CN103912966B - A kind of earth source heat pump refrigeration system optimal control method - Google Patents

A kind of earth source heat pump refrigeration system optimal control method Download PDF

Info

Publication number
CN103912966B
CN103912966B CN201410125301.7A CN201410125301A CN103912966B CN 103912966 B CN103912966 B CN 103912966B CN 201410125301 A CN201410125301 A CN 201410125301A CN 103912966 B CN103912966 B CN 103912966B
Authority
CN
China
Prior art keywords
refrigeration
prediction
load
cooling water
chilled
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201410125301.7A
Other languages
Chinese (zh)
Other versions
CN103912966A (en
Inventor
周传辉
赵亚洲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University of Science and Engineering WUSE
Original Assignee
Wuhan University of Science and Engineering WUSE
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University of Science and Engineering WUSE filed Critical Wuhan University of Science and Engineering WUSE
Priority to CN201410125301.7A priority Critical patent/CN103912966B/en
Publication of CN103912966A publication Critical patent/CN103912966A/en
Application granted granted Critical
Publication of CN103912966B publication Critical patent/CN103912966B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Abstract

The present invention relates to a kind of earth source heat pump refrigeration system energy conservation optimizing method, belong to the optimal control for energy saving field of air conditioning system.Several models are optimized combination by the present invention, even if the forecast model of a poor effect, as long as it is containing systematic opposition information, after it carries out associated prediction with one and several good forecast model, remain able to the prediction characteristic of improvement system, for improving final precision of prediction, control system uses combination forecasting method to comprehensively utilize the information that various methods provide, avoid Individual forecast model to lose useful information, reduce randomness, improve precision of prediction.The optimal value of the present invention is under guaranteeing the premise meeting end workload demand, and the energy consumption making system is minimum.After optimal setting is determined, control CS central refrigeration system and operate in optimum setting operating mode.

Description

A kind of earth source heat pump refrigeration system optimal control method
Technical field
The present invention relates to a kind of earth source heat pump refrigeration system energy conservation optimizing method, belong to the optimal control for energy saving field of air conditioning system.
Background technology
National economy energy resource consumption is constituted mainly industrial energy consumption, traffic energy consumption and building energy consumption, along with moving forward steadily of well-off society of China, the paces of Process of Urbanization Construction also can not be ignored in this building energy consumption problem brought continuing to increase, and 2/3rds of building energy consumption for air conditioning system consume energy, under the ever-increasing present situation of ratio that building energy consumption accounts for whole energy resource consumption, the air conditioner system energy saving in building has become as an emphasis in energy-saving field and focus.In the today advocating green low-carbon life energetically, the dynamics strengthening air conditioner energy saving has huge practical significance undoubtedly to saving the energy, ground source heat pump technology is the energy-saving building technology that country widelys popularize in recent decades, cooling tower equipment is replaced with underground pipe compared with traditional air conditioning system, geothermal energy resources can be made full use of, realize taking summer heat-obtaining in cold winter, be a kind of real green air conditioner technology.
Application and development along with building automation system and variable-frequency control technique, automatic control technology is promoted at field of heating ventilation air conditioning gradually and is come, but central air conditioner system is a complication system with time lag, time-varying, non-linear and big inertia, its complexity causes that central air conditioner system is difficult to describe with accurately mathematical model or method, this to control system to realize accurately controlling to bring no small difficulty, is still largely dependent upon labor management in Practical Project;Simultaneously because lack advanced control technological means and equipment, central air conditioner system still adopts traditional labor management mode and easy switching control device mostly, air conditioner refrigerating can not be realized follow the change of end load and dynamically regulate, cause energy waste very big when operation at part load, make energy for building inefficiency, air conditioning system automaticity is not high, directly influences HVAC management level.
On the other hand, it is achieved refrigeration system running operating point and refrigerating capacity mate for controlling system it is critical that, accomplish this point, it is desirable to control system must take a set of practicable control algolithm as guidance;But in most cases HVAC controls system is utilize certain experience or semiempirical formula to be adjusted, this method is only based on the thought of data fitting, consider it appeared that the coefficient just with curvilinear regression acquisition refrigeration system Energy Efficiency Ratio with each influence factor is one group of static parameter further, actually refrigeration system Energy Efficiency Ratio not only with some relating to parameters also with other relating to parameters, therefore the parameter in fit correlation formula is not actually changeless but one group of slow time-varying coefficient, namely under different running statuses, coefficient is different, from the angle of Self Adaptive Control, the semiempirical formula of matching is not particularly suited for the inline diagnosis of control system.Additionally, whether this empirical equation can disclose the mechanism of action affecting heat pump Energy Efficiency Ratio well, its accuracy not yet has the foundation of science to be verified, therefore in actual motion there is very big roughening in regulating of heat pump, if system can not be run under a good operating point, necessarily cause that its Energy Efficiency Ratio is had a greatly reduced quality, be unfavorable for energy-conservation.
In sum, time-varying behavioral characteristics due to central air conditioner system, traditional Energy Saving Control strategy can not in refrigeration system running real-time online accumulation and comprehensively relevant information, carry out the control parameter of instant correction or the system of adjustment, air conditioning system more can not be made to be in optimum or the duty close to optimum all the time.
One, the background that air-conditioning Real-time Load is determined
The premise run that optimizes realizing earth-source hot-pump system is the air conditioner load that control system can predict to degree of precision subsequent time building, to ensure that system is cold quantitatively equal for cold-peace load, the time synchronizes.But building air-conditioning load variations has the typical non linear feature of the stochastic behaviours such as dynamic, time variation, the amount of disturbing, uncertainty more, for ease of controlling system advancement, the prediction load value of subsequent time building must being obtained, so seeking a kind of load forecasting method effectively accurately, optimization operation and the control of air conditioning system being significant.In actual motion, if the method for carry calculation is determined when air-conditioning hourly load is according to lectotype selection, necessarily waste bigger manpower and materials, be not therefore used in Engineering Operation.nullAt present,The determination of air conditioner load is actually the method for a kind of " remedying ",Not from the building enclosure of building and indoor heat gain、The angle of personnel's moisture dispersed amount considers,But consider from low-temperature receiver side refrigerating capacity,Control system by Temperature Humidity Sensor record indoor by time humiture,Owing to the humiture of air-conditioned room has certain allowable fluctuation range,As long as indoor actual humiture is little with setting humiture deviation,Just it is believed that refrigerating capacity substantially meets the air conditioner load in this moment,Then control systemic effect to regulate refrigerating capacity in the compressor etc. of pump and main frame and make refrigerating capacity meet room conditioning load,Although this means to save the situation meets the requirement of engineering to a certain extent,But due to the existence of stickiness time this,Being not precluded from a certain moment causes that owing to cold station refrigerating capacity can not meet air conditioner load the indoor temperature and humidity bigger phenomenon of fluctuation exists,So or the too high requirement that can not meet comfort level of indoor temperature,The too low waste energy of temperature,It is unfavorable for building energy conservation.It may be further contemplated, this remedial measure actually there is also the defect of following 3 aspects:
1), the accuracy of refrigeration capacity test
Existing cold station refrigerating capacity determines that employing following formula calculates: Q=cpρqVΔ t, if but a certain moment needs open a pump or stop a pump, cause flow to strengthen suddenly or reduce suddenly a lot, necessarily cause that temperature has bigger fluctuation, if the moment of temperature test is in the scope of this larger fluctuation, what necessarily cause cold calculates the error that unreasonable existence is bigger, and additionally flow transducer reliability for temperature sensor is also poor, and it is very big that the load that profit calculates in this way is likely to deviation actual value under certain conditions.
2), the equilibrium problem of the chilling requirement of the refrigerating capacity of cold and end
Even if the results contrast of cold test is accurate, the refrigerating capacity of cold is also a very big problem with mating of chilling requirement: if supply water temperature fixes on setting value, chilled-water flow is sufficiently large, this indicates that current refrigerating capacity disclosure satisfy that the requirement of building end, therefore the actual cold measured is chilling requirement, when supply water temperature is bigger than normal higher than setting value or supply backwater temperature difference, the cold probably currently provided when flow is relatively low is lower than the cold needed, building does not reach the operating mode needed, it is also possible to be that system is in suitable running status, how differentiation is enough and inadequate, water temperature and flow according only to the test of cold station are often difficult to judge.
3), " remedy " adjustment and lack theoretical foundation
Control system, by testing indoor temperature and humidity, determines increasing or the minimizing of cold, current owing to lacking certain adjustment foundation, often anthropic factor is bigger, regulated by the start and stop etc. of the unlatching of hand control pump or main frame, owing to control algolithm is indefinite, control relatively rough.
Two, the background that refrigeration system optimization of operating parameters regulates
