CN104898422A - Hierarchical optimization method for united cold supply system - Google Patents

Abstract

The invention provides a hierarchical optimization method for a united cold supply system consisting of a conventional electric refrigerator and an ice storage system. The method comprises the steps of optimizing the on-state and the optimal power setting value of each cold source by adopting a mixed integer programming method according to the economic model of each cold source and the time-of-use power price policy, designing a target-coupled coordinated distributed prediction control method to better improve the dynamic performance of the system in view of the dynamic adjusting characteristic and physical constraint of each cold source, and newly optimizing the setting value of each cold source under a distributed framework, so that each cold source ensures the total load in the dynamic process and tracks the optimal refrigerating power setting value as much as possible at the same time. The method quantitatively solves the steady and dynamic scheduling problem of the united cold supply system by adopting mixed integer programming and distributed prediction control methods respectively, and provides a rational suggestion for energy management and optimization scheduling of the cold supply system.

Description

The hierarchical optimal algorithm of United system

Technical field

The present invention relates to energy network regulation degree field, particularly a kind of hierarchical optimal algorithm of the United system be made up of conventional electric refrigerating machine and ice-storage system.

Background technology

Due to the problem of large size city ubiquity land used anxiety, in order to utilize limited land area to increase urban capacity as far as possible, skyscraper becomes the desirable selection of urban planners.Skyscraper often has independently cooling network, provides that entire building is various uses cold demand.Along with the appearance of various emerging technology and the popularization of tou power price policy, only use the cold supply system economic benefit of conventional electrical chillers can not be satisfactory, in order to improve the economic benefit of skyscraper cold supply system, the complementary cooperative system of many coolings unit is applied more and more general.The complementary cooperative system of typical many coolings unit generally comprises: ice conserve cold, conventional electric refrigerating machine, cold, heat and electricity triple supply and earth source heat pump etc., the wherein refrigeration duty of conventional electric refrigerating machine and the general shared more than 90% of ice conserve cold.Owing to from main control energy storage and can use the energy period, ice-storage system is more and more subject to the favor of people, and effectively carrying out cooperation control to itself and conventional electric refrigerating machine is guarantee that this cooperative system is saved energy and reduce the cost and the key of increasing economic efficiency.

Current ice-storage system and jointly controlling of conventional electrical chillers rest on qualitative aspect more, the i.e. cold storage of ice making when electricity price paddy, the cool thermal discharge when electricity price peak, lack the analysis & control quantitative to whole system, this results in its control method insuperior, and have a dynamic process, so only consider that the control method of stable state is difficult to guarantee that the actual cooling power of whole system catches up with rapidly prediction load from starting the cooling power reaching setting due to electric refrigerating machine.Therefore, need to study a kind of optimal control method make system in dynamic process, ensure the prerequisite of total load under high-efficiency and economic run.

For the such distributed system of skyscraper cold supply system, usually general steady state optimization method can be adopted to dispatch it, but due to electric refrigerating machine dynamic response problem, adopt Distributed Predictive Control to optimize electric refrigerating machine cooling power dynamic property further.For the Dynamic controlling of distributed system, transformed to distributed control structure by centralized at present, because method can be caused complicated for centerized fusion and calculated amount is bigger than normal.In practical application in industry, Model Predictive Control (MPC) can introduce process time lag naturally by the pulse of controlled device or step response sequence prediction object change in future, can obtain the control effects more outstanding than conventional control methods for multi-variable system.In large scale system, traditional Model Predictive Control derives Distributed Predictive Control (DMPC) this form, large scale system is resolved into multiple subsystem, between subsystem with subsystem, data exchange mutually, each subsystem utilizes the data of self and its sub-systems to complete solving of the MPC problem of self, significantly reduces calculated amount and can reach the optimum state of whole system.

Therefore, the present invention intends using for reference the above coordination approach to coupled system, and goal in research is coupled but the dynamic optimization of the distributed system be not coupled between subsystem, and whole system high-efficiency and economic under guaranteeing to provide the prerequisite of total prediction refrigeration duty is run.

Summary of the invention

For prior art above shortcomings, the present invention proposes a kind of hierarchical optimal algorithm of United system, the method carries out economy modeling to electric refrigerating machine and Ice Storage Tank and cooling power under using the method for mixed integer programming to obtain economic target optimal conditions during the steady-state operation of each electric refrigerating machine and Ice Storage Tank; Lower floor use Distributed Predictive Control method to ensure each electric refrigerating machine and Ice Storage Tank follow the tracks of the stable state setting value and total cooling power tracking prediction load that upper strata economic optimization draws as far as possible in dynamic process, solve target coupling but the Distributed Predictive Control problem that is not coupled of subsystem.

For realizing above-mentioned object, the invention provides a kind of hierarchical optimal algorithm of United system, described United system is the United system be made up of conventional electric refrigerating machine and ice-storage system, and described method specifically comprises the steps:

Step S1: utilize its power consumption of electric refrigerating machine service data matching about the function of refrigeration work consumption and cooling water inlet temperature;

Step S2: utilize the electric refrigerating machine power consumption that obtains in previous step about the function of its refrigeration work consumption and cooling water inlet temperature set up electric refrigerating machine cooling economic model and and then set up the economic model of ice-storage system cooling, the objective function of final formation steady-state economy optimization problem and constraint condition also solve, and obtain start and stop state under optimal economic benefit condition of each electric refrigerating machine and Ice Storage Tank and cooling power;

Step S3: on the basis of previous step, the coordination Distributed Predictive Control method of design object coupling optimizes the real-time cooling set value of the power of whole cold supply system at each electric refrigerating machine of each sampling instant again to improve the dynamic property of whole system load-responsive;

Step S4: utilize alternative manner to solve above-mentioned Distributed Predictive Control problem, obtain the cooling set value of the power of each each sampling instant of electric refrigerating machine.

Preferably, in described step S1: the most important performance parameter Energy Efficiency Ratio of centrifugal electric refrigerating machine is the ratio of refrigeration work consumption and power consumption.In engineering practice, needing the data such as the cooling water inlet temperature according to reality, power consumption and refrigeration work consumption to be represented by power consumption by regretional analysis becomes cooling water inlet temperature and refrigeration work consumption Binary quadratic functions form to utilize mathematical programming instrument to carry out solving of subsequent step optimization problem.

Preferably, in described step S2: every platform electric refrigerating machine has opening and closing two duties, therefore 0-1 discrete variable Y is adopted to describe the duty of electric refrigerating machine.Differentiation for Double-working-condition electric refrigerating machine air conditioning condition and ice making operating mode then needs use two 0-1 discrete variables to describe respectively.The refrigeration work consumption of electric refrigerating machine then describes by continuous variable.According to the regretional analysis carried out in step S1, for conventional electric refrigerating machine, its power consumption can be expressed as the Binary quadratic functions about refrigeration work consumption and cooling water inlet temperature:

$P_{\mathrm{in}}=g(P_{\mathrm{out}},T_c^{\mathrm{in}})---\left(1\right)$

So the electricity charge cost C that electric refrigerating machine cooling produces is:

C＝Y×g(P _out,T)×p×t (2)

Wherein p is cooling period electricity price, and t is cooling period of time length.Y is electric refrigerating machine start and stop marks, for the electric refrigerating machine power consumption that simulates in step S1 is about the function of refrigeration work consumption and cooling water inlet temperature.

