CN110850717A - Neural network heat pump defrosting control device and control method by utilizing fan current - Google Patents
Neural network heat pump defrosting control device and control method by utilizing fan current Download PDFInfo
- Publication number
- CN110850717A CN110850717A CN201911132291.9A CN201911132291A CN110850717A CN 110850717 A CN110850717 A CN 110850717A CN 201911132291 A CN201911132291 A CN 201911132291A CN 110850717 A CN110850717 A CN 110850717A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- defrosting
- current
- heat pump
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F25—REFRIGERATION OR COOLING; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS; MANUFACTURE OR STORAGE OF ICE; LIQUEFACTION SOLIDIFICATION OF GASES
- F25B—REFRIGERATION MACHINES, PLANTS OR SYSTEMS; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS
- F25B30/00—Heat pumps
- F25B30/06—Heat pumps characterised by the source of low potential heat
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F25—REFRIGERATION OR COOLING; COMBINED HEATING AND REFRIGERATION SYSTEMS; HEAT PUMP SYSTEMS; MANUFACTURE OR STORAGE OF ICE; LIQUEFACTION SOLIDIFICATION OF GASES
- F25D—REFRIGERATORS; COLD ROOMS; ICE-BOXES; COOLING OR FREEZING APPARATUS NOT OTHERWISE PROVIDED FOR
- F25D21/00—Defrosting; Preventing frosting; Removing condensed or defrost water
- F25D21/06—Removing frost
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
- G05B13/027—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Mechanical Engineering (AREA)
- Thermal Sciences (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Combustion & Propulsion (AREA)
- Physiology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Chemical & Material Sciences (AREA)
- Air Conditioning Control Device (AREA)
Abstract
The invention provides a neural network heat pump defrosting control device and a control method by utilizing fan current, wherein the method comprises the following steps: s1, the current measuring device collects the current signal of the air source heat pump evaporator fan in real time; s2, the data conversion module filters, demodulates, decomposes and reconstructs the current signal to obtain data information, and then the time domain and frequency domain analysis is carried out by utilizing a feature extraction algorithm to extract a feature data value; s3, the data diagnosis module carries out diagnosis analysis processing on the characteristic data value by using an ELM neural network based on a GA algorithm to obtain the diagnosis information of frosting faults; and S4, the defrosting control module carries out defrosting control on the air source heat pump evaporator according to the diagnosis information. The invention utilizes the ELM neural network model based on the GA algorithm to improve the convergence speed and the convergence time and increase the stability of the system; by diagnosing the change condition of the fan current, defrosting is quickly and accurately controlled according to the diagnosis result, and the operation efficiency of the heat pump system is improved.
Description
Technical Field
The invention belongs to the technical field of heat pump defrosting, and particularly relates to a neural network heat pump defrosting control device and method utilizing fan current.
Background
The air source heat pump technology is an energy-saving and environment-friendly heating technology established based on the reverse Carnot cycle principle. The air source heat pump system obtains a low-temperature heat source through natural energy (air heat storage), and becomes a high-temperature heat source after the system efficiently collects heat and integrates, so as to supply heat or hot water. The air source heat pump has the advantages of wide application range, low operation cost, no environmental pollution and good energy-saving and emission-reducing effects, and is widely applied to the fields of chemical industry, heat energy, heating ventilation and the like.
Because the air contains moisture and has high relative humidity, when the refrigerating system normally operates, the refrigerant in the evaporator tube of the air source heat pump absorbs the heat of the outdoor air, the surface temperature of the evaporator is far lower than the dew point temperature of the air, and the moisture in the air can be separated out and condensed on the tube wall of the evaporator. When the temperature of the tube wall is lower than 0 ℃, water dew is condensed into frost, so that the evaporator is frosted. Although the air source heat pump is widely applied to urban development in China, the frosting phenomenon exists when the outdoor heat exchanger of the air source heat pump operates in winter, so that the operating condition of the air source heat pump during heating in winter is not ideal.
The frosting of the outdoor heat exchanger mainly causes adverse effects on the operation of the heat pump unit from two aspects. On one hand, the formation of the frost layer increases the surface heat conduction resistance of the outdoor heat exchanger and reduces the heat transfer coefficient of the outdoor heat exchanger; on the other hand, the existence of the frost layer increases the resistance of air flowing through the outdoor heat exchanger, and reduces the air flow, thereby reducing the heating performance of the unit. Along with the increase of the frost layer on the wall surface of the outdoor heat exchanger, the evaporation temperature of the outdoor heat exchanger is reduced, the heating capacity of a unit is reduced, the performance of a fan is attenuated, the input current is increased, the heat supply performance coefficient is reduced, and the compressor is stopped in severe cases, so that the unit cannot work normally. Therefore, the change of the current of the fan motor in the frosting state is researched, the frosting phenomenon can be found in time through the real-time monitoring of the current flowing through the fan, and the defrosting is realized in time. The existing technology for controlling defrosting based on fan current only adopts macroscopic features (amplitude and sine frequency) to compare threshold values, so that the recognition rate is low, and the control effect is poor.
Disclosure of Invention
The traditional method for controlling the defrosting of the evaporator is to measure the temperature of a coil pipe of a heat exchanger through a temperature sensor or defrost at regular time, and the defrosting control mode has low efficiency, and can not find faults in time and effectively clear the faults. The invention aims to provide a neural network heat pump defrosting control device and a control method by utilizing fan current, which can accurately and quickly judge the current state of an air source heat pump by utilizing a neural network system according to a current signal of a fan so as to find and clear faults as early as possible, reduce the low efficiency of the heat pump in a frosting state and ensure that an air source heat pump unit runs in a high-efficiency state.
In order to achieve the purpose, the invention adopts the following technical scheme:
the neural network heat pump defrosting control device using the fan current includes:
the current measuring device is used for measuring a current signal of a fan of the air source heat pump evaporator in real time;
the data conversion module is used for filtering, demodulating, decomposing and reconstructing the current signal to obtain data information, then analyzing the time domain and the frequency domain by utilizing a characteristic extraction algorithm to extract a characteristic data value,
the data diagnosis module is used for carrying out diagnosis analysis processing on the characteristic data value by utilizing an ELM neural network based on a GA algorithm to obtain the diagnosis information of the frosting fault;
and the defrosting control module is used for carrying out defrosting control on the air source heat pump evaporator according to the diagnosis information.
As one of the preferable schemes of the present invention, in the data conversion module, filtering and demodulation operations are performed by using Hilbert transform, and decomposition and reconstruction operations are performed by using wavelet packet analysis, frequency domain analysis, or CZT transform.
As one preferable aspect of the present invention, the feature extraction algorithm employs any one of wavelet packet decomposition, fast fourier transform, and principal component analysis.
The current measuring device further comprises a data acquisition module for collecting and transmitting the acquired data of the current measuring device to the data conversion module.
The invention also provides a neural network heat pump defrosting control method by utilizing the fan current, and the control device comprises the following steps:
s1, the current measuring device collects the current signal of the air source heat pump evaporator fan in real time;
s2, the data conversion module filters, demodulates, decomposes and reconstructs the current signal to obtain data information, and then the time domain and frequency domain analysis is carried out by utilizing a feature extraction algorithm to extract a feature data value;
s3, the data diagnosis module carries out diagnosis analysis processing on the characteristic data value by using an ELM neural network based on a GA algorithm to obtain the diagnosis information of frosting faults;
and S4, the defrosting control module carries out defrosting control on the air source heat pump evaporator according to the diagnosis information.
Compared with the prior art, the invention has the beneficial effects that:
the convergence speed and the convergence time are improved by utilizing the ELM neural network model based on the GA algorithm, and the stability of the system is improved. By diagnosing the change condition of the current flowing through the fan and quickly and accurately controlling defrosting according to the diagnosis result, the operation efficiency of the heat pump system is improved, mechanical damage caused by overlong frosting time is reduced, and the purpose of saving cost is achieved.
Drawings
FIG. 1 is a schematic diagram of the control device of the present invention;
fig. 2 is a flow chart of the control method of the present invention.
Reference numerals:
the system comprises an 11-air source heat pump circulating system, a 12-air source heat pump evaporator fan, a 13-alternating current power supply, a 14-data acquisition module, a 15-data conversion module, a 16-data diagnosis module, a 17-defrosting control module and a 21-current measuring device.
Detailed Description
The technical solution of the present invention will be further explained below.
Example one
As shown in fig. 1, the neural network heat pump defrosting control device using fan current in this embodiment includes an air source heat pump circulation system 11, a fan 12 of an air source heat pump evaporator, an ac power supply 13, a current measuring device 21, a data acquisition module 14, a data conversion module 15, a data diagnosis module 16, and a defrosting control module 17. The system is connected in a mode that a current measuring device 21 is connected with a motor lead of a fan 12 of an air source heat pump evaporator, the fan 12 of the air source heat pump evaporator is connected with an alternating current power supply 13, the current measuring device 21, a data acquisition module 14, a data conversion module 15, a data diagnosis module 16 and a defrosting control module 17 are sequentially connected, and the defrosting control module 17 is connected with an air source heat pump circulating system 11 and used for controlling a heat pump to defrost.
The fan 12 of the air source evaporator is used for driving outdoor air to flow through the outdoor evaporator of the air source heat pump to obtain heat in the air, if the evaporator frosts, the heating capacity is reduced, the running state of the fan changes, the input current for driving the fan to run changes, and higher harmonics are generated.
The current measuring device 21 is used for measuring the change of the fan current of the outdoor heat exchanger of the air source heat pump in real time and comprises a switching power supply and a current sensor. The current sensor can adopt various types such as electromagnetic type, electronic type and the like, and the installation mode adopts an induction type installation or an access type installation method according to the different types of the sensor and the measured current.
The data acquisition module 14 automatically acquires non-electric quantity or electric quantity signals in analog and digital measurement units of a sensor or other devices to be measured, and sends the non-electric quantity or electric quantity signals to a processor for analysis and processing. The system mainly comprises a data acquisition card and can be accessed to a data acquisition system through buses such as PXI, serial ports, USB, Ethernet or wireless networks.
The data conversion module 15 performs data preprocessing and feature extraction on the acquired original data information, the data preprocessing includes filtering, demodulating, decomposing and reconstructing the data to acquire useful data information, the feature extraction refers to performing time domain and frequency domain analysis on the data by using a feature extraction algorithm, and the feature extraction algorithm can adopt wavelet packet decomposition, fast fourier transform, principal component analysis and the like.
The data diagnosis module 16 performs signal processing on the extracted feature signals by using a machine learning intelligent learning method, and classifies data information to diagnose frosting faults, wherein the machine learning method comprises various learning methods such as an artificial neural network, deep learning, a support vector machine and the like. The invention provides an improved RBF neural network for carrying out diagnosis analysis processing on characteristic signals.
The defrosting control module 17 implements defrosting control according to the diagnostic information of the data diagnostic module, and performs defrosting if a frosting fault is diagnosed.
The defrosting control device provided by the embodiment of the invention is used for preprocessing fan current data through filtering, demodulation, decomposition, reconstruction and the like, extracting a characteristic data value through a characteristic extraction algorithm, and diagnosing the frosting condition of the outdoor heat exchanger of the air source heat pump by combining with data analysis such as an ELM neural network model based on a GA algorithm and the like, so that more accurate judgment data can be obtained, faults can be found and eliminated as soon as possible, the low efficiency of the heat pump in the frosting state is reduced, and the operation of the air source heat pump unit in the high-efficiency state is ensured.
Example two
As shown in fig. 2, the present embodiment provides a neural network heat pump defrosting control method using fan current, which at least includes the following steps:
s1, the current measuring device collects the current signal of the air source heat pump evaporator fan in real time;
s2, the data conversion module filters, demodulates, decomposes and reconstructs the current signal to obtain data information, and then the time domain and frequency domain analysis is carried out by utilizing a feature extraction algorithm to extract a feature data value;
s3, the data diagnosis module carries out diagnosis analysis processing on the characteristic data value by using an ELM neural network based on a GA algorithm to obtain the diagnosis information of frosting faults;
and S4, the defrosting control module carries out defrosting control on the air source heat pump evaporator according to the diagnosis information.
In step S2, the data conversion module performs filtering, demodulation, decomposition, and reconstruction on the current signal to obtain data information, specifically:
s2.1, carrying out Hilbert conversion on data:
s (t) is the given time domain signal, Hilbert transforms into a convolution of s (t) with h (t) 1/(tt),
in the formula, H is a Hilbert transform operator, and the transformed signal is processedProcessing;
and S2.2, carrying out decomposition and reconstruction operation by utilizing wavelet packet analysis, frequency domain analysis or CZT conversion.
In step S2.2, the wavelet packet analysis step is as follows:
1) decomposition of signals using wavelet packets
Carrying out n-layer wavelet packet decomposition on the collected current signals, and respectively extracting the nth layer from low frequency to high frequency 2nWavelet packet coefficients at the individual nodes; 2nEach node is (i, j) which represents the jth node of the ith layer, where i is n and j is 0,1,2,3 …,2n-1;
2) Reconstructing wavelet packet decomposition coefficient, and extracting signal characteristics of each frequency band range
Let each node wavelet packet coefficient Hi,jThe corresponding reconstructed signal is Si,j(ii) a Analyzing all nodes of the nth layer, the total signal S can be represented by the following formula:
3) calculating the total energy of each frequency band signal
Suppose Sn,j(j=0,1,2,3…,2n-1) corresponding energy En,j(j=0,1,2,3…,2n-1), then energy Sn,jCan be represented by the following formula:
wherein: h isj,k(j=0,1,2,3…,2n-1; k-1, 2, …, n) represents the reconstructed signal Sn,jThe amplitude of the discrete points of (a);
4) constructing feature vectors
Defining the total energy of the signal asThe relative wavelet packet energy of a certain frequency band isThen the relative wavelet packet energy eigenvector is
Step S3 is specifically as follows:
s3.1, ELM neural network initialization
Normalizing the input characteristic quantity of the ELM neural network to be between [0 and 1], wherein the normalization formula is as follows:
where X is the normalized value, X is the normalized sample datamin、XmaxRespectively the minimum value and the maximum value of the normalized data of the sample;
s3.2ELM neural network parameter initialization
Setting training times N, training errors e, a connection weight omega from an input layer to a hidden layer, a connection weight β from the hidden layer to an output layer and a threshold b of a neuron of the hidden layer, wherein an input characteristic value is an energy characteristic value x decomposed from fault current to an nth layeriThe number of neurons in the hidden layer is calculated according to the formula:determining, wherein n and l are the number of input neurons and the number of output neurons respectively, a is usually 1-10, and the output value is y;
dividing data into a test set and a prediction set;
s3.3 data testing
Importing test set data, namely fault characteristic values extracted by wavelet packet decomposition, and calculating the fault identification success rate;
s3.4GA Algorithm parameter initialization
Setting size N of population1Degree of evolution G, crossover probability PcAnd the mutation probability Pm;
Forming genetic codes by the weight values and the threshold values of the ELM neural network, repeatedly operating the neural network established in S3.3 and recording the genetic codes completed in each operation to establish a population;
s3.5 establishing an evaluation function
Calculating and recording corresponding fitness values according to the running result of the neural network constructed by each group of chromosomes;
s3.6GA Algorithm selection, crossover and mutation
1) Selecting a part of population to generate next generation population with a certain probability;
2) selecting a plurality of chromosomes from the parent generation population to carry out the operation of the next generation of chromosomes, and generating new individuals through the cross combination of the chromosomes;
3) selecting one individual from the population, and carrying out mutation on a certain code in the selected chromosome to generate more excellent individuals;
s3.7 fitness value calculation
Decoding the genetic code in the newly obtained chromosome individual, calculating the fitness and comparing the fitness with the population; repeating S3.5-S3.6, continuously performing the operations on the population and calculating the fitness value until the maximum evolution times are reached, wherein the chromosome corresponding to the fitness optimal solution is the weight and the threshold of the established ELM neural network;
s3.8 error calculation for neural networks
Calculating the difference between the predicted output and the expected output, if the error meets the specified value, completing the construction of the ELM neural network based on the GA algorithm, and if the error does not meet the specified value, returning to S3.3 to continue learning until the error meets the specified value;
3.9 diagnosis of frosting Fault
And (3) taking the feature vector obtained by decomposing the wavelet packet as the input of the ELM neural network, and identifying by the ELM neural network trained by S3.8 and based on the GA algorithm:
if the recognition result needs defrosting and the current non-defrosting mode is adopted, defrosting is started; if defrosting is needed and the current mode is the defrosting mode, the running state is unchanged;
if the recognition result does not need defrosting and the current non-defrosting mode is adopted, the running state is unchanged;
if defrosting is not needed and the current defrosting mode is already, defrosting is stopped, and the diagnosis information is transmitted to the defrosting control module.
EXAMPLE III
Compared with the second embodiment, the neural network heat pump defrosting control method using the fan current is different only in the feature extraction mode adopted in the step S2.2, and in order to reduce resource waste, for the deficiency of fourier transform, the present embodiment adopts Chirp-Z transform (CZT) transform:
for a known signal x (N), when 0< N < N-1, the CZT expression is
Equidistant sampling in a spiral fashion in the Z plane, ZkA sampling point of Z, expressed as
zk=AW-k,k=0,1……M-1(2)
Where M is the desired number of analysis points, M is not necessarily equal to n, A and M are arbitrary complex numbers, the expression is
Will zkIs substituted into the expression (1) of CZT to obtain
Using Bluestein equation
The Bluestein equation (5) is substituted into (4)
From the convolution formula, the converted data is finally obtained as follows
Training the converted data in an ELM network, and performing frosting diagnosis on the trained network, wherein the method specifically comprises the following steps:
s3.1, ELM neural network initialization
Normalizing the input characteristic quantity of the ELM neural network to be between [0 and 1], wherein the normalization formula is as follows:
where X is the normalized value, X is the normalized sample datamin、XmaxRespectively the minimum value and the maximum value of the normalized data of the sample;
s3.2ELM neural network parameter initialization
Setting training times N, training errors e, a connection weight omega from an input layer to a hidden layer, a connection weight β from the hidden layer to an output layer and a threshold b of a neuron of the hidden layer, wherein an input characteristic value is an energy characteristic value x decomposed from fault current to an nth layeriThe number of neurons in the hidden layer is calculated according to the formula:determining, wherein n and l are the number of input neurons and the number of output neurons respectively, a is usually 1-10, and the output value is y;
dividing data into a test set and a prediction set;
s3.3 data testing
Importing test set data, namely CZT converted data obtained by formula (8), and calculating the success rate of fault identification;
s3.4GA Algorithm parameter initialization
Setting size N of population1Degree of evolution G, crossover probability PcAnd the mutation probability Pm;
Forming genetic codes by the weight values and the threshold values of the ELM neural network, repeatedly operating the neural network established in S3.3 and recording the genetic codes completed in each operation to establish a population;
s3.5 establishing an evaluation function
Calculating and recording corresponding fitness values according to the running result of the neural network constructed by each group of chromosomes;
s3.6GA Algorithm selection, crossover and mutation
1) Selecting a part of population to generate next generation population with a certain probability;
2) selecting a plurality of chromosomes from the parent generation population to carry out the operation of the next generation of chromosomes, and generating new individuals through the cross combination of the chromosomes;
3) selecting one individual from the population, and carrying out mutation on a certain code in the selected chromosome to generate more excellent individuals;
s3.7 fitness value calculation
Decoding the genetic code in the newly obtained chromosome individual, calculating the fitness and comparing the fitness with the population; repeating S3.5-S3.6, continuously performing the operations on the population and calculating the fitness value until the maximum evolution times are reached, wherein the chromosome corresponding to the fitness optimal solution is the weight and the threshold of the established ELM neural network;
s3.8 error calculation for neural networks
Calculating the difference between the predicted output and the expected output, if the error meets the specified value, completing the construction of the ELM neural network based on the GA algorithm, and if the error does not meet the specified value, returning to S3.3 to continue learning until the error meets the specified value;
3.9 diagnosis of frosting Fault
And (3) taking the data obtained by the CZT conversion obtained by the formula (8) as the input of the ELM neural network, and identifying by the ELM neural network trained by S3.8 and based on the GA algorithm:
if the recognition result needs defrosting and the current non-defrosting mode is adopted, defrosting is started; if defrosting is needed and the current mode is the defrosting mode, the running state is unchanged;
if the recognition result does not need defrosting and the current non-defrosting mode is adopted, the running state is unchanged;
if defrosting is not needed and the current defrosting mode is already, defrosting is stopped, and the diagnosis information is transmitted to the defrosting control module.
The conventional defrosting method for the evaporator is to utilize a temperature sensor to measure the temperature of a coil pipe of a heat exchanger to judge whether frost is formed or set time intervals for defrosting at regular time, and the methods are low in precision and efficiency, cannot find faults quickly and cannot eliminate the faults timely. In order to rapidly find the frosting of the outdoor evaporator of the air source heat pump in real time, the defrosting method and the defrosting device control the heat pump system to defrost in time by utilizing a data processing and fault diagnosis device according to the current of the measuring fan. The method realizes the control of the defrosting system by utilizing the acquired current signals through wavelet packet decomposition and an ELM neural network fitting input-output mapping relation model based on a GA algorithm, thereby improving the operating efficiency of the heat pump system, reducing the mechanical damage caused by overlong frosting time and achieving the purpose of saving cost.
It should be noted that the above embodiments can be freely combined as necessary. The foregoing has outlined rather broadly the preferred embodiments and principles of the present invention and it will be appreciated that those skilled in the art may devise variations of the present invention that are within the spirit and scope of the appended claims.
Claims (9)
1. Utilize neural network heat pump defrosting control device of fan current, its characterized in that includes:
the current measuring device is used for measuring a current signal of a fan of the air source heat pump evaporator in real time;
the data conversion module is used for filtering, demodulating, decomposing and reconstructing the current signal to obtain data information, then analyzing the time domain and the frequency domain by utilizing a characteristic extraction algorithm to extract a characteristic data value,
the data diagnosis module is used for carrying out diagnosis analysis processing on the characteristic data value by utilizing an ELM neural network based on a GA algorithm to obtain the diagnosis information of the frosting fault;
and the defrosting control module is used for carrying out defrosting control on the air source heat pump evaporator according to the diagnosis information.
2. The control device according to claim 1, characterized in that: in the data conversion module, Hilbert conversion is used for filtering and demodulating, and wavelet packet analysis, frequency domain analysis or CZT conversion is used for decomposing and reconstructing.
3. The control device according to claim 2, characterized in that: the feature extraction algorithm adopts any one of wavelet packet decomposition, fast Fourier transform and principal component analysis.
4. The control device according to claim 3, characterized in that: the device also comprises a data acquisition module which is used for collecting the acquired data of the current measuring device and transmitting the acquired data to the data conversion module.
5. The neural network heat pump defrosting control method using the fan current, using the control device of claim 4, characterized in that: the method comprises the following steps:
s1, the current measuring device collects the current signal of the air source heat pump evaporator fan in real time;
s2, the data conversion module filters, demodulates, decomposes and reconstructs the current signal to obtain data information, and then the time domain and frequency domain analysis is carried out by utilizing a feature extraction algorithm to extract a feature data value;
s3, the data diagnosis module carries out diagnosis analysis processing on the characteristic data value by using an ELM neural network based on a GA algorithm to obtain the diagnosis information of frosting faults;
and S4, the defrosting control module carries out defrosting control on the air source heat pump evaporator according to the diagnosis information.
6. The control method according to claim 5, characterized in that: in step S2, the data conversion module performs filtering, demodulation, decomposition, and reconstruction on the current signal to obtain data information, specifically:
s2.1, carrying out Hilbert conversion on data:
s (t) is the given time domain signal, Hilbert transforms into a convolution of s (t) with h (t) 1/(tt),
in the formula, H is a Hilbert transform operator, and the transformed signal is processedProcessing;
and S2.2, carrying out decomposition and reconstruction operation by utilizing wavelet packet analysis, frequency domain analysis or CZT conversion.
7. The control method according to claim 6, characterized in that: in step S2.2, the wavelet packet analysis step is as follows:
1) decomposition of signals using wavelet packets
Carrying out n-layer wavelet packet decomposition on the collected current signals, and respectively extracting the nth layer from low frequency to high frequency 2nWavelet packet coefficients at the individual nodes; 2nEach node is (i, j) which represents the jth node of the ith layer, where i is n and j is 0,1,2,3 …,2n-1;
2) Reconstructing wavelet packet decomposition coefficient, and extracting signal characteristics of each frequency band range
Let each node wavelet packet coefficient Hi,jThe corresponding reconstructed signal is Si,j(ii) a Analyzing all nodes of the nth layer, the total signal S can be represented by the following formula:
3) calculating the total energy of each frequency band signal
Suppose Sn,j(j=0,1,2,3…,2n-1) corresponding energy En,j(j=0,1,2,3…,2n-1), then energy Sn,jCan be represented by the following formula:
wherein: h isj,k(j=0,1,2,3…,2n-1; k-1, 2, …, n) represents the reconstructed signal Sn,jThe amplitude of the discrete points of (a);
4) constructing feature vectors
8. The control method according to claim 6, characterized in that: in step S2.2, the CZT conversion analysis step is as follows:
for a known signal x (N), when 0< N < N-1, the CZT expression is
Equidistant sampling in a spiral fashion in the Z plane, ZkA sampling point of Z, expressed as
zk=AW-k,k=0,1……M-1(2)
Wherein M is the number of pre-analysis points, A and M are arbitrary complex numbers, and the expression is
Will zkIs substituted into the expression (1) of CZT to obtain
Using Bluestein equation
The Bluestein equation (5) is substituted into (4)
Order to
Obtaining the converted data by a convolution formula:
9. the control method according to claim 7 or 8, characterized in that: step S3 is specifically as follows:
s3.1, ELM neural network initialization
Normalizing the input characteristic quantity of the ELM neural network to be between [0 and 1], wherein the normalization formula is as follows:
where X is the normalized value, X is the normalized sample datamin、XmaxRespectively the minimum value and the maximum value of the normalized data of the sample;
S3.2ELM neural network parameter initialization
Setting training times N, training errors e, a connection weight omega from an input layer to a hidden layer, a connection weight β from the hidden layer to an output layer and a threshold b of a neuron of the hidden layer, wherein an input characteristic value is an energy characteristic value x decomposed from fault current to an nth layeriThe number of neurons in the hidden layer is calculated according to the formula:determining, wherein n and l are the number of input neurons and the number of output neurons respectively, a is usually 1-10, and the output value is y;
dividing data into a test set and a prediction set;
s3.3 data testing
Importing test set data, namely data obtained by frequency domain analysis or eigenvector extracted by wavelet packet decomposition or data after CZT conversion, and calculating the success rate of fault identification;
S3.4GA Algorithm parameter initialization
Setting size N of population1Degree of evolution G, crossover probability PcAnd the mutation probability Pm;
Forming genetic codes by the weight values and the threshold values of the ELM neural network, repeatedly operating the neural network established in S3.3 and recording the genetic codes completed in each operation to establish a population;
s3.5 establishing an evaluation function
Calculating and recording corresponding fitness values according to the running result of the neural network constructed by each group of chromosomes;
S3.6GA selection, intersection and variation of algorithms
1) Selecting a part of population to generate next generation population with a certain probability;
2) selecting a plurality of chromosomes from the parent generation population to carry out the operation of the next generation of chromosomes, and generating new individuals through the cross combination of the chromosomes;
3) selecting one individual from the population, and carrying out mutation on a certain code in the selected chromosome to generate more excellent individuals;
s3.7 fitness value calculation
Decoding the genetic code in the newly obtained chromosome individual, calculating the fitness and comparing the fitness with the population; repeating S3.5-S3.6, continuously performing the operations on the population and calculating the fitness value until the maximum evolution times are reached, wherein the chromosome corresponding to the fitness optimal solution is the weight and the threshold of the established ELM neural network;
s3.8 error calculation for neural networks
Calculating the difference between the predicted output and the expected output, if the error meets the specified value, completing the construction of the ELM neural network based on the GA algorithm, and if the error does not meet the specified value, returning to S3.3 to continue learning until the error meets the specified value;
3.9 diagnosis of frosting Fault
And (3) taking data obtained by frequency domain analysis or eigenvector obtained by wavelet packet decomposition or data obtained by CZT conversion as input of an ELM neural network, and identifying by the ELM neural network trained by S3.8 and based on the GA algorithm:
if the recognition result needs defrosting and the current non-defrosting mode is adopted, defrosting is started; if defrosting is needed and the current mode is the defrosting mode, the running state is unchanged;
if the recognition result does not need defrosting and the current non-defrosting mode is adopted, the running state is unchanged;
if defrosting is not needed and the current defrosting mode is already, defrosting is stopped, and the diagnosis information is transmitted to the defrosting control module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911132291.9A CN110850717B (en) | 2019-11-19 | 2019-11-19 | Neural network heat pump defrosting control device and control method by utilizing fan current |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911132291.9A CN110850717B (en) | 2019-11-19 | 2019-11-19 | Neural network heat pump defrosting control device and control method by utilizing fan current |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110850717A true CN110850717A (en) | 2020-02-28 |
CN110850717B CN110850717B (en) | 2022-05-20 |
Family
ID=69602508
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911132291.9A Active CN110850717B (en) | 2019-11-19 | 2019-11-19 | Neural network heat pump defrosting control device and control method by utilizing fan current |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110850717B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113536989A (en) * | 2021-06-29 | 2021-10-22 | 广州博通信息技术有限公司 | Refrigerator frosting monitoring method and system based on camera video frame-by-frame analysis |
WO2022095499A1 (en) * | 2020-11-06 | 2022-05-12 | 青岛海尔电冰箱有限公司 | Defrosting control method and refrigeration appliance having same |
WO2022248229A1 (en) * | 2021-05-28 | 2022-12-01 | Ebm-Papst Mulfingen Gmbh & Co. Kg | Method and system for defrosting a refrigeration unit |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004350496A (en) * | 2003-04-30 | 2004-12-09 | Matsushita Electric Ind Co Ltd | Motor driving device |
US20170131005A1 (en) * | 2014-07-01 | 2017-05-11 | Sinjin Enertec Co., Ltd. | Heat pump heating-cooling system using hybrid heat source and control method thereof |
EP3239626A1 (en) * | 2016-04-27 | 2017-11-01 | PLUM spólka z ograniczona odpowiedzialnoscia | Method for controlling heat pump operation |
CN108644879A (en) * | 2018-04-01 | 2018-10-12 | 太原理工大学 | Air source heat pump couples collection control heating system and its method with solar water heater |
CN109210840A (en) * | 2018-09-12 | 2019-01-15 | 珠海格力电器股份有限公司 | Air-cooled heat pump unit and defrosting control method thereof |
CN109386966A (en) * | 2018-09-21 | 2019-02-26 | 珠海格力电器股份有限公司 | Defrosting control method and device for heat pump unit, storage medium and heat pump unit |
CN109886448A (en) * | 2019-03-14 | 2019-06-14 | 浙江工业大学 | Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm |
CN109974201A (en) * | 2018-12-18 | 2019-07-05 | 青岛经济技术开发区海尔热水器有限公司 | Defrosting control method and air source heat pump system |
CN110068086A (en) * | 2019-04-15 | 2019-07-30 | 青岛海尔空调电子有限公司 | Heat pump air conditioning system and its defrosting control method |
-
2019
- 2019-11-19 CN CN201911132291.9A patent/CN110850717B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004350496A (en) * | 2003-04-30 | 2004-12-09 | Matsushita Electric Ind Co Ltd | Motor driving device |
US20170131005A1 (en) * | 2014-07-01 | 2017-05-11 | Sinjin Enertec Co., Ltd. | Heat pump heating-cooling system using hybrid heat source and control method thereof |
EP3239626A1 (en) * | 2016-04-27 | 2017-11-01 | PLUM spólka z ograniczona odpowiedzialnoscia | Method for controlling heat pump operation |
CN108644879A (en) * | 2018-04-01 | 2018-10-12 | 太原理工大学 | Air source heat pump couples collection control heating system and its method with solar water heater |
CN109210840A (en) * | 2018-09-12 | 2019-01-15 | 珠海格力电器股份有限公司 | Air-cooled heat pump unit and defrosting control method thereof |
CN109386966A (en) * | 2018-09-21 | 2019-02-26 | 珠海格力电器股份有限公司 | Defrosting control method and device for heat pump unit, storage medium and heat pump unit |
CN109974201A (en) * | 2018-12-18 | 2019-07-05 | 青岛经济技术开发区海尔热水器有限公司 | Defrosting control method and air source heat pump system |
CN109886448A (en) * | 2019-03-14 | 2019-06-14 | 浙江工业大学 | Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm |
CN110068086A (en) * | 2019-04-15 | 2019-07-30 | 青岛海尔空调电子有限公司 | Heat pump air conditioning system and its defrosting control method |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2022095499A1 (en) * | 2020-11-06 | 2022-05-12 | 青岛海尔电冰箱有限公司 | Defrosting control method and refrigeration appliance having same |
WO2022248229A1 (en) * | 2021-05-28 | 2022-12-01 | Ebm-Papst Mulfingen Gmbh & Co. Kg | Method and system for defrosting a refrigeration unit |
DE102021113802A1 (en) | 2021-05-28 | 2022-12-01 | Ebm-Papst Mulfingen Gmbh & Co. Kg | Method and system for defrosting a refrigeration unit |
CN113536989A (en) * | 2021-06-29 | 2021-10-22 | 广州博通信息技术有限公司 | Refrigerator frosting monitoring method and system based on camera video frame-by-frame analysis |
Also Published As
Publication number | Publication date |
---|---|
CN110850717B (en) | 2022-05-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110850717B (en) | Neural network heat pump defrosting control device and control method by utilizing fan current | |
Fan et al. | A framework for knowledge discovery in massive building automation data and its application in building diagnostics | |
CN109633301B (en) | Non-invasive electrical appliance load identification method based on quantum genetic optimization | |
CN106646026A (en) | Non-intrusive household appliance load identification method | |
CN112215405B (en) | Non-invasive resident electricity load decomposition method based on DANN domain adaptive learning | |
CN110659693A (en) | K-nearest neighbor classification-based rapid topology identification method and system for power distribution network and readable storage medium | |
CN113036759B (en) | Fine granularity identification method and identification system for power consumer load | |
CN110416995B (en) | Non-invasive load decomposition method and device | |
CN110779264B (en) | Defrosting control system and method based on fan vibration and information fusion | |
CN110503153A (en) | Photovoltaic system method for diagnosing faults based on differential evolution algorithm and support vector machines | |
CN111156657A (en) | Air conditioner frosting state determining method and device | |
CN211261438U (en) | Defrosting control system based on fan vibration and information fusion | |
CN113076985B (en) | Power consumer load identification method and device and computer readable medium | |
CN112804292B (en) | Ammeter distribution area identification method | |
CN117315434A (en) | Non-invasive load identification method based on time sequence imaging and deep learning | |
CN110779265A (en) | Remote monitoring system and method for air source heat pump defrosting based on fan current | |
CN110849050B (en) | Heat pump defrosting control device and control method by utilizing fan vibration and neural network | |
CN117235617A (en) | ML-RFKNN-based photovoltaic array fault diagnosis method in sand and dust weather | |
Cheng et al. | Fault detection and isolation for chiller system based on deep autoencoder | |
CN111881793A (en) | Non-invasive load monitoring method and system based on capsule network | |
CN211261439U (en) | Remote monitoring system for air source heat pump defrosting based on fan current | |
CN115684908A (en) | State monitoring method for power GIS equipment, storage medium and electronic device | |
CN114357412A (en) | Efficiency self-control optimizing control system of heating ventilation air conditioning system | |
CN116500386A (en) | Wind power plant collector line cable partial discharge signal acquisition and processing method | |
CN112149726B (en) | Totally-enclosed compressor fault diagnosis method based on knowledge sharing and model migration |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |