (10) is as follows:

${\stackrel{\&OverBar;}{Y}}_r\left(k\right)={\left(\begin{array}{cccc}{Y}_r(k+1|k)& Y_r(k+2|k)& ...& Y_r(k+M|k)\end{array}\right)}^T.$

$\stackrel{\&OverBar;}{X}\left(k\right)=\left(\begin{array}{c}\mathrm{\&Phi;X}\left(k\right)\\ {\mathrm{\&Phi;}}^{2}X\left(k\right)\\ .\\ .\\ .\\ {\mathrm{\&Phi;}}^{M}X\left(k\right)\end{array}\right)$

$U_p\left(k\right)=\left(\begin{array}{c}U\left(k\right|k)\\ U(k+1|k)\\ .\\ .\\ .\\ U(k+N-1|k)\end{array}\right),$

$W_c\left(k\right)={\left(\begin{array}{ccccccc}{T}_a^1& {\mathrm{\&Phi;}}_s^1& O& ...& T_a^F& {\mathrm{\&Phi;}}_s^F& O\end{array}\right)}^T,$

$\stackrel{\&OverBar;}{\mathrm{\&Psi;}}=\mathrm{diag}\left(\begin{array}{ccc}\mathrm{\&Psi;}& ...& \mathrm{\&Psi;}\end{array}\right),$

Wherein, M represents prediction length, Y _r(k+M|k) status predication of M sampling interval after a kth time is represented; X (k+M|k) represents the status predication of M sampling interval after a kth time, and M represents prediction length, and N represents prediction length, and U (k+N-1|k) represents the status predication of N-1 sampling interval after a kth time, represent the external air temperature in F region, X (k) represents k moment system state, R represents two interregional equivalent resistances, Ψ represents F zone heater power input rate matrix, G represents separate manufacturing firms model parameter, H represents separate manufacturing firms model parameter, and Φ represents separate manufacturing firms model parameter represent the total power budget number in buildings, represent the power input rate vector of F zone heater;

${\min }_{U_p\left(k\right)}{||\begin{array}{c}{S}_Q\stackrel{\&OverBar;}{G}{U}_p\left(k\right)-S_Q{\stackrel{\&OverBar;}{Y}}_r\left(k\right)\\ S_R{U}_p\left(k\right)\end{array}||}^2---\left(11\right)$

Wherein, $S_Q^T{S}_Q=Q,S_R^T{S}_R=R,$ S _qrepresent On Square-Rooting Matrices, represent S _qtransposed matrix, S _rrepresent On Square-Rooting Matrices, represent S _rtransposed matrix, U _pthe new system input variable used after (k) expression (10) distortion; Q and R is used to the compensation matrix to time error and power input respectively.

2. in intelligent building according to claim 1 based on the consumption control method of heat prediction management, it is characterized in that, described heat model, specific as follows:

For the jth region having n well heater, by law of conservation of energy, obtaining heat balance equation is

$\begin{array}{c}\frac{{\mathrm{dT}}_j}{\mathrm{dt}}=\frac{1}{R_j^a{C}_j}(T_a^j-T_j)+\frac{1}{C_j}\underset{i=1,i\&NotEqual;j}{\overset{F}{\mathrm{\&Sigma;}}}\frac{1}{R_{\mathrm{ij}}^r}(T_i-T_j)\\ +\frac{1}{C_j}{A}_j^w{\mathrm{\&Phi;}}_s^j+\frac{1}{C_j}\underset{n=1}{\overset{N_j}{\mathrm{\&Sigma;}}}{\mathrm{\&Phi;}}_j^n+\frac{1}{C_j}{\mathrm{\&sigma;}}_j{\mathrm{\&omega;}}_j,\end{array}---\left(1\right)$

T represents the time, and F represents areal, N _jrepresent well heater number in the j of region, σ _jrepresent that dimension receives noise variance, T _jrepresent a jth intra-zone temperature, represent that a jth region is with outside thermal resistance, C _jrepresent the thermal capacitance in a jth region, represent a jth region exterior temperature, represent the thermal resistance in i-th region and a jth region, T _irepresent i-th intra-zone temperature, represent solar radiation quantity Φ _seffective window areas that can enter, represent the energy variation that solar radiation produces, represent the power of the n-th well heater in a jth region, ω _jexpression standard dimension receives noise;

In the system model in a jth region, system state is indoor air temperature, and system input is heater power; Disturbance factor comprises three aspects, and ambient temperature, solar radiation heat, standard dimension receives noise.