Ignoring under the prerequisite of Ice Storage Tank storage cold with storage time loss, in order to consider the cost of Ice Storage Tank cooling, introduce Ice Storage Tank cold average price p _sconcept, describes the price of unit cold, this price to fill cold, to fill cold period electricity price relevant with Double-working-condition electric refrigerating machine ice making operating mode input/output relation, defined by following formula:

$p_s=\frac{p_s{W}_{\mathrm{init}}+{\mathrm{pt}}^{\mathrm{in}}{P}_m-p_{\mathrm{init}}{t}^{\mathrm{out}}{P}_s{Y}_s}{W_{\mathrm{init}}+t^{\mathrm{in}}{P}_m-t^{\mathrm{out}}{P}_s{Y}_s}---\left(3\right)$

Wherein p _initfor cold average price initial value, W _initfor Ice Storage Tank cold initial value, t ⁱⁿfor rushing cool time, t ^outfor Discharging time, P _mfor the total ice making power of Double-working-condition electric refrigerating machine, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark.As can be seen from the definition of Ice Storage Tank cold average price, not considering under the prerequisite of storing up cold loss in time, whether average price and Ice Storage Tank let cool irrelevant, only may change when Ice Storage Tank is rushed cold.

Ice Storage Tank storage cold increment is:

ΔW＝t ⁱⁿP-t ^outP _sY _s(4)

Objective function is electricity charge sum needed for various electric refrigerating machine air conditioning condition cooling and Ice Storage Tank cooling:

$\underset{Y_i,Y^s,P_{\mathrm{out},i},P^s}{\min }J=\mathrm{pt\Σ}{Y}_i{g}_i(P_{\mathrm{out},i},T_{c,i}^{\mathrm{in}})+p_{s}t{P}_s{Y}_s---\left(5\right)$

P is cooling period electricity price, and t is cooling period of time length, Y _ibe i-th electric cold-producing medium start and stop mark, be i-th electric refrigerating machine power consumption about the function of refrigeration work consumption and cooling water inlet temperature, wherein P _{out, i}be i-th electric refrigerating machine cooling power, T _c,i ⁱⁿbe i-th electric refrigerating machine cooling water inlet temperature, Y _sfor Ice Storage Tank cooling start and stop mark, p _sfor Ice Storage Tank cold average price, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark.

Constraint condition comprises cooling power and prediction balancing the load, each electric refrigerating machine rated power limits, Ice Storage Tank stores up cold bound and ice-storage system mode of operation is unique:

$\left\{\begin{array}{c}{Y}_i{P}_{\mathrm{out},i}+P_s{Y}_s=P_p\\ {\mathrm{\α}}_i{P}_{i,\mathrm{rated}}\≤P_{\mathrm{out},i}\≤P_{i,\mathrm{rated}}\\ W_{\min }\≤W{\≤W}_{\max }\\ Y_s\×Y_{\mathrm{ice}}=0\\ Y_{\mathrm{ice}}\×Y_{\mathrm{air},i}=0,i=\mathrm{1,2},...,N\end{array}\right.---\left(6\right)$

Wherein Y _ibe i-th electric cold-producing medium start and stop mark, P _{out, i}be i-th electric refrigerating machine cooling power, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark, P _pfor prediction load, α _i∈ (0,1) is i-th electric refrigerating machine minimum output power ratio, P _{i, rated}be i-th electric refrigerating machine output rating, W is Ice Storage Tank storage cold, W _maxand W _minfor Ice Storage Tank storage cold bound; Y _icefor refrigerating unit with dual duty ice making operating mode start and stop mark, during acquiescence ice making operating mode, all Double-working-condition cold are opened simultaneously, accelerate ice making speed; Y _{air, i}be i-th refrigerating unit with dual duty air conditioning condition start and stop mark, N platform refrigerating unit with dual duty altogether;

Finally, under the prerequisite of the objective function and constraint condition that form steady-state economy optimization problem, above-mentioned optimization problem is solved, obtain start and stop state under optimal economic benefit condition of each electric refrigerating machine and Ice Storage Tank and cooling power.

Preferably, in step S3, the coordination Distributed Predictive Control method of described mark coupling, specific as follows:

Dcs is divided into N number of subsystem according to air conditioning condition electric refrigerating machine quantity, each subsystem is made up of the electric refrigerating machine of Model Predictive Control (MPC) and corresponding control, the data of different sub-systems MPC use communicator to carry out real-time exchange and complete calculation task, thus make the cooling power under the optimal economic benefit condition obtained in each air conditioning condition electric refrigerating machine cooling power tracking step S2 and cooling power sum under total cooling power tracking each air conditioning condition electric refrigerating machine optimal economic benefit condition;

Electric refrigerating machine refrigeration work consumption dynamic process model transport function is as follows:

$H\left(s\right)=\frac{1}{1+\mathrm{\τs}}{e}^{-{\mathrm{\τ}}_{d}s}---\left(7\right)$

τ is first order inertial loop time constant, τ _dfor delay time, then s subsystem separate manufacturing firms model form is:

$\begin{array}{c}\left\{\begin{array}{c}{x}_s(k+1)=A_s{x}_s\left(k\right)+B_s{u}_s\left(k\right)\\ p_s\left(k\right)=C_s{x}_s\left(k\right)\end{array}\right.\\ s=\mathrm{1,2},L,N\end{array}---\left(8\right)$

Wherein u _sk () is k moment electric refrigerating machine output power setting value, be the optimized variable of optimization problem in step S3, p _sk () is k+1 moment electric refrigerating machine real output, x _s(k)=[p _s(k) p _s(k+1) L p _s(k+n _d)] ^tfor state variable, n _dfor electric refrigerating machine delay time is relative to the multiple in discrete system sampling time; A _s, B _s, C _sfor the matrix of coefficients in state space equation, wherein determined by discrete system sampling time and electric refrigerating machine dynamic process time constant, R represents set of real numbers, if the sampling time is Δ t, then has:

$A_s=\left[\begin{array}{cccc}0& 1& L& 0\\ 0& 0& O& M\\ M& O& O& 1\\ 0& L& 0& e^{-\mathrm{\Δt}/\mathrm{\τ}}\end{array}\right]$

B _s＝[0 L 0 1-e ^-Δt/τ] ^T(9)

C _s＝[1 0 L 0]

