CN106765959A - Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms - Google Patents

Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms Download PDF

Info

Publication number
CN106765959A
CN106765959A CN201611227073.XA CN201611227073A CN106765959A CN 106765959 A CN106765959 A CN 106765959A CN 201611227073 A CN201611227073 A CN 201611227073A CN 106765959 A CN106765959 A CN 106765959A
Authority
CN
China
Prior art keywords
conditioning system
genetic algorithm
heating ventilation
ventilation air
neural network
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.)
Pending
Application number
CN201611227073.XA
Other languages
Chinese (zh)
Inventor
杨定义
阮云峰
郑鹏
胡志勇
阳许军
邱显龙
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 Hong Xin Technological Service Co Ltd
Original Assignee
Wuhan Hong Xin Technological Service Co Ltd
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 Hong Xin Technological Service Co Ltd filed Critical Wuhan Hong Xin Technological Service Co Ltd
Priority to CN201611227073.XA priority Critical patent/CN106765959A/en
Publication of CN106765959A publication Critical patent/CN106765959A/en
Pending legal-status Critical Current

Links

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/10—Temperature
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/20—Humidity
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/50—Air quality properties
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
    • F24F11/63—Electronic processing
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/10—Temperature
    • F24F2110/12—Temperature of the outside air
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/20—Humidity
    • F24F2110/22—Humidity of the outside air
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/30—Velocity
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/30—Velocity
    • F24F2110/32—Velocity of the outside air
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/50—Air quality properties
    • F24F2110/64—Airborne particle content

Abstract

The present invention provides a kind of heating ventilation air-conditioning system automation control method based on genetic algorithm and depth B P neural network algorithms, to solve the problems, such as that heating ventilation air-conditioning system day-to-day operation process sees active Energy Saving Control, and the body-sensing experience for improving intelligent building indoor environment comfort level.The present invention builds the BP neural network based on intelligent building Monitoring Data, in order to improve system control effect, realize predictable precise control, by the use of heating ventilation air-conditioning system day-to-day operation data as training sample, carry out neural metwork training, heat-air conditioner energy-saving control forecasting model is set up, heating ventilation air-conditioning system is solved in itself because operating mode is complicated, the modeling problem with the complex control characteristic such as non-linear, large time delay and ambient influnence；Genetic algorithm is introduced into solve the problems, such as to be absorbed in local optimum in BP neural network training process and convergence is slow；The present invention carries out operational factor and adjusts control using neutral net, and the purpose of environment sensing and energy-conservation is actively improved to reach heating ventilation air-conditioning system.

