WO2019052241A1 - 用于空调的冷媒泄漏检测方法和装置 - Google Patents

用于空调的冷媒泄漏检测方法和装置 Download PDF

Info

Publication number
WO2019052241A1
WO2019052241A1 PCT/CN2018/091070 CN2018091070W WO2019052241A1 WO 2019052241 A1 WO2019052241 A1 WO 2019052241A1 CN 2018091070 W CN2018091070 W CN 2018091070W WO 2019052241 A1 WO2019052241 A1 WO 2019052241A1
Authority
WO
WIPO (PCT)
Prior art keywords
refrigerant
air conditioner
neural network
network model
data
Prior art date
Application number
PCT/CN2018/091070
Other languages
English (en)
French (fr)
Inventor
陈翀
黄辉
宋德超
田涛
刘佰兰
Original Assignee
格力电器（武汉）有限公司
珠海格力电器股份有限公司
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 格力电器（武汉）有限公司, 珠海格力电器股份有限公司 filed Critical 格力电器（武汉）有限公司
Priority to EP18856758.0A priority Critical patent/EP3637328A4/en
Priority to US16/633,580 priority patent/US11614249B2/en
Publication of WO2019052241A1 publication Critical patent/WO2019052241A1/zh

Links

Images

Classifications

    • 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
    • F24F11/32—Responding to malfunctions or emergencies
    • F24F11/36—Responding to malfunctions or emergencies to leakage of heat-exchange fluid
    • 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
    • F24F11/50—Control or safety arrangements characterised by user interfaces or communication
    • F24F11/56—Remote control
    • F24F11/58—Remote control using Internet communication
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
    • G06F17/145—Square transforms, e.g. Hadamard, Walsh, Haar, Hough, Slant transforms
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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

Definitions

  • the present invention relates to the field of artificial intelligence, and in particular to a refrigerant leakage detecting method and apparatus for an air conditioner.
  • the method of refrigerant leakage detection in the air-conditioning industry is mainly to analyze the operating parameters of the air-conditioner in the case of refrigerant leakage, and summarize a series of control rules based on expert experience.
  • the air-conditioning main board judges according to this rule. When it is detected that the air-conditioning operating parameters meet the predetermined rules, it is considered that the air-conditioner has a refrigerant leak, and then prompts the user to repair.
  • the current control method has the following problems:
  • control rules are poorly adapted, and the control parameters of different models are not uniform;
  • the main object of the present application is to provide a refrigerant leakage detecting method and apparatus for an air conditioner to solve the problem of determining whether the air conditioner has an inaccurate refrigerant leakage by manual experience.
  • a refrigerant leakage detecting method for an air conditioner comprising: acquiring current operating parameters of an air conditioner and environmental information of an environment in which the air conditioner is located; The operating parameter and the environmental information are input to the trained neural network model, and the residual amount of the refrigerant output by the neural network model is obtained; and whether the air conditioner has a refrigerant leakage is determined according to the remaining amount of the refrigerant.
  • the method further includes: acquiring operating parameters and environment information of each type of air conditioner when the refrigerant leaks;
  • the air conditioner trains the neural network in the operating parameters and environmental information of the refrigerant leakage, and obtains the trained neural network model, wherein the input of the model is an operating parameter and environmental information of the air conditioner when the refrigerant leaks, and the model The output is the remaining amount of refrigerant.
  • the method further includes: normalizing processing parameters of the air conditioners in the respective types of air conditioners when the refrigerant leaks, and obtaining standardized parameters. And wherein the normalization processing includes linear processing; transforming the normalized parameters to obtain transformed data, wherein the transform processing includes nonlinear processing.
  • the method further includes: classifying the transformed data to obtain data of multiple categories; and from the plurality of categories The data is extracted as the training sample data according to the preset interval, wherein the preset interval includes a preset time interval or a preset number interval; and the training sample data is trained by the neural network model to obtain the training. Neural network model.
  • the method further includes: inputting test data into the trained neural network model, to obtain the An output of the trained neural network model; determining whether an error between the output result and the test result corresponding to the test data is less than a target minimum error; and an error between the output result and the test result corresponding to the test data is greater than or equal to In the case of the target minimum error, adjusting the parameters of the neural network model by updating the weight and offset of the neural network model until the error between the output result and the test result corresponding to the test data is less than the Target minimum error.
  • obtaining operating parameters of each type of air conditioner when the refrigerant leaks includes: receiving operating parameters and environmental information reported by each type of air conditioner when the refrigerant leaks; and/or acquiring various types of air conditioners when the user actually uses the Internet of Things. Operating parameters and environmental information when the refrigerant leaks.
  • determining whether the air conditioner has a refrigerant leakage according to the remaining amount of the refrigerant includes: acquiring an original refrigerant amount of the air conditioner; determining whether the refrigerant remaining amount is smaller than the original refrigerant amount; and if the refrigerant remaining amount is less than The amount of the original refrigerant determines that the air conditioner has a refrigerant leak.
  • the neural network model includes any one of the following: a BP neural network model; a CNN convolutional neural network model; and a residual neural network model, after determining that the air conditioner has a refrigerant leak according to the remaining amount of the refrigerant, Set the way to send a reminder.
  • a refrigerant leakage detecting device for an air conditioner, the device comprising: a first acquiring unit configured to acquire a current operating parameter of the air conditioner and the air conditioner is located Environment information; an input unit configured to input the current operating parameter and environmental information into the trained neural network model to obtain a residual amount of refrigerant output by the neural network model; and a determining unit configured to be based on the refrigerant The remaining amount determines whether the air conditioner has a refrigerant leak.
  • the device further includes: a second acquiring unit, configured to acquire operating parameters and environmental information of each type of air conditioner when the refrigerant leaks before inputting the operating parameter and the environment information into the trained neural network model a first training unit configured to perform neural network training according to operating parameters of each type of air conditioner when the refrigerant leaks, to obtain the trained neural network model, wherein the input of the model is an air conditioner when the refrigerant leaks The operating parameters and environmental information, the output of the model is the remaining amount of refrigerant.
  • a second acquiring unit configured to acquire operating parameters and environmental information of each type of air conditioner when the refrigerant leaks before inputting the operating parameter and the environment information into the trained neural network model
  • a first training unit configured to perform neural network training according to operating parameters of each type of air conditioner when the refrigerant leaks, to obtain the trained neural network model, wherein the input of the model is an air conditioner when the refrigerant leaks
  • the operating parameters and environmental information, the output of the model is the remaining amount of ref
  • the device further includes: a processing unit configured to, after acquiring operating parameters and environmental information of each type of air conditioner when the refrigerant leaks, operating parameters and environmental information of the respective types of air conditioners when the refrigerant leaks
  • the parameter is subjected to a normalization process to obtain a standardized parameter, wherein the normalization process includes linear processing; and a transform unit configured to perform transform processing on the normalized parameter to obtain transformed data, wherein the transform process includes nonlinear processing.
  • the apparatus further includes: a classification unit configured to: after transforming the normalized parameter to obtain the transformed data, classifying the transformed data to obtain data of multiple categories; and extracting unit And the data is extracted from the data of the plurality of categories according to the preset interval, as the training sample data, wherein the preset interval includes a preset time interval or a preset number interval; the second training unit is set to Performing a neural network model training on the training sample data to obtain the trained neural network model.
  • a classification unit configured to: after transforming the normalized parameter to obtain the transformed data, classifying the transformed data to obtain data of multiple categories
  • extracting unit And the data is extracted from the data of the plurality of categories according to the preset interval, as the training sample data, wherein the preset interval includes a preset time interval or a preset number interval
  • the second training unit is set to Performing a neural network model training on the training sample data to obtain the trained neural network model.
  • a storage medium comprising a stored program, wherein, when the program is running, controlling a device in which the storage medium is located to execute a refrigerant for an air conditioner of the present application Leak detection method.
  • a processor for executing a program, wherein the program is executed to execute the refrigerant leak detecting method for an air conditioner of the present application.
  • the present application obtains the current operating parameters of the air conditioner and the environmental information of the environment in which the air conditioner is located; inputs the current operating parameters and environmental information into the trained neural network model to obtain the remaining amount of the refrigerant output by the neural network model; and judges according to the remaining amount of the refrigerant Whether there is refrigerant leakage in the air conditioner solves the problem of judging whether the air conditioner has inaccurate refrigerant leakage through manual experience, and thus achieves the effect of detecting the leakage of the air conditioner refrigerant through the artificial neural network algorithm.
  • FIG. 1 is a flow chart of a refrigerant leak detecting method for an air conditioner according to an embodiment of the present application
  • FIG. 2 is a schematic diagram of a neural network algorithm for performing calculations according to an embodiment of the present application
  • FIG. 3 is a schematic diagram of a structural design of a BP network according to an embodiment of the present application.
  • FIG. 4 is a schematic diagram showing the structural design of a CNN convolutional neural network according to an embodiment of the present application.
  • FIG. 5 is a schematic diagram of a structural design of a residual neural network according to an embodiment of the present application.
  • FIG. 6 is a schematic diagram of a neural network learning target according to an embodiment of the present application.
  • FIG. 7 is a schematic diagram of a smart device running an artificial neural network algorithm according to an embodiment of the present application.
  • FIG. 8 is a schematic diagram of a refrigerant leakage detecting device for an air conditioner according to an embodiment of the present application.
  • Embodiments of the present application provide a refrigerant leakage detecting method for an air conditioner.
  • FIG. 1 is a flow chart of a refrigerant leakage detecting method for an air conditioner according to an embodiment of the present application. As shown in FIG. 1, the method includes the following steps:
  • Step S102 Acquire current operating parameters of the air conditioner and environment information of the environment where the air conditioner is located.
  • Step S104 Input the current running parameter and the environment information into the trained neural network model, and obtain the residual amount of the refrigerant output by the neural network model.
  • Step S106 It is determined whether there is a refrigerant leak in the air conditioner according to the remaining amount of the refrigerant.
  • the current operating parameters of the air conditioner and the environment information of the environment in which the air conditioner is located are obtained; the current operating parameters and environmental information are input into the trained neural network model, and the remaining amount of the refrigerant output by the neural network model is obtained; according to the remaining amount of the refrigerant Judging whether there is refrigerant leakage in the air conditioner solves the problem of judging whether the air conditioner has inaccurate refrigerant leakage through manual experience, and thus achieving the effect of detecting the leakage of the air conditioner refrigerant through the artificial neural network algorithm.
  • the current operating parameters of the air conditioner include various types of parameters, for example, the operating speed of the compressor of the air conditioner, the running time of the compressor, the temperature of the outdoor condenser, etc., and the environmental information includes the ambient temperature of the indoor and outdoor, etc.
  • the residual amount of refrigerant determined by the model can be obtained, and then according to the residual amount of the refrigerant and the amount of the original refrigerant of the air conditioner, it is judged whether there is a refrigerant leakage, and if there is a refrigerant leakage, the refrigerant can be passed.
  • the air conditioner sends a reminder, and can also send a reminder through the preset smart terminal device, and can also control the air conditioner to perform a predetermined operation to prevent or reduce leakage of the air conditioner refrigerant.
  • the trained neural network model Before inputting the operating parameter and the environmental information into the trained neural network model, acquiring operating parameters and environmental information of each type of air conditioner when the refrigerant leaks; performing, according to operating parameters of each type of air conditioner when the refrigerant leaks Neural network training, the trained neural network model is obtained, wherein the input of the model is the operating parameter and environmental information of the air conditioner when the refrigerant leaks, and the output of the model is the residual amount of the refrigerant. Due to different air conditioner types of different air conditioner models, the operating parameters and environmental information of each type of air conditioner in the case of refrigerant leakage can be obtained according to the type of air conditioner, and then the data of each type of parameters and environmental information are separately sorted, and each type is obtained. Air-conditioned trained network neural model.
  • the processing includes linear processing; transforming the normalized parameters to obtain transformed data, wherein the transform processing includes nonlinear processing.
  • the linear processing may be a normalization process or the like, and the nonlinear processing may be a logarithmic transformation, a square root transformation, a cube root transformation, or the like.
  • the collected sample data can be more consistent with the requirements of the model training, and the results obtained by the model training can be more accurate.
  • the neural network model training is performed on the training sample data to obtain a trained neural network model.
  • the data can be classified, and then a certain amount of data is selected as sample data from each class.
  • the selection method may be one of every 50 numbers as sample data, or may be every certain time.
  • the time for example, 24 hours to take a piece of data as sample data, can also be other ways of spacing.
  • the test data is input into the trained neural network model, and the output result of the trained neural network model is obtained; and the output result is judged. Whether the error between the test results corresponding to the test data is less than the target minimum error; when the error between the output result and the test result corresponding to the test data is greater than or equal to the target minimum error, updating the weight and bias of the neural network model The parameters of the neural network model are adjusted until the error between the output result and the test result corresponding to the test data is less than the target minimum error.
  • the model has been trained to be used and can be used for actual prediction.
  • obtaining operating parameters and environmental information of each type of air conditioner when the refrigerant leaks includes: receiving operating parameters and environmental information reported by each type of air conditioner when the refrigerant leaks; and/or obtaining the actual use of the user through the Internet of Things.
