CN110553376A - CQPSO algorithm-based VAV air-conditioning system temperature control method - Google Patents

Abstract

The invention relates to a temperature control method of a VAV air conditioning system based on a CQPSO algorithm, which comprises the following steps of S1 initializing a particle swarm, S2 calculating the average optimal position of the particle swarm, S3 executing processing operation on each particle in the particle swarm, S4 judging whether an algorithm ending condition is met, executing the step S5 if the algorithm ending condition is met, otherwise, returning to the step S2 after the step S is carried out, S5 obtaining and outputting an optimal solution, and S6 substituting the obtained optimal solution into a PID equation to obtain a PID parameter optimization control model based on the CQPSO and establishing a pressure-independent variable air volume end device cascade controller.

Description

CQPSO algorithm-based VAV air-conditioning system temperature control method

Technical Field

The invention relates to a temperature control method of a VAV air conditioning system based on a CQPSO algorithm.

Background

The proportion of building energy consumption in China is increasing day by day, and in the building energy consumption, the air conditioner energy consumption occupies a part of proportion, and the air conditioner energy saving of the building is urgent. The accurate regulation and control of the air conditioning load of the building is one of the important bases for the energy-saving optimization design of the air conditioner of the building.

A Variable Air Volume Air conditioning System (VAV Air conditioning System) is an all-Air conditioning System that adjusts the indoor temperature and humidity by fixing the Air supply temperature and changing the Air supply Volume. Due to the characteristics of flexibility, comfort, energy conservation and the like of the variable air volume air conditioning system, the variable air volume air conditioning system is widely applied at home and abroad in recent years. At present, the conventional PID control is generally adopted for the temperature control of the VAV air conditioning system. Because the VAV air conditioner control system has the characteristics of nonlinearity, large hysteresis, time-varying property and the like, the traditional PID control method has the problems of low control precision and poor anti-jamming capability.

Disclosure of Invention

In view of the above, the present invention is directed to a method for controlling a temperature of a VAV air conditioning system based on a cloud-adaptive quantum-behaved particle swarm optimization (CQPSO) algorithm,

In order to achieve the purpose, the invention adopts the following technical scheme:

A temperature control method of a VAV air conditioning system based on a CQPSO algorithm comprises the following steps:

Step S1: initializing a particle swarm, setting the population number as M, the particle dimension N, the maximum iteration number T and the proportionality coefficient K_Pintegral coefficient K_IAnd a differential coefficient K_DInitial value of (K)_i(0)＝[K_P(0)，K_I(0)，K_D(0)]And an individual optimum position P is set_i(0)＝f(X_i(0))；

Step S2: calculating the average optimal position of the population;

Step S3: for each particle i (1. ltoreq. i. ltoreq.M) in the population, a processing operation is performed:

step S4: judging whether the preset iteration number is met or the error precision is met, if so, executing the step S5, otherwise, setting t to be t +1, and returning to the step S2;

step (ii) ofS5: obtaining and outputting an optimal solution K_i(t)

Step S6: will find K_iAnd (t) substituting into a PID equation to obtain a PID parameter optimization control model based on the CQPSO, and establishing a pressure-independent variable air volume end device cascade controller.

further, the CQPSO algorithm specifically includes:

Suppose that in an N-dimensional target search space, there is a population of M particles, where the attractor of the ith particle is p_i＝(p_i，1，p_i，2，…，p_i，N) The coordinates are shown as formula (1); in each dimension with P_iCoordinate P_i，jAnd establishing a one-dimensional DETLA potential well for the center, and then, the basic evolution equation of the coordinates of the j-th dimension is shown as the formula (2).

p_i，j(t)＝φ_i，j(t)·P_i，j(t)+[1-φ_i，j(t)]·G_j(t) (1)

For L_i，j(t) evaluation, which introduces the average best position C (t), i.e. the average of the best positions of all individual particles, is defined as formula (3):

L_i，j(t) is evaluated by the formula (4):

L_i，j(t)＝2α·|C_j(t)-X_i，j(t)| (4)

The evolution process of the CQPSO algorithm can be obtained according to the formulas (2) and (4) as shown in formula (5):

X_i，j(t+1)＝p_i，j(t)±α·|C_j(t)-X_i，j(t)|·ln[1/u_i，j(t)] (5)

In the formula u_i，j(t) to U (0, 1); t is the number of iterations; α is the contraction-expansion coefficient; when u is_i，jwhen (t) < 0.5, the preceding symbol is "-", when u is_i，jWhen (t) is more than or equal to 0.5, taking a plus sign before alpha; and (4) calculating to obtain the average optimal position of the population according to the formula (3).

Further, the contraction-expansion coefficient α is solved as follows

Dividing a quantum particle swarm into three subgroups, and generating strategies by adopting different inertia weights respectively; setting the size of the particle swarm to be M, and setting the particle X in the t iteration_iHas a fitness value of f_i(ii) a The average fitness value of the particle swarm isThe fitness value is superior to f_avgIs calculated by averaging the fitness values of f_avg', the fitness value is inferior to f_avgIs averaged to obtain f_avgand respectively adopting different contraction-expansion coefficients alpha, wherein the generation rule is as follows:

f_i＞f_avg″

Fitness value greater than f_avg"is the poorer particle in the population, alpha is 1.0

f_i＜f_avg′

Fitness value less than f_avgThe particles of' are the more excellent particles in the population, and α is 0.5

f_avg′≤f_i≤f_avg″

Non-linear dynamic adjustment of particle X according to particle fitness value by X condition cloud generator_ithe contraction-expansion coefficient α of (a).

Further, the step S3 is specifically:

Step S31: calculating the current position X of the particle i_i(t) and converting X_i(t) fitness value and previous iteration P_i(t-1) comparison of fitness values, if X_i(t) has a fitness value superior to P_i(t-1) fitness value, P_i(t)＝X_i(t), otherwise P_i(t)＝P_i(t-1)；

Step S32: calculating the current global optimum position of the population, i.e. P_i(t) comparing the fitness value with the global optimum G (t-1) of the previous iteration iff[X_i(t)]＜f[G(t-1)]then G (t) ═ P_i(t-1)；

Step S33: calculating the position of a random point according to the formula (1) for each dimension of the particle i;

Step S34: and adopting different contraction-expansion coefficients alpha according to different particle fitness values, wherein the ordinary particle swarm adaptively adjusts the contraction-expansion coefficients alpha by using the cloud, and the new positions of the particles are updated according to the formula (5).

Compared with the prior art, the invention has the following beneficial effects:

The invention optimizes the PID controller based on the CQPSO, has good effects of improving the performance of the system, increasing the control precision and adaptability, can better inhibit the interference of external factors on the variable air volume air conditioning system, and quickens the speed of recovering the steady state again, thereby obtaining more ideal control effect when being applied to the nonlinear and uncertain VAV air conditioning system.

Drawings

FIG. 1 is a flow chart of the algorithm of the present invention;

FIG. 2 is a normal cloud model according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating a CQPSO-based PID parameter optimization according to an embodiment of the invention;

Fig. 4 is a cascade control diagram of a pressure independent variable air volume end device in accordance with an embodiment of the present invention.

Detailed Description

The invention is further explained below with reference to the drawings and the embodiments.

referring to fig. 1, the present invention provides a method for controlling a temperature of a VAV air conditioning system based on a CQPSO algorithm, comprising the steps of:

Step S1: initializing a particle swarm, setting the population number as M, the particle dimension N, the maximum iteration number T and the proportionality coefficient K_PIntegral coefficient K_IAnd a differential coefficient K_Dinitial value of (K)_i(0)＝[K_P(0)，K_I(0)，K_D(0)]And an individual optimum position P is set_i(0)＝f(X_i(0))；

Step S2: calculating the average optimal position of the population;

step S3: for each particle i (1. ltoreq. i. ltoreq.M) in the population, a processing operation is performed:

Step S4: judging whether the preset iteration times or the error precision is met; if yes, executing step S5, otherwise, setting t to t +1, and returning to step S2;

Step S5: obtaining and outputting an optimal solution K_i(t)

Step S6: will find K_iAnd (t) substituting into a PID equation to obtain a PID parameter optimization control model based on the CQPSO, and establishing a pressure-independent variable air volume end device cascade controller.

In this embodiment, the CQPSO algorithm specifically includes:

Suppose that in an N-dimensional target search space, there is a population of M particles, where the attractor of the ith particle is p_i＝(p_i，1，p_i，2，…，p_i，N) The coordinates are shown as formula (1); in each dimension with P_iCoordinate P_i，jand establishing a one-dimensional DETLA potential well for the center, and then, the basic evolution equation of the coordinates of the j-th dimension is shown as the formula (2).

p_i，j(t)＝φ_i，j(t)·P_i，j(t)+[1-φ_i，j(t)]·G_j(t) (1)

For L_i，j(t) evaluation, which introduces the average best position C (t), i.e. the average of the best positions of all individual particles, is defined as formula (3):

L_i，j(t) is evaluated by the formula (4):

L_i，j(t)＝2α·|C_j(t)-X_i，j(t)| (4)

The evolution process of the CQPSO algorithm can be obtained according to the formulas (2) and (4) as shown in formula (5):

X_i，j(t+1)＝p_i，j(t)±α·|C_j(t)-X_i，j(t)|·ln[1/u_i，j(t)] (5)

In the formula u_i，j(t) to U (0, 1); t is the number of iterations; α is the contraction-expansion coefficient; when u is_i，jWhen (t) < 0.5, the preceding symbol is "-", when u is_i，jWhen (t) is more than or equal to 0.5, taking a plus sign before alpha; and (4) calculating to obtain the average optimal position of the population according to the formula (3).