3. in intelligent building according to claim 1 based on the consumption control method of heat prediction management, it is characterized in that, described thermodynamical model, specific as follows:

X(k)＝ΦX(k-1)+GU(k-1)+HW(k-1)，

Y(k)＝X(k)， ⑷

X (k) represents k moment system state, Φ represents separate manufacturing firms model parameter, X (k-1) represents k-1 moment system state, G represents separate manufacturing firms model parameter, etching system input quantity when U (k-1) represents k-1, H represents separate manufacturing firms model parameter, and W (k-1) represents k-1 moment system disturbance amount, and Y (k) represents k moment Systematical control output quantity;

Wherein, kth time sampling interval T _slower system state, discrete-time system model matrix is expressed as $\mathrm{\&Phi;}=e^{\mathrm{ATs}},G={\&Integral;}_0^{\mathrm{Ts}}{e}^{\mathrm{As}}\mathrm{Bds},$ With $H={\&Integral;}_0^{\mathrm{Ts}}{e}^{\mathrm{As}}\mathrm{Dds}.$

Wherein, T _srepresent the time interval, e represents constant e, and s represents integration item, and A represents state-space model parameter, and B represents state-space model parameter, and D represents state-space model parameter, and G represents separate manufacturing firms model parameter, and H represents separate manufacturing firms model parameter.

4. in intelligent building according to claim 1 based on the consumption control method of heat prediction management, it is characterized in that, described step 3 comprises the steps:

Step 3.1: by schedulability testing algorithm, obtains the minimum load budget budget of demand fulfillment _min;

Step 3.2: the load budget budget < budget obtained if current _min, then perform step 3.3, otherwise, perform step 3.4;

Step 3.3: request increases energy consumption load budget, obtains a new load budget budget ', re-executes step 3.1, or loosen caloric restriction, obtains a new minimum load budget ' _min, re-execute step 3.2;

Step 3.4: perform step 4.

5. in intelligent building according to claim 4 based on the consumption control method of heat prediction management, it is characterized in that, described schedulability testing algorithm, is specially:

First, the design of schedulability problem is become following optimization problem:

$\underset{\mathrm{\&eta;}}{\min }\underset{j=1}{\overset{F}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{N^j}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j^i{P}_j^i,$

s.t.-CA ^-1(DW+Bη)∈int(Safe). (7)

Wherein, s.t. represent be tied in;

There is η=[η in definition _j] ∈ [0,1] ^f, make ${\mathrm{\&Sigma;}}_{j=1}^F{\mathrm{\&Sigma;}}_{i=1}^{N^j}{\mathrm{\&eta;}}_j^i{P}_j^i\&le;\mathrm{budget}$

η _jrepresent the number in [0, the 1] interval existed, [] ^frepresent F dimensional vector, F represents F-schedulability, N ^jrepresent a jth zone heater number, represent the η value of the existence of i-th well heater in a jth region, represent power input rate;

A represents state-space model parameter, and B represents state-space model parameter, and D represents state-space model parameter, and W represents system disturbance, and C represents state-space model parameter;

S set afe is defined as

$[{Y_l}^1,Y_u^1]\&times;[{Y_l}^2,Y_u^2]\&times;...\&times;[{Y_l}^F,Y_u^F].$

Wherein, Y _l ^frepresent the lower bound vector of F zone heater, represent the upper bound vector of F zone heater;

Minimum load budget budget _minobtained by this optimization problem; Budget _minshould be the load being not less than target function value to count in advance budget, arrange

Restriction in optimization problem is linear, is represented as:

Y _l+CA ^-1DW＜-CA ^-1BPη＜Y _u+CA ^-1DW.⑻

Wherein, Y _lrepresent the lower bound controlling output variable, Y _urepresent the upper bound controlling output variable, P represents power input vector;

Wherein

Y _l＝[Y _l ¹，Y _l ²，…，Y _l ^F] ^T，

$Y_u={[Y_u^1,Y_u^2,...,Y_u^F]}^T.$

This optimization problem passes through linear programming for solution.