Air conditioning device

Abstract

An air conditioning device includes: a refrigerant circuit connecting a compressor, an outdoor heat exchanger, a throttling element, and an indoor heat exchanger to circulate a refrigerant therein; the processing part is configured to be capable of collecting operation parameters of the air conditioning device; a first probability that refrigerant flow sound higher than a set abnormal sound threshold value is generated and a second probability that refrigerant flow sound higher than the set abnormal sound threshold value is not generated in the refrigerant loop are predicted based on a neural network model, and input data of the neural network model comprise the collected operation parameters; the method further comprises one or more of the refrigerant charge, the refrigerant impurity proportion and the aging degree of a valve element arranged in the refrigerant loop. When the first probability is higher than the second probability, an intervention signal is output to perform a noise reduction pre-processing. According to the method, the neural network model is used for learning, and the generation of the refrigerant flow sound is captured and predicted, so that automatic prediction and active intervention are realized, and the condition of passive response after a complaint is received is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioning, and in particular to an air conditioning device. Background Art

[0002] As the use of air-conditioning systems becomes more and more common, various non-standard installations are increasingly being installed in the market due to various human factors or geographical factors. The problems caused by the comprehensive installation, some users overload the air-conditioning system, which will lead to problems such as aging of components and dust accumulation. These factors will further cause abnormal sounds in the air-conditioning system unit (abnormal sound for short). When dealing with noise problems, the subjective factor of human hearing is relatively large, and the most difficult to judge is the sound of refrigerant flow. Some users with sensitive hearing will complain to the manufacturer about the noise problems caused by the friction between the refrigerant and pipes, valves and other components in the air-conditioning system when they are resting at night.

[0003] At present, the manufacturer does not have a good solution. Most of the time, it can only respond passively. After receiving complaints, it conducts on-site re-debugging and modifies the general control logic of the unit to try to solve the problem. The solution is not ideal, and the efficiency is low and the cost is very high. There are some methods for calculating the flow sound of refrigerants disclosed in the prior art, such as the technical solution disclosed in the Chinese patent application (CN116245034A): "The invention discloses a joint calculation method for refrigerant flow sound based on computational fluid dynamics and computational acoustics, which belongs to the field of refrigerant flow sound simulation analysis. The flow field of the refrigerant when flowing through the throttling device is calculated by computational fluid dynamics, and then the flow field calculation result is used as the sound source to import into the acoustic calculation software for calculation, and the fluid-solid coupling calculation method is used to make the calculation result closer to the actual result. It is of great value for studying the generation mechanism of refrigerant flow sound and developing a low-noise system."

[0004] It can be seen that the comparative documents still use the traditional refrigeration system cycle benchmarks and formulas and use simulation to identify and process the refrigerant flow sound, but all processing methods are limited to experience and formulas, which makes it extremely difficult to apply them in products and has poor generalization. Summary of the invention

[0005] In order to solve the problem that abnormal noises in air-conditioning systems, especially abnormal noises in the flow of refrigerant, are difficult to judge, and the processing methods provided by the existing technology are limited to experience and formulas, are difficult to apply in products and have poor generalization, the first aspect of this application designs and provides an air conditioning device.

[0006] An air conditioning device includes a refrigerant circuit which connects a compressor, an outdoor heat exchanger, a throttling element and an indoor heat exchanger so that the refrigerant circulates therein.

[0007] In one or more embodiments of the present application, the air conditioning device also includes: a processing unit, which is configured to collect operating parameters of the air conditioning device; predicting a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold in the refrigerant circuit based on a neural network model, the input data of the neural network model including the collected operating parameters, and also including one or more of the refrigerant filling amount, the refrigerant impurity ratio, and the degree of aging of the valve element arranged in the refrigerant circuit; outputting an intervention signal to perform noise reduction preprocessing when the first probability is higher than the second probability.

[0008] In one or more embodiments of the present application, input data of the neural network model includes collected operating parameters and refrigerant charge, and the neural network model is trained based on the refrigerant charge data set.

[0009] In one or more embodiments of the present application, a refrigerant filling amount data set is obtained by the following method: selecting multiple simulated air-conditioning systems, each of which has an outdoor unit with a different refrigeration capacity; configuring multiple preset refrigerant filling states and preset indoor unit combination opening and closing states, and configuring each simulated air-conditioning system to operate under different preset refrigerant filling states and different preset indoor unit combination opening and closing states; detecting the refrigerant flow sound of each simulated air-conditioning system under different preset refrigerant filling states and different preset indoor unit combination opening and closing states, and comparing the measured refrigerant flow sound with the set abnormal sound threshold, and recording the test results through binary tags; collecting the operating parameters, refrigerant filling state, indoor unit combination opening and closing state and binary tag detection results of the simulated air-conditioning system to establish a refrigerant filling amount data set.

[0010] In one or more embodiments of the present application, the preset refrigerant charging state includes: a standard refrigerant charging state, a refrigerant under-charging state and a refrigerant over-charging state.

[0011] In one or more embodiments of the present application, the combined on-off state of the indoor units includes: operating one indoor unit, operating some indoor units, and operating all indoor units.

[0012] In one or more embodiments of the present application, the input data of the neural network model includes the collected operating parameters and refrigerant impurity ratios, and the neural network model is trained based on the refrigerant impurity data set.

[0013] In one or more embodiments of the present application, a refrigerant impurity data set is obtained by the following method: mixing impurities of different set proportions into a set amount of refrigerant; detecting the refrigerant flow sound when a simulated air-conditioning system is running with impurities of different set proportions mixed therein, and comparing the measured refrigerant flow sound with a set abnormal sound threshold, and recording the test results through a binary tag; collecting the operating parameters of the simulated air-conditioning system, the set impurity ratio and the binary tag detection results to establish a refrigerant impurity data set.

[0014] In one or more embodiments of the present application, a refrigerant impurity data set is obtained by the following method: mixing a set proportion of impurities into a set amount of refrigerant; detecting the refrigerant flow sound of a simulated air-conditioning system when the refrigerant is mixed with the set proportion of impurities; comparing the measured refrigerant flow sound with the set abnormal sound threshold, recording the test results through binary tags, and establishing a data group including operating parameters, refrigerant impurity ratio and binary tag detection results; increasing the impurity ratio in the set refrigerant according to the set ratio; estimating whether the set impurity ratio upper limit is reached, if the set impurity ratio upper limit is not reached, then detecting the refrigerant flow sound of the simulated air-conditioning system when the refrigerant is running with the impurities increased in the set proportion again, and comparing the measured refrigerant flow sound with the set abnormal sound threshold, recording the test results through binary tags, and establishing a data group including operating parameters, the increased refrigerant impurity ratio and binary tag detection results, until the set impurity ratio upper limit is reached, and using all data groups to establish a refrigerant impurity data set.

[0015] In one or more embodiments of the present application, input data of the neural network model includes collected operating parameters and aging degree of valve components in the refrigerant circuit, and the neural network model is trained based on the valve component aging degree data set.

[0016] In one or more embodiments of the present application, a valve element aging degree data set is obtained by the following method: respectively installing valve elements with different aging cycles in a simulated air-conditioning system, detecting the refrigerant flow sound of the simulated air-conditioning system when valve elements with different aging cycles are installed, and comparing the measured refrigerant flow sound with the set abnormal sound threshold, and recording the test results through binary tags; collecting the operating parameters of the simulated air-conditioning system, different aging cycles and binary tag detection results to establish a valve element aging degree data set.

[0017] In one or more embodiments of the present application, in an initial state, the valve element has a first aging cycle; after collecting the operating parameters of the simulated air-conditioning system, the first aging cycle and the binary tag detection results, the valve element is placed in an aging test environment for multiple times until the timing cycle ends, and after each timing cycle ends, the valve element is placed in a simulated air-conditioning system, and the operating parameters of the simulated air-conditioning system are collected again, and the aging cycle and the binary tag detection results are updated, wherein the updated aging cycle is the sum of the timing cycle and the duration of the updated aging cycle of the previous cycle; the timing cycle is generated based on the designed service life of the valve element; the temperature, humidity and salinity of the aging test environment are higher than the temperature, humidity and salinity of the air conditioning device use environment.