Central refrigerating system includes three subsystems, i.e. refrigerator system, cooling water system and chilled water system.These three subsystem operationally all consumes certain energy, influences each other between them, interacts.Generally when the energy consumption of a subsystem reduces, the energy consumption of another subsystem will increase.And, the increase amount of a sub-system energy consumption is generally also not equal to the minimizing amount of another subsystem energy consumption.Therebetween relation is continually changing with the change of operating condition.Therefore, the target that whole central refrigerating system is optimized is not make the power consumption of wherein some single subsystem minimum, but makes the total power consumption of whole system minimum.When adopting system optimization method, it is necessary to the operation of whole central refrigerating system is regarded as an overall coordination process.The basic thought of system optimization is exactly that the total energy consumption making above three subsystem is minimum under meeting given operating mode (premise of all of end workload demand),
In actual applications, subsystems is carried out Performance Evaluation relatively difficult.This is because subsystems is coupling rather than independent, therefore refrigeration system operational factor is optimized the task of control include following some:
1) building refrigeration duty and associated external environment, are determined;
2) performance of refrigeration unit, is determined;
3), identify and determine that the optimal setting of control variable, these optimal settings should make the energy consumption of whole system minimum;
4), control system and subsystem operate in the adjustment of Optimal Setting value;
For a certain specific system, carry out the main task of system optimization and include finding the cooling water of optimum to enter coolant-temperature gage, chilled water temperature, cooling water flow, chilled-water flow, conveying equipment load sharing rate etc..These optimal values are under guaranteeing the premise meeting end workload demand, and the energy consumption making system is minimum.After optimal setting is determined, control CS central refrigeration system and operate in optimum setting operating mode.
Should be noted that, these optimal settings are not invariable, and are as the change of building load and operating condition (performance etc. such as outdoor temperature humidity and each subsystem) and realize Optimized Matching.All of these factors taken together will make vehicle air-conditioning become relatively difficult in practice.
Intercouple additionally, due to whole refrigeration system, refrigeration system is optimized and is necessary for whole system closed-Loop Analysis, some subsystem can not be departed from entirety to be individually analyzed, this strong coupling first step that optimal control is worked is difficult to find that breach just, thus bringing very big difficulty to the optimal control of refrigeration system.But the power consumption of unit accounts for the overwhelming majority in refrigeration system, the Energy Efficiency Ratio of cold source system and the correlation maximum of unit, therefore the operational energy efficiency ratio of unit reaches optimum, and the Energy Efficiency Ratio of whole system substantially also reaches optimum, therefore the breach of Optimization Work from refrigeration unit, can be obtained.
Summary of the invention
For the deficiency that existing refrigeration system energy-saving control method exists, the present invention proposes a kind of intelligent optimized control method for earth source heat pump refrigeration system.The basic thought of the present invention is, automatic building control system Lon-works platform carries out secondary development and realizes the integrated of single-chip microcomputer Lon-works, Single-chip Controlling language adopts MATLAB to write, single-chip microcomputer is as one-level control unit, including Air-conditioning Load Prediction module and refrigeration system optimization of operating parameters, module is set, Lon-works realizes the real-time control of the collection of refrigeration system real-time running state data and the transmission of one-level control signal and refrigeration system relevant hardware devices as Two-stage control unit, ensure that system is cold quantitatively equal for cold-peace load on the one hand, time synchronizes, on the other hand each parametric variable of refrigeration system can dynamically be regulated by control system in advance, guarantee that refrigeration system is in optimum or the duty close to optimum all the time.
Realizing refrigeration system optimum control is when giving with air conditioner load, and control system provides one group of optimized parameter and system runs optimum operating condition point, and the Energy Efficiency Ratio COP value that system is run under this group parameter is maximum.In actual motion, owing to changing the aperiodic of the thermal inertia of building, outside air temperature and the factor such as solar radiation is not occur immediately, but delayed a period of time, therefore for ensureing the humiture that air conditioning system requires, thermal source (low-temperature receiver) by time heat supply (cold) amount be the amount of a kind of dynamic change, in order to realize on-demand heat (cold) better, by dynamic methods analyst thermodynamic status, and by the method for Prediction Parameters, cold supply system dynamically must be regulated.Therefore, it is that prediction air conditioner load is so that inline diagnosis that control system realizes the premise of optimum control.Then the optimizing regulation of refrigeration system running state parameter is completed.
In Air-conditioning Load Prediction, due to air conditioner load change and external environment, in disturb and the factor such as thermal inertia of building is relevant, it can be difficult to find Influencing Mechanism.Building load Forecasting Methodology is a lot, and the condition of each self application is different with feature, in one period, building load has the feature of this kind of model sometimes, sometimes there is the feature of another kind of model, sometimes both have both, so several models can be optimized combination, even if the forecast model of a poor effect, as long as it is containing systematic opposition information, after it carries out associated prediction with one and several good forecast model, remain able to the prediction characteristic of improvement system, for improving final precision of prediction, control system uses combination forecasting method to comprehensively utilize the information that various methods provide, Individual forecast model is avoided to lose useful information, reduce randomness, improve precision of prediction.Pluses and minuses based on various forecast models, select Intelligent Forecasting: gray prediction method, GRNN neural net prediction method and LSSVM Forecasting Methodology, finally the method grey data integration technology of the utilization Optimal Combination Forecasting that predicts the outcome of these three forecast model is carried out fusion and obtain final Air-conditioning Load Prediction value.
In refrigeration system optimization of operating parameters adjustment, owing to the factor of refrigeration system running performance parameters mainly has following 6: the refrigerating capacity of handpiece Water Chilling Units, chilled water outlet temperature, cooling water inlet temperature, cooling water flow, chilled-water flow, the load sharing rate of equipment group;The equipment such as equipment group here is except unit for the whole refrigeration system of earth source heat pump, also has cooling water pump, chilled water pump;The determination of load sharing rate has very big associating with the sample properties curve of equipment, for unit, the meaning of sharing of load is minimum at overall refrigerating effect one its total power consumption of timing, for water pump, what be responsible for due to them is the conveying of the water yield, and the factor affecting their power consumption is this factor of flow, so, the premise that these equipment optimizations are controlled is the total flow size obtaining conveying, after flow is determined, the load sharing rate of each equipment can be determined according to the sample curve of equipment, realized certain flow conveying task and equipment total power consumption is minimum.
Therefore when whole refrigeration system is optimized, it is possible to be undertaken in two steps:
1., according to air-conditioning total load, under meeting refrigeration unit optimum COP premise, determine cooling water (chilled water) flow and (cooling water inlet, chilled water outlet) temperature of correspondence, complete first step optimization.
2., complete the load sharing rate of refrigeration unit according to total refrigerating capacity, determined its load sharing rate and then the optimization that completion system is final according to total flow by water pump curve.
The technical scheme is that a kind of earth source heat pump refrigeration system optimal control method, implement step as follows:
The first step, operation along with refrigeration system, Lon-works parametric controller constantly gathers the operational factor of refrigeration unit, dimensionless number according to the employing of Jitian's function model is according to processing form, refrigeration unit Energy Efficiency Ratio, refrigerating capacity, cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature are carried out nondimensionalization process respectively, constantly expands refrigeration unit runtime database;
Second step, use three kinds of Intelligent Forecastings: air conditioner load is predicted in gray scale prediction, generalized regression nerve networks prediction, least square method supporting vector machine prediction respectively, mutual supplement with each other's advantages feature for three kinds of forecast models, use grey data Fusion Model to integrate predicting the outcome of three kinds of models and obtain the predictive value of air conditioner load, reach the purpose of the relatively accurate prediction of air conditioner load.
3rd step, use the least square method forgotten with index that the refrigeration unit Energy Efficiency Ratio empirical equation based on Jitian's function model carries out the auto-adapted fitting of coefficient, realize the purpose that empirical equation coefficient dynamically adjusts, according to the principle that refrigerating capacity is equal with on air conditioner load in time synchronous number, air conditioner load refrigerating capacity in Jitian's function model predicted substitutes, to accurate description refrigeration unit Energy Efficiency Ratio under certain refrigerating capacity and cooling water flow, chilled-water flow, cooling water inlet temperature, 4 parameters of chilled water outlet temperature relation.
4th step, utilize extremum principle, to 4 parameters (cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature), Jitian's function model is carried out partial derivative calculating, and 4 equation group that simultaneous obtains obtain the subsequent time refrigeration unit 4 operational factor operating points close to optimum.
5th step, simplex method is used to carry out the optimal combinatorial search of 4 parameters of refrigeration unit, the evaluation function adopted in simplex method generalized regression nerve networks (GRNN) calculates, the input of GRNN neutral net is that (cooling water flow and cooling water inlet temperature merge into a main gene to three main genes, chilled-water flow is a main gene, chilled water outlet temperature is a main gene), neutral net is output as refrigeration unit Energy Efficiency Ratio.
6th step, carries out the optimization of other operational factors of refrigeration system: the unlatching number of units of refrigeration unit, cooling water pump and cooling water pump and rate of load condensate distribution.
7th step, combines according to the optimized operation parameter of subsequent time refrigeration system, each parametric variable is controlled in advance, and during to ensure that the moment to be measured arrives, refrigeration system is in the duty of optimum.
Accompanying drawing explanation
Fig. 1 is Lon-works and single-chip microcomputer integrating control schematic diagram in earth source heat pump refrigeration system;
Fig. 2 is Lon-works monitor supervision platform data acquisition and control signal transmission system diagram;
Fig. 3 is earth source heat pump refrigeration system control flow chart;
Fig. 4 is that (this caption: if the flow after optimizing is more than measured discharge, flow indicator is red to earth-source hot-pump system flow Single-chip Controlling basic circuit diagram, and strengthens total water current amount;It is otherwise green, reduces total water current amount);
Fig. 5 is earth-source hot-pump system load rate of plant distribution Single-chip Controlling basic circuit diagram (this caption: equipment group here refers to main frame and cooling water pump, chilled water pump.What when equipment is water pump, relay drove is the water knockout drum between main frame and water pump;When equipment is main frame, what relay drove is the compressor of main frame);
Fig. 6 is based on GRNN neutral net Air-conditioning Load Prediction schematic diagram (this caption: in a hour, sampling time interval is 10 minutes, relevant parameter sequence is 6);
Fig. 7 is four operational factor principal component analysis flow charts of refrigeration unit;
Fig. 8 is based on GRNN neural computing refrigeration unit Energy Efficiency Ratio schematic diagram;
Fig. 9 is based on 4 optimization of operating parameters flow charts of refrigeration unit of simplex method;
Figure 10 is principle of genetic algorithm figure;
Figure 11 is genetic algorithm chromosome coding mode;
Figure 12 is the genetic algorithm chromosome code segment data region of search;
Figure 13 is based on the refrigeration plant of Genetic Simulated Annealing Algorithm and runs number of units and rate of load condensate distribution global optimization flow chart.
Detailed description of the invention
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is carried out system to explain.
With reference to accompanying drawing 1, a kind of earth source heat pump refrigeration system optimal control method of the present invention mainly includes following step:
Step one, operation along with refrigeration system, Lon-works parametric controller constantly gathers the operational factor of refrigeration unit: refrigeration unit Energy Efficiency Ratio, refrigerating capacity, cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature, refrigeration unit open number of units and rate of load condensate, chilled water pump open number of units and rate of load condensate, cooling water pump open number of units and rate of load condensate.
Refrigeration unit Energy Efficiency Ratio, refrigerating capacity, cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature are carried out nondimensionalization process respectively, constantly expands refrigeration unit runtime database;
r c o p = c o p cop r , r Q = Q e - Q e r Q e r , r T c i = T c i - T c i , r T c i , r - T e o , r , r T e o = T e o - T e o , r T c i , r - T e o , r , r e w = w e - w e , r w e , r , r c w = w c - w c , r w c , r
Teo-chilled water outlet temperature, Tci-cooling water inlet temperature;we-chilled-water flow;Qe-refrigeration duty, wc-cooling water flow, subscript r represents value during this parameter declared working condition;
Step 2, three kinds of Intelligent Forecastings of utilization: air conditioner load is predicted in gray scale prediction, generalized regression nerve networks prediction, least square method supporting vector machine prediction respectively, finally use grey data Fusion Model to integrate predicting the outcome of three kinds of models and obtain the predictive value of air conditioner load, reach the purpose of relatively accurate prediction air conditioner load.
(1), adopt generalized regression nerve networks (GRNN) to carry out Air-conditioning Load Prediction as shown in Figure 6, be described as follows:
1), GRNN neural network structure is input layer, mode layer, summation layer, output layer:
2), the ground floor of network be input layer, neuron number is equal to the dimension 8 of the input vector of learning sample, and each neuron is simple distribution unit, directly input variable is passed to mode layer.
3), input parameter X be upper one hour indoor epidemic disaster sequence, upper one hour outdoor temperature humidity sequence, upper one hour occupancy sequence, this time be engraved in one day time numbering, season type, what day totally 8 parameter, its output Y is the air conditioner load of prediction time, and computing formula is:
Y ^ ( X ) = Σ i = 1 n Y i exp [ - ( X - X i ) T ( X - X i ) 2 σ 2 ] Σ i = 1 n exp [ - ( X - X i ) T ( X - X i ) 2 σ 2 ]
4), in order to eliminate the discordance of different parameters physics dimension, all of parameter is all normalized.
5), the second layer of network is mode layer, neuron number is equal to number of training (present invention is taken as 500), the weight function of this layer is Euclidean distance function (| | dist | |), and its effect is the weights IW of computing network input and ground floor1,iBetween distance, b1For mode layer threshold value.The transmission function of mode layer is RBFAs the transmission function of network, σiFor smoothing factor.
6), the defining method of smoothing factor σ is: make parameter σ with increment Delta σ at certain limit [σminmax] interior incremental variations, here:
σ m i n = ( - D min 2 2 l n ξ ) 1 2 , σ m a x = 1
Wherein DminFor the minima of Euclid distance between input sample each in learning sample;The minimum positive number that ξ > 0 can recognise that for computer.In learning sample, remove a sample, by remaining sample architecture generalized regression nerve networks, this sample is estimated, obtain the error between estimated value and sample value;Each sample is repeated this process, obtains error sequence, by the mean-square value of error sequence:
As the evaluation index of network performance, smoothing parameter corresponding for minimum error is used for last GRNN neutral net.
7), the third layer of network be summation layer, summation layer comprises two kinds of neuron, one of which neuron calculating formulaDenominator, the neuronic output of all mode layers is carried out arithmetic summation, each neuron of mode layer is 1 with these neuronic weights that are connected, its transmission function be:
S D = Σ i = 1 n p i
The molecule neuronic transmission function of summation is
S N j = Σ i = 1 n y i j p i , j = 1 , 2 ... p
Its weight function is standardization dot product weight function, the vector n of computing network3, its each element is by vector α2With weight matrix IW2,iIn the dot product of every row element again divided by vector α2Each element sum obtain,
8), last layer of network be linear convergent rate layer, by result n3It is supplied to linear transfer function α4=purelin (n3), the output of computing network.
9), GRNN neutral net training sample number defining method:
Training sample number primarily determines that method is:
n = N / 6 N = 20 ~ 40 N / ln N N = 40 ~ 100 N / 2 N = 100 ~ 400
Here N is total number of samples, after primarily determining that training sample number, next scans near n value, after determining smoothing parameter for each n value, determines training sample number according to the minima of BIC criterion evaluation index:
BIC (n)=NlnE+nlnN
10), GRNN network connects the correction employing BP algorithm of weights.
11) neutral net that training just can be utilized after, completing GRNN neural metwork training ripe carries out the prediction of air conditioner load.
(2), the concrete grammar of employing gray scale prediction is:
