CN105159085B - The real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning - Google Patents

Abstract

The invention discloses a kind of real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning, the second order equivalent heat parameter model of the affiliated building of air-conditioning is set up in real time, foundation and control in real time to carry out air conditioner load equivalent aggregation model；Set up the second order differential equations of the affiliated building of air conditioner load, determine input and output amount and need the parameter of identification, real-time data acquisition is carried out to input and output amount by the senior measuring terminal of intelligent grid, and data transfer is carried out to Load aggregation business, Load aggregation business carries out real-time parameter identification using cooperative particle swarm algorithm.Research in terms of the present invention compensate for Domestic Air-condition load Real-time modeling set is not enough, constructs more accurate air conditioner load model, and participating in demand response for it provides technical support.

Description

The real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning

Technical field

The invention belongs to the parameter identification technique of electric power system dispatching part, more particularly to Load aggregation business is to belonging to air-conditioning The real-time parameter identification of the second order equivalent heat parameter model of building.

Background technology

The senior measuring system of intelligent grid provides technical support, Load aggregation for the real-time data acquisition of relevant departments Business can not only carry out real-time parameter identification by intelligent grid, and can carry out online polymerization model to the air-conditioning controlled Set up and control in real time.Rely on the hardware supported and related on-line identification technology of intelligent grid, the air-conditioning of Load aggregation business Load on-line identification participates in demand response to air conditioner load and plays vital effect.

Air conditioner load as a kind of important demand response resource, the equivalent heat parameter model of its affiliated building by It is widely used in the every field of air conditioner load control, but some parameters and thickness, the window of the wall of building in its model The factors such as family area, volume size are closely related, it is impossible to obtained by measuring, it is therefore desirable to pass through certain parameter identification means Parameter to air conditioner load model is identified, and China starts late in terms of air-conditioning participates in demand response, does not have also at present The related parameter identification technique on air conditioner load model.At the same time, for the purpose for simplifying calculating, many documents and reality Border engineering project is using the single order equivalent heat parameter model simplified, and its precision is poor, therefore needs equivalent to more accurate second order Physochlaina infudibularis exponential model carries out real-time parameter identification, to realize the accurate control of air conditioner load.

Cooperative particle swarm algorithm can by one it is more complicated the problem of decomposed, pass through several simple subproblems Consult to substitute asking for global optimum with the comprehensive optimal solution for finding problem, the local optimum for not only solving standard particle group's algorithm Topic, and fast convergence rate, are a kind of practical parameter identification methods.

The content of the invention

Goal of the invention：In order to make up the affiliated building second order equivalent heat parameter model ginseng of air-conditioning during existing demand response The blank of number on-line real-time test, the present invention provides a kind of air-conditioning model real-time parameter identification side based on cooperative particle swarm algorithm Method, carries out real-time data acquisition to input and output amount, and carry out to Load aggregation business by the senior measuring terminal of intelligent grid Data transfer, Load aggregation business is recognized using cooperative particle swarm algorithm to the real-time parameter of air conditioner load model, sets up negative in real time Air conditioner load model in lotus polymerization business's compass of competency, real-time control and scheduling to carry out air conditioner load.

Technical scheme：To achieve the above object, the technical solution adopted by the present invention is：

A kind of real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning, comprises the following steps：

(1) the second order equivalent heat parameter model of the affiliated building of air-conditioning is built, it is T to determine input quantity_o, Q and t, output quantity For T_iAnd T_m, parameter to be identified is R₁And R₂：

In formula：T_oRepresent outdoor temperature (unit：DEG C), T_iRepresent indoor air temperature (unit：DEG C), T_mRepresent indoor solid Temperature (unit：DEG C), Q represents refrigerating/heating amount, C_aFor air specific heat capacity (unit：J/ DEG C), C_mIt is (single for specific heat of solid and heat Position：J/ DEG C), R₁For the inverse (unit of air heat loss factor：DEG C/W), R₂For the inverse (unit of solid heat loss factor：℃/ W)；T represents time (unit：s).

(2) Load aggregation business gathers input data by the senior measuring terminal of intelligent grid, and to the input number of collection N groups input array is formed according to pretreatment is carried out；

(3) set up the design parameter of the affiliated building of air-conditioning and the build-in attribute of empty gas and solid determines R₁And R₂Value model Enclose, by R₁And R₂Span as cooperative particle swarm solution room；Determine the population N of cooperative particle swarm, sub- population number M, communication frequency function F (t) and iterations H；In the initialization procedure to particle, the position initial value of one of particle For the parameter identification value of previous time period, the position initial value of remaining N-1 particle divides in solution room according to mean random Initialized with principle, while the translational speed to each particle is initialized；

(4) particle j current location is designated as x_j=(R_{1_j},R_{2_j}), by (R_{1_j},R_{2_j}) bring formula (1) and formula (2) into, according to The method for solving of ODE is solved to formula (1) and formula (2), n groups input array (T_o, Q, t) calculate obtain n solve (T_{i_j},T_{m_j})；Wherein, R_{1_j}Represent particle j R₁Value, R_{2_j}Represent particle j R₂Value, T_{i_j}Represent particle j T_iValue, T_{m_j}Table Show particle j T_mValue；

(5) the fitness function f of cooperative particle swarm is set_fitness, the fitness value of each all solutions of particle is calculated, with every Location updating its optimal location pbest corresponding to the minimum fitness value of individual particle, and select suitable from every sub- population The optimal location for answering the minimum particle of angle value is the global optimum position gbest of the sub- population；

(6) g (t)=rand (0,1) is made,And judge g (t) and F (t) magnitude relationship：If g (t) ＜ F (t) the minimum global optimum position gbest of fitness value in all sub- populations, is selected_best, and update the overall situation of other sub- populations Optimal location is gbest=gbest_best；If g (t) >=F (t), the global optimum position gbest of every sub- population is kept not Become；After the globally optimal solution of all sub- populations has been updated, current translational speed and the position of each particle are regularly refreshed；

(7) return to step (4), until iterations h reaches H；

(8) the global optimum position gbest of all sub- populations is contrasted, the minimum global optimum of selection fitness value Position gbest_bestIt is used as final optimum results.

Specifically, in the step (2), being pre-processed to the inputoutput data of collection, specifically including herein below：

(a1) because outdoor temperature changes slower, according to statistical law, five points value changes of continuous acquisition should This less calculates the average value of outdoor temperature as followsWith standard deviation S：

In formula：t_iRepresent i-th of outdoor temperature of continuous acquisition；

(a2) judgeWhether set up：If not, then illustrate t_iIt is normal data；Otherwise t is illustrated_iIt is bad number According to, it is necessary to be modified；

(a3) bad data correction formula is：

In formula：t_iRepresent revised i-th of outdoor temperature, α, β is customized correction factor, and alpha+beta=2.

Specifically, in the step (3), sub- population number M value principle is as follows：Assuming that the number of particles of population is certain；M Value is bigger, and the sub- population invariable number for representing whole population is more, and its diversity is better, is more conducive to find globally optimal solution；But M values increase While big, the computer memory space required for during parameter identification is bigger, calculates the time also longer；Examined both comprehensive Consider, M=3 can be taken.

Specifically, in the step (4), directly calculating T using following formula_{i_j}And T_{m_j}：

Wherein：

A=C_mC_aR_{2_j} (4-3)

In formula：T_i(0) indoor air temperature initial value is represented；T_m(0) indoor solid initial temperature is represented.

Specifically, in the step (5), the fitness function f of cooperative particle swarm_fitnessFor：

In formula：T_{i_j}Represent the indoor temperature by calculating obtained particle j, T_{m_j}Represent by calculating obtained particle j Indoor solid temperature,Expression actually measures obtained particle j indoor temperature,Represent actually to measure obtained grain Sub- j indoor solid temperature.

Specifically, in the step (6), updating the present speed of each particle and the rule of position being：

In formula：c₁And c₂For nonnegative constant, accelerated factor is represented；rand₁And rand₂For the random number between 0~1；W tables Show Inertia Weight, w=w_max-(w_max-w_min)×h/H；Represent the translational speed of particle j during the h times iteration；Represent the h times Particle j position during iteration；pbest_jRepresent particle j optimal location.

Beneficial effect：The real-time parameter identification side for the affiliated building second order equivalent heat parameter model of air-conditioning that the present invention is provided Method, the second order equivalent heat parameter model of the affiliated building of relatively accurately air-conditioning can be built in real time, can be provided for relevant departments The parameter of real-time air conditioner load, is peak clipping, frequency modulation, standby, suppression regenerative resource fluctuation that air conditioner load participates in power system Deng there is provided foundation；Research in terms of the present invention compensate for Domestic Air-condition load Real-time modeling set simultaneously is not enough, constructs more smart True air conditioner load model, participates in demand response for it and provides technical support.

Brief description of the drawings

Fig. 1 is the general flow chart of the inventive method；

Fig. 2 is the hardware composition figure that Load aggregation quotient system is united.