For s subsystem, s=1,2 ..., N; MPC coordination strategy comprise electric refrigerating machine in the prediction time domain that following P sampling instant forms, predict the single air conditioner operating mode electric refrigerating machine optimal economic benefit condition that output power obtains close to step S2 as far as possible under cooling power r _sand be different from general MPC strategy, predict that output power catches up with the cooling power sum r under each air conditioning condition electric refrigerating machine optimal economic benefit condition as far as possible to control all electric refrigerating machines, in objective function, introduce the quadratic sum of the cooling power sum gap in prediction time domain under each the total predicted power of air conditioning condition electric refrigerating machine and optimal economic benefit condition, being therefore coupled appears in different sub-systems MPC objective function; To sum up, s subsystem MPC objective function is as follows:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}---\left(10\right)$

Wherein, U _s=[u _s(k) u _s(k+1) ... u _s(k+P-1)] ^tfor electric refrigerating machine output power setting value sequence in prediction time domain, u _sk () is k moment electric refrigerating machine output power setting value, be the optimized variable of optimization problem; And Q, R ∈ [0,1] is weight coefficient, is used for weighing separate unit electric refrigerating machine respectively and predicts that Steam Generator in Load Follow degree and all electric refrigerating machine total loads follow the tracks of the importance of degree; r _s(k+i) be cooling power under (k+i) moment s platform air conditioning condition electric refrigerating machine optimal economic benefit condition, r (k+i) is the cooling power sum of (k+i) moment each air conditioning condition electric refrigerating machine under optimal economic benefit condition; be its output power in (k+i) moment that s platform air conditioning condition electric refrigerating machine was predicted by forecast model in the k moment, for its output power sum in (k+i) moment that all the other the air conditioning condition electric refrigerating machines except s platform air conditioning condition electric refrigerating machine were predicted by forecast model in the k moment;

MPC forecast model is:

$\begin{array}{c}\left\{\begin{array}{c}{\hat{x}}_s(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_s{u}_s(k+j-1)\right)\\ \stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)\end{array}\right.\\ i=\mathrm{1,2},L,P\end{array}---\left(11\right)$

for state variable in (k+i) moment s platform air conditioning condition electric refrigerating machine state space equation of being calculated by forecast model in the k moment, it is its output power in (k+i) moment that s platform air conditioning condition electric refrigerating machine was predicted by forecast model in the k moment; A _s ^i-jand A _s ⁱfor the corresponding power of matrix of coefficients in state space equation.

For s subsystem MPC controller, whole optimization problem form is:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}$

$\begin{array}{c}\begin{array}{cc}s.t.& {\hat{x}}_s\end{array}(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_s{u}_s(k+j-1)\right)\\ \stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)\\ u_{s,\min }\≤u_s(k+i)\≤u_{s,\max }\\ p_{s,\min }\≤{\hat{p}}_s(k+i|k)\≤p_{s,\max }\\ i=\mathrm{1,2},L,P\end{array}---\left(12\right)$

Wherein, { u _{s, min}, u _{s, max}, { p _{s, min}, p _{s, max}it is the boundary constraint of control variable.

Preferably, described step S4 specifically comprises:

Step S41: in the k moment, according to the cooling power r under the every platform air conditioning condition electric refrigerating machine optimal economic benefit condition obtained in step S3 _s, the electric refrigerating machine output power setting value sequence U of each subsystem of initialization in prediction time domain _s, make iterations l=0:

$\begin{array}{c}{U}_s^l={\left[u_s^l(k+1)u_s^l(k+2)K{u}_s^l(k+P)\right]}^T\\ ={\left[r_s(k+1)r_s(k+2)K{r}_s(k+P)\right]}^T\\ s=\mathrm{1,2},L,N\end{array}$

Above-mentioned be the l time iteration electric refrigerating machine output power setting value in (k+1) moment, P is prediction time domain length;

Step S42: according to MPC forecast model, calculates each subsystem MPC and controls output quantity predicted value in time domain ${\hat{p}}_s={[\stackrel{^^^}{p_s}(k+1|k)p_s(k+2|k)...p_s(k+P|k)]}^T,$ And inform other N-1 subsystem MPC;

Step S43: for each subsystem, the optimization problem in solution procedure S3, obtains optimum solution, i.e. optimum refrigeration machine output power setting value sequence

Step S44: check whether the condition of convergence of all subsystem MPC meets, namely to given precision ε _swhether ∈ R (s=1,2, L, N), all have if all subsystem conditions of convergence meet, make each subsystem optimum control amount with optimum refrigeration machine output power setting value sequence equal, namely $U_s^{*}=U_s^{(l*)},$ Go to step S45; Otherwise, order $U_s^{l+1}={\mathrm{\α}}_s^{(l*)}+(1-\mathrm{\α})U_s^l(s=\mathrm{1,2},L,N),$ α is the constant between 0 and 1, introduces the situation that this parameter does not restrain in order to avoiding method of trying one's best, chooses numerical values recited, l=l+1, go to step S42 by demand;

Step S45: choose controlled quentity controlled variable in the k moment $u_s^{*}=\left[\begin{array}{cccc}1& 0& L& 0\end{array}\right]U_s^{*}(s=\mathrm{1,2},...,N)$ Act on corresponding subsystem.

Step S46: barrel shift, to next sampling instant, i.e. k+1 → k, returns step S41, repeats said process.

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

The present invention is directed in the hierarchical optimal algorithm of the United system be made up of conventional electric refrigerating machine and ice-storage system by establishing the economic model of electric refrigerating machine and Ice Storage Tank, and use mixed integer programming and Distributed Predictive Control to solve steady-state optimization problem and optimization problems respectively, determine the dispatching method of United system quantitatively, whole system high-efficiency and economic under guaranteeing to provide the prerequisite of total prediction refrigeration duty is run, for cold supply system energy management and Optimized Operation provide rational suggestion.

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 the United system structural drawing be made up of conventional electric refrigerating machine and ice-storage system;

Fig. 2 is the process flow diagram of one embodiment of the invention;

Fig. 3 is the method schematic diagram of one embodiment of the invention;

Fig. 4 is the Distributed Predictive Control iterative method flow diagram of one embodiment of the invention;

Fig. 5 is 3900kW cold regretional analysis figure in one embodiment of the invention;

Fig. 6 is 2150kW cold regretional analysis figure in one embodiment of the invention;

Fig. 7 is 6329kW cold regretional analysis figure in one embodiment of the invention;

Fig. 8 is load prediction figure in one embodiment of the invention;

Fig. 9 is two kinds of every halfhour electricity charge comparison diagrams of strategy in one embodiment of the invention;

Figure 10 is the Dynamic performance Optimization simulation result of one embodiment of the invention, wherein: (a) is overall, b () is First 3900kW electric refrigerating machine, (c) is second 3900kW electric refrigerating machine, and (d) is First 2150kW electric refrigerating machine.

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.Below in conjunction with accompanying drawing, the United system hierarchical optimal algorithm be made up of conventional electric refrigerating machine and ice-storage system of the present invention is described in further detail.