Description

Heat-air conditioner energy-saving control based on genetic algorithm and depth B P neural network algorithms Method
Technical field
The present invention relates to intelligent building field, and in particular to the Automated condtrol of heating ventilation air-conditioning system in intelligent building, profit With genetic algorithm and depth B P neural network algorithms realizing the side to hvac equipment intelligentized control method and energy-conservation in building Method.
Background technology
Genetic algorithm (Genetic Algorithm) is natural selection and the science of heredity machine for simulating Darwinian evolutionism The computation model of the biological evolution process of reason, is a kind of method by simulating natural evolution process searches optimal solution.
Artificial neural network (Artificial Neural Network, i.e. ANN), it is from information processing angle to human brain Neuroid carries out abstract, the simple analog model of foundation, and different networks are constituted by different connected modes.More than nearest ten Nian Lai, the research work of artificial neural network deepens continuously, and has been achieved for very big progress, and it is in pattern-recognition, intelligent machine Device people, automatically control, predicted estimate, biology, medical science, economic dispatch field have successfully solved many modern computers and have been difficult to solve Practical problem certainly, shows good intelligent characteristic.
BP (Back Propagation, backpropagation) neutral net is a kind of by many of Back Propagation Algorithm training Layer feedforward network, is one of current most widely used neural network model.BP networks can learn and store substantial amounts of input-defeated Exit pattern mapping relations, and without disclosing the math equation of this mapping relations of description in advance.Its learning rules are to use ladder Degree descent method, the weights and threshold value of network are constantly adjusted by backpropagation, make the error sum of squares of network minimum.BP nerves Network model topological structure includes input layer (input), hidden layer (hidden layer) and output layer (output layer).
Sigmoid functions, are a function for common S types in biology, also referred to as S sigmoid growth curves.In information In science, due to properties such as its list increasing and the increasings of inverse function list, Sigmoid functions are often used as the threshold function table of neutral net, By between variable mappings to 0,1.
BAS (Building Automation System), i.e. building automation system or equipments of building automation system, It is by the electric power in building or groups of building, illumination, air-conditioning, plumbing, fire-fighting, transport, security personnel, garage management equipment or is System, the integrated system constituted for the purpose of centralized watch, control and management.
IBMS (Intelligent Building Management System), i.e. intelligent building/building management system. IBMS is the building collection for further realizing higher level with communications network system, information network system on the basis of BAS Into management system.The core of IBMS is that each subsystem in building construction is integrated into the integrated system of " organic ", and it connects Mouth Interface Standard, standardization, the information exchange and communications protocol for completing each subsystem are changed, and realize five function collection of aspect Into：The integrated and integrated management of all subsystem informations, centralized watch and control to all subsystems, the pipe of clobal Reason, process automation management.The final function of realizing centralized watch control and integrated management.
HVAC (Heating, Ventilation and Air Conditioning, abbreviation HVAC) refer to it is indoor or The system or relevant device of the responsible heating of in-car, ventilation and air adjustment.
With the development of society, intelligent building conservation is global megatrend and main trend, is 21 century Chinese architecture One emphasis and focus of career development, while be also an urgent demand of Chinese reform and development, and energy-conservation and environmental protection are to realize The key of sustainable development.From sustained development theory, the crucial of building energy conservation is again to improve energy efficiency, therefore should Using raising energy efficiency as building energy conservation the starting point, realize intelligent building highly modernize, high comfort while energy Realize that energy resource consumption is greatly lowered, with the purpose for reaching building construction energy efficiency He cut operating costs.
Heating ventilation air-conditioning system is widely used in intelligent building, and it is larger in the overall energy consumption proportion of building, exceedes 25%.And the current energy saving means of heating ventilation air-conditioning system are limited, it is typically only possible by rational engineering design, improves building guarantor Warm nature energy, improve the means such as controlled level to realize the purpose of energy-conservation.
The control mode of current heating ventilation air-conditioning system has two kinds：Manually control and arrange parameter are automatically controlled.This two The experience that kind control mode is all based on people is pre-set, in actual moving process, heating ventilation air-conditioning system Energy Saving Control Perceptually there is very big subjectivity random and limitation with environment body-sensing comfort level.
The content of the invention
Based on above mentioned problem, the present invention proposes to lift heating ventilation air-conditioning system based on genetic algorithm and neural network algorithm The quantized data that some links such as Based Intelligent Control level, sensor, actuator, controller to heating ventilation air-conditioning system are collected enters Row analysis and prediction, so that heating ventilation air-conditioning system Optimum operating control parameter is exported, it is continuable that operating system is carried out Control adjustment, reaches the purpose of optimal overall efficiency, environmental degree of comfort sensing optimal and active energy-conservation.