[0018] In one or more embodiments of the present application, the operating parameters include one or more of a set temperature, a return air temperature, a supply air temperature, a set wind speed, a gas pipe temperature, a liquid pipe temperature, an indoor throttling element opening, an outdoor temperature, a compressor operating frequency, and an outdoor throttling element opening.

[0019] In one or more embodiments of the present application, a processing unit is configured to collect operating parameters of the air conditioning device; predict a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold based on a neural network model, wherein input data of the neural network model include the collected operating parameters, refrigerant charge, refrigerant impurity ratio, and aging degree of valve elements in the refrigerant circuit; infer the cause of the refrigerant flow sound to be the refrigerant charge, the refrigerant impurity ratio, or the aging degree of the valve elements based on the neural network model, and output an intervention signal to perform noise reduction preprocessing when the first probability is higher than the second probability.

[0020] In the air conditioning device provided in the present application, a processing unit configured with a neural network model is used to predict the refrigerant flow sound, which can learn and capture nonlinear relationships in the data and extract key features from the original data, better capture the pattern of the refrigerant flow sound, and through data collection, better generalization can be achieved, complex relationships can be captured and effectively integrated and processed, thereby improving the prediction performance of the flow sound. Manufacturers can intervene in advance before the flow sound occurs to avoid passive responses after receiving complaints.

[0021] After reading the specific embodiments of the present invention in conjunction with the accompanying drawings, other features and advantages of the present invention will become more clear. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0023] Figure 1 A schematic diagram of the structure of an air conditioning device in one or more embodiments of the air conditioning device provided in this application;

[0024] Figure 2 A schematic diagram of the structure of an air conditioning device in one or more embodiments of the air conditioning device provided in this application;

[0025] Figure 3 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0026] Figure 4 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0027] Figure 5 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0028] Figure 6 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0029] Figure 7 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0030] Figure 8 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0031] Fig. 9 An example of a refrigerant charge dataset;

[0032] Fig.10 An example of a refrigerant charge dataset;

[0033] Fig.11 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0034] Fig.12 is an example of a refrigerant impurity dataset;

[0035] Fig.13 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0036] Fig.14 This is an example of a valve component aging degree dataset;

[0037] Fig.15 A flow chart of one or more embodiments of the air conditioning device provided in this application;

[0038] In the figure: 100, air conditioning device; 10, outdoor unit; 11, indoor unit 101, compressor; 102, outdoor heat exchanger; 103, outdoor throttling element; 104a, indoor throttling element; 104b, indoor throttling element; 105, indoor heat exchanger; 105a, indoor heat exchanger; 105b, indoor heat exchanger; 106, outdoor fan; 107, indoor fan; 108, gas-liquid separator; 109, gas-side shut-off valve; 110, liquid-side shut-off valve; 20, processing unit. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0040] In the description of the present application, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0041] The terms "first", "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0042] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0043] In the present invention, unless otherwise clearly stipulated and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes the first feature being directly above and obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes the first feature being directly below and obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0044] The disclosure below provides many different embodiments or examples to realize different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplicity and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed. In addition, the present invention provides various specific examples of processes and materials, but those of ordinary skill in the art can be aware of the application of other processes and / or the use of other materials.

[0045] Figure 1 This is a schematic diagram of the structure of the air conditioning device provided in this application. An air conditioning device refers to a device or system that can heat, cool, dehumidify, and other processes on the air, including household air conditioners, commercial air conditioners, industrial air conditioners, air handling units, and centralized air conditioning systems. An air conditioning device can independently perform cooling operations or independently perform heating operations.