As shown in Figure 1, the United system structure for being made up of conventional electric refrigerating machine and ice-storage system.

Fig. 2 is the process flow diagram of the United system hierarchical optimal algorithm be made up of conventional electric refrigerating machine and ice-storage system of a specific embodiment of the present invention.First according to each low-temperature receiver economic model and tou power price policy, adopt Integer programming, optimize opening and the optimal power setting value of each low-temperature receiver, consider dynamic adjustment performance and the physical constraint of each low-temperature receiver simultaneously, in order to improve dynamic performance better, devise the coordination Distributed Predictive Control method of a kind of target coupling, under Distributed Architecture, again optimize the setting value of each low-temperature receiver, while making each low-temperature receiver ensure total load in dynamic process, follow the tracks of optimum refrigeration work consumption setting value as far as possible.

Concrete, as shown in Figure 2, the United system hierarchical optimal algorithm be made up of conventional electric refrigerating machine and ice-storage system of the present invention, comprises the following steps:

Step S1: utilize its power consumption of electric refrigerating machine service data matching about the function of refrigeration work consumption and cooling water inlet temperature;

The ratio that the most important performance parameter Energy Efficiency Ratio (COP) of centrifugal electric refrigerating machine is refrigeration work consumption and power consumption.In engineering practice, needing the data such as the cooling water inlet temperature according to reality, power consumption and refrigeration work consumption to be represented by power consumption by regretional analysis becomes cooling water inlet temperature and refrigeration work consumption Binary quadratic functions form to utilize mathematical programming instrument to carry out solving of subsequent step optimization problem.

Step S2: the economic model setting up electric refrigerating machine and ice-storage system cooling, forms the objective function of steady-state economy optimization problem and constraint condition and solves;

Every platform electric refrigerating machine has opening and closing two duties, therefore adopts 0-1 discrete variable Y to describe the duty of electric refrigerating machine.Differentiation for Double-working-condition electric refrigerating machine air conditioning condition and ice making operating mode then needs use two 0-1 discrete variables to describe respectively.The refrigeration work consumption of electric refrigerating machine then describes by continuous variable.According to the regretional analysis carried out in step S1, for conventional electric refrigerating machine, its power consumption can be expressed as the Binary quadratic functions about refrigeration work consumption and cooling water inlet temperature:

$P_{\mathrm{in}}=g(P_{\mathrm{out}},T_c^{\mathrm{in}})---\left(1\right)$

So the electricity charge cost C that electric refrigerating machine cooling produces is:

C＝Y×g(P _out,T)×p×t (2)

Wherein p is cooling period electricity price, and t is cooling period of time length.

Ignoring under the prerequisite of Ice Storage Tank storage cold with storage time loss, in order to consider the cost of Ice Storage Tank cooling, introduce Ice Storage Tank cold average price p _sconcept, describes the price of unit cold, this price to fill cold, to fill cold period electricity price relevant with Double-working-condition electric refrigerating machine ice making operating mode input/output relation, defined by following formula:

$p_s=\frac{p_s{W}_{\mathrm{init}}+{\mathrm{pt}}^{\mathrm{in}}{P}_m-p_{\mathrm{init}}{t}^{\mathrm{out}}{P}_s{Y}_s}{W_{\mathrm{init}}+t^{\mathrm{in}}{P}_m-t^{\mathrm{out}}{P}_s{Y}_s}---\left(3\right)$

Wherein p _initfor cold average price initial value, W _initfor Ice Storage Tank cold initial value, t ⁱⁿfor rushing cool time, t ^outfor Discharging time, P _mfor the total ice making power of Double-working-condition electric refrigerating machine, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark.As can be seen from the definition of Ice Storage Tank cold average price, not considering under the prerequisite of storing up cold loss in time, whether average price and Ice Storage Tank let cool irrelevant, only may change when Ice Storage Tank is rushed cold.

Ice Storage Tank storage cold increment is:

ΔW＝t ⁱⁿP-t ^outP _sY _s(4)

Objective function is electricity charge sum needed for various electric refrigerating machine air conditioning condition cooling and Ice Storage Tank cooling:

$\underset{Y_i,Y^s,P_{\mathrm{out},i},P^s}{\min }J=\mathrm{pt\Σ}{Y}_i{g}_i(P_{\mathrm{out},i},T_{c,i}^{\mathrm{in}})+p_{s}t{P}_s{Y}_s---\left(5\right)$

P _{out, i}be i-th electric refrigerating machine cooling power, T _c,i ⁱⁿbe i-th electric refrigerating machine cooling water inlet temperature.

Main constraint condition comprises cooling power and prediction balancing the load, each electric refrigerating machine rated power limits, Ice Storage Tank stores up cold bound and ice-storage system mode of operation is unique:

$\left\{\begin{array}{c}{Y}_i{P}_{\mathrm{out},i}+P_s{Y}_s=P_p\\ {\mathrm{\α}}_i{P}_{i,\mathrm{rated}}\≤P_{\mathrm{out},i}\≤P_{i,\mathrm{rated}}\\ W_{\min }\≤W{\≤W}_{\max }\\ Y_s\×Y_{\mathrm{ice}}=0\\ Y_{\mathrm{ice}}\×Y_{\mathrm{air},i}=0,i=\mathrm{1,2},...,N\end{array}\right.---\left(6\right)$

Wherein P _pfor prediction load, α _i∈ (0,1) is i-th electric refrigerating machine minimum output power ratio, P _{i, rated}be i-th electric refrigerating machine output rating, W _maxand W _minfor Ice Storage Tank storage cold bound.Y _icefor refrigerating unit with dual duty ice making operating mode start and stop mark, during acquiescence ice making operating mode, all Double-working-condition cold are opened simultaneously, accelerate ice making speed.Y _{air, i}be i-th refrigerating unit with dual duty air conditioning condition start and stop mark, N platform refrigerating unit with dual duty altogether.

Step S3: the coordination Distributed Predictive Control method basis of the optimum refrigeration work consumption of each electric refrigerating machine obtained in the optimization of previous step steady-state economy being designed the coupling of a kind of target optimizes the dynamic property of whole cold supply system load-responsive;

Dcs is divided into N number of subsystem according to air conditioning condition electric refrigerating machine quantity, each subsystem is made up of the electric refrigerating machine of MPC and corresponding control, the data of different sub-systems MPC use communicator to carry out real-time exchange and complete calculation task thus make the expectation value that calculates in each air conditioning condition electric refrigerating machine cooling power tracking step S2 and total cooling power tracking total expected value, and concrete distributed control system structure as shown in Figure 3.

Electric refrigerating machine refrigeration work consumption dynamic process model transport function is as follows:

$H\left(s\right)=\frac{1}{1+\mathrm{\τs}}{e}^{-{\mathrm{\τ}}_{d}s}---\left(7\right)$

τ is first order inertial loop time constant, τ _dfor delay time, then s subsystem separate manufacturing firms model form is:

$\begin{array}{c}\left\{\begin{array}{c}{x}_s(k+1)=A_s{x}_s\left(k\right)+B_s{u}_s\left(k\right)\\ p_s\left(k\right)=C_s{x}_s\left(k\right)\end{array}\right.\\ s=\mathrm{1,2},L,N\end{array}---\left(8\right)$

Wherein u (k) is k moment electric refrigerating machine output power setting value, is the optimized variable of optimization problem, p _sk () is k+1 moment electric refrigerating machine real output, n _dfor electric refrigerating machine delay time is relative to the multiple in discrete system sampling time.X _s(k)=[p _s(k) p _s(k+1) L p _s(k+n _d)] ^tfor state variable. determine if the sampling time is Δ t, then have by sampling time and electric refrigerating machine time constant:

$A_s=\left[\begin{array}{cccc}0& 1& L& 0\\ 0& 0& O& M\\ M& O& O& 1\\ 0& L& 0& e^{-\mathrm{\Δt}/\mathrm{\τ}}\end{array}\right]$

B _s＝[0 L 0 1-e ^-Δt/τ] ^T(9)

C _s＝[1 0 L 0]

For s subsystem (s=1,2 ..., N), MPC coordination strategy comprises electric refrigerating machine and predict output power as far as possible close to the expectation power r that step S2 calculates in the prediction time domain that following P sampling instant forms _sand be different from general MPC strategy, total expectation power r is caught up with as far as possible in order to control all electric refrigerating machines prediction output power, this strategy introduces total predicted power and the quadratic sum always expecting difference power distance in prediction time domain in objective function, and therefore coupling appears in different sub-systems MPC objective function.To sum up, s subsystem MPC objective function is as follows:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}---\left(10\right)$

Wherein, U _s=[u (k) u (k+1) ... u (k+P-1)] ^tfor electric refrigerating machine output power setting value sequence in prediction time domain, and Q, R ∈ [0,1] is weight coefficient, is used for weighing separate unit electric refrigerating machine respectively and predicts that Steam Generator in Load Follow degree and all electric refrigerating machine total loads follow the tracks of the importance of degree.

MPC forecast model is:

$\begin{array}{c}\left\{\begin{array}{c}{\hat{x}}_s(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_s{u}_s(k+j-1)\right)\\ \stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)\end{array}\right.\\ i=\mathrm{1,2},L,P\end{array}---\left(11\right)$

For s subsystem MPC controller, whole optimization problem form is:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}$

$\begin{array}{c}\begin{array}{cc}s.t.& {\hat{x}}_s\end{array}(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_s{u}_s(k+j-1)\right)\\ \stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)\\ u_{s,\min }\≤u_s(k+i)\≤u_{s,\max }\\ p_{s,\min }\≤{\hat{p}}_s(k+i|k)\≤p_{s,\max }\\ i=\mathrm{1,2},L,P\end{array}---\left(12\right)$

Wherein, { u _{s, min}, u _{s, max}, { p _{s, min}, p _{s, max}it is the boundary constraint of control variable.

Step S4: utilize alternative manner to solve above-mentioned Distributed Predictive Control problem, obtain the cooling set value of the power of each each sampling instant of electric refrigerating machine;

Alternative manner process flow diagram as shown in Figure 4.

Step S41: in the k moment, according to the desirable refrigeration work consumption r of the every platform electric refrigerating machine solved in step S3 _sthe U of each subsystem of initialization in prediction time domain _s, make iterations l=0,

$\begin{array}{c}{U}_s^l={\left[u_s^l(k+1)u_s^l(k+2)K{u}_s^l(k+P)\right]}^T\\ ={\left[r_s(k+1)r_s(k+2)K{r}_s(k+P)\right]}^T\\ s=\mathrm{1,2},L,N\end{array}$

Step S42: according to MPC forecast model, calculates each subsystem MPC and controls output quantity predicted value in time domain ${\hat{p}}_s={[\stackrel{^^^}{p_s}(k+1|k)p_s(k+2|k)...p_s(k+P|k)]}^T,$ And inform other N-1 subsystem MPC.

Step S43: the optimization problem in each subsystem step S3, obtains optimum solution

Step S44: check whether the condition of convergence of all subsystem MPC meets, namely to given precision ε _swhether ∈ R (s=1,2, L, N), all have if all subsystem conditions of convergence meet, make each subsystem optimum control amount go to step S45; Otherwise, order α is the constant between 0 and 1, introduces the situation that this parameter does not restrain in order to avoiding method of trying one's best, chooses numerical values recited, l=l+1, go to step S42 by demand.

Step S45: choose controlled quentity controlled variable in the k moment $u_s^{*}=\left[\begin{array}{cccc}1& 0& L& 0\end{array}\right]U_s^{*}(s=\mathrm{1,2},...,N)$ Act on corresponding subsystem.

Step S46: barrel shift, to next sampling instant, i.e. k+1 → k, returns step S41, repeats said process.

Based on above-mentioned implementation process, the present invention has carried out simulating, verifying for Shanghai high-rise building low district cold supply system.In this cold supply system, conventional electric refrigerating machine totally three, two specified refrigeration work consumption 3900kW, a specified refrigeration work consumption 2150kW, Double-working-condition electric refrigerating machine totally three, the specified refrigeration work consumption 6329kW of air conditioning condition, the specified refrigeration work consumption 3868kW of ice making operating mode.

First, run according to electric refrigerating machine the matching that measured data carries out input/output relation function by step S1.

Fig. 5, Fig. 6, Fig. 7 are respectively 3900kW, 2150kW and 6329kW centrifugal electric refrigerating machine Regression Analysis Result figure.In each figure, X1 is cold cooling power perunit value, and X2 is cold cooling water inlet temperature, and Y1 is cold power consumption perunit value.The fiducial interval of region corresponding to confidence level 99% that in figure, red curve is surrounded

As can be seen from Fig. 5-7 figure, dualistic and quadric regression analysis quite accurately can set up electric refrigerating machine input electric power and the funtcional relationship exporting cold power and cooling water inlet temperature, concrete function expression as following various, P _infor electric refrigerating machine power consumption, P _outfor electric refrigerating machine refrigeration work consumption, for electric refrigerating machine cooling water inlet temperature.