  • Operating parameters and environmental information for each type of air conditioner in the event of refrigerant leakage can be a variety of
  • determining whether the air conditioner has a refrigerant leakage according to the remaining amount of the refrigerant includes: obtaining an original refrigerant amount of the air conditioner; determining whether the remaining amount of the refrigerant is less than the original refrigerant amount; and if the remaining amount of the refrigerant is less than the original refrigerant amount, determining that the air conditioner has a refrigerant leakage .
  • the amount of raw refrigerant can be the amount of refrigerant at the time of shipment from the air conditioner, or the amount of refrigerant at each time the air conditioner is turned on.
  • the neural network model includes any one of the following: a BP neural network model; a CNN convolutional neural network model; and a residual neural network model, which sends a reminder in a preset manner after determining that the air conditioner has a refrigerant leak according to the remaining amount of the refrigerant.
  • the preset method can be of various types.
  • the air conditioner emits an alarm sound or displays an indicator light, and can also be an alarm sound from a mobile phone, a remote controller, or an intelligent control center in the home to prompt the user to check and repair in time.
  • there are other types of neural network models there are other types of neural network models.
  • the embodiment of the present application utilizes an artificial neural network algorithm to learn and train a neural network by using a large number of operating parameter samples when the air conditioner refrigerant leaks.
  • the neural network is fitted to the relationship between the air conditioning operating parameters, and finally the neural network can accurately detect the amount of refrigerant leakage.
  • the method has strong nonlinear mapping ability, self-learning and self-adaptive ability, generalization ability and fault tolerance. Compared with the traditional rule control method, it does not rely on expert experience. Through the study of a large amount of sample data, the network can automatically correct its own parameters, and finally achieve the following effects:
  • control algorithm can be automatically applied to different air conditioner models, and the versatility is good;
  • the technical solution of the embodiment of the present application includes the following steps:
  • the specific collection methods include, but are not limited to, operating parameters of the air conditioner in the laboratory simulation environment, and collection of air conditioning operating parameters when the actual user uses the Internet of Things technology.
  • the input parameters include, but are not limited to, indoor and outdoor ambient temperature, outdoor condenser temperature, outdoor humidity, compressor operating speed, compressor running time, and the like.
  • the input parameters are not only a single parameter, but also an input parameter matrix composed according to a certain rule. For example, the outdoor condenser temperature is collected once every second, and the temperature parameter of 10 minutes is continuously collected as an input parameter matrix.
  • the data processing methods used in the embodiments of the present application include, but are not limited to, linear processing such as normalization of data, and nonlinear processing such as logarithmic transformation, square root transformation, and cubic root transformation.
  • the data samples are classified according to certain rules by analyzing the collected and labeled data and combining expert knowledge. Data is extracted evenly from different categories of samples as training samples. Training samples must not only contain the law of refrigerant leakage, but also reflect diversity and uniformity. In the embodiment of the present application, the air conditioning room, the internal and external environmental temperature, and humidity may be combined to list all the sample data, and then read as training sample data at a certain interval; after the training sample is extracted, the remaining data may be used as a test. data.
  • FIG. 2 is a schematic diagram of a neural network algorithm for performing calculation according to an embodiment of the present application. As shown in FIG. 2, a refrigerant residual amount can be obtained by inputting a plurality of input parameters into a neural network algorithm.
  • the specific artificial neural network structure includes, but is not limited to, the following three network structures.
  • FIG. 3 is a schematic diagram of a structural design of a BP network according to an embodiment of the present application.
  • the neural network algorithm mainly needs to solve the problem of setting several hidden layers and several hidden nodes.
  • the determination of hidden and hidden nodes needs to be constantly adjusted during network training.
  • the design first sets a hidden layer, and improves the network performance by adjusting the number of hidden layer nodes. When there are too many hidden nodes and too many fittings, consider adding hidden layers and reducing hidden nodes to improve network performance. In practical applications, the input layer, hidden layer, output layer node number, and hidden layer number can be adjusted as needed.
  • FIG. 4 is a schematic diagram showing the structural design of a CNN convolutional neural network according to an embodiment of the present application. As shown in FIG. 4, a multi-layer convolution network is continuously extracted from low-dimensional features to obtain higher-dimensional features and thus can be used. Classify or related tasks.
  • the raw data in the embodiment of the present application is continuously collected, which is intuitively related to time.
  • a certain amount of data can be combined into one strip, which is regarded as the same data form as the image.
  • the input data can be extracted by the convolutional neural network, and the residual amount of the refrigerant can be accurately detected.
  • the network structure can be adjusted according to actual conditions.
  • FIG. 5 is a schematic diagram of a structural design of a residual neural network according to an embodiment of the present application.
  • the method of deepening the number of network layers and changing the size of the convolution kernel does not improve the network performance. . Adding residual blocks can better connect the data before and after, and enhance the ability to express features, so it can enhance the learning ability of the convolutional network.
  • the training sample data is needed to train the network.
  • the training method can be adjusted according to the actual network structure and the problems found in the training. Only one of the methods of the embodiments of the present application is exemplified as follows:
  • the training is ended. If it is not satisfied, the weight W k of the network is updated according to the following method, and the offset b l is :
  • C(w,b) is the error energy function (taking the standard variance function as an example)
  • n is the total number of training samples, and the summation is performed on the total training sample x.
  • W k is the initial weight
  • b l is the initial bias
  • the partial derivative of the bias for the error energy function The value can be obtained by the chain derivation rule until the output error of the network is less than ⁇ .
  • test sample is used to test the network.
  • test error does not meet the requirements, repeat the above steps to retrain the network; if the test error meets the requirements, the network training test is completed.
  • FIG. 7 is a schematic diagram of a smart device running an artificial neural network algorithm according to an embodiment of the present application.
  • an air conditioner with wireless communication when an air conditioner with wireless communication is in operation, an operating parameter of the air conditioner is uploaded to the smart device.
  • the smart device inputs the operating parameters into the algorithm, and after determining the refrigerant leakage condition, sends a control command to the air conditioner.
  • the smart device includes but is not limited to a wireless communication module, a router, a server, a smart phone, and the like.
  • the embodiment of the present application provides a refrigerant leakage detecting device for an air conditioner, and the device can be used to execute a refrigerant leakage detecting method for an air conditioner according to an embodiment of the present application.
  • FIG. 8 is a schematic diagram of a refrigerant leakage detecting device for an air conditioner according to an embodiment of the present application. As shown in FIG. 8, the device includes:
  • the first obtaining unit 10 is configured to acquire current operating parameters of the air conditioner and environment information of an environment where the air conditioner is located;
  • the input unit 20 is configured to input current operating parameters and environmental information into the trained neural network model, and obtain a residual amount of refrigerant output by the neural network model;
  • the determining unit 30 is configured to determine whether the air conditioner has a refrigerant leak according to the remaining amount of the refrigerant.
  • the first acquisition unit 10 acquires the current operating parameters of the air conditioner and the environment information of the environment in which the air conditioner is located; the input unit 20 inputs the current operating parameters and environment information into the trained neural network model, and obtains the The residual amount of the refrigerant outputted by the neural network model is determined; the determining unit 30 determines whether the air conditioner has a refrigerant leak according to the remaining amount of the refrigerant, and solves the problem of determining whether the air conditioner has an inaccurate refrigerant leakage through manual experience, thereby achieving the manual Neural network algorithm to detect the effect of air conditioning refrigerant leakage to improve accuracy.
  • the device further includes: a second acquiring unit, configured to acquire operating parameters of each type of air conditioner when the refrigerant leaks before inputting the operating parameter and the environment information into the trained neural network model; the first training unit It is used to perform neural network training according to the operating parameters of each type of air conditioner in the case of refrigerant leakage, and obtain a trained neural network model.
  • the input of the model is the operating parameter of the air conditioner when the refrigerant leaks, and the output of the model is the residual amount of the refrigerant. .
  • the device further includes: a processing unit, configured to perform standardization processing on operating parameters of each type of air conditioner when the refrigerant leaks after obtaining operating parameters of each type of air conditioner when the refrigerant leaks, to obtain standardized parameters, wherein
  • the normalization process includes linear processing
  • a transform unit for transforming the normalized parameters to obtain transformed data, wherein the transform processing includes nonlinear processing.
  • the apparatus further includes: a classification unit, configured to perform transformation processing on the normalized parameters to obtain the transformed data, and then classify the transformed data to obtain data of multiple categories; and extracting units for The data of the plurality of categories is respectively extracted according to the preset interval as the training sample data, wherein the preset interval includes a preset time interval or a preset number interval; and the second training unit is configured to perform a neural network model on the training sample data. Train and get a trained neural network model.
  • the first obtaining unit 10, the input unit 20 and the determining unit 30 may be operated in a computer terminal as part of the device, and the functions implemented by the above module may be executed by a processor in the computer terminal, the computer terminal It can also be a smart phone (such as Android phone, iOS phone, etc.), tablet, PDA, and mobile Internet devices (MID), PAD and other terminal devices.
  • the refrigerant leakage detecting device for an air conditioner includes a processor and a memory, and the first acquiring unit, the input unit, the determining unit, and the like are all stored as a program unit in a memory, and the processor executes the program unit stored in the memory. Implement the corresponding functions.
  • the processor contains a kernel, and the kernel removes the corresponding program unit from the memory.
  • the kernel can be set to one or more, and the artificial neural network algorithm can be used to detect the air conditioner refrigerant leakage to improve the accuracy by adjusting the kernel parameters.
  • the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory (flash RAM), the memory including at least one Memory chip.
  • RAM random access memory
  • ROM read only memory
  • flash RAM flash memory
  • the embodiment of the present application provides a device, including a processor, a memory, and a program stored on the memory and operable on the processor.
  • the processor executes the program, the following steps are implemented: obtaining current operating parameters of the air conditioner and the air conditioner Environment information of the environment; inputting the current operating parameter and environment information into the trained neural network model, obtaining a residual amount of refrigerant output by the neural network model; determining whether the air conditioner exists according to the remaining amount of the refrigerant Refrigerant leaks.
  • the devices in this document can be servers, PCs, PADs, mobile phones, and the like.
  • the present application also provides a computer program product, when executed on a data processing device, adapted to perform a process of initializing the following method steps: obtaining current operating parameters of the air conditioner and environmental information of the environment in which the air conditioner is located; The current operating parameters and environmental information are input to the trained neural network model, and the remaining amount of refrigerant outputted by the neural network model is obtained; and whether the air conditioner has a refrigerant leakage is determined according to the remaining amount of the refrigerant.
  • embodiments of the present application can be provided as a method, system, or computer program product.
  • the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment in combination of software and hardware.
  • the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
  • embodiments of the present application can be provided as a method, system, or computer program product.
  • the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment in combination of software and hardware.
  • the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • the present application obtains the current operating parameters of the air conditioner and the environmental information of the environment in which the air conditioner is located; inputs the current operating parameters and environmental information into the trained neural network model to obtain the remaining amount of the refrigerant output by the neural network model; and judges according to the remaining amount of the refrigerant Whether there is refrigerant leakage in the air conditioner solves the problem of judging whether the air conditioner has inaccurate refrigerant leakage through manual experience, and thus achieves the effect of detecting the leakage of the air conditioner refrigerant through the artificial neural network algorithm.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Fuzzy Systems (AREA)
  • Databases & Information Systems (AREA)
  • Algebra (AREA)
  • Human Computer Interaction (AREA)
  • Air Conditioning Control Device (AREA)