In the present embodiment, the contraction-expansion coefficient α is solved as follows

Dividing a quantum particle swarm into three subgroups, and generating strategies by adopting different inertia weights respectively; setting the size of the particle swarm to be M, and setting the particle X in the t iteration_iHas a fitness value of f_i(ii) a The average fitness value of the particle swarm isThe fitness value is superior to f_avgis calculated by averaging the fitness values of f_avg', the fitness value is inferior to f_avgis averaged to obtain f_avgAnd respectively adopting different contraction-expansion coefficients alpha, wherein the generation rule is as follows:

f_i＞f_avg″

fitness value greater than f_avg"is the poorer particle in the population, alpha is 1.0

f_i＜f_avg′

Fitness value less than f_avgthe particles of' are the more excellent particles in the population, and α is 0.5

f_avg′≤f_i≤f_avg″

Non-linear dynamic adjustment of particle X according to particle fitness value by X condition cloud generator_iThe contraction-expansion coefficient α of (a). The improved contraction-expansion coefficient alpha generation algorithm of the cloud self-adaptive quantum particle swarm is as follows:

as is well understood from the high-order limit,Thereby ensuring alpha is equal to 0.5, 1.0]as can be seen from the above equation, the contraction-expansion coefficient α decreases with a decrease in the particle fitness value, thereby achieving a smaller contraction-expansion coefficient α than that of a superior particle.

let T be the linguistic value on discourse u, u to [0, 1]Mapping C of_T(u)：U→[0，1]； u→C_T(u) then C_TThe distribution of (u) over u is called the membership cloud of T, cloud for short. When C is present_T(u) when obeying a normal distribution, it is called a normal cloud model. The random number set follows a normal distribution rule and has a stable tendency, and is characterized by an expected value Ex, an entropy En and a super-entropy He; as shown in fig. 2;

Parameter selection:

En affects the distribution of a normal cloud. According to the "3 En" rule, 99.74% of the contributing quantitative values for linguistic values on domain u fall on c₁The above. The larger En, the larger the horizontal width of cloud coverage. Combining the speed and the precision of the algorithm, and taking c from the algorithm₁＝2.8。

He determines the degree of dispersion of the cloud droplets. When He is too small, randomness is lost to some extent; too much He will lose the "stability tendency", c in the algorithm₂＝8。

In this embodiment, the step S3 specifically includes:

Step S31: calculating the current position X of the particle i_i(t) and converting X_i(t) fitness value and previous iteration P_i(t-1) comparison of fitness values, if X_i(t) has a fitness value superior to P_i(t-1) fitness value, P_i(t)＝X_i(t), otherwise P_i(t)＝P_i(t-1)；

Step S32: calculating the current global optimum position of the population, i.e. P_i(t) the fitness value is compared with the global optimum G (t-1) of the previous iteration if f [ X ]_i(t)]＜f[G(t-1)]Then G (t) ═ P_i(t-1)；

Step S33: calculating the position of a random point according to the formula (1) for each dimension of the particle i;

Step S34: and adopting different contraction-expansion coefficients alpha according to different particle fitness values, wherein the ordinary particle swarm adaptively adjusts the contraction-expansion coefficients alpha by using the cloud, and the new positions of the particles are updated according to the formula (5).

In this embodiment, a CQPSO-based PID parameter optimization structure is shown in fig. 3, and according to a pressure-independent variable air volume end control principle, a PID cascade control is adopted to control the temperature of the variable air volume air conditioning system, and the principle is shown in fig. 4: the main loop is a temperature control loop, the temperature of the area collected by the temperature sensor is compared with the set temperature and then input into the temperature controller, and the set air volume value is calculated and used as the set value of the auxiliary loop; the auxiliary loop is an air volume control loop and is used for controlling the air valve actuator, the error between the tail end actual air volume acquired by the air volume sensor and the set air volume is input into the air volume controller, and then the opening degree of the valve is sent to the air valve actuator, so that the tail end actual air volume is adjusted to be kept at the set value. In the pressure-independent type end cascade control system, the main control object is the indoor temperature and has a large capacity lag, so the PID control is adopted for the main loop controller for controlling the temperature, and the PI control is adopted for the sub-loop controller for controlling the air volume.

the above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.

Claims (4)

1. A temperature control method of a VAV air conditioning system based on a CQPSO algorithm is characterized by comprising the following steps:

step S1: initializing a particle swarm, setting the population number as M, the particle dimension N, the maximum iteration number T and the proportionality coefficient K_PIntegral coefficient K_Iand differentialCoefficient K_Dinitial value of (K)_i(0)＝[K_P(0)，K_I(0)，K_D(0)]And an individual optimum position P is set_i(0)＝f(X_i(0))；

step S2: calculating the average optimal position of the population;

Step S3: for each particle i (1. ltoreq. i. ltoreq.M) in the population, a processing operation is performed:

Step S4: judging whether the error precision is met or a preset iteration number is reached; if yes, executing step S5, otherwise, setting t to t +1, and returning to step S2;

Step S5: obtaining and outputting an optimal solution K_i(t)

Step S6: will find K_iAnd (t) substituting into a PID equation to obtain a PID parameter optimization control model based on the CQPSO, and establishing a pressure-independent variable air volume end device cascade controller.

2. The CQPSO algorithm-based VAV air conditioning system temperature control method of claim 1, wherein: the CQPSO algorithm specifically comprises the following steps:

suppose that in an N-dimensional target search space, there is a population of M particles, where the attractor of the ith particle is p_i＝(p_i，1，p_i，2，…，p_i，N) The coordinates are shown as formula (1); in each dimension with P_iCoordinate P_i，jAnd establishing a one-dimensional DETLA potential well for the center, and then, the basic evolution equation of the coordinates of the j-th dimension is shown as the formula (2).

p_i，j(t)＝φ_i，j(t)·P_i，j(t)+[1-φ_i，j(t)]·G_j(t) (1)

for L_i，j(t) evaluation, which introduces the average best position C (t), i.e. the average of the best positions of all individual particles, is defined as formula (3):

L_i，i(t) is evaluated by the formula (4):

L_i，j(t)＝2α·|C_j(t)-X_i，j(t)| (4)

The evolution process of the CQPSO algorithm can be obtained according to the formulas (2) and (4) as shown in formula (5):

X_i，j(t+1)＝p_i，j(t)±α·|C_j(t)-X_i，j(t)|·ln[1/u_i，j(t)] (5)

In the formula u_i，j(t) to U (0, 1); t is the number of iterations; α is the contraction-expansion coefficient; when u is_i，jWhen (t) < 0.5, the preceding symbol is "-", when u is_i，jwhen (t) is more than or equal to 0.5, taking a plus sign before alpha; and (4) calculating to obtain the average optimal position of the population according to the formula (3).

3. the CQPSO algorithm-based VAV air conditioning system temperature control method of claim 2, wherein: the contraction-expansion coefficient α is solved as follows

dividing a quantum particle swarm into three subgroups, and generating strategies by adopting different inertia weights respectively; setting the size of the particle swarm to be M, and setting the particle X in the t iteration_iHas a fitness value of f_i(ii) a The average fitness value of the particle swarm isThe fitness value is superior to f_avgis calculated by averaging the fitness values of f_avg', the fitness value is inferior to f_avgIs averaged to obtain f_avgand respectively adopting different contraction-expansion coefficients alpha, wherein the generation rule is as follows:

f_i＞f_avg″

Fitness value greater than f_avg"is the poorer particle in the population, alpha is 1.0

f_i＜f_avg′

Fitness value less than f_avgThe particles of' are the more excellent particles in the population,α＝0.5

f_avg′≤f_i≤f_avg″

Non-linear dynamic adjustment of particle X according to particle fitness value by X condition cloud generator_iThe contraction-expansion coefficient α of (a).

4. The CQPSO algorithm-based VAV air conditioning system temperature control method of claim 2, wherein the step S3 specifically is:

step S31: calculating the current position X of the particle i_i(t) and converting X_i(t) fitness value and previous iteration P_i(t-1) comparison of fitness values, if X_i(t) has a fitness value superior to P_i(t-1) fitness value, P_i(t)＝X_i(t), otherwise P_i(t)＝P_i(t-1)；

Step S32: calculating the current global optimum position of the population, i.e. P_i(t) the fitness value is compared with the global optimum G (t-1) of the previous iteration if f [ X ]_i(t)]＜f[G(t-1)]then G (t) ═ P_i(t-1)；

step S33: calculating the position of a random point according to the formula (1) for each dimension of the particle i;

step S34: and adopting different contraction-expansion coefficients alpha according to different particle fitness values, wherein the ordinary particle swarm adaptively adjusts the contraction-expansion coefficients alpha by using the cloud, and the new positions of the particles are updated according to the formula (5).