The centrifugal electric refrigerating machine of 3900kW is had:

$P_{\mathrm{in}}=146.7+0.033P_{\mathrm{out}}-{7.3T}_c^{\mathrm{in}}+{0.17}_c^{\mathrm{in}2}+3.5\×{10}^{-3}{P}_{\mathrm{out}}{T}_c^{\mathrm{in}}+3.2\×{10}^{-6}{P}_{\mathrm{out}}^2---\left(13\right)$

The centrifugal electric refrigerating machine of 2150kW is had:

$P_{\mathrm{in}}=88.9+0.042P_{\mathrm{out}}-{4.6T}_c^{\mathrm{in}}+{0.10T}_c^{\mathrm{in}2}+3.3\×{10}^{-3}{P}_{\mathrm{out}}{T}_c^{\mathrm{in}}+1.8\×{10}^{-6}{P}_{\mathrm{out}}^2---\left(14\right)$

Double-working-condition electric refrigerating machine 6329kW air conditioning condition is had:

$P_{\mathrm{in}}=270.9+0.030P_{\mathrm{out}}-{9.1T}_c^{\mathrm{in}}+{0.19T}_c^{\mathrm{in}2}+3.9\×{10}^{-3}{P}_{\mathrm{out}}{T}_c^{\mathrm{in}}+4.7\×{10}^{-6}{P}_{\mathrm{out}}^2---\left(15\right)$

For the ice making operating mode of Double-working-condition electric refrigerating machine, in order to accelerate ice making speed, the electric refrigerating machine total power ice making when distributing ice making, so there is not the operating mode of Partial Power ice making, now ice making power cold and power consumption elec is only respectively about cooling water inlet temperature T _c ⁱⁿone-place 2-th Order function:

$\left\{\begin{array}{c}\mathrm{cold}=-{1.87T}_c^{\mathrm{in}2}+{17.63}_c^{\mathrm{in}}+5239.50\\ \mathrm{elec}=-{0.50T}_c^{\mathrm{in}2}+23.07T_c^{\mathrm{in}}+766.22\end{array}\right.---\left(16\right)$

Then, carry out steady-state economy optimization Simulation by step S2, prediction load and electrovalence policy as follows.As shown in Figure 8.

Table 1 tou power price policy

Period (time)	Electricity price (unit/kWh)
		22:00～06:00	0.234

06:00～08:00	0.706
		08:00～11:00	1.037
11:00～13:00	0.706
		13:00～15:00	1.037
15:00～18:00	0.706
		18:00～21:00	1.037
21:00～22:00	0.706

In order to economic optimization strategy joint takes effect, choose a kind of stable state scheduling strategy of routine as a comparison, this strategy preferential Ice Storage Tank when electricity price valley is rushed cold, and the preferential electric refrigerating machine cooling when electricity price level values and peak value, each electric refrigerating machine is enabled by the order that rated power is ascending.Two kinds of per electricity charge half an hour of strategy are to such as Fig. 9.

Through calculating, steady-state economy optimisation strategy is under given prediction loading condiction, and in 6 hours, total electricity charge cost is 3970 yuan, and the total electricity charge cost of Comparing method is 4053 yuan, and in 6 hours, strategy of the present invention saves expense 2% relatively.

Subsequently, the design of Distributed Predictive Control optimisation strategy and problem solving is carried out according to step S3 and S4.Dynamic optimization sampling interval is 2.5 minutes, namely after every 30 minutes of the stable state scheduling strategy of step S2 calculates once each electric refrigerating machine and the optimum cooling power of Ice Storage Tank, and Dynamic performance Optimization upgrades once each electric refrigerating machine cooling set value of the power for every 2.5 minutes to guarantee each electric refrigerating machine and the Ice Storage Tank stable state setting value that draws of tracking step S2 economic optimization and total cooling power tracking predicts load as far as possible in dynamic process.Ice groove cooling first order modeling time constant is set to 2 minutes, and each electric refrigerating machine air conditioning condition dynamic model time constant and delay time are set as follows shown in table:

Table 2 air conditioning condition electric refrigerating machine dynamic parameter

The expectation of the air conditioning condition electric refrigerating machine that overall He Getai enabled is born load and is applied respectively and do not apply the tactful real-time cooling powertrace caused of DMPC as shown in picture group 10, because under current predictive load condition, 6329kW electric refrigerating machine is not opened, so only include 3900kW and 2150kW electric refrigerating machine in picture group.

The air conditioning condition electric refrigerating machine real-time cooling power that when adopting in order to quantitative description or do not adopt DMPC tactful, whole system and every platform were enabled is with the difference of the optimum cooling power degree of closeness under the economic optimum condition calculated in step S2, calculate the quadratic sum of each sampling instant deviation in 6 hours, numerical value is as shown in table 3, can find out that after adopting Dynamic performance Optimization strategy, sum of square of deviations obviously reduces, achieve the rapid response of overall system and each subsystem, embody good control effects.

Table 3 dynamic optimization effect

Load object	Adopt DMPC strategy	Do not adopt DMPC strategy
			Totally	3.40×10 7	1.03×10 8
First 3900kW electric refrigerating machine air conditioning condition	4.91×10 7	1.11×10 8
			Second 3900kW electric refrigerating machine air conditioning condition	8.71×10 6	3.82×10 7
First 2150kW electric refrigerating machine air conditioning condition	8.71×10 6	3.82×10 7

Comprehensive simulating result, the hierarchical optimal algorithm of the United system be made up of conventional electric refrigerating machine and ice-storage system that the present invention proposes has good economic worth and Dynamic controlling effect concurrently, and running for actual cold supply system economic optimization has certain directive significance.

In sum, by establishing the economic model of electric refrigerating machine and Ice Storage Tank in the hierarchical optimal algorithm of the United system for being made up of conventional electric refrigerating machine and ice-storage system of the present invention, and use mixed integer programming and Distributed Predictive Control to solve steady-state optimization problem and optimization problems respectively, determine the dispatching method of United system quantitatively.

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. a hierarchical optimal algorithm for United system, described United system is the United system be made up of conventional electric refrigerating machine and ice-storage system, it is characterized in that, described method comprises the steps:

Step S1: utilize its power consumption of electric refrigerating machine service data matching about the function of refrigeration work consumption and cooling water inlet temperature;

Step S2: utilize the electric refrigerating machine power consumption that obtains in previous step about the function of its refrigeration work consumption and cooling water inlet temperature set up electric refrigerating machine cooling economic model and and then set up the economic model of ice-storage system cooling, the objective function of final formation steady-state economy optimization problem and constraint condition also solve, and obtain start and stop state under optimal economic benefit condition of each electric refrigerating machine and Ice Storage Tank and cooling power;

Step S3: on the basis of previous step, the coordination Distributed Predictive Control method of design object coupling optimizes the real-time cooling set value of the power of whole cold supply system at each electric refrigerating machine of each sampling instant again to improve the dynamic property of whole system load-responsive;

Step S4: utilize alternative manner to solve above-mentioned Distributed Predictive Control problem, obtain the cooling set value of the power of each each sampling instant of electric refrigerating machine.

2. the hierarchical optimal algorithm of United system as claimed in claim 1, it is characterized in that, in described step S1: the most important performance parameter Energy Efficiency Ratio of centrifugal electric refrigerating machine is the ratio of refrigeration work consumption and power consumption, by regretional analysis power consumption to be represented according to the cooling water inlet temperature of reality, power consumption and refrigeration work consumption data and become cooling water inlet temperature and refrigeration work consumption Binary quadratic functions form, to utilize mathematical programming instrument to carry out solving of subsequent step optimization problem.

3. the hierarchical optimal algorithm of United system as claimed in claim 2, is characterized in that, in described step S2: every platform electric refrigerating machine has opening and closing two duties, adopts 0-1 discrete variable Y to describe the duty of electric refrigerating machine; Differentiation for Double-working-condition electric refrigerating machine air conditioning condition and ice making operating mode then uses two 0-1 discrete variables to describe respectively; The refrigeration work consumption of electric refrigerating machine then describes by continuous variable; According to the regretional analysis carried out in step S1, for conventional electric refrigerating machine, its power consumption P _inbe expressed as about refrigeration work consumption P _outwith cooling water inlet temperature binary quadratic functions:

$P_{\mathrm{in}}=g(P_{\mathrm{out}},T_c^{\mathrm{in}})---\left(1\right)$

So the electricity charge cost C that electric refrigerating machine cooling produces is:

$C=Y\×g(P_{\mathrm{out}},T_c^{\mathrm{in}})\×p\×t---\left(2\right)$

Wherein p is cooling period electricity price, and t is cooling period of time length, and Y is electric refrigerating machine start and stop marks, for the electric refrigerating machine power consumption that simulates in step S1 is about the function of refrigeration work consumption and cooling water inlet temperature;

Ignoring under the prerequisite of Ice Storage Tank storage cold with storage time loss, in order to consider the cost of Ice Storage Tank cooling, introduce Ice Storage Tank cold average price p _sconcept, describes the price of unit cold, this price to fill cold, to fill cold period electricity price relevant with Double-working-condition electric refrigerating machine ice making operating mode input/output relation, defined by following formula:

$p_s=\frac{p_s{W}_{\mathrm{init}}+{\mathrm{pt}}^{\mathrm{in}}{P}_m-p_{\mathrm{init}}{t}^{\mathrm{out}}{P}_s{Y}_s}{W_{\mathrm{init}}+t^{\mathrm{in}}{P}_m-t^{\mathrm{out}}{P}_s{Y}_s}---\left(3\right)$

Wherein p _initfor cold average price initial value, W _initfor Ice Storage Tank cold initial value, t ⁱⁿfor rushing cool time, t ^outfor Discharging time, P _mfor the total ice making power of Double-working-condition electric refrigerating machine, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark; Being found out by the definition of Ice Storage Tank cold average price, not considering under the prerequisite of storing up cold loss in time, whether average price and Ice Storage Tank let cool irrelevant, only may change when Ice Storage Tank is rushed cold;

Ice Storage Tank storage cold increment △ W is:

△W＝t ⁱⁿP-t ^outP _sY _s(4)

Objective function is electricity charge sum needed for various electric refrigerating machine air conditioning condition cooling and Ice Storage Tank cooling:

$\underset{Y_i,Y^s,P_{\mathrm{out},i},P^s}{\min }J=\mathrm{pt\Σ}{Y}_i{g}_i(P_{\mathrm{out},i},T_{c,i}^{\mathrm{in}})+p_s{\mathrm{tP}}_s{Y}_s---\left(5\right)$

P is cooling period electricity price, and t is cooling period of time length, Y _ibe i-th electric cold-producing medium start and stop mark, be i-th electric refrigerating machine power consumption about the function of refrigeration work consumption and cooling water inlet temperature, wherein P _{out, i}be i-th electric refrigerating machine cooling power, T _c,i ⁱⁿbe i-th electric refrigerating machine cooling water inlet temperature, Y _sfor Ice Storage Tank cooling start and stop mark, p _sfor Ice Storage Tank cold average price, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark;

Constraint condition comprises cooling power and prediction balancing the load, each electric refrigerating machine rated power limits, Ice Storage Tank stores up cold bound and ice-storage system mode of operation is unique:

$\left\{\begin{array}{c}{Y}_i{P}_{\mathrm{out},i}+P_s{Y}_s=P_p\\ {\mathrm{\α}}_i{P}_{i,\mathrm{rated}}\≤P_{\mathrm{out},i}\≤P_{i,\mathrm{rated}}\\ W_{\min }\≤W\≤W_{\max }\\ Y_s\×Y_{\mathrm{ice}}=0\\ Y_{\mathrm{ice}}\×Y_{\mathrm{air},i}=0,i=\mathrm{1,2},...,N\end{array}\right.---\left(6\right)$

Wherein Y _ibe i-th electric cold-producing medium start and stop mark, P _{out, i}be i-th electric refrigerating machine cooling power, P _sfor Ice Storage Tank cooling power, Y _sfor Ice Storage Tank cooling start and stop mark, P _pfor prediction load, α _i∈ (0,1) is i-th electric refrigerating machine minimum output power ratio, P _{i, rated}be i-th electric refrigerating machine output rating, W is Ice Storage Tank storage cold, W _maxand W _minfor Ice Storage Tank storage cold bound; Y _icefor refrigerating unit with dual duty ice making operating mode start and stop mark, during acquiescence ice making operating mode, all Double-working-condition cold are opened simultaneously, accelerate ice making speed; Y _{air, i}be i-th refrigerating unit with dual duty air conditioning condition start and stop mark, N platform refrigerating unit with dual duty altogether;

Finally, under the prerequisite of the objective function and constraint condition that form steady-state economy optimization problem, above-mentioned optimization problem is solved, obtain start and stop state under optimal economic benefit condition of each electric refrigerating machine and Ice Storage Tank and cooling power.

4. the hierarchical optimal algorithm of United system as claimed in claim 1, is characterized in that, in step S3, and the coordination Distributed Predictive Control method of described mark coupling, specific as follows:

Dcs is divided into N number of subsystem according to air conditioning condition electric refrigerating machine quantity, each subsystem is made up of the electric refrigerating machine of Model Predictive Control (MPC) and corresponding control, the data of different sub-systems MPC use communicator to carry out real-time exchange and complete calculation task, thus make the cooling power under the optimal economic benefit condition obtained in each air conditioning condition electric refrigerating machine cooling power tracking step S2 and cooling power sum under total cooling power tracking each air conditioning condition electric refrigerating machine optimal economic benefit condition;

Electric refrigerating machine refrigeration work consumption dynamic process model transport function is as follows:

$H\left(s\right)=\frac{1}{1+\mathrm{\τs}}{e}^{-{\mathrm{\τ}}_{d}s}---\left(7\right)$

τ is first order inertial loop time constant, τ _dfor delay time, then s subsystem separate manufacturing firms model form is:

$\left\{\begin{array}{c}{x}_s(k+1)=A_s{x}_s\left(k\right)+B_s{u}_s\left(k\right)\\ p_s\left(k\right)=C_s{x}_s\left(k\right)\end{array}\right.---\left(8\right)$

s＝1,2,L,N

Wherein u _sk () is k moment electric refrigerating machine output power setting value, be the optimized variable of optimization problem in step S3, p _sk () is k+1 moment electric refrigerating machine real output, x _s(k)=[p _s(k) p _s(k+1) L p _s(k+n _d)] ^tfor state variable, n _dfor electric refrigerating machine delay time is relative to the multiple in discrete system sampling time; A _s, B _s, C _sfor the matrix of coefficients in state space equation, wherein determined by discrete system sampling time and electric refrigerating machine dynamic process time constant, R represents set of real numbers, if the sampling time is △ t, then has:

$A_s=\left[\begin{array}{cccc}0& 1& L& 0\\ 0& 0& O& M\\ M& O& O& 1\\ 0& L& 0& e^{-\mathrm{\Δt}/\mathrm{\τ}}\end{array}\right]$

B _s＝[0 L 0 1-e ^-△t/τ] ^T(9)

C _s＝[1 0 L 0]

For s subsystem, s=1,2, N; MPC coordination strategy comprise electric refrigerating machine in the prediction time domain that following P sampling instant forms, predict the single air conditioner operating mode electric refrigerating machine optimal economic benefit condition that output power obtains close to step S2 as far as possible under cooling power r _sand predict that output power catches up with the cooling power sum r under each air conditioning condition electric refrigerating machine optimal economic benefit condition to control all electric refrigerating machines, in objective function, introduce the quadratic sum of the cooling power sum gap in prediction time domain under each the total predicted power of air conditioning condition electric refrigerating machine and optimal economic benefit condition, being therefore coupled appears in different sub-systems MPC objective function; To sum up, s subsystem MPC objective function is as follows:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}---\left(10\right)$

Wherein, U _s=[u _s(k) u _s(k+1) ... u _s(k+P-1)] ^tfor electric refrigerating machine output power setting value sequence in prediction time domain, u _sk () is k moment electric refrigerating machine output power setting value, be the optimized variable of optimization problem; And Q, R ∈ [0,1] is weight coefficient, is used for weighing separate unit electric refrigerating machine respectively and predicts that Steam Generator in Load Follow degree and all electric refrigerating machine total loads follow the tracks of the importance of degree; r _s(k+i) be cooling power under (k+i) moment s platform air conditioning condition electric refrigerating machine optimal economic benefit condition, r (k+i) is the cooling power sum of (k+i) moment each air conditioning condition electric refrigerating machine under optimal economic benefit condition; be its output power in (k+i) moment that s platform air conditioning condition electric refrigerating machine was predicted by forecast model in the k moment, for its output power sum in (k+i) moment that all the other the air conditioning condition electric refrigerating machines except s platform air conditioning condition electric refrigerating machine were predicted by forecast model in the k moment;

MPC forecast model is:

$\left\{\begin{array}{c}{\hat{x}}_s(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_s{u}_s(k+j-1)\right)\\ \stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)\end{array}\right.---\left(11\right)$

i＝1,2,L,P

for state variable in (k+i) moment s platform air conditioning condition electric refrigerating machine state space equation of being calculated by forecast model in the k moment, it is its output power in (k+i) moment that s platform air conditioning condition electric refrigerating machine was predicted by forecast model in the k moment; A _s ^i-jand A _s ⁱfor the corresponding power of matrix of coefficients in state space equation;

For s subsystem MPC controller, whole optimization problem form is:

$\begin{array}{c}\begin{array}{cc}\underset{U_s}{\min }& J_s=Q\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r_s(k+i)-{\hat{p}}_s(k+i|k))}^2\end{array}\\ +R\underset{i=1}{\overset{P}{\mathrm{\Σ}}}{(r(k+i)-\stackrel{^^}{p_s}(k+i|k)-\underset{j\≠s}{\overset{N}{\underset{j=1}{\mathrm{\Σ}}}}{p}_j(k+i|k))}^2\end{array}$

$\begin{array}{cc}s.t.& {\hat{x}}_s(k+i|k)={A_s}^i{x}_s\left(k\right)+\underset{j=1}{\overset{i}{\mathrm{\Σ}}}\left({A_s}^{i-j}{B}_u{u}_s(k+j-1)\right)\end{array}$

$\stackrel{^^}{p_s}(k+i|k)=C_s{x}_s(k+i|k)---\left(12\right)$

u _s,min≤u _s(k+i)≤u _s,max

$p_{s,\min }\≤{\hat{p}}_s(k+i|k)\≤p_{s,\max }$

i＝1,2,L,P

Wherein, { u _{s, min}, u _{s, max}, { p _{s, min}, p _{s, max}it is the boundary constraint of control variable.

5. the hierarchical optimal algorithm of the United system as described in any one of claim 1-4, is characterized in that, in described step S4, take the Distributed Predictive Control problem in alternative manner solution procedure S3, concrete steps are as follows:

Step S41: in the k moment, according to the cooling power r under the every platform air conditioning condition electric refrigerating machine optimal economic benefit condition obtained in step S3 _s, the electric refrigerating machine output power setting value sequence U of each subsystem of initialization in prediction time domain _s, make iterations l=0:

$\begin{array}{c}{U}_s^l={\left[u_s^l(k+1)u_s^l(k+2)K{u}_s^l(k+P)\right]}^T\\ {=\left[r_s(k+1)r_s(k+2)K{r}_s(k+P)\right]}^T\end{array}$

s＝1,2,L,N

Above-mentioned be the l time iteration electric refrigerating machine output power setting value in (k+1) moment, P is prediction time domain length;

Step S42: according to MPC forecast model, calculates each subsystem MPC and controls output quantity predicted value in time domain ${\hat{p}}_s={[\stackrel{^^^}{p_s}(k+1|k)p_s(k+2|k)...p_s(k+P|k)]}^T,$ And inform other N-1 subsystem MPC;

Step S43: for each subsystem, the optimization problem in solution procedure S3, obtains optimum solution, i.e. optimum refrigeration machine output power setting value sequence

Step S44: check whether the condition of convergence of all subsystem MPC meets, namely to given precision ε _swhether ∈ R (s=1,2, L, N), all have if all subsystem conditions of convergence meet, make each subsystem optimum control amount with optimum refrigeration machine output power setting value sequence equal, namely $U_s^{*}=U_a^{(l*)},$ Go to step S45; Otherwise, order $U_s^{l+1}={\mathrm{\αU}}_s^{(l*)}+(1-\mathrm{\α})U_s^l(s=\mathrm{1,2},L,N),$ α is the constant between 0 and 1, and l=l+1, goes to step S42;

Step S45: choose controlled quentity controlled variable in the k moment $u_s^{*}=\left[\begin{array}{cccc}1& 0& L& 0\end{array}\right]$ $U_s^{*}(s=\mathrm{1,2},..,N)$ Act on corresponding subsystem;

Step S46: barrel shift, to next sampling instant, i.e. k+1 → k, returns step S41, repeats said process.