Abstract

本申请公开了一种用于空调的冷媒泄漏检测方法和装置。该方法包括：获取空调当前的运行参数和空调所处环境的环境信息；将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量；根据冷媒剩余量判断空调是否存在冷媒泄漏。通过本申请，达到了通过人工神经网络算法来检测空调冷媒泄漏提高准确性的效果。

Description

用于空调的冷媒泄漏检测方法和装置 技术领域
本发明涉及人工智能领域，具体而言，涉及一种用于空调的冷媒泄漏检测方法和装置。
背景技术
空调长期在冷媒泄漏的情况下运行而不及时报修，将会导致压缩机不可逆的损坏。因此空调在使用过程中，若出现冷媒泄漏，应立即停止运转压缩机，并提醒用户报修。
目前，在空调行业冷媒泄漏检测的方法主要是，对空调在冷媒泄漏的情况下的运行参数进行分析，根据专家经验，总结出一系列的控制规则。空调主板按此规则进行判断，当检测到空调运行参数符合预定的规则时，则认为空调出现冷媒泄漏，进而提示用户报修。但由于空调种类较多，不同型号的空调运行参数差别较大，且各运行参数之间的耦合关系和变化规律难以完全掌握，且专家经验存在一定的主观性等原因。因此，目前的控制方法存在以下问题：
1、控制规则适应性差，不同机型的控制参数不统一；
2、冷媒泄漏检测准确率低，存在较多的误检测。
针对相关技术中通过人工经验来判断空调是否存在冷媒泄漏不准确的问题，目前尚未提出有效的解决方案。
发明内容
本申请的主要目的在于提供一种用于空调的冷媒泄漏检测方法和装置，以解决通过人工经验来判断空调是否存在冷媒泄漏不准确的问题。
为了实现上述目的，根据本申请的一个方面，提供了一种用于空调的冷媒泄漏检测方法，该方法包括：获取空调当前的运行参数和所述空调所处环境的环境信息；将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。
进一步地，在将所述运行参数和环境信息输入到训练好的神经网络模型之前，所述方法还包括：获取各个类型的空调在冷媒泄漏时的运行参数和环境信息；根据所述各个类型的空调在冷媒泄漏时的运行参数和环境信息进行神经网络训练，得到所述训 练好的神经网络模型，其中，所述模型的输入为空调在冷媒泄漏时的运行参数和环境信息，所述模型的输出为冷媒剩余量。
进一步地，在获取各个类型的空调在冷媒泄漏时的运行参数之后，所述方法还包括：对所述各个类型的空调在冷媒泄漏时的运行参数和环境信息的参数进行标准化处理，得到标准化参数，其中，所述标准化处理包括线性处理；对所述标准化参数进行变换处理，得到变换后的数据，其中，所述变换处理包括非线性处理。
进一步地，在对所述标准化参数进行变换处理，得到变换后的数据之后，所述方法还包括：对所述变换后的数据进行分类，得到多个类别的数据；从所述多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，其中，所述预设间隔包括预设时间间隔或预设数量间隔；对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型。
进一步地，在对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型之后，所述方法还包括：将测试数据输入到所述训练好的神经网络模型，得到所述训练好的神经网络模型的输出结果；判断所述输出结果与测试数据对应的测试结果之间的误差是否小于目标最小误差；在所述输出结果与测试数据对应的测试结果之间的误差大于等于所述目标最小误差的情况下，通过更新所述神经网络模型的权值和偏置调整所述神经网络模型的参数，直至所述输出结果与测试数据对应的测试结果之间的误差小于所述目标最小误差。
进一步地，获取各个类型的空调在冷媒泄漏时的运行参数包括：接收各个类型的空调在冷媒泄漏时上报的运行参数和环境信息；和/或通过物联网获取用户实际使用时的各个类型的空调在冷媒泄漏时的运行参数和环境信息。
进一步地，根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏包括：获取所述空调的原始冷媒量；判断所述冷媒剩余量是否小于所述原始冷媒量；如果所述冷媒剩余量是否小于所述原始冷媒量，则判断出所述空调存在冷媒泄漏。
进一步地，所述神经网络模型包括以下任意一项：BP神经网络模型；CNN卷积神经网络模型；残差神经网络模型，在根据所述冷媒剩余量判断所述空调存在冷媒泄漏之后，通过预设的方式发出提醒。
为了实现上述目的，根据本申请的另一方面，还提供了一种用于空调的冷媒泄漏检测装置，该装置包括：第一获取单元，设置为获取空调当前的运行参数和所述空调所处环境的环境信息；输入单元，设置为将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；判断单元，设置为根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。
进一步地，所述装置还包括：第二获取单元，设置为在将所述运行参数和环境信息输入到训练好的神经网络模型之前，获取各个类型的空调在冷媒泄漏时的运行参数和环境信息；第一训练单元，设置为根据所述各个类型的空调在冷媒泄漏时的运行参数进行神经网络训练，得到所述训练好的神经网络模型，其中，所述模型的输入为空调在冷媒泄漏时的运行参数和环境信息，所述模型的输出为冷媒剩余量。
进一步地，所述装置还包括：处理单元，设置为在获取各个类型的空调在冷媒泄漏时的运行参数和环境信息之后，对所述各个类型的空调在冷媒泄漏时的运行参数和环境信息的参数进行标准化处理，得到标准化参数，其中，所述标准化处理包括线性处理；变换单元，设置为对所述标准化参数进行变换处理，得到变换后的数据，其中，所述变换处理包括非线性处理。
进一步地，所述装置还包括：分类单元，设置为在对所述标准化参数进行变换处理，得到变换后的数据之后，对所述变换后的数据进行分类，得到多个类别的数据；提取单元，设置为从所述多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，其中，所述预设间隔包括预设时间间隔或预设数量间隔；第二训练单元，设置为对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型。
为了实现上述目的，根据本申请的另一方面，还提供了一种存储介质，包括存储的程序，其中，在所述程序运行时控制所述存储介质所在设备执行本申请的用于空调的冷媒泄漏检测方法。
为了实现上述目的，根据本申请的另一方面，还提供了一种处理器，用于运行程序，其中，所述程序运行时执行本申请的用于空调的冷媒泄漏检测方法。
本申请通过获取空调当前的运行参数和空调所处环境的环境信息；将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量；根据冷媒剩余量判断空调是否存在冷媒泄漏，解决了通过人工经验来判断空调是否存在冷媒泄漏不准确的问题，进而达到了通过人工神经网络算法来检测空调冷媒泄漏提高准确性的效果。
附图说明
构成本申请的一部分的附图用来提供对本发明的进一步理解，本发明的示意性实施例及其说明用于解释本发明，并不构成对本发明的不当限定。在附图中：
图1是根据本申请实施例的用于空调的冷媒泄漏检测方法的流程图；
图2是根据本申请实施例的神经网络算法进行计算的示意图；
图3是根据本申请实施例的BP网络的结构设计的示意图；
图4是根据本申请实施例的CNN卷积神经网络的结构设计的示意图；
图5是根据本申请实施例的残差神经网络的结构设计的示意图；
图6是根据本申请实施例的一种神经网络学习目标的示意图；
图7是根据本申请实施例的一种运行人工神经网络算法的智能装置的示意图；
图8是根据本申请实施例的用于空调的冷媒泄漏检测装置的示意图。
具体实施方式
需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。下面将参考附图并结合实施例来详细说明本申请。
为了使本技术领域的人员更好地理解本申请方案，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分的实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
需要说明的是，本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的本申请的实施例。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
本申请实施例提供了一种用于空调的冷媒泄漏检测方法。
图1是根据本申请实施例的用于空调的冷媒泄漏检测方法的流程图，如图1所示，该方法包括以下步骤：
步骤S102：获取空调当前的运行参数和空调所处环境的环境信息。
步骤S104：将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量。
步骤S106：根据冷媒剩余量判断空调是否存在冷媒泄漏。
该实施例采用获取空调当前的运行参数和空调所处环境的环境信息；将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量；根据冷媒剩余量判断空调是否存在冷媒泄漏，解决了通过人工经验来判断空调是否存在冷媒泄漏不准确的问题，进而达到了通过人工神经网络算法来检测空调冷媒泄漏提高准确性的效果。
在本申请实施例中，空调当前的运行参数包括多种类型的参数，例如，空调的压缩机运行转速、压缩机运行时间、室外冷凝器温度等，环境信息包括室内外的环境温度等，将当前运行参数和环境信息输入到训练好的神经网络模型中就可以得到通过模型确定的冷媒剩余量，然后根据冷媒剩余量和空调原有的冷媒量判断是否存在冷媒泄漏，如果存在冷媒泄漏可以通过空调发出提醒，也可以通过预设的智能终端设备发出提醒，还可以控制空调执行预定操作防止或减少空调冷媒的泄漏。
可选地，在将运行参数和环境信息输入到训练好的神经网络模型之前，获取各个类型的空调在冷媒泄漏时的运行参数和环境信息；根据各个类型的空调在冷媒泄漏时的运行参数进行神经网络训练，得到训练好的神经网络模型，其中，模型的输入为空调在冷媒泄漏时的运行参数和环境信息，模型的输出为冷媒剩余量。由于不同空调型号的空调类型不同，可以根据空调类型分别获取各个类型的空调在冷媒泄漏时的运行参数和环境信息，然后分别对各个类型的参数和环境信息进行数据整理，得到基于每个类型的空调的训练好的网络神经模型。
可选地，在获取各个类型的空调在冷媒泄漏时的运行参数和环境信息之后，对各个类型的空调在冷媒泄漏时的运行参数和环境信息的参数进行标准化处理，得到标准化参数，其中，标准化处理包括线性处理；对标准化参数进行变换处理，得到变换后的数据，其中，变换处理包括非线性处理。
线性处理可以是归一化处理等，非线性处理可以是对数变换、平方根变换、立方根变换等处理。通过标准化处理和变换处理可以使采集到的样本数据更加符合模型训练的要求，也可以使模型训练得到的结果更加准确。
可选地，在对标准化参数进行变换处理，得到变换后的数据之后，对变换后的数据进行分类，得到多个类别的数据；从多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，其中，预设间隔包括预设时间间隔或预设数量间隔；对训练样本数据进行神经网络模型训练，得到训练好的神经网络模型。
除了数据处理之外，还可以对数据进行分类，然后从每个类中选取一定数量的数据作为样本数据，选取的方式可以是每隔50个数取一个作为样本数据，也可以是每隔一定的时间，例如24小时取一个数据作为样本数据，还可以是其他方式的间隔。
可选地，在对训练样本数据进行神经网络模型训练，得到训练好的神经网络模型之后，将测试数据输入到训练好的神经网络模型，得到训练好的神经网络模型的输出结果；判断输出结果与测试数据对应的测试结果之间的误差是否小于目标最小误差；在输出结果与测试数据对应的测试结果之间的误差大于等于目标最小误差的情况下，通过更新神经网络模型的权值和偏置调整神经网络模型的参数，直至输出结果与测试数据对应的测试结果之间的误差小于目标最小误差。
如果训练好的神经网络模型的识别结果与真实结果之间的差距较大，则可以继续增加样本数量或者通过修改神经网络模型的权值和偏置来调整神经网络模型的参数，直到得到的输出结果和测试数据对应的测试结果之间的误差小于设定的目标最小误差，在这种情况下，说明模型已经训练完成，可以用于实际预测。
可选地，获取各个类型的空调在冷媒泄漏时的运行参数和环境信息包括：接收各个类型的空调在冷媒泄漏时上报的运行参数和环境信息；和/或通过物联网获取用户实际使用时的各个类型的空调在冷媒泄漏时的运行参数和环境信息。空调的运行参数的来源可以是多种
可选地，根据冷媒剩余量判断空调是否存在冷媒泄漏包括：获取空调的原始冷媒量；判断冷媒剩余量是否小于原始冷媒量；如果冷媒剩余量是否小于原始冷媒量，则判断出空调存在冷媒泄漏。原始冷媒量可以是空调出厂时的冷媒量，也可以是每次空调开机时的冷媒量。
可选地，神经网络模型包括以下任意一项：BP神经网络模型；CNN卷积神经网络模型；残差神经网络模型，在根据冷媒剩余量判断空调存在冷媒泄漏之后，通过预设的方式发出提醒。预设的方式可以是多种类型，例如，空调发出报警声或者显示指示灯，也可以是手机、遥控器或者家中的智能控制中心发出报警声，以提示用户及时检查和维修。除了列举的这三种神经网络模型之外，还可以是其他类型的神经网络模型。
本申请实施例还提供了一种优选实施方式，下面结合优选实施方式对本申请实施例的技术方案进行说明。
本申请实施例利用人工神经网络算法，运用大量空调冷媒泄漏时的运行参数样本，对神经网络进行学习和训练。通过调整网络结构及网络节点间的权值，使神经网络拟合空调运行参数之间的关系，最终使神经网络能准确检测出冷媒泄漏量。
该方法具有很强的非线性映射能力、自学习和自适应能力、泛化能力以及容错能力。相对于传统的规则控制方法，其不依赖于专家经验，通过大量样本数据的学习，网络可以自动不断修正自身参数，最终达到以下效果：
1、该控制算法可自动适用不同的空调型号，通用性好；
2、随着学习样本数据的增加，该控制算法的检测准确率可无限接近于100％。
具体地，本申请实施例的技术方案包括以下几个步骤：
1、原始数据搜集。
首先，搜集空调在冷媒泄漏时，在所有可能的运行环境下的运行参数，并进行详细标注。具体搜集方式包括但不限于空调在实验室模拟环境下的运行参数、通过物联网技术搜集实际用户使用时的空调运行参数等方式。
2、输入、输出参数选择及预处理。
通过对原始数据的分析和结合专家知识，选取对冷媒泄漏检测影响较大且易检测的参数作为输入参数，将冷媒剩余量作为输出量。本申请实施例中，输入参数包括但不限于室内外的环境温度、室外冷凝器温度、室外湿度、压缩机的运行转速、压缩机运行时间等。输入参数不仅为单一参数，也包括按一定规律组成的输入参数矩阵，比如：每秒采集一次室外冷凝器温度，连续采集10分钟的温度参数作为一个输入参数矩阵。
因各参数具有不同的物理意义和量纲，还需要对输入参数作标准化处理，然后再进行统一的变换处理。本申请实施例中用到的数据处理方法包括但不限于，对数据进行归一化等线性处理及对数变换、平方根变换、立方根变换等非线性处理。
3、训练和测试数据样本集选取。
通过对已搜集并标注数据的分析和结合专家知识，对数据样本按一定的规则进行分类。从不同类别的样本中，均匀提取数据，作为训练样本。训练样本不仅要蕴含冷媒泄漏的规律，还要体现出多样性和均匀性。本申请实施例中，可以按空调室、内外环境温度、湿度进行组合，列出所有的样本数据，然后按一定的间隔读取作为训练样本数据；提取出训练样本后，剩余的数据可作为测试数据。
4、网络结构设计。
根据冷媒泄漏的数据特性及其所蕴含的规律，可初步确定网络的基本结构、网络的输入、输出节点数、网络隐层数、隐节点数、网络初始权值等。图2是根据本申请实施例的神经网络算法进行计算的示意图，如图2所示，通过多个输入参数输入到神经网络算法中，可以得到冷媒剩余量。具体人工神经网络结构包括但不限于以下三种网络结构。
4.1BP神经网络。
图3是根据本申请实施例的BP网络的结构设计的示意图，如图3所示，该神经网络算法主要需要解决设几个隐层和几个隐节点的问题。隐层和隐节点的确定需在网络训练时不断的调整。设计时先设置一个隐层，通过调整隐层节点数来改善网络性能；当隐节点数过多，出现过多拟合时，再考虑增加隐层，减少隐节点，来改善网络性能。实际应用时可以根据需要调整输入层、隐层、输出层节点数及隐层层数。
4.2CNN卷积神经网络。
图4是根据本申请实施例的CNN卷积神经网络的结构设计的示意图，如图4所示，多层卷积网络就是从低维度的特征不断提取合并得到更高维的特征从而可以用来进行分类或相关任务。
在本申请实施例中的原始数据是连续采集得到，直观上是和时间相关的。但是可以将一定量的数据一条条组合起来，将其看成和图像一样的数据形式，行与行、列与列之间存在空间连续关系，这些关系影响着最终数据的“标签”，也就是冷媒剩余量。这样就可以通过卷积神经网络对输入数据进行特征提取，准确检测出冷媒剩余量。在实际应用时网络结构可根据实际情况调整。
4.3残差神经网络。
图5是根据本申请实施例的残差神经网络的结构设计的示意图，如图5所示，在调试CNN网络时，加深网络层数和改变卷积核大小的方法并不能使得网络表现得到提升。加入残差块可以更好的连接前后数据，加强特征表达能力，所以其能够加强卷积网络的学习能力。图6是根据本申请实施例的一种神经网络学习目标的示意图，如图6所示，神经网络的输入为x，期望输出为H(x)，把输入x传入到输出作为初始结构后，需要学习的目标就变为F(x)＝H(x)-x。实际应用时网络结构可根据实际情况调整。
5、网络训练与测试。
网络设计完成后，需用训练样本数据，对网络进行训练。训练方法可根据实际的网络结构及训练中发现的问题进行调整。此处仅针对本申请实施例的其中一种方法举例说明如下：
导入输入数据x，根据激活函数、初始化的权值及偏置计算出网络的实际输出a(x)，即a(x)＝1/(1+e -z),其中Z＝W k*x+b l
判断网络的期望输出y(x)与实际输出a(x)是否满足输出精度要求即：
判断是否满足‖y(x)-a(x)‖<∈，∈为目标最小误差，
如果满足精度要求则结束训练，如不满足则根据以下方式更新网络的权值W k,偏 置b l：
C(w,b)为误差能量函数(以标准方差函数为例)，n为训练样本的总数量，求和是在总的训练样本x上进行，
Figure PCTCN2018091070-appb-000001
更新各层权值：
Figure PCTCN2018091070-appb-000002
更新各层偏置：
Figure PCTCN2018091070-appb-000003
其中：W k为初始权值，
Figure PCTCN2018091070-appb-000004
为误差能量函数对权值的偏导数，b l为初始偏置，
Figure PCTCN2018091070-appb-000005
为误差能量函数对偏置的偏导数，
Figure PCTCN2018091070-appb-000006
的值可通过链式求导法则获得，直至网络的输出误差小于∈为止。
网络训练完成后，再用测试样本正向测试网络。当测试误差不满足要求时，则重复以上步骤，重新训练网络；若测试误差满足要求，则网络训练测试完成。
6、人工神经网络算法检测冷媒泄漏的实现
图7是根据本申请实施例的一种运行人工神经网络算法的智能装置的示意图，如图7所示，带无线通讯的空调器运行时，将空调的运行参数上传至智能装置。智能装置将运行参数输入到算法中，判断出冷媒泄漏情况后，向空调发送控制指令。本申请实施例中，智能装置包括但不限于无线通讯模块、路由器、服务器、智能手机等。
需要说明的是，在附图的流程图示出的步骤可以在诸如一组计算机可执行指令的计算机系统中执行，并且，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
本申请实施例提供了一种用于空调的冷媒泄漏检测装置，该装置可以用于执行本申请实施例的用于空调的冷媒泄漏检测方法。
图8是根据本申请实施例的用于空调的冷媒泄漏检测装置的示意图，如图8所示，该装置包括：
第一获取单元10，用于获取空调当前的运行参数和空调所处环境的环境信息；
输入单元20，用于将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量；
判断单元30，用于根据冷媒剩余量判断空调是否存在冷媒泄漏。
该实施例采用第一获取单元10获取空调当前的运行参数和所述空调所处环境的环境信息；输入单元20将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；判断单元30根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏，解决了通过人工经验来判断空调是否存在冷媒泄漏不准确的问题，进而达到了通过人工神经网络算法来检测空调冷媒泄漏提高准确性的效果。
可选地，该装置还包括：第二获取单元，用于在将运行参数和环境信息输入到训练好的神经网络模型之前，获取各个类型的空调在冷媒泄漏时的运行参数；第一训练单元，用于根据各个类型的空调在冷媒泄漏时的运行参数进行神经网络训练，得到训练好的神经网络模型，其中，模型的输入为空调在冷媒泄漏时的运行参数，模型的输出为冷媒剩余量。
可选地，该装置还包括：处理单元，用于在获取各个类型的空调在冷媒泄漏时的运行参数之后，对各个类型的空调在冷媒泄漏时的运行参数进行标准化处理，得到标准化参数，其中，标准化处理包括线性处理；变换单元，用于对标准化参数进行变换处理，得到变换后的数据，其中，变换处理包括非线性处理。
可选地，该装置还包括：分类单元，用于在对标准化参数进行变换处理，得到变换后的数据之后，对变换后的数据进行分类，得到多个类别的数据；提取单元，用于从多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，其中，预设间隔包括预设时间间隔或预设数量间隔；第二训练单元，用于对训练样本数据进行神经网络模型训练，得到训练好的神经网络模型。
此处需要说明的是，上述第一获取单元10、输入单元20和判断单元30可以作为装置的一部分运行在计算机终端中，可以通过计算机终端中的处理器来执行上述模块实现的功能，计算机终端也可以是智能手机(如Android手机、iOS手机等)、平板电脑、掌上电脑以及移动互联网设备(Mobile Internet Devices，MID)、PAD等终端设备。
所述用于空调的冷媒泄漏检测装置包括处理器和存储器，上述第一获取单元、输入单元、判断单元等均作为程序单元存储在存储器中，由处理器执行存储在存储器中的上述程序单元来实现相应的功能。
处理器中包含内核，由内核去存储器中调取相应的程序单元。内核可以设置一个 或以上，通过调整内核参数来通过人工神经网络算法来检测空调冷媒泄漏提高准确性。
存储器可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)，存储器包括至少一个存储芯片。
本申请实施例提供了一种设备，设备包括处理器、存储器及存储在存储器上并可在处理器上运行的程序，处理器执行程序时实现以下步骤：获取空调当前的运行参数和所述空调所处环境的环境信息；将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。本文中的设备可以是服务器、PC、PAD、手机等。
本申请还提供了一种计算机程序产品，当在数据处理设备上执行时，适于执行初始化有如下方法步骤的程序：获取空调当前的运行参数和所述空调所处环境的环境信息；将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请是参照根据本申请实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或 方框图一个方框或多个方框中指定的功能的步骤。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括要素的过程、方法、商品或者设备中还存在另外的相同要素。
本领域技术人员应明白，本申请的实施例可提供为方法、系统或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
以上仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。
工业实用性
本申请通过获取空调当前的运行参数和空调所处环境的环境信息；将当前的运行参数和环境信息输入到训练好的神经网络模型，得到神经网络模型输出的冷媒剩余量；根据冷媒剩余量判断空调是否存在冷媒泄漏，解决了通过人工经验来判断空调是否存在冷媒泄漏不准确的问题，进而达到了通过人工神经网络算法来检测空调冷媒泄漏提高准确性的效果。

Claims (14)

  1. 一种用于空调的冷媒泄漏检测方法，包括：
    获取空调当前的运行参数和所述空调所处环境的环境信息；
    将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；
    根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。
  2. 根据权利要求1所述的方法，其中，在将所述运行参数和环境信息输入到训练好的神经网络模型之前，所述方法还包括：
    获取各个类型的空调在冷媒泄漏时的运行参数和环境信息；
    根据所述各个类型的空调在冷媒泄漏时的运行参数和环境信息进行神经网络训练，得到所述训练好的神经网络模型，其中，所述模型的输入为空调在冷媒泄漏时的运行参数和环境信息，所述模型的输出为冷媒剩余量。
  3. 根据权利要求2所述的方法，其中，在获取各个类型的空调在冷媒泄漏时的运行参数和环境信息之后，所述方法还包括：
    对所述各个类型的空调在冷媒泄漏时的运行参数和环境信息的参数进行标准化处理，得到标准化参数，其中，所述标准化处理包括线性处理；
    对所述标准化参数进行变换处理，得到变换后的数据，其中，所述变换处理包括非线性处理。
  4. 根据权利要求3所述的方法，其中，在对所述标准化参数进行变换处理，得到变换后的数据之后，所述方法还包括：
    对所述变换后的数据进行分类，得到多个类别的数据；
    从所述多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，
    其中，所述预设间隔包括预设时间间隔或预设数量间隔；
    对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型。
  5. 根据权利要求4所述的方法，其中，在对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型之后，所述方法还包括：
    将测试数据输入到所述训练好的神经网络模型，得到所述训练好的神经网络模型的输出结果；
    判断所述输出结果与测试数据对应的测试结果之间的误差是否小于目标最小误差；
    在所述输出结果与测试数据对应的测试结果之间的误差大于等于所述目标最小误差的情况下，通过更新所述神经网络模型的权值和偏置调整所述神经网络模型的参数，直至所述输出结果与测试数据对应的测试结果之间的误差小于所述目标最小误差。
  6. 根据权利要求2所述的方法，其中，获取各个类型的空调在冷媒泄漏时的运行参数和环境信息包括：
    接收各个类型的空调在冷媒泄漏时上报的运行参数和环境信息；和/或
    通过物联网获取用户实际使用时的各个类型的空调在冷媒泄漏时的运行参数和环境信息。
  7. 根据权利要求1所述的方法，其中，根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏包括：
    获取所述空调的原始冷媒量；
    判断所述冷媒剩余量是否小于所述原始冷媒量；
    如果所述冷媒剩余量是否小于所述原始冷媒量，则判断出所述空调存在冷媒泄漏。
  8. 根据权利要求1所述的方法，其中，所述神经网络模型包括以下任意一项：
    BP神经网络模型；
    CNN卷积神经网络模型；
    残差神经网络模型，
    在根据所述冷媒剩余量判断所述空调存在冷媒泄漏之后，通过预设的方式发出提醒。
  9. 一种用于空调的冷媒泄漏检测装置，包括：
    第一获取单元，设置为获取空调当前的运行参数和所述空调所处环境的环境信息；
    输入单元，设置为将所述当前的运行参数和环境信息输入到训练好的神经网络模型，得到所述神经网络模型输出的冷媒剩余量；
    判断单元，设置为根据所述冷媒剩余量判断所述空调是否存在冷媒泄漏。
  10. 根据权利要求9所述的装置，其中，所述装置还包括：
    第二获取单元，设置为在将所述运行参数和环境信息输入到训练好的神经网络模型之前，获取各个类型的空调在冷媒泄漏时的运行参数和环境信息；
    第一训练单元，设置为根据所述各个类型的空调在冷媒泄漏时的运行参数和环境信息进行神经网络训练，得到所述训练好的神经网络模型，其中，所述模型的输入为空调在冷媒泄漏时的运行参数和环境信息，所述模型的输出为冷媒剩余量。
  11. 根据权利要求10所述的装置，其中，所述装置还包括：
    处理单元，设置为在获取各个类型的空调在冷媒泄漏时的运行参数和环境信息之后，对所述各个类型的空调在冷媒泄漏时的运行参数和环境信息的参数进行标准化处理，得到标准化参数，其中，所述标准化处理包括线性处理；
    变换单元，设置为对所述标准化参数进行变换处理，得到变换后的数据，其中，所述变换处理包括非线性处理。
  12. 根据权利要求11所述的装置，其中，所述装置还包括：
    分类单元，设置为在对所述标准化参数进行变换处理，得到变换后的数据之后，对所述变换后的数据进行分类，得到多个类别的数据；
    提取单元，设置为从所述多个类别的数据中分别按照预设间隔提取数据，作为训练样本数据，其中，所述预设间隔包括预设时间间隔或预设数量间隔；
    第二训练单元，设置为对所述训练样本数据进行神经网络模型训练，得到所述训练好的神经网络模型。
  13. 一种存储介质，所述存储介质包括存储的程序，其中，在所述程序运行时控制所述存储介质所在设备执行权利要求1至8中任意一项所述的用于空调的冷媒泄漏检测方法。
  14. 一种处理器，所述处理器用于运行程序，其中，所述程序运行时执行权利要求1至8中任意一项所述的用于空调的冷媒泄漏检测方法。
PCT/CN2018/091070 2017-09-15 2018-06-13 用于空调的冷媒泄漏检测方法和装置 WO2019052241A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP18856758.0A EP3637328A4 (en) 2017-09-15 2018-06-13 METHOD AND DEVICE FOR DETECTING LEAKAGE OF REFRIGERANT FLUID FOR AIR CONDITIONER
US16/633,580 US11614249B2 (en) 2017-09-15 2018-06-13 Refrigerant leak detection method and device for air conditioner

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710839637.3A CN107730000A (zh) 2017-09-15 2017-09-15 用于空调的冷媒泄漏检测方法和装置
CN201710839637.3 2017-09-15

Publications (1)

Publication Number Publication Date
WO2019052241A1 true WO2019052241A1 (zh) 2019-03-21

Family

ID=61206585

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/091070 WO2019052241A1 (zh) 2017-09-15 2018-06-13 用于空调的冷媒泄漏检测方法和装置

Country Status (4)

Country Link
US (1) US11614249B2 (zh)
EP (1) EP3637328A4 (zh)
CN (1) CN107730000A (zh)
WO (1) WO2019052241A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110348398A (zh) * 2019-07-15 2019-10-18 上海点积实业有限公司 一种神经网络模型确定方法和系统

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107730000A (zh) * 2017-09-15 2018-02-23 珠海格力电器股份有限公司 用于空调的冷媒泄漏检测方法和装置
CN110454904B (zh) * 2018-05-07 2020-12-11 珠海格力电器股份有限公司 除湿机中冷媒含量的处理方法和装置
CN109282418A (zh) * 2018-11-26 2019-01-29 珠海格力电器股份有限公司 冷风机缺氟保护报警模型的构建方法、报警方法及冷风机
CN109764476B (zh) * 2018-12-18 2020-09-04 珠海格力电器股份有限公司 冷媒泄漏监控方法、监控系统及空调机组
CN110657552A (zh) * 2019-09-04 2020-01-07 特灵空调系统(中国)有限公司 故障检测方法、换热系统及计算机可读存储介质
US11231198B2 (en) 2019-09-05 2022-01-25 Trane International Inc. Systems and methods for refrigerant leak detection in a climate control system
JP6791429B1 (ja) * 2019-09-09 2020-11-25 ダイキン工業株式会社 冷媒量判定装置、方法、およびプログラム
CN110671779B (zh) * 2019-10-24 2022-01-25 宁波奥克斯电气股份有限公司 空调器的冷媒余量检测方法、装置、服务器及存储介质
CN111750491B (zh) * 2020-05-14 2022-01-18 海信（山东）空调有限公司 一种空调及基于神经网络的空调控制方法
CN112288057A (zh) * 2020-11-04 2021-01-29 广州粤创富科技有限公司 一种多模式宠物实时定位的方法、设备及存储介质
CN112665107B (zh) * 2020-12-21 2023-03-14 广东美的暖通设备有限公司 制冷剂泄漏检测方法、装置及设备
CN112665790B (zh) * 2020-12-21 2023-03-14 广东美的暖通设备有限公司 冷媒泄漏检测方法、装置及设备
CN113091233B (zh) * 2021-04-01 2023-04-25 特灵空调系统(中国)有限公司 空调冷媒量确定方法、系统和可读存储介质
CN113357773A (zh) * 2021-05-31 2021-09-07 青岛海尔空调器有限总公司 用于空调控制的方法、装置和空调
CN113919395A (zh) * 2021-10-12 2022-01-11 大连理工大学 基于一维卷积神经网络的给水管网渗漏事故诊断方法
CN114234360B (zh) * 2021-11-09 2023-06-23 海信（广东）空调有限公司 空调器及其制冷剂泄漏检测方法
CN114088212A (zh) * 2021-11-29 2022-02-25 浙江天铂云科光电股份有限公司 基于温度视觉的诊断方法及诊断装置
KR20230088078A (ko) * 2021-12-10 2023-06-19 삼성전자주식회사 전자 장치 및 그 제어 방법
CN114838496B (zh) * 2022-04-22 2024-02-23 北京百车宝科技有限公司 基于人工智能的空调消声器性能检测方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000222046A (ja) * 1999-02-01 2000-08-11 Isao Takahashi ニューラルネット温度コントロール装置
CN106322656A (zh) * 2016-08-23 2017-01-11 海信（山东）空调有限公司 一种空调控制方法及服务器和空调系统
CN106705384A (zh) * 2017-02-09 2017-05-24 美的集团股份有限公司 冷媒泄漏的提醒方法及装置和空调器
CN107730000A (zh) * 2017-09-15 2018-02-23 珠海格力电器股份有限公司 用于空调的冷媒泄漏检测方法和装置

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100432224B1 (ko) * 2002-05-01 2004-05-20 삼성전자주식회사 공기 조화기의 냉매 누설 검출 방법
US7111469B2 (en) * 2004-09-22 2006-09-26 Horan Christopher J Process for refrigerant charge level detection using a neural net
US20110112814A1 (en) * 2009-11-11 2011-05-12 Emerson Retail Services, Inc. Refrigerant leak detection system and method
CN104566771A (zh) * 2013-10-23 2015-04-29 珠海格力电器股份有限公司 除湿机的冷媒控制方法及装置
CN105372087B (zh) 2015-11-30 2018-06-22 南通大学 基于多传感器信号分析的故障诊断方法
CN105928156B (zh) * 2016-05-12 2018-08-07 广东美的制冷设备有限公司 冷媒泄漏检测方法、冷媒泄漏检测装置及空调器
DE102016110585A1 (de) * 2016-06-08 2017-12-14 Truma Gerätetechnik GmbH & Co. KG Klimasystem und Verfahren zur Leckageerkennung in einem Klimasystem

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000222046A (ja) * 1999-02-01 2000-08-11 Isao Takahashi ニューラルネット温度コントロール装置
CN106322656A (zh) * 2016-08-23 2017-01-11 海信（山东）空调有限公司 一种空调控制方法及服务器和空调系统
CN106705384A (zh) * 2017-02-09 2017-05-24 美的集团股份有限公司 冷媒泄漏的提醒方法及装置和空调器
CN107730000A (zh) * 2017-09-15 2018-02-23 珠海格力电器股份有限公司 用于空调的冷媒泄漏检测方法和装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3637328A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110348398A (zh) * 2019-07-15 2019-10-18 上海点积实业有限公司 一种神经网络模型确定方法和系统
CN110348398B (zh) * 2019-07-15 2022-12-13 上海点积实业有限公司 一种神经网络模型确定方法和系统

Also Published As

Publication number Publication date
EP3637328A4 (en) 2020-07-15
CN107730000A (zh) 2018-02-23
EP3637328A1 (en) 2020-04-15
US11614249B2 (en) 2023-03-28
US20200208861A1 (en) 2020-07-02

Similar Documents

Publication Publication Date Title
WO2019052241A1 (zh) 用于空调的冷媒泄漏检测方法和装置
US11677576B2 (en) Systems and approaches for establishing relationships between building automation system components
US11385611B2 (en) Advanced identification and classification of sensors and other points in a building automation system
Li et al. Application of pattern matching method for detecting faults in air handling unit system
Fan et al. A study on semi-supervised learning in enhancing performance of AHU unseen fault detection with limited labeled data
Yuwono et al. Unsupervised feature selection using swarm intelligence and consensus clustering for automatic fault detection and diagnosis in heating ventilation and air conditioning systems
CN110610275A (zh) 一种基于acqpso-elm的变风量空调负荷预测方法及系统
US11371741B2 (en) Air conditioning apparatus and method for controlling using learned sleep modes
Umoh et al. Support vector machine-based fire outbreak detection system
CN114217025B (zh) 评估空气质量浓度预测中气象数据对其影响的分析方法
CN113552855B (zh) 工业设备动态阈值设定方法、装置、电子设备及存储介质
US20180293441A1 (en) Method and apparatus for real-time detection of a scene
CN113989632A (zh) 一种遥感图像桥梁检测方法、装置、电子设备及存储介质
CN114236276B (zh) 一种电器远程测试的方法和系统
Ferreira et al. Unsupervised learning of load signatures to estimate energy-related building features using surrogate modelling techniques
US20240012175A1 (en) Apparatus and method for calculating sensible temperature in consideration of outdoor ground heating and heatwave warning apparatus and method based on sensible temperature in consideration of outdoor ground heating
Park Data-driven building metadata inference
CN117392614B (zh) 实验室安全风险智能检测方法、装置以及应急安全柜
US20210405631A1 (en) Machine Learning Based Direct Method of Determining Status of Facility Control Loop Components
CN110454904B (zh) 除湿机中冷媒含量的处理方法和装置
Vorger et al. Reconstruction of Building Occupation by Machine Learning
Vaithilingam et al. Intelligent Air Conditioning Systems: Enhancing Energy Efficiency and Indoor Air Quality through IoT and Air Conditioning Unit Using Machine Learning
Jia et al. Building Thermal Load Prediction Using Spatio-Temporal Graph Neural Networks
CN118014184A (zh) 一种基于数字孪生的猪场预警方法、系统及存储介质
CN113516875A (zh) 一种多媒体教学设备智能控制方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18856758

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2018856758

Country of ref document: EP

Effective date: 20200107

NENP Non-entry into the national phase

Ref country code: DE