Embodiment

The present invention is further described below in conjunction with the accompanying drawings.

A kind of real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning, comprises the following steps：

(1) the second order equivalent heat parameter model of the affiliated building of air-conditioning is built, it is T to determine input quantity_o, Q and t, output quantity For T_iAnd T_m, parameter to be identified is R₁And R₂：

In formula：T_oRepresent outdoor temperature (unit：DEG C), T_iRepresent indoor air temperature (unit：DEG C), T_mRepresent indoor solid Temperature (unit：DEG C), Q represents refrigerating/heating amount, C_aFor air specific heat capacity (unit：J/ DEG C), C_mIt is (single for specific heat of solid and heat Position：J/ DEG C), R₁For the inverse (unit of air heat loss factor：DEG C/W), R₂For the inverse (unit of solid heat loss factor：℃/ W)；T represents time (unit：s).

(2) Load aggregation business gathers input data by the senior measuring terminal of intelligent grid, and to the input number of collection N groups input array is formed according to pretreatment is carried out；The specific method of pretreatment is：

(a1) because outdoor temperature changes slower, according to statistical law, five points value changes of continuous acquisition should This less calculates the average value of outdoor temperature as followsWith standard deviation S：

In formula：t_iRepresent i-th of outdoor temperature of continuous acquisition；

(a2) judgeWhether set up：If not, then illustrate t_iIt is normal data；Otherwise t is illustrated_iIt is bad number According to, it is necessary to be modified；

(a3) bad data correction formula is：

In formula：t_iRepresent revised i-th of outdoor temperature, α, β is customized correction factor, and alpha+beta=2.

(3) set up the design parameter of the affiliated building of air-conditioning and the build-in attribute of empty gas and solid determines R₁And R₂Value model Enclose, by R₁And R₂Span as cooperative particle swarm solution room；Determine the population N of cooperative particle swarm, sub- population number M, communication frequency function F (t) and iterations H；In the initialization procedure to particle, the position initial value of one of particle For the parameter identification value of previous time period, the position initial value of remaining N-1 particle divides in solution room according to mean random Initialized with principle, while the translational speed to each particle is initialized.

Sub- population number M value principle is as follows：Assuming that the number of particles of population is certain；The bigger whole population of expression of M values Sub- population invariable number is more, and its diversity is better, is more conducive to find globally optimal solution；But while M values increase, parameter identification process In required for computer memory space it is bigger, calculate the time it is also longer；The two comprehensive consideration, can take M=3.

(4) particle j current location is designated as x_j=(R_{1_j},R_{2_j}), by (R_{1_j},R_{2_j}) bring formula (1) and formula (2) into, according to The method for solving of ODE is solved to formula (1) and formula (2), n groups input array (T_o, Q, t) calculate obtain n solve (T_{i_j},T_{m_j})；Wherein, R_{1_j}Represent particle j R₁Value, R_{2_j}Represent particle j R₂Value, T_{i_j}Represent particle j T_iValue, T_{m_j}Table Show particle j T_mValue；By fortran, T directly can be calculated using following formula_{i_j}And T_{m_j}：

Wherein：

A=C_mC_aR_{2_j} (4-3)

In formula：T_i(0) indoor air temperature initial value is represented；T_m(0) indoor solid initial temperature is represented.

(5) the fitness function f of cooperative particle swarm is set_fitness, the fitness value of each all solutions of particle is calculated, with every Location updating its optimal location pbest corresponding to the minimum fitness value of individual particle, and select suitable from every sub- population The optimal location for answering the minimum particle of angle value is the global optimum position gbest of the sub- population.

The fitness function f of cooperative particle swarm_fitnessFor：

In formula：T_{i_j}Represent the indoor temperature by calculating obtained particle j, T_{m_j}Represent by calculating obtained particle j Indoor solid temperature,Expression actually measures obtained particle j indoor temperature,Represent actually to measure obtained grain Sub- j indoor solid temperature.

(6) g (t)=rand (0,1) is made,And judge g (t) and F (t) magnitude relationship：If g (t) ＜ F (t) the minimum global optimum position gbest of fitness value in all sub- populations, is selected_best, and update the overall situation of other sub- populations Optimal location is gbest=gbest_best；If g (t) >=F (t), the global optimum position gbest of every sub- population is kept not Become；After the globally optimal solution of all sub- populations has been updated, current translational speed and the position of each particle are regularly refreshed.