1), gray scale prediction adopt gray system theory GM (1,1) forecast model, take the air conditioner load time series (every 10 minutes sampling once totally 6 groups of data) of previous hour of prediction time: x(0)=[x(0)(1),x(0)(2),x(0)(3),x(0)(4),x(0)(5),x(0)], and accumulation interval sequence: x (6)(1)=[x(1)(1),x(1)(2),x(1)(3),x(1)(4),x(1)(5),x(1)(6)], accumulation interval sequential element
2), prediction time air conditioner load is obtained by following formula:
x ( 0 ) ( 7 ) = [ x ( 0 ) ( 1 ) - u a ] e - 6 a + u a - x ( 1 ) - - - ( 6 )
Wherein, parameter vectorY=[x(0)(2),x(0)(3),x(0)(4),x(0)(5),x(0)(6)],
B = - 1 2 ( x ( 1 ) ( 1 ) + x ( 1 ) ( 2 ) ) - 1 2 ( x ( 1 ) ( 2 ) + x ( 1 ) ( 3 ) ) - 1 2 ( x ( 1 ) ( 3 ) + x ( 1 ) ( 4 ) ) - 1 2 ( x ( 1 ) ( 4 ) + x ( 1 ) ( 5 ) ) - 1 2 ( x ( 1 ) ( 5 ) + x ( 1 ) ( 6 ) ) 1 1 1 1 1 T
(3), the concrete grammar of employing least square method supporting vector machine prediction is:
1) the air conditioner load time series (totally 6 groups of data of sampling once every 10 minutes) of previous hour of prediction time, is taken:
2), following matrix equation is solved:
0 e T e Ω + C - 1 I 7 × 7 b α = 0 Y
Wherein,
3), kernel function is taken as gaussian kernel:
K ( x i , x j ) = exp ( - | | x i - x j | | 2 2 σ 2 )
4), least square method supporting vector machine Air-conditioning Load Prediction computing formula is:
f ( x ) = Σ i = 1 n α i exp ( - | | x i - x | | 2 2 σ 2 ) + b
Here σ is that to be taken as 1, C be that penalty factor is taken as 50, x to nuclear parameteriFor air conditioner load sampling instant in previous hour in one day time numbering, x be prediction time in one day time numbering.
(4), the implementation of grey data fusion method is:
1)、a1,a2,a3The air conditioner load value of the prediction time for being obtained by above-mentioned 3 forecast models, the distance between definition any two value is as follows:
dij=| ai-aj|, i, j ∈ { 1,2,3}
2), the function for support between two data of structure:
r i j = c o s ( πd i j 2 × m a x { d i j } ) ; i , j ∈ { 1 , 2 , 3 }
3) matrix R={r, is tried to achieveijEigenvalue of maximum λ, corresponding special syndrome vectorTake:Obtain after fusion:It is grey data fusion forecasting value.
Step 3, use the least square method forgotten with index that the refrigeration unit Energy Efficiency Ratio empirical equation based on Jitian's function model carries out the auto-adapted fitting of coefficient.
(1), Jitian's function be one and there is the good empirical model for describing refrigeration unit Energy Efficiency Ratio COP and refrigerating capacity, cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature returning characteristic, its expression formula is:
r c o p = a 1 ( a 2 r T c i 2 + a 3 r T c i + 1 ) ( a 4 r Q 4 + a 5 r Q 3 + a 6 r Q 2 + a 7 r Q + 1 ) ( a 8 r T e o 2 + a 9 r T e o + 1 ) ( a 10 r e w 2 + a 11 r e w + 1 ) ( a 12 r c w 2 + a 13 r c w + 1 ) + a 14
Wherein, a1~aiFor fitting constant.
(2), based on forget with index method of least square definition residual error V (a, t) be:
λ-forgetting factor, 0 < λ≤1, a=[a1,a2,...a14]
Wherein, forgetting factor is taken as 0.75, rcopThe i actual Energy Efficiency Ratio size of refrigeration unit that () is nondimensionalization,For using the actual Energy Efficiency Ratio of refrigeration unit of the nondimensionalization of Jitian's function model matching.
(3), by residual error V, (a, t) to a1~a14Carry out partial derivative calculating respectively and just can determine that not t in the same time, 14 fitting constants in Jitian's function model:
&part; V ( a , t ) &part; a i = 0 , ( i = 1 ~ 14 )
Step 4, prediction time air conditioner load is substituted into Jitian's function model as refrigerant system capacity, according to extremum principle, by Jitian's function model, 4 parameters (cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature) are carried out partial derivative calculating:
&part; r c o p &part; r T c i = 0 , &part; r c o p &part; r T e o = 0 , &part; r c o p &part; r e w = 0 , &part; r c o p &part; r c w = 0
4 equation group that simultaneous obtains obtain 4 operational factor operating points of subsequent time refrigeration unit near-optimization:
( r ~ T c i o p t i , r ~ T e o o p t i , r ~ e w o p t i , r ~ c w o p t i )
Step 5, uses simplex method to carry out the optimal combinatorial search of 4 parameters of refrigeration unit;
Specifically include:
(1), principal component analytical method is used to be analyzed finding main gene to 4 parameters (cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature);
In the present invention, 4 parameters relevant to refrigeration unit Energy Efficiency Ratio being carried out principal component analysis, obtain main gene, with reference to accompanying drawing 7, the concrete grammar of principal component analysis is:
1), to 4 parameters (cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature) sample being standardized processing, computing formula is:N is total sample number;
2), the sample variable matrix X after standardization is carried out correspondent transform again:
By row summation X . j = &Sigma; i = 1 n x i j , j = 1 , 2 , 3 , 4 , By row summation X i . = &Sigma; j = 1 4 x i j , i = 1 , 2 ... n , Summation T = &Sigma; i = 1 n &Sigma; j = 1 4 x i j .
According to above-mentioned calculating, the element in new matrix Z after obtaining correspondent transform:
z i j = x i j - x i . x . j T x i . x . j , ( i = 1 , 2 ... n , j = 1 , 2 , 3 , 4 )
In formula, xijFor the element in raw data matrix;zijFor the element in the new matrix after conversion;
3) the covariance matrix R of matrix Z, is calculated
R=zTZ=(rij)4×4, the element in matrix R in formula
4) eigenvalue and the characteristic of correspondence vector thereof of matrix R, are determined
The eigenvalue tried to achieve by Jacobi algorithm and characteristic vector, by eigenvalue by descending order arrangement: λ1≥λ2≥λ3≥λ4>=0, its characteristic of correspondence vector is
ei=(e1i,e2i,e3i,e4i)T, (i=1,2...n, j=1,2,3,4)
5), Factor load-matrix A is calculated
First calculating the accumulation contribution rate of principal component, when cumulative percentage is more than 85%, taking k composition above is principal component, being calculated as follows of accumulation contribution rate:
&Sigma; i = 1 k &lambda; i &Sigma; i = 1 4 &lambda; i &GreaterEqual; 85 %
Thus calculate R type Factor load-matrix;
A = e 11 &lambda; 1 e 12 &lambda; 2 ... e 1 k &lambda; k e 21 &lambda; 1 e 22 &lambda; 2 ... e 2 k &lambda; k e 31 &lambda; 1 e 32 &lambda; 1 ... e 3 k &lambda; 1 e 41 &lambda; 1 e 42 &lambda; 2 ... e 4 k &lambda; k
Every string in matrix is exactly corresponding characteristic vector and the subduplicate product of eigenvalue.
6), mapping classification:
Choose two maximum and secondary big eigenvalue λ of R type12And corresponding characteristic vector e1,e2, at RmIn space withConstruct two coordinate axess respectively, and be designated as F1And F2.So, each factor of influence is at plane F1-F2A upper corresponding point, is classified as a class by contiguous factor of influence, represents that they can merge into an integrated contributory factor.
7), utilizing works project measured data of the present invention, 4 parameters are carried out principal component analysis, obtain first three main gene and soluble total data sample 94% information, first principal component mainly reflects the information of chilled-water flow, Second principal component, mainly reflects the information of chilled water outlet temperature, and the 3rd main constituent reflects the relation of cooling water flow, cooling water inlet temperature.
8), these 4 parameters explanation letter ratio in three main genes is reset to: αij(i=1,2,3;J=1,2,3,4), according to the conclusion that principal component analysis draws, these 4 parameters are explained that quantity of information is as the weight of respective dimensionless number, obtains three main gene λ in main gene1、λ2、λ3Computing formula:
(2), the detailed process of simplex search;
1), the structure of initial simplex
Construct a simplex having 4 summits, initial point: X(0)=(x1,x2,x3), all the other 3 points are elected as: X(1)=(x1+p,x2+q,x3+q),X(2)=(x1+q,x2+p,x3+ q), X(3)=(x1+q,x2+q,x3+ p), wherein choose: p=0.94a, q=0.24a, a are the length of side of simplex, be set as step-length when searching for.Then calculating one by one at search variables is X(0), X(1), X(2), X(3)Evaluation function f (X), and compare.
2), initial point: X(0)=(x1,x2,x3) computing formula be:
x 1 = &Sigma; j = 1 4 &alpha; 1 j r j , x 2 = &Sigma; j = 1 4 &alpha; 2 j r j , x 3 = &Sigma; j = 1 4 &alpha; 3 j r j
Wherein r = ( r 1 , r 2 , r 3 , r 4 ) = ( r ~ T c i o p t i , r ~ T e o o p t i , r ~ e w o p t i , r ~ c w o p t i ) .
3), the defining method of the length of side a of simplex is:
A=max{abs (x1-x2),abs(x1-x3),abs(x2-x3)}
4), evaluation function f (X) adopts generalized regression nerve networks:
Topological structure is as shown in Figure 8 for generalized regression nerve networks (GRNN neutral net), the input of neutral net is that (cooling water flow and cooling water inlet temperature merge into a main gene to 3 main genes, chilled-water flow is a main gene, chilled water outlet temperature is a main gene), it is output as the refrigeration unit Energy Efficiency Ratio COP of correspondence, all the other explanations are identical with the GRNN neutral net carrying out Air-conditioning Load Prediction employing, repeat no more.
5), the iterative process of simplex method
Its flow process as shown in Figure 9, including following operation:
1., reflection
Seek worst point X(h)Pip, even X(R)=Xc+α(Xc-X(h)), wherein XcIt is removing X in the summit of simplex(h)The barycenter oftriangle of 3 summits composition in addition is namely:
X c = 1 3 &lsqb; &Sigma; i = 0 3 ( X ( i ) - X ( h ) ) &rsqb; + X ( h )
So X(R)It is X(h)About centre of form XcPip, the functional value of this point is f (X(R)).The reflection coefficient that wherein α is given is taken as 0.3.
2., extend
Comparison function value f (X(R)) and f (X(L)), if f is (X(R))≤f(X(L)), namely represent that the pip of new choosing the most better fortunately then suitably can extend to find better point than original in the original direction of search again, by X(R)-XCExtend:
X(E)=XC+γ(X(R)-XC)
Wherein γ is given lengthening coefficient, is taken as 1.5, if f is (X(E))≤f(X(L)) then with X(E)Replace X(h), otherwise with X(R)Replace X(h)
3., shrink
If for all removing worst point X(h)Other points (X in addition(i)≠X(h)) on desired value have f (X(i)) < f (X(R)) < f (X(h)) or f (X(G))≤f(X(R))(f(X(G)) for f (X(R)) outside 3 functional values in second largest value, be time bad value point), illustrates that selected pip is not so good, then by vector contraction, can make:
X(N)=Xc+β(X(h)-Xc)
Wherein 0 < β < 1 is given constriction coefficient, is taken as 0.8, at f (X(R)) < f (X(h)) when, with X(R)Replace X(h)After shrink again.As f (X(N)) < f (X(h)) time, with X(N)Replace X(h)
4., whole simplex is compressed
If f is (X(N))≥f(X(h)), even if after illustrating to have done above-mentioned contraction, target function value does not improve, then original simplex can be reduced half to best point, by all of vector X(i)-X(h)Reduce half, order:
X ( i ) = X ( L ) + 1 2 ( X ( i ) - X ( L ) ) = 1 2 ( X ( i ) + X ( L ) ) , i = 1 , 2 , 3 ,
5., obtain new simplex, constantly repeat above-mentioned iterative process, until meeting certain finish condition:
max{f(X(i))-f(X(C)) < ε
ε is that the precision set is taken as 0.001.
Step 6, carries out the optimization of other operational factors of refrigeration system: the unlatching number of units of refrigeration unit, cooling water pump and cooling water pump and rate of load condensate distribution, mainly includes herein below:
(1), when air-conditioning total load is certain, obtains the rate of load condensate of unit in whole refrigeration system and make all unit total power consumptions minimum:
Owing to refrigeration unit performance map is generally the Energy Efficiency Ratio sample curve f (α) with rate of load condensate, when according to its sample properties curve, air-conditioning total load one timing can determine that the load sharing rate of unit makes its power consumption minimum.If overall refrigerating effect is Q, the specified refrigerating capacity of separate unit refrigeration unit is Qr, in system, refrigeration unit number of units amounts to N platform, and the mathematical description of its load sharing rate is:
m i n &Sigma; x i &alpha; i f ( x i &alpha; i )
s . t &Sigma;x i &alpha; i = Q Q r
xi=0,1i=1,2...N
0<αi≤1
(2), the power consumption of given pump with the sample curve f (α) of rate of load condensate, when according to its sample properties curve, total flow one timing determines that the load sharing rate of conveying equipment makes its power consumption minimum:
Total flow is w, and the metered flow of single pump is wr, in system, pump number of units amounts to N platform, then the optimization of conveying equipment is under certain flow, and the power consumption of pump is minimum, and its mathematical description is:
m i n &Sigma; i = 1 N f ( x i &alpha; i )
s.txi=0,1i=1,2...N
0<αi≤1
&Sigma; i = 1 N x i &alpha; i = w w e , r
(3), the double optimization method of refrigeration unit and cooling water pump, the unlatching number of units of chilled water pump and rate of load condensate:
The sequencing contro of the equipment (unit and water pump) of refrigeration system includes the distribution of the rate of load condensate of number of units and the unlatching number of units opened, and the nonlinear programming problem in (1) and (2) is completed by double optimization.
1), a suboptimization is the number of units in order to determine start and stop, unit and water pump all adopt the optimization method based on performance map, for refrigeration unit (conveying equipment), when overall refrigerating effect Q (total flow w) is certain, (for refrigeration unit, to be Energy Efficiency Ratio COP maximum for maximal efficiency to obtain maximal efficiency by refrigeration unit (water pump) performance curve;For conveying equipment, maximal efficiency is that feed flow is maximum with the ratio of power consumption) run lower corresponding refrigerating capacity Qop(flow wop), in a suboptimization consider refrigerating capacity (flow) mean allocation, the unlatching number of units n of refrigeration unit (water pump) primarily determine that into:
For refrigeration unit: n = &lsqb; Q Q o p &rsqb; + 1 Q Q o p - &lsqb; Q Q o p &rsqb; > 0.5 &lsqb; Q Q o p &rsqb; Q Q o p - &lsqb; Q Q o p &rsqb; < 0.5 , For water pump: n = &lsqb; w w o p &rsqb; + 1 w w o p - &lsqb; w w o p &rsqb; > 0.5 &lsqb; w w o p &rsqb; w w o p - &lsqb; w w o p &rsqb; < 0.5 ,
2), double optimization be to complete final optimized distribution, obtain preferably opening after number of units by a suboptimization, double optimization is a constrained nonlinear programming problem,
1., have for refrigeration unit:
m i n &Sigma; &alpha; i f ( &alpha; i )
s . t &Sigma;&alpha; i = Q Q r
0<αi≤ 1i=1,2.....m
In order to constrained planning problem is converted into Unconstrained optimization, it is considered to majorized function:
f ( &alpha; ) = m i n &Sigma; &alpha; i f ( &alpha; i ) + &lambda; ( &Sigma;&alpha; i - Q Q r ) 2 i = 1 , 2.... m
2., for pump, Unconstrained optimization majorized function is:
f ( &alpha; ) = m i n &Sigma; f ( &alpha; i ) + &lambda; ( &Sigma;&alpha; i - w w r ) 2 i = 1 , 2 ... . m
Wherein m is for opening number of units, and λ is that weight factor is taken as 0.8~1.2, and α is rate of load condensate allocation vector (m dimension), and number of units m is taken as n-1 respectively, n, and n+1 carries out the contrast of f (α) value.
3), in order to determine final optimum results, rate of load condensate αiDetermination adopt based on genetic algorithm such as accompanying drawing 13 of simulated annealing, wherein mainly include following key technology based on the genetic algorithm of simulated annealing:
1., chromosome coding, such as accompanying drawing 11, employing binary coding mode, chromosome is m section (the unlatching number of units of the refrigeration unit (water pump) in refrigeration system) altogether.
2., such as accompanying drawing 12, αiThe region of search be defined as: [αop-Δα,αop+ Δ α], wherein
For refrigeration unit
&Delta; &alpha; = m a x { | &alpha; o p - Q ( n + 1 ) Q r | , | &alpha; o p - Q ( n - 1 ) Q r | } , &alpha; o p = Q o p Q r
For water pump
&Delta; &alpha; = m a x { | &alpha; o p - w ( n + 1 ) w r | , | &alpha; o p - w ( n - 1 ) w r | } , &alpha; o p = w o p w r
3., the structure of fitness function,
The principle of genetic algorithm optimizing such as accompanying drawing 10, its target finds most suitable m parameter combination exactly, but easily occur when genetic algorithm is evolved in early days that population precocity is absorbed in the situation of local optimum, in order to avoid this situation is based on simulated annealing Metropolis criterion, randomly selecting individual i and j in filial generation group, individual i competition enters follow-on selected probability and is:
f ( i ) &le; f ( j ) p ( i ) = 1 f ( i ) &GreaterEqual; f ( j ) p ( i ) = 1 exp ( f ( j ) - f ( i ) T ) &GreaterEqual; &alpha; p ( i ) = 0 exp ( f ( j ) - f ( i ) T ) < &alpha;
Wherein T is annealing temperature, α=0.5.
4., the setting that annealing temperature is interval:
The fitness function of structure in three, annealing temperature interval is taken as T=300~0 here, and each circulating temperature is changed to T=α T, α=0.95.
5., the setting of every generation population number and evolutionary generation in genetic algorithm:
Every generation population at individual is set to 500, and evolutionary generation was set to for 200 generations.
6. 3 circulations, in genetic algorithm, such as accompanying drawing 13:
Ground floor circulation (in one layer of circulation) is the searching process between every generation population at individual, second layer circulation is simulated annealing process, this layer of circulation reduces along with the continuous of Simulated annealing, third layer circulation is that (namely number of units m is taken as n-1 to change chromosome length m respectively, n, n+1), it is achieved there is the process optimized further between the optimum individual after the Evolution of Population of coloured differently body length.
Step 7, combines according to the optimized operation parameter of subsequent time refrigeration system, and with reference to accompanying drawing 3, each parametric variable is controlled by control system in advance, and during to ensure that the moment to be measured arrives, refrigeration system is in the duty of optimum.
The present invention carries out secondary development on automatic building control system Lon-works platform and realizes the integrated of single-chip microcomputer Lon-works, single-chip microcomputer is as one-level control unit, including Air-conditioning Load Prediction module and refrigeration system optimization of operating parameters, module is set, Lon-works realizes the real-time control of the collection of refrigeration system real-time running state data and the transmission of one-level control signal and refrigeration system relevant hardware devices as Two-stage control unit, ensure that system is cold quantitatively equal for cold-peace load on the one hand, time synchronizes, on the other hand each parametric variable of refrigeration system can dynamically be regulated by control system in advance, guarantee that refrigeration system is in optimum or the duty close to optimum all the time.