[0046] The air conditioning device includes an outdoor unit 10 and an indoor unit 11, and a liquid-side connecting pipe 112 and a gas-side connecting pipe 113 connecting the outdoor unit 10 and the indoor unit to form a vapor compression refrigeration cycle. Figure 1As shown, the refrigeration cycle is mainly realized by a refrigerant circuit, in which components such as a compressor 101, a condenser, a throttling element and an evaporator are used. The refrigeration cycle includes a series of processes involving compression, condensation, expansion and evaporation to cool or heat the indoor space.

[0047] From a principle perspective, low-temperature, low-pressure refrigerant enters the compressor 101, which compresses it into high-temperature, high-pressure refrigerant gas and discharges the compressed refrigerant gas. The discharged refrigerant gas flows into the condenser. The condenser condenses the compressed refrigerant into a liquid phase, and heat is released to the surrounding environment through the condensation process.

[0048] The throttling element expands the high-temperature and high-pressure liquid refrigerant condensed in the condenser into a low-pressure liquid refrigerant. The evaporator evaporates the refrigerant expanded in the throttling device and returns the refrigerant gas in a low-temperature and low-pressure state to the compressor 101. The evaporator can achieve a refrigeration effect by utilizing the latent heat of evaporation of the refrigerant to exchange heat with the material to be cooled. In the whole cycle, the split air conditioner 100 can adjust the temperature of the indoor space.

[0049] The outdoor unit 10 is introduced as follows: the outdoor unit 10 is arranged in an outdoor natural environment, and is a part of the refrigeration cycle including a compressor 101, a switching valve 114 (usually a four-way valve), an outdoor heat exchanger 102, an outdoor throttling element 103 (usually an electronic expansion valve), a liquid-side shutoff valve 110, and a gas-side shutoff valve 109. The outdoor unit 10 can perform heating operation or cooling operation on the outdoor side to provide the indoor unit 11 with energy for increasing the indoor temperature or reducing the indoor temperature.

[0050] The compressor 101 is used to suck in and compress a low-pressure gas refrigerant into a high-pressure gas refrigerant, and then discharge it. The compressor 101 may be a reciprocating compressor, a screw compressor, a centrifugal compressor, etc. The compressor 101 is driven by a motor, and the speed of the motor may be variably controlled by a frequency converter. The number of compressors 101 may be one or more.

[0051] The switching valve 114 is used to switch the flow direction of the refrigerant between the cooling mode and the heating mode. The switching valve 114 is usually a four-way valve.

[0052] The outdoor heat exchanger 102 is configured to be used as a condenser in cooling operation and as an evaporator in heating operation. The outdoor heat exchanger 102 can perform heat exchange with the air guided by the outdoor fan 106 so that the refrigerant flowing in the outdoor heat exchanger 102 undergoes a phase change (condensation or evaporation). The outdoor fan 106 is driven by a motor, and the speed of the motor can be controlled to change the flow rate of the air heat exchanged with the outdoor heat exchanger 102 by adjusting the speed. The outdoor fan 106 can be an axial flow fan, a cross flow fan or other optional fan types. The outdoor fan 106 is arranged near the outdoor heat exchanger 102.

[0053] The outdoor throttling element 103 (electronic expansion valve) is used to reduce the pressure of the high-pressure refrigerant in the cooling mode and the heating mode.

[0054] The liquid-side shutoff valve 110 and the gas-side shutoff valve 109 are located at the interface connecting the external equipment and the pipeline.

[0055] The outdoor unit 10 is also provided with a gas-liquid separator 108 (liquid storage device). The gas-liquid separator 108 is provided on the suction side of the compressor 101 and is a shell-shaped component for separating and storing the refrigerant into gas and liquid. The gas-liquid separator 108 can store excess refrigerant.

[0056] In the heating mode, a refrigerant circuit for heating operation can be formed in the outdoor unit 10, which is referred to as a heating cycle. The outdoor throttling element 103, the outdoor heat exchanger 102, the switching valve 114 (one path in the four-way valve), the gas-liquid separator 108, the compressor 101 and the switching valve 114 (the other path in the four-way valve) are connected in sequence from the liquid-side connecting pipe 112 to the gas-side connecting pipe 113.

[0057] In the cooling mode, a refrigerant circuit for cooling operation can be formed in the outdoor unit 10, which is referred to as a refrigeration cycle. The direction from the gas-side connecting pipe 113 to the liquid-side connecting pipe 112 is connected in sequence with a switching valve 114 (one path in the four-way valve), a gas-liquid separator 108, a compressor 101, a switching valve 114 (another path in the four-way valve), an outdoor heat exchanger 102 and an outdoor throttling element 103.

[0058] The entire outdoor unit 10 is controlled by an outdoor controller, which is arranged in an electrical box with good sealing performance and heat dissipation function. The outdoor controller includes components such as a processor, a storage unit, an input / output interface, and a communication interface. The processor can be a dedicated processor, a central processing unit (CPU), etc. The processor can access the storage unit to execute instructions or applications stored in the storage unit to realize related functions. The storage unit can include a volatile memory and / or a non-volatile memory. The input / output interface can be connected to various sensors arranged in the outdoor unit 10 to receive the detection values ​​of various sensors arranged in the outdoor unit 10. The input / output interface can also be connected to devices such as a frequency converter, a compressor 101, an outdoor fan 106, a switching valve 114, and an outdoor throttling element 103 to output control instructions generated by the processor to them. The communication interface can support different wireless communication protocols, such as Wi-Fi, Bluetooth, near-field communication, NB-IoT, etc., to communicate with other electronic devices, including but not limited to cloud servers, computers (host computers), smart phones, tablets, PDAs, smart control tools, wearable devices and vehicle-mounted equipment, etc.

[0059] The indoor unit 11 is described below. The indoor unit 11 performs cooling operation or heating operation using energy generated by the outdoor unit 10 to increase the indoor temperature or energy to decrease the indoor temperature.

[0060] The indoor unit 11 includes an indoor fan 107 and an indoor heat exchanger 105. The indoor unit 11 generally further includes a casing to form an air duct to guide air flow.

[0061] The indoor fan 107 is used to introduce air into the air-conditioned room from the return air port of the shell, and after heat exchange with the indoor heat exchanger 105, the air is sent into the air-conditioned room through the air supply port on the shell.

[0062] The indoor heat exchanger 105 is used as a condenser in heating operation and as an evaporator in cooling operation. The refrigerant in the indoor heat exchanger 105 exchanges heat with the air guided by the indoor fan 107, so that the refrigerant therein undergoes a phase change (condensation or evaporation).

[0063] An indoor controller is provided in the indoor unit 11. The indoor controller includes components such as a processor, a storage unit, an input / output interface, and a communication interface. The processor may be a dedicated processor, a central processing unit (CPU), etc. The processor may access the storage unit to execute instructions or applications stored in the storage unit to implement related functions. The storage unit may include a volatile memory and / or a non-volatile memory. The input / output interface may be communicatively connected to various sensors provided in the indoor unit 11 to receive detection values ​​of various sensors provided in the indoor unit 11. The communication interface may support different wireless communication protocols, such as Wi-Fi, Bluetooth, etc.

[0064] The indoor controller is communicatively connected to a control terminal, which may be a remote controller, a wired controller, or other mobile terminals such as a smart phone, a tablet computer, a computer, a wearable device, and the like.

[0065] The number of indoor units can be one or more, for example Figure 2 As shown, an indoor unit 11a and an indoor unit 11b are provided in the air conditioning device. The indoor unit may be a ceiling-mounted indoor unit, a wall-mounted indoor unit, a floor-standing indoor unit, a duct-type indoor unit, etc. The indoor unit 11b is also provided with components such as an indoor heat exchanger and an indoor fan. Figure 2 The indoor heat exchangers 105a, 105b and the indoor fans 107a, 107b are shown.

[0066] In one or more embodiments of the present application, when multiple indoor units are provided, the indoor units are also provided with indoor throttling elements (such as Figure 2 The indoor throttling element is configured to reduce the pressure of the refrigerant and expand it, and the opening of the indoor throttling element is adjustable, for example, an electronic expansion valve can be selected.

[0067] The flow of refrigerant in the refrigerant circuit may cause some problems. If the refrigerant flow sound is too significant, it will cause noise pollution in the indoor or surrounding environment, interfere with the living or working environment, and affect people's quality of life and work efficiency. In addition, abnormal refrigerant flow sound may also indicate that there are fluid mechanics problems in the refrigeration system, such as pipe blockage, valve failure, or unstable operation of the refrigeration system, resulting in decreased system efficiency, increased energy consumption, and reduced air-conditioning effect. It may also cause increased mechanical loss and shortened equipment life. In order to immediately identify this series of problems and perform noise reduction pre-processing, in the air conditioning device provided in one or more embodiments of the present application, a processing unit 20 is constructed based on a neural network model.

[0068] The processing unit 20 provided in the air conditioning device is configured to collect the operating parameters of the air conditioning device, and predict the first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit and the second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold based on the neural network model. In one or more embodiments of the present application, the input data of the neural network model includes the collected operating parameters, and also includes one or more of the refrigerant charge amount, the refrigerant impurity ratio, and the aging degree of the valve element in the refrigerant circuit. When the first probability is higher than the second probability, an intervention signal is output to perform noise reduction preprocessing.

[0069] The processing unit 20 can be implemented by an outdoor controller, or by an indoor controller, or by an indoor controller and an outdoor controller that are connected to each other in communication, or by a cloud server that is connected to the indoor controller and / or the outdoor controller in communication. The neural network model in the processing unit 20 is a mathematical model based on the structure of a biological nervous system. The neural network model automatically adjusts the connection weights and achieves accurate prediction of the output by learning the patterns and features of the input data. The neural network model can adapt to nonlinear relationships and can learn abstract representations from the data.

[0070] In one or more embodiments of the present application, the abnormal sound threshold may be set according to a refrigeration equipment noise standard, a building noise standard, an environmental noise standard and / or a refrigeration equipment performance standard, for example, set to 30-60 dB.

[0071] In one or more embodiments of the present application, the neural network model selects operating parameters and refrigerant charge as input data. The refrigerant charge is one of the causes of refrigerant flow noise. If the refrigerant charge in the refrigerant circuit is insufficient, the refrigerant may not flow smoothly in the pipeline, causing turbulence in the pipeline and causing noise. Excessive refrigerant charge may also increase the pressure in the pipeline and increase the flow velocity, resulting in refrigerant flow noise.

[0072] Based on this, Figure 3 As shown, in one or more embodiments of the present application, the processing unit 20 is configured to perform the following Figure 3 Multiple steps shown.

[0073] Step S101: Collecting the operating parameters of the air conditioning device. In one or more embodiments of the present application, the operating parameters of the air conditioning device include one or more of the set temperature, return air temperature, supply air temperature, set wind speed (speed of the indoor fan), gas pipe temperature, liquid pipe temperature, indoor throttling element opening, outdoor temperature, compressor operating frequency, and outdoor throttling element opening.

[0074] Step S102: Acquire the real-time refrigerant charge of the air conditioning device. The real-time refrigerant charge can be estimated based on the real-time pressure in the refrigerant circuit, or can be estimated using other algorithms in the prior art, or can be detected using other detection methods in the prior art.

[0075] Step S103: predicting a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold in the refrigerant circuit based on the neural network model.

[0076] Step S104: The first probability predicted by the neural network model is higher than the second probability.

[0077] Step S105: The processing unit 20 outputs an intervention signal to perform noise reduction pre-processing.

[0078] In one or more embodiments of the present application, the neural network model selects operating parameters and refrigerant impurity ratio as input data. The refrigerant impurity ratio is one of the causes of refrigerant flow noise. Impurities in the refrigerant, especially at a high ratio, may cause the refrigerant to flow unsteadily, generating turbulence or irregular flow, thereby causing flow noise. It may also cause the refrigerant to form an uneven flow field in the pipeline, generating fluid mechanics effects such as eddies, thereby causing flow noise. Impurities in the refrigerant will also increase friction and vibration during the flow of the refrigerant, thereby increasing the possibility of generating flow noise.

[0079] Based on this, Figure 4 As shown, in one or more embodiments of the present application, the processing unit 20 is configured to perform the following Figure 4 Multiple steps shown.

[0080] Step S201: Collecting the operating parameters of the air conditioning device. The selection of the operating parameters is described in the above embodiment and will not be repeated here.

[0081] Step S202: Obtain the refrigerant impurity ratio in the refrigerant circuit.

[0082] In one or more embodiments of the present application, a filter is provided in the refrigerant circuit, and a differential pressure sensor is provided at the filter. The differential pressure sensor measures the pressure difference when the refrigerant passes through the filter. When metal particles or other impurities accumulate at the filter, the resistance will increase, resulting in a change in the differential pressure. The detection of the differential pressure change can detect the accumulation of metal particles at the filter. The more metal particles accumulate at the filter, the higher the proportion of refrigerant impurities in the refrigerant circuit. Thus, the proportion of refrigerant impurities in the refrigerant circuit can be obtained by the pressure difference (differential pressure) measured by the differential pressure sensor. In addition to the differential pressure sensor, the proportion of refrigerant impurities in the refrigerant circuit can also be obtained by a vibration sensor, a resistance sensor, or a magnetic sensor.

[0083] In one or more embodiments of the present application, the refrigerant impurity ratio in the refrigerant circuit may also be acquired by a mass spectrometer sensor.

[0084] Step S203: predicting, based on the neural network model, a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold.

[0085] Step S204: The first probability predicted by the neural network model is higher than the second probability.

[0086] Step S205: The processing unit 20 outputs the intervention signal to perform noise reduction pre-processing.

[0087] In one or more embodiments of the present application, the neural network model selects operating parameters and the degree of aging of valve elements as input data. In the refrigerant circuit, the valve elements include electronic expansion valves as indoor throttling elements and outdoor throttling elements, and solenoid valves as shut-off valves. The aging of the valve elements may cause the sealing performance to deteriorate and leak. The leak may cause the refrigerant flow to be unstable and produce abnormal noise. The structure inside the valve element may cause the refrigerant flow to be blocked due to corrosion or blockage, which may also cause flow sound. The aging of the valve element may also cause irregular vibrations during operation, which are transmitted through the pipeline and generate noise when the refrigerant flows. It may also cause inflexible movement or inability to close completely, causing the refrigerant to bypass the valve part, resulting in abnormal flow and flow sound.

[0088] Based on this, Figure 5 As shown, in one or more embodiments of the present application, the processing unit 20 is configured to perform the following Figure 5 Multiple steps shown.

[0089] Step S301: Collecting the operating parameters of the air conditioning device. The selection of the operating parameters is described in the above embodiment and will not be repeated here.

[0090] Step S302: Obtain the aging degree of valve elements in the refrigerant circuit.

[0091] In one or more embodiments of the present application, the aging degree of the valve element can be represented by the ratio of the service life of the valve element to the designed service life. The service life of the valve element can be stored in the indoor controller or the outdoor controller, or in the cloud server for easy access at any time.

[0092] Step S303: predicting a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold in the refrigerant circuit based on the neural network model.

[0093] Step S304: The first probability predicted by the neural network model is higher than the second probability.

[0094] Step S305: The processing unit 20 outputs the intervention signal to perform noise reduction pre-processing.

[0095] In one or more embodiments of the present application, the neural network model selects operating parameters, refrigerant charge, refrigerant impurity ratio and valve element aging degree as input data. Figure 6 As shown, in one or more embodiments of the present application, the processing unit 20 is configured to perform the following Figure 6 Multiple steps shown.

[0096] Step S401: Collecting the operating parameters of the air conditioning device. The selection of the operating parameters is described in the above embodiment and will not be repeated here.

[0097] Step S402: Obtaining the refrigerant charge, the refrigerant impurity ratio and the aging degree of the valve element in the refrigerant circuit. The manner of obtaining the refrigerant charge, the refrigerant impurity ratio and the aging degree of the valve element in the refrigerant circuit is described