Update the present speed of each particle and the rule of position is：

In formula：c₁And c₂For nonnegative constant, accelerated factor is represented；rand₁And rand₂For the random number between 0~1；W tables Show Inertia Weight, w=w_max-(w_max-w_min)×h/H；Represent the translational speed of particle j during the h times iteration；Represent the h times Particle j position during iteration；pbest_jRepresent particle j optimal location.

(7) return to step (4), until iterations h reaches H.

(8) the global optimum position gbest of all sub- populations is contrasted, the minimum global optimum of selection fitness value Position gbest_bestIt is used as final optimum results.

Described above is only the preferred embodiment of the present invention, it should be pointed out that：For the ordinary skill people of the art For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should It is considered as protection scope of the present invention.

Claims (6)

1. a kind of real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning, it is characterised in that：Including Following steps：

(1) the second order equivalent heat parameter model of the affiliated building of air-conditioning is built, it is T to determine input quantity_o, Q and t, output quantity is T_iWith T_m, parameter to be identified is R₁And R₂：

$\frac{{\mathrm{dT}}_i}{dt}=-\frac{1}{C_a}(\frac{1}{R_1}+\frac{1}{R_2})T_i+\frac{1}{R_2}\frac{1}{C_a}{T}_m+\frac{1}{R_1}\frac{1}{C_a}{T}_o+\frac{1}{C_a}Q---\left(1\right)$

$\frac{{\mathrm{dT}}_m}{dt}=\frac{1}{R_2}\frac{1}{C_m}{T}_i-\frac{1}{R_2}\frac{1}{C_m}{T}_m---\left(2\right)$

In formula：T_oRepresent outdoor temperature, T_iRepresent indoor air temperature, T_mIndoor solid temperature is represented, Q represents refrigerating/heating Amount, C_aFor air specific heat capacity, C_mFor specific heat of solid and heat, R₁For the inverse of air heat loss factor, R₂For solid heat loss factor's It is reciprocal；T represents the time；

(2) Load aggregation business gathers input data by the senior measuring terminal of intelligent grid, and the input data of collection is entered Row pretreatment forms n groups input array；

(3) set up the design parameter of the affiliated building of air-conditioning and the build-in attribute of empty gas and solid determines R₁And R₂Span, By R₁And R₂Span as cooperative particle swarm solution room；Determine the population N of cooperative particle swarm, sub- population number M, Communication frequency function F (t) and iterations H；In the initialization procedure to particle, the position initial value of one of particle is The parameter identification value of previous time period, the position initial value of remaining N-1 particle is distributed in solution room according to mean random Principle is initialized, while the translational speed to each particle is initialized；

(4) particle j current location is designated as x_j=(R_{1_j},R_{2_j}), by (R_{1_j},R_{2_j}) bring formula (1) and formula (2) into, according to ordinary differential The method for solving of equation is solved to formula (1) and formula (2), n groups input array (T_o, Q, t) calculate obtain n solve (T_{i_j}, T_{m_j})；Wherein, R_{1_j}Represent particle j R₁Value, R_{2_j}Represent particle j R₂Value, T_{i_j}Represent particle j T_iValue, T_{m_j}Represent particle J T_mValue；

(5) the fitness function f of cooperative particle swarm is set_fitness, the fitness value of each all solutions of particle is calculated, each grain is used Location updating its optimal location pbest corresponding to the minimum fitness value of son, and select fitness from every sub- population The optimal location for being worth minimum particle is the global optimum position gbest of the sub- population；

(6) g (t)=rand (0,1) is made,And judge g (t) and F (t) magnitude relationship：If g (t) ＜ F (t), choosing Select the minimum global optimum position gbest of fitness value in all sub- populations_best, and update the global optimum position of other sub- populations It is set to gbest=gbest_best；If g (t) >=F (t), the global optimum position gbest of every sub- population keeps constant；More After the globally optimal solution of new complete all sub- populations, current translational speed and the position of each particle are regularly refreshed；

(7) return to step (4), until iterations h reaches H；

(8) the global optimum position gbest of all sub- populations is contrasted, the minimum global optimum position of selection fitness value gbest_bestIt is used as final optimum results.

2. the real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning according to claim 1, It is characterized in that：In the step (2), the inputoutput data of collection is pre-processed, herein below is specifically included：

(a1) average value of outdoor temperature is calculated as followsWith standard deviation S：

$\stackrel{\‾}{t}=\frac{1}{5}(t_{i-2}+t_{i-1}+t_i+t_{i+1}+t_{i+2})---(2-1)$

$S=\sqrt{\frac{1}{5}\underset{i=-2}{\overset{2}{\Σ}}{(t_i-\stackrel{\‾}{t})}^2}---(2-2)$

In formula：t_iRepresent i-th of outdoor temperature of continuous acquisition；

(a2) judgeWhether set up：If not, then illustrate t_iIt is normal data；Otherwise t is illustrated_iIt is bad data, Need to be modified；

(a3) bad data correction formula is：

$t_i=\frac{\mathrm{\α}}{4}(t_{i-2}+t_{i+2})+\frac{\mathrm{\β}}{4}(t_{i-1}+t_{i+1})---(2-3)$

In formula：t_iRepresent revised i-th of outdoor temperature, α, β is customized correction factor, and alpha+beta=2.

3. the real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning according to claim 1, It is characterized in that：In the step (3), sub- population number M=3.

4. the real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning according to claim 1, It is characterized in that：In the step (4), directly T is calculated using following formula_{i_j}And T_{m_j}：

$T_{i\_j}={\mathrm{\α}}_1{e}^{r_{1}t}+{\mathrm{\α}}_2{e}^{r_{2}t}+\frac{d}{c}---(4-1)$

$T_{m\_j}=({\mathrm{\α}}_1{r}_1{C}_a{R}_{2\_j}c+{\mathrm{\α}}_1+R_{2\_j}{\mathrm{\α}}_1)e^{r_{1}t}+({\mathrm{\α}}_2{r}_2{C}_a{R}_{2\_j}+{\mathrm{\α}}_2+R_{2\_j}c)e^{r_{2}t}+\frac{d}{c}---(4-2)$

Wherein：

A=C_mC_aR_{2_j} (4-3)

$b=\frac{C_m(R_{1\_j}+R_{2\_j})+R_{1\_j}{C}_a}{R_{1\_j}}---(4-4)$

$c=\frac{1}{R_{1\_j}}---(4-5)$

$d=Q+\frac{T_o}{R_{1\_j}}---(4-6)$

$r_1=\frac{-b+\sqrt{b^2-4ac}}{2a}---(4-7)$

$r_2=\frac{-b-\sqrt{b^2-4ac}}{2a}---(4-8)$

${\mathrm{\α}}_1=\frac{r_2{T}_i\left(0\right)-{\stackrel{\·}{T}}_i\left(0\right)-r_2\frac{d}{c}}{r_2-r_1}---(4-9)$

${\mathrm{\α}}_2=T_i\left(0\right)-\frac{d}{c}-{\mathrm{\α}}_1---(4-10)$

${\stackrel{\·}{T}}_i\left(0\right)=\frac{1}{R_{2\_j}{C}_a}{T}_m\left(0\right)-\frac{R_{1\_j}+R_{2\_j}}{R_{1\_j}{R}_{2\_j}{C}_a}T\left(0\right)+\frac{R_{1\_j}Q+T_o}{R_{1\_j}{C}_a}---(4-11)$

In formula：T_i(0) indoor air temperature initial value is represented；T_m(0) indoor solid initial temperature is represented.

5. the real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning according to claim 1, It is characterized in that：In the step (5), the fitness function f of cooperative particle swarm_fitnessFor：

$f_{fitness}=\underset{j=1}{\overset{n}{\Σ}}\[({\hat{T}}_{i\_j}-T_{i\_j})+{({\hat{T}}_{m\_j}-T_{m\_j})}^2\]---(5-1)$

In formula：T_{i_j}Represent the indoor temperature by calculating obtained particle j, T_{m_j}Represent the room by calculating obtained particle j Interior solid temperature,Expression actually measures obtained particle j indoor temperature,Represent actually to measure obtained particle j's Indoor solid temperature.

6. the real-time parameter discrimination method of the affiliated building second order equivalent heat parameter model of air-conditioning according to claim 1, It is characterized in that：In the step (6), update the present speed of each particle and the rule of position is：

$v_j^{h+1}=w\×v_j^h+c_1\×{\mathrm{rand}}_1\×({\mathrm{pbest}}_j-x_j^h)+c_2\×{\mathrm{rand}}_2\×(gbest-x_j^h)---(6-1)$

$x_j^{h+1}=x_j^h+v_j^{h+1}---(6-2)$

In formula：c₁And c₂For nonnegative constant, accelerated factor is represented；rand₁And rand₂For the random number between 0~1；W represents inertia Weights, w=w_max-(w_max-w_min)×h/H；Represent the translational speed of particle j during the h times iteration；When representing the h times iteration Particle j position；pbest_jRepresent particle j optimal location.