Claims (4)

1. an earth source heat pump refrigeration system optimal control method, comprises the following steps:
Step one, gathering the operational factor of refrigeration unit, including refrigeration unit Energy Efficiency Ratio, refrigerating capacity, cooling water flow, chilled-water flow, cooling water inlet temperature, chilled water outlet temperature, refrigeration unit opens number of units and rate of load condensate, chilled water pump opens number of units and rate of load condensate, cooling water pump open number of units and rate of load condensate;
Step 2, three kinds of Intelligent Forecastings of utilization: air conditioner load is predicted in gray scale prediction, generalized regression nerve networks prediction, least square method supporting vector machine prediction respectively, finally use grey data Fusion Model predicting the outcome of three kinds of models of integration to obtain the predictive value of air conditioner load;
Step 3, use the least square method forgotten with index that the refrigeration unit Energy Efficiency Ratio empirical equation based on Jitian's function model carries out the auto-adapted fitting of coefficient;
Step 4, prediction time air conditioner load is substituted into Jitian's function model as refrigerant system capacity, according to extremum principle, by Jitian's function model, cooling water flow, chilled-water flow, cooling water inlet temperature, 4 parameters of chilled water outlet temperature are carried out partial derivative calculating, obtain the subsequent time refrigeration unit 4 operational factor operating points close to optimum;
Step 5, uses simplex method to carry out the optimal combinatorial search of refrigeration unit cooling water flow, chilled-water flow, cooling water inlet temperature, 4 parameters of chilled water outlet temperature;
Step 6, carries out the optimization of other operational factors of refrigeration system, including unlatching number of units and the rate of load condensate distribution of refrigeration unit, cooling water pump and cooling water pump;
Step 7, combines according to the optimized operation parameter of subsequent time refrigeration system, and each parametric variable is controlled by control system in advance, and during to ensure that the moment to be measured arrives, refrigeration system is in the duty of optimum.
2. earth source heat pump refrigeration system optimal control method as claimed in claim 1, it is characterised in that: in described step 2, the neural network structure of generalized regression nerve networks prediction is input layer, mode layer, summation layer, output layer.
3. earth source heat pump refrigeration system optimal control method as claimed in claim 1, it is characterised in that: the implementation of described grey data fusion method is:
A), predict for gray scale, generalized regression nerve networks prediction, least square method supporting vector machine predict the air conditioner load value of prediction time that these 3 forecast models obtain, the distance between definition any two value is as follows: dij=| ai-aj|, i, j ∈ { 1,2,3}
B), the function for support between two data of structure: r i j = c o s ( &pi;d i j 2 &times; m a x { d i j } ) , i , j &Element; { 1 , 2 , 3 } ;
C) matrix R={r, is tried to achieveijEigenvalue of maximum λ, corresponding special syndrome vectorTake:Obtain after fusion:It is grey data fusion forecasting value.
4. earth source heat pump refrigeration system optimal control method as claimed in claim 1, it is characterised in that: the optimization step of described step 6 is:
A), when air-conditioning total load is certain, obtains the rate of load condensate of unit in whole refrigeration system and make all unit total power consumptions minimum;
B), the power consumption of given pump with the sample curve of rate of load condensate, when according to its sample properties curve, total flow one timing determines that the load sharing rate of conveying equipment makes its power consumption minimum;
C), the double optimization method of refrigeration unit and cooling water pump, the unlatching number of units of chilled water pump and rate of load condensate.
CN201410125301.7A 2014-03-31 2014-03-31 A kind of earth source heat pump refrigeration system optimal control method Expired - Fee Related CN103912966B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410125301.7A CN103912966B (en) 2014-03-31 2014-03-31 A kind of earth source heat pump refrigeration system optimal control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410125301.7A CN103912966B (en) 2014-03-31 2014-03-31 A kind of earth source heat pump refrigeration system optimal control method

Publications (2)

Publication Number Publication Date
CN103912966A CN103912966A (en) 2014-07-09
CN103912966B true CN103912966B (en) 2016-07-06

Family

ID=51038845

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410125301.7A Expired - Fee Related CN103912966B (en) 2014-03-31 2014-03-31 A kind of earth source heat pump refrigeration system optimal control method

Country Status (1)

Country Link
CN (1) CN103912966B (en)

Families Citing this family (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6476818B2 (en) * 2014-10-15 2019-03-06 オムロン株式会社 Heat demand estimation device, heat demand estimation method, equipment control device, equipment control method, equipment control system, control program, and recording medium
CN104654690B (en) * 2014-11-18 2017-01-11 深圳职业技术学院 Method and system for controlling water chilling unit
CN107429930B (en) * 2015-04-01 2019-12-20 三菱电机株式会社 Air conditioning system control device
CN105156327B (en) * 2015-09-30 2017-01-25 深圳德尔科机电环保科技有限公司 Energy-saving control method for compressed air industrial screw-type air compressor group control system
CN105259752B (en) * 2015-11-13 2018-08-03 北京博锐尚格节能技术股份有限公司 Load distribution method, device and system for equipment group
CN105937823B (en) * 2016-03-31 2019-03-29 中国农业大学 A kind of earth source heat pump control method and system
TWI604162B (en) * 2016-06-21 2017-11-01 Chunghwa Telecom Co Ltd Automatic air conditioner operation capacity adjustment system and method
JP6618860B2 (en) * 2016-06-27 2019-12-11 荏原冷熱システム株式会社 Heat source system and control method thereof
CN107143981B (en) * 2017-05-24 2019-06-28 山东师范大学 A kind of controlling system of central air conditioner and method
CN107816828B (en) * 2017-10-19 2018-12-04 英特尔产品(成都)有限公司 For determining the method, apparatus and system of the temperature set-point of the refrigeration machine of refrigeration system
CN109323425B (en) * 2018-11-15 2021-05-25 广东美的制冷设备有限公司 Control method and device of air conditioner and readable storage medium
CN109726896A (en) * 2018-12-05 2019-05-07 新奥数能科技有限公司 The calculation method and device of energy efficiency, storage medium, electronic device
CN109634121B (en) * 2018-12-28 2021-08-03 浙江工业大学 Multi-parent genetic algorithm air source heat pump multi-objective optimization control method based on radial basis function neural network
CN110057045B (en) * 2019-03-20 2021-10-29 青岛海尔空调器有限总公司 Control method for air conditioner
CN110118382B (en) * 2019-04-15 2020-12-29 天津大学 General operation regulation strategy identification and evaluation method for heat exchange station
CN110239721B (en) * 2019-06-23 2020-11-27 北京航空航天大学 Optimization design method for electric air circulation refrigeration system
CN112365029B (en) * 2019-09-03 2021-08-17 深圳市得益节能科技股份有限公司 Missing value processing method for air conditioner load prediction and air conditioner load prediction system
CN111043720B (en) * 2019-10-21 2021-05-14 天津大学 Low-cost robustness adjustment strategy making method of refrigeration system under load uncertainty
CN111158264B (en) * 2020-01-09 2021-06-29 吉林大学 Model prediction control rapid solving method for vehicle-mounted application
CN111256294B (en) * 2020-01-17 2021-01-05 深圳市得益节能科技股份有限公司 Model prediction-based optimization control method for combined operation of water chilling unit
CN111442480A (en) * 2020-04-08 2020-07-24 广东美的暖通设备有限公司 Operation control method and system for air conditioning equipment, air conditioning equipment and storage medium
CN113051737B (en) * 2021-03-16 2022-05-20 天津大学 Ground source heat pump data driving modeling method considering thermal measurement data abnormity
CN113821902A (en) * 2021-06-17 2021-12-21 李壮举 Active disturbance rejection control system for static optimization of central air-conditioning refrigeration station

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5579993A (en) * 1995-01-06 1996-12-03 Landis & Gyr Powers, Inc. HVAC distribution system identification
CN101251291A (en) * 2008-04-03 2008-08-27 上海交通大学 Central air conditioning system global optimization energy-saving control method and device based on model
CN103234256A (en) * 2013-04-17 2013-08-07 上海达希能源科技有限公司 Dynamic load tracking central air conditioner cold source global optimum energy-saving control method
CN103453623A (en) * 2013-09-13 2013-12-18 天津大学建筑设计研究院 Water source heat pump air-conditioning system operating parameter optimization control method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5579993A (en) * 1995-01-06 1996-12-03 Landis & Gyr Powers, Inc. HVAC distribution system identification
CN101251291A (en) * 2008-04-03 2008-08-27 上海交通大学 Central air conditioning system global optimization energy-saving control method and device based on model
CN103234256A (en) * 2013-04-17 2013-08-07 上海达希能源科技有限公司 Dynamic load tracking central air conditioner cold source global optimum energy-saving control method
CN103453623A (en) * 2013-09-13 2013-12-18 天津大学建筑设计研究院 Water source heat pump air-conditioning system operating parameter optimization control method

Also Published As

Publication number Publication date
CN103912966A (en) 2014-07-09

Similar Documents

Publication Publication Date Title
CN103912966B (en) A kind of earth source heat pump refrigeration system optimal control method
Kumar et al. Energy analysis of a building using artificial neural network: A review
CN106920006B (en) Subway station air conditioning system energy consumption prediction method based on ISOA-LSSVM
Wang et al. Supervisory and optimal control of building HVAC systems: A review
JP5572799B2 (en) Air conditioning system controller
CN107860102B (en) Method and device for controlling central air conditioner
CN105719028B (en) A kind of air conditioner load dynamic prediction method based on multifactor chaos support vector machines
CN112001439A (en) GBDT-based shopping mall building air conditioner cold load prediction method, storage medium and equipment
Xikai et al. Comparison of regression models for estimation of carbon emissions during building's lifecycle using designing factors: a case study of residential buildings in Tianjin, China
Wang et al. Evaluation of operation performance of a multi-chiller system using a data-based chiller model
Hwang et al. Using deep learning approaches with variable selection process to predict the energy performance of a heating and cooling system
CN111649457A (en) Dynamic predictive machine learning type air conditioner energy-saving control method
Ceballos-Fuentealba et al. A simulation and optimisation methodology for choosing energy efficiency measures in non-residential buildings
Zhang et al. Development and evaluation of cooling load prediction models for a factory workshop
CN111580382A (en) Unit-level heat supply adjusting method and system based on artificial intelligence
Wang et al. Multi-criteria comprehensive study on predictive algorithm of heating energy consumption of district heating station based on timeseries processing
KR20180138371A (en) Method for evaluating data based models and conducting predictive control of capsule type ice thermal storage system using the same
CN110598923A (en) Air conditioner load prediction method based on support vector regression optimization and error correction
Perera et al. Prediction of space heating energy consumption in cabins based on multivariate regression modelling
CN113778215A (en) Method for realizing data center PUE prediction and consumption reduction strategy based on big data
Zhao et al. Data analysis and modeling of chilled water loops in air conditioning systems
Souayfane et al. A weather-clustering and energy-thermal comfort optimization methodology for indoor cooling in subtropical desert climates
Bogdanovs et al. Impact of Intelligence System of Building Management System for Energy Efficiency in the Test Facility
CN113028610B (en) Method and device for global optimization and energy-saving control of dynamic load of central air conditioner
CN114200839A (en) Office building energy consumption intelligent control model for dynamic monitoring of coupled environmental behaviors

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20160706

Termination date: 20170331