The present invention provides a kind of heating ventilation air-conditioning system automation control based on genetic algorithm and depth B P neural network algorithms Method processed, to solve the problems, such as heating ventilation air-conditioning system day-to-day operation process active Energy Saving Control, and improves ring in intelligent building room The body-sensing experience of border comfort level.
The present invention adopts the following technical scheme that realization：
A kind of heating ventilation air-conditioning system automation control method based on genetic algorithm and depth B P neural network algorithms, passes through The control parameter and environmental index parameter of the heating ventilation air-conditioning system collected to IBMS carry out data quantization, utilize the data as Training sample, carries out neural metwork training, sets up heat-air conditioner energy-saving control forecasting model, using the characteristic reality of BP neural network The Complex Modeling of existing heating ventilation air-conditioning system；Genetic algorithm is wherein introduced into solve to be absorbed in local optimum in BP neural network training process The problem slow with convergence；
After the completion of neural metwork training, IBMS passes through air-conditioning system configuration data and the environmental monitoring data for collecting, From amenity and the angle of energy-conservation, periodically attempt calculating air conditioner operation parameters, and be input in neutral net It is predicted, according to the feedback result of neutral net, IBMS carries out automatic operating adjustment, is finally reached to heating ventilation air-conditioning system Optimal and active energy-conservation the purpose of environment sensing.
Comprise the following steps that；
Step S201, IBMS system completes number by collecting current HVAC service data and environmental monitoring data According to normalized；
Step S202, IBMS system calculates HVAC combustion adjustment of lower a moment ginseng from environmental degree of comfort and Energy Angle Number；
Step S203, combustion adjustment parameter of lower a moment is input to carries out validity in the neutral net for having trained and comment Estimate；
Step S204, the neutral net for having trained output assessment result, if assessment result is not up to desirable level, IBMS systems recalculate adjusting parameter, re-start assessment；
Step S205, HVAC combustion adjustment parameter evaluation reach desirable level, IMBS systems then output control signal, Operational factor adjustment is carried out to HVAC；
Step S206, heating ventilation air-conditioning system combustion adjustment control flow terminate.
The process of the neural metwork training includes：
Determine |input paramete；
It is determined that after |input paramete, carrying out |input paramete normalized, deviation standardization, is the line to initial data Property conversion, numerical value is mapped between [0-1.0]；Transfer function is as follows：
Wherein, max is sample data codomain maximum, and min is sample data codomain minimum value, and x samples for sample data Value, x* is the data value after linear transformation；
After normalized, start to build the neutral net on basis, determine the neutral net number of plies and each layer nerve The quantity of unit；
It is determined that after complete each layer of neuronal quantity, setting excitation function：
Wherein, e is natural constant；
The amount of bias Bias values of each neuron, the i.e. default weight value of each neuron are set；
One group of weight coefficient of satisfaction requirement is found by learning using genetic algorithm, makes given error function minimum； In learning process, judge whether the error amount of network meets requirement and use following computational methods：
Wherein, tjIt is desired value, yjIt is the output valve of each neuron, EpBe an error for nervous layer and, E is whole The error of neutral net and；
By the study of many wheels, deconditioning is just started until error reaches expected accuracy.
The Weight Training step adjusted by genetic algorithm between the initial neuron of BP neural network is as follows；
Step S102, BP neural network parameter initialization, are normalized to sample data, again to output after training Carry out renormalization treatment, training sample, by using genetic algorithm and depth B P Neural Network Predictions after normalized Model is trained；
Step S103, initialization population, population is obtained using roulette, and Population Size can determine based on experience value, base Present principles are that calculating is sampled to sample, and the purpose of optimization is reached by crossover operation；
Step S104, calculating Population adaptation value, the error between prediction data and expected data is used as fitness function；
Step S105, crossover operation, hybrid rate are adjusted based on experience value, and are determined by testing；
In step S106, mutation operation, aberration rate is adjusted based on experience value, and is determined by testing；
Step S107, circulation are selected, intersected, being made a variation, being calculated adaptation Value Operations, until reaching evolution number of times, are obtained Optimal initial weight；
Step S108, acquisition neutral net initial weight, using genetic algorithm initialization weight, threshold value, work as genetic algorithm When having performed a generation, the weight of a new generation reinserts neutral net；
Step S109, the output valve for calculating hidden layer, exporting node layer；
Step S110, calculating hidden layer, the output error value of output layer；
Step S111, using error calculation method calculation error, judge whether to reach acceptable error range, be subjected to Error then train and terminate；
Step S112, constantly adjust each layer connection weight and threshold value；
Step S113, neural metwork training terminate.
Control method described in the invention needs to be integrated with intelligent building management system (hereinafter referred to as IBMS), whole Individual invention is divided into two parts：One is that heating ventilation air-conditioning system day-to-day operation data and indoor and outdoor surroundingses monitoring number are obtained from IBMS According to, using obtain sample data neutral net is trained, period can using genetic algorithm carry out weights it is miscellaneous plus variation, enter Row Multiple Training is until neutral net is ripe；Two is IBMS according to obtaining heating ventilation air-conditioning system in the neutral net for having trained maturation Control assessment parameter, adjustment is controlled to heating ventilation air-conditioning system operational factor.
Due to heating ventilation air-conditioning system operating mode complexity in itself, with the complicated control such as non-linear, large time delay and ambient influnence Characteristic processed so that basis modeling is extremely difficult, in order to improve system control effect, realizes predictable precise control, the present invention By the use of heating ventilation air-conditioning system day-to-day operation data as training sample, neural metwork training is carried out, set up heat-air conditioner energy-saving control Forecast model processed.
Realize heating ventilation air-conditioning system automation control method in intelligent building using the method described in the present invention, with Under several advantages：
(1) forecast model is trained based on heating ventilation air-conditioning system day-to-day operation data and environmental monitoring data, realizes air-conditioning Full automation Based Intelligent Control, and preferable environmental degree of comfort can be obtained perceive.
(2) realize that the automatic control air conditioner of HVAC is opened, closed, automatically adjust temperature, humidity, ventilation value, can reach To preferable energy-saving effect.
Brief description of the drawings
Fig. 1 is the relation schematic diagram in embodiment between neural computing unit and IBMS systems；
Fig. 2 trains flow chart to combine the BP neural network of genetic algorithm in embodiment；
Fig. 3 realizes intelligent control to assess heating ventilation air-conditioning system control parameter validity using neutral net in embodiment The flow chart of system.
Specific embodiment
Highly preferred embodiment of the present invention needs to use IBMS systems, and IBMS systems will have heating ventilation air-conditioning system data to adopt Collection is accessed and control ability.Wherein IBMS can gather the data such as heating ventilation air-conditioning system, environment temperature, humidity detection.The present invention The control method of proposition belongs to the part in IBMS systems, with the relation of IBMS as shown in Figure 1.In having this implementation process also Need to have BP neural network and the corresponding rudimentary knowledge of genetic algorithm.
The present invention proposes that heating ventilation air-conditioning system control method is mainly every by the heating ventilation air-conditioning system collected to IBMS Control parameter and environmental index parameter carry out data quantization, and the complexity of heating ventilation air-conditioning system is realized using the characteristic of BP neural network Modeling.
First, |input paramete is determined.IBMS systems have been usually directed to HVAC empty in building construction information system management is carried out Adjusting system control parameter, environmental monitoring data, season (spring, summer, autumn, winter), period, people stream counting etc..Wherein HVAC control Parameter processed mainly includes current air-conditioning temperature, blasting humidity, air-supply wind speed；Environmental monitoring data mainly includes outdoor temp Degree, indoor temperature, outside humidity, indoor humidity, outdoor wind speed, indoor air velocity, gas concentration lwevel, outdoor PM2.5 concentration, room Interior PM2.5 concentration.
It is determined that after |input paramete, carrying out |input paramete normalized, deviation standardization, is the line to initial data Property conversion, numerical value is mapped between [0,1.0].Transfer function is as follows：
In transfer function, max is sample data codomain maximum, and min is sample data codomain minimum value, and x is sample number According to sample value.After completing to calculate, x* is the data value after linear transformation, and codomain is in [0,1.0] interval.
For the conversion of discontinuous change input variable, such as season, can be quantified as：[spring, summer, autumn, winter]=[0.0, 0.33,0.66,1.0]；24 hours can be quantified as whole day by the period：[0,1/24,…,23/24].Use above-mentioned input variable Transform method come complete input variable numerical transformation work.
It is determined that after |input paramete, starting to build the neutral net on basis.Select to build 3 layers of nerve net in implementation process Network, includes 1 input layer, 1 output layer and 1 hidden layer.The selection of the neutral net number of plies under normal circumstances needs to be based on Experience judges, for facility on describing, it is 3 that the number of plies is selected here.
It is determined that after the neutral net number of plies, starting to determine the quantity of each layer neuron.Input layer god is selected in the present embodiment It is 5 through first quantity, because the |input paramete respectively obtained from IBMS systems has been divided into 5 classes totally 15 quantizating index：HVAC system System control parameter, environmental monitoring data, season (spring, summer, autumn, winter), period, people stream counting, therefore basis in the present embodiment Input data categorizing selection input layer quantity is 5.The selection of output layer neuron number is 3, then middle hidden layer Neuron number is 4, and computational methods can use following empirical equation：
Wherein, in is input layer number, and out is output layer neuron number, and h is individual for hidden layer neuron Number；
It is determined that being that each nerve sets Sigmoid excitation functions after complete each layer of neuronal quantity.BP neural network Sigmoid functions or linear function are generally used as excitation function, can be in the hope of BP using Sigmoid functions or its derivative The summation of certain neuron, desired value and error amount in neutral net.Formula is described as follows：
Wherein, e is natural constant e (about 2.71828), and the codomain of f (x) is interval [0,1] after calculating, is judged with this Neuron is state of activation or holddown.
The amount of bias Bias values of each neuron are set, that is, each neuron default weight value.Neutral net Essence is a non-linear expressions for complexity, and bias is equivalent to its zero degree term coefficient.1 is set in the present embodiment.
So far can set up substantially and contain a BP neural network for the standard of hidden layer.But standard BP neural network Have the following disadvantages：Convergence rate is slow, is absorbed in local optimum, it is difficult to determine the implicit number of plies and hidden layer neuron number.Therefore Each layer weight is hybridized present invention introduces genetic algorithm, it is complete to reach Fast Convergent and solve the problems, such as that local optimum is realized The optimal purpose of office.Because being related to the complexity of heating ventilation air-conditioning system, it is impossible to determine the concrete structure and weight of neutral net Coefficient, can only be by obtaining a model for meeting day-to-day operation demand after study, its cardinal principle is, by study, to find One group of weight coefficient of satisfaction requirement, makes given error function minimum.In sample data learning process, the mistake of network is judged Whether difference meets requirement mainly uses following computational methods：
Wherein, tjIt is the desired value of each neuron, yjBe the output valve of each neuron, m be each layer of neuron Number, p is the number of plies of neutral net, EpFor an error for nervous layer and, E be whole neutral net error and.
By the study of many wheels, deconditioning is just started until error reaches expected accuracy.
After the completion of network training, during heating ventilation air-conditioning system day-to-day operation, IBMS systems pass through the sky for collecting Adjusting system configuration data and environmental monitoring data, from amenity and the angle of energy-conservation, periodically attempt calculating empty Allocation and transportation line parameter, and be input in neutral net and be predicted, according to the feedback result of neutral net, IBMS is to HVAC system System carries out automatic operating adjustment, is finally reached optimal and active energy-conservation the purpose of environment sensing.
Next, just two idiographic flows of the invention are described：
The Weight Training flow S100 adjusted by genetic algorithm between the initial neuron of BP neural network in the present invention As shown in Fig. 2 flow S100 originates in S101.
In step s 102, BP neural network parameter initialization, is normalized to sample data, right again after training Output carries out renormalization treatment, training sample, by using genetic algorithm and depth B P neural network algorithms after normalized Forecast model is trained.
In step s 103, population is initialized, population is obtained using roulette, Population Size can be true based on experience value Fixed, general principle is that calculating is sampled to sample, and the purpose of optimization is reached by crossover operation.
In step S104, Population adaptation value is calculated, the error between prediction data and expected data is used as fitness letter Number.
In step S105, crossover operation, hybrid rate can be adjusted based on experience value, and be determined by testing.
In step s 106, mutation operation, aberration rate can be adjusted based on experience value, and be determined by testing.
In step s 107, circulation is selected, intersected, being made a variation, being calculated adaptation Value Operations, until reaching evolution number of times, Obtain optimal initial weight.
In step S108, neutral net initial weight is obtained, using genetic algorithm initialization weight, threshold value, work as heredity During the complete generation of algorithm performs, the weight of a new generation reinserts neutral net, and weight can be evolved into arbitrary size, not by any The limitation of form.
In step S109, hidden layer, the output valve of output node layer are calculated.
In step s 110, hidden layer, the output error value of output layer are calculated.
In step S111, using error calculation method calculation error, judge whether to reach acceptable error range, can The error of receiving is then trained and terminated.
In step S112, each layer connection weight and threshold value are constantly adjusted.
In step S113, neural metwork training terminates.
So far, the BP neural network training flow that genetic algorithm is combined in the present invention is described.
The IBMS of realization of the invention assesses heating ventilation air-conditioning system control parameter validity by neutral net, realizes intelligence Change the flow of control method as shown in figure 3, starting from S200.
In step s 201, IBMS systems are by collecting current HVAC service data and environmental monitoring data and complete Into data normalization treatment.
In step S202, IBMS systems calculate the operation of HVAC of lower a moment and adjust from environmental degree of comfort and Energy Angle Whole parameter.
In step S203, combustion adjustment parameter of lower a moment is input in the neutral net for having trained carries out validity Assessment.
In step S204, the neutral net output assessment result for having trained, if assessment result is not up to preferable water Flat, then IBMS systems recalculate adjusting parameter, re-start assessment.
In step S205, HVAC combustion adjustment parameter evaluation reaches desirable level, IMBS systems then output control Signal, operational factor adjustment is carried out to HVAC.
In step S206, heating ventilation air-conditioning system combustion adjustment control flow terminates.
So far, heating ventilation air-conditioning system control parameter validity is assessed using neutral net in the present embodiment, is realized intelligent The flow of control is described.
Using the method described in the present invention, using heating ventilation air-conditioning system day-to-day operation data as training sample, carry out Neural metwork training, sets up heat-air conditioner energy-saving control forecasting model, finally realizes that heating ventilation air-conditioning system is automatic in intelligent building Change control method, with following advantage：
(1) forecast model is trained based on heating ventilation air-conditioning system day-to-day operation data and environmental monitoring data, realizes air-conditioning Full automation Based Intelligent Control, and preferable environmental degree of comfort can be obtained perceive.
(2) realize that the automatic control air conditioner of HVAC is opened, closed, automatically adjust temperature, humidity, ventilation value, can reach To preferable energy-saving effect.
The description of above-described embodiment flow only for clear explanation basic skills of the invention, but on the present invention is not limited in State embodiment；Every any simple modification, equivalent variations and modification made according to embodiment in technical spirit of the invention, Fall within the protection domain of technical scheme.

Claims (4)

1. a kind of heating ventilation air-conditioning system automation control method based on genetic algorithm and depth B P neural network algorithms, its feature It is：Data quantization is carried out by the control parameter and environmental index parameter of the heating ventilation air-conditioning system collected to IBMS, is utilized The data carry out neural metwork training as training sample, heat-air conditioner energy-saving control forecasting model are set up, using BP nerve nets The characteristic of network realizes the Complex Modeling of heating ventilation air-conditioning system；Genetic algorithm is wherein introduced into solve to be fallen into BP neural network training process Enter local optimum and the slow problem of convergence；
After the completion of neural metwork training, IBMS passes through air-conditioning system configuration data and the environmental monitoring data for collecting, from ring The angle of border comfortableness and energy-conservation is set out, and periodically attempts calculating air conditioner operation parameters, and is input in neutral net and is carried out Prediction, according to the feedback result of neutral net, IBMS carries out automatic operating adjustment, is finally reached environment to heating ventilation air-conditioning system Perceive the purpose of optimal and active energy-conservation.
2. a kind of heating ventilation air-conditioning system based on genetic algorithm and depth B P neural network algorithms according to claim 1 from Dynamicization control method, it is characterised in that：Comprise the following steps that；
Step S201, IBMS system is returned by collecting current HVAC service data and environmental monitoring data, and completing data One change is processed；
Step S202, IBMS system calculates HVAC combustion adjustment parameter of lower a moment from environmental degree of comfort and Energy Angle；
Step S203, combustion adjustment parameter of lower a moment is input in the neutral net for having trained carries out efficiency assessment；
Step S204, the neutral net for having trained output assessment result, if assessment result is not up to desirable level, IBMS System recalculates adjusting parameter, re-starts assessment；
Step S205, HVAC combustion adjustment parameter evaluation reach desirable level, IMBS systems then output control signal, to warm Logical air-conditioning carries out operational factor adjustment；
Step S206, heating ventilation air-conditioning system combustion adjustment control flow terminate.
3. a kind of heating ventilation air-conditioning system based on genetic algorithm and depth B P neural network algorithms according to claim 2 from Dynamicization control method, it is characterised in that；The process of the neural metwork training includes：
Determine |input paramete；
It is determined that after |input paramete, carrying out |input paramete normalized, deviation standardization, is the linear change to initial data Change, numerical value is mapped between [0-1.0]；Transfer function is as follows：
x * = x - m i n max - m i n
Wherein, max is sample data codomain maximum, and min is sample data codomain minimum value, and x is sample data sample value, x* It is the data value after linear transformation；
After normalized, start to build the neutral net on basis, determine the neutral net number of plies and each layer neuron Quantity；
It is determined that after complete each layer of neuronal quantity, setting excitation function：
f ( x ) = 1 1 + e - x
Wherein, e is natural constant；
The amount of bias Bias values of each neuron, the i.e. default weight value of each neuron are set；
One group of weight coefficient of satisfaction requirement is found by learning using genetic algorithm, makes given error function minimum；Learning During habit, judge whether the error amount of network meets requirement and use following computational methods：
E = Σ p = 1 p E p = Σ j = 1 m 1 2 [ t j - y j ] 2
Wherein, tjIt is desired value, yjIt is the output valve of each neuron, EpBe an error for nervous layer and, E is whole nerve The error of network and；
By the study of many wheels, deconditioning is just started until error reaches expected accuracy.
4. a kind of heating ventilation air-conditioning system based on genetic algorithm and depth B P neural network algorithms according to claim 3 from Dynamicization control method, it is characterised in that：The Weight Training adjusted by genetic algorithm between the initial neuron of BP neural network is walked It is rapid as follows；
Step S102, BP neural network parameter initialization, are normalized to sample data, and output is carried out again after training Renormalization treatment, training sample, by using genetic algorithm and depth B P Neural Network Prediction models after normalized It is trained；
Step S103, initialization population, population is obtained using roulette, and Population Size can determine based on experience value, substantially former Reason is that calculating is sampled to sample, and the purpose of optimization is reached by crossover operation；
Step S104, calculating Population adaptation value, the error between prediction data and expected data is used as fitness function；
Step S105, crossover operation, hybrid rate are adjusted based on experience value, and are determined by testing；
In step S106, mutation operation, aberration rate is adjusted based on experience value, and is determined by testing；
Step S107, circulation are selected, intersected, being made a variation, being calculated adaptation Value Operations, until reaching evolution number of times, obtain optimal Initial weight；
Step S108, acquisition neutral net initial weight, using genetic algorithm initialization weight, threshold value, when genetic algorithm is performed During a complete generation, the weight of a new generation reinserts neutral net；
Step S109, the output valve for calculating hidden layer, exporting node layer；
Step S110, calculating hidden layer, the output error value of output layer；
Step S111, using error calculation method calculation error, judge whether to reach acceptable error range, acceptable mistake Poor then training terminates；
Step S112, constantly adjust each layer connection weight and threshold value；
Step S113, neural metwork training terminate.
CN201611227073.XA 2016-12-27 2016-12-27 Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms Pending CN106765959A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611227073.XA CN106765959A (en) 2016-12-27 2016-12-27 Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611227073.XA CN106765959A (en) 2016-12-27 2016-12-27 Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms

Publications (1)

Publication Number Publication Date
CN106765959A true CN106765959A (en) 2017-05-31

Family

ID=58921635

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611227073.XA Pending CN106765959A (en) 2016-12-27 2016-12-27 Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms

Country Status (1)

Country Link
CN (1) CN106765959A (en)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107044710A (en) * 2016-12-26 2017-08-15 深圳达实智能股份有限公司 Energy-saving control method for central air conditioner and system based on joint intelligent algorithm
CN107655160A (en) * 2017-11-27 2018-02-02 中山路得斯空调有限公司 A kind of central air-conditioning regulating system based on neural network prediction
CN108361927A (en) * 2018-02-08 2018-08-03 广东美的暖通设备有限公司 A kind of air-conditioner control method, device and air conditioner based on machine learning
CN108388175A (en) * 2018-02-24 2018-08-10 浙江盾安自控科技有限公司 Self-optimizing efficiency managing and control system based on artificial neural network and method
CN109946968A (en) * 2019-03-27 2019-06-28 北京快电科技有限公司 Distributed energy matching system and method in a kind of building
CN109959123A (en) * 2019-03-11 2019-07-02 浙江工业大学 A kind of energy-saving method for air conditioner based on genetic algorithm and shot and long term memory Recognition with Recurrent Neural Network
CN110069690A (en) * 2019-04-24 2019-07-30 成都市映潮科技股份有限公司 A kind of theme network crawler method, apparatus and medium
CN110094837A (en) * 2019-04-30 2019-08-06 珠海格力电器股份有限公司 Intelligent control apparatus for air-conditioner and method
CN110805997A (en) * 2019-11-14 2020-02-18 中金新源（天津）科技有限公司 Energy-saving control method for central air-conditioning system
CN112212481A (en) * 2019-07-10 2021-01-12 中华电信股份有限公司 System and method for controlling environmental comfort by deep reinforcement learning
CN112443943A (en) * 2019-08-30 2021-03-05 珠海格力电器股份有限公司 Model training method based on small amount of labeled data, control system and air conditioner
CN113011080A (en) * 2020-12-22 2021-06-22 浙江农林大学 Negative oxygen ion concentration inversion method
CN113654143A (en) * 2021-08-18 2021-11-16 四川省建筑科学研究院有限公司 Air-conditioning water system pump valve linkage control method based on neural network
CN113758520A (en) * 2021-08-18 2021-12-07 贵州众创仪云科技有限公司 Internet of things laboratory environment monitoring method and system based on neural network
CN115654697B (en) * 2022-11-21 2023-04-25 四川旷谷信息工程有限公司 Temperature control method and device for semi-enclosed space and computer readable storage medium
CN116383676A (en) * 2023-06-05 2023-07-04 北京建筑材料检验研究院股份有限公司 Intelligent detection method for heating and ventilation equipment based on data processing
CN116558082A (en) * 2023-06-16 2023-08-08 广东新拓展建筑工程有限公司 Energy-saving ventilation system
CN113011080B (en) * 2020-12-22 2024-04-19 浙江农林大学 Negative oxygen ion concentration inversion method

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11108415A (en) * 1997-10-07 1999-04-23 Sharp Corp Control device for air-conditioner
CN102980272A (en) * 2012-12-08 2013-03-20 珠海派诺科技股份有限公司 Air conditioner system energy saving optimization method based on load prediction
CN104008427A (en) * 2014-05-16 2014-08-27 华南理工大学 Central air conditioner cooling load prediction method based on BP neural network
CN104484715A (en) * 2014-11-28 2015-04-01 江苏大学 Neural network and particle swarm optimization algorithm-based building energy consumption predicting method
CN104633829A (en) * 2013-11-06 2015-05-20 上海思控电气设备有限公司 Building cooling station energy-saving control device and method thereof

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11108415A (en) * 1997-10-07 1999-04-23 Sharp Corp Control device for air-conditioner
CN102980272A (en) * 2012-12-08 2013-03-20 珠海派诺科技股份有限公司 Air conditioner system energy saving optimization method based on load prediction
CN104633829A (en) * 2013-11-06 2015-05-20 上海思控电气设备有限公司 Building cooling station energy-saving control device and method thereof
CN104008427A (en) * 2014-05-16 2014-08-27 华南理工大学 Central air conditioner cooling load prediction method based on BP neural network
CN104484715A (en) * 2014-11-28 2015-04-01 江苏大学 Neural network and particle swarm optimization algorithm-based building energy consumption predicting method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
赵妍: "《面向大数据的挖掘方法研究》", 31 July 2016, 电子科技大学出版社 *

Cited By (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107044710A (en) * 2016-12-26 2017-08-15 深圳达实智能股份有限公司 Energy-saving control method for central air conditioner and system based on joint intelligent algorithm
CN107655160A (en) * 2017-11-27 2018-02-02 中山路得斯空调有限公司 A kind of central air-conditioning regulating system based on neural network prediction
CN108361927A (en) * 2018-02-08 2018-08-03 广东美的暖通设备有限公司 A kind of air-conditioner control method, device and air conditioner based on machine learning
CN108388175A (en) * 2018-02-24 2018-08-10 浙江盾安自控科技有限公司 Self-optimizing efficiency managing and control system based on artificial neural network and method
CN109959123A (en) * 2019-03-11 2019-07-02 浙江工业大学 A kind of energy-saving method for air conditioner based on genetic algorithm and shot and long term memory Recognition with Recurrent Neural Network
CN109946968A (en) * 2019-03-27 2019-06-28 北京快电科技有限公司 Distributed energy matching system and method in a kind of building
CN109946968B (en) * 2019-03-27 2022-03-25 北京快电科技有限公司 Distributed energy matching system and method in building
CN110069690A (en) * 2019-04-24 2019-07-30 成都市映潮科技股份有限公司 A kind of theme network crawler method, apparatus and medium
CN110094837A (en) * 2019-04-30 2019-08-06 珠海格力电器股份有限公司 Intelligent control apparatus for air-conditioner and method
CN112212481A (en) * 2019-07-10 2021-01-12 中华电信股份有限公司 System and method for controlling environmental comfort by deep reinforcement learning
CN112443943A (en) * 2019-08-30 2021-03-05 珠海格力电器股份有限公司 Model training method based on small amount of labeled data, control system and air conditioner
CN110805997A (en) * 2019-11-14 2020-02-18 中金新源（天津）科技有限公司 Energy-saving control method for central air-conditioning system
CN113011080A (en) * 2020-12-22 2021-06-22 浙江农林大学 Negative oxygen ion concentration inversion method
CN113011080B (en) * 2020-12-22 2024-04-19 浙江农林大学 Negative oxygen ion concentration inversion method
CN113654143A (en) * 2021-08-18 2021-11-16 四川省建筑科学研究院有限公司 Air-conditioning water system pump valve linkage control method based on neural network
CN113758520A (en) * 2021-08-18 2021-12-07 贵州众创仪云科技有限公司 Internet of things laboratory environment monitoring method and system based on neural network
CN115654697B (en) * 2022-11-21 2023-04-25 四川旷谷信息工程有限公司 Temperature control method and device for semi-enclosed space and computer readable storage medium
CN116383676A (en) * 2023-06-05 2023-07-04 北京建筑材料检验研究院股份有限公司 Intelligent detection method for heating and ventilation equipment based on data processing
CN116383676B (en) * 2023-06-05 2023-08-25 北京建筑材料检验研究院股份有限公司 Intelligent detection method for heating and ventilation equipment based on data processing
CN116558082A (en) * 2023-06-16 2023-08-08 广东新拓展建筑工程有限公司 Energy-saving ventilation system
CN116558082B (en) * 2023-06-16 2023-11-17 广东新拓展建筑工程有限公司 Energy-saving ventilation system

Similar Documents

Publication Publication Date Title
CN106765959A (en) Heat-air conditioner energy-saving control method based on genetic algorithm and depth B P neural network algorithms
CN102629106B (en) Water supply control method and water supply control system
CN107909220A (en) Electric heating load prediction method
CN106920006A (en) A kind of subway station air conditioning energy consumption Forecasting Methodology based on ISOA LSSVM
CN106951611A (en) A kind of severe cold area energy-saving design in construction optimization method based on user's behavior
CN107909149A (en) A kind of Temperature in Greenhouse Forecasting Methodology based on Genetic BP Neutral Network
CN112415924A (en) Energy-saving optimization method and system for air conditioning system
CN108921339A (en) Genetic Support Vector Machine photovoltaic power interval prediction method based on quantile estimate
CN109164707A (en) A kind of indoor environment negative-feedback regu- lation system based on artificial neural network algorithm
CN105320184A (en) Intelligent monitoring system of indoor environment of building
CN110443418A (en) Urban water consumption prediction technique based on GA-BP neural network
CN106371316B (en) Water island dosing On-Line Control Method based on PSO LSSVM
CN114322199B (en) Digital twinning-based ventilation system autonomous optimization operation regulation and control platform and method
Song et al. An indoor temperature prediction framework based on hierarchical attention gated recurrent unit model for energy efficient buildings
CN113033917A (en) Sewage treatment plant prediction planning operation management method based on peripheral data
Wu et al. Intelligent optimization framework of near zero energy consumption building performance based on a hybrid machine learning algorithm
CN107818340A (en) Two-stage Air-conditioning Load Prediction method based on K value wavelet neural networks
CN114811713B (en) Two-level network inter-user balanced heat supply regulation and control method based on mixed deep learning
CN115423301A (en) Intelligent electric power energy management and control method, device and system based on Internet of things
CN113743538A (en) Intelligent building energy consumption prediction method, equipment and medium based on IPSO-BP neural network
CN109242136A (en) A kind of micro-capacitance sensor wind power Chaos-Genetic-BP neural network prediction technique
CN113237135B (en) SD-LSTMs-based apartment independent heating system heat demand prediction and energy-saving control method
CN101833281A (en) Control method for saving energy of aeration in sewage treatment
CN112070317A (en) Hotel air conditioner energy consumption prediction method
Ali et al. Building power control and comfort management using genetic programming and fuzzy logic

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20170531

RJ01 Rejection of invention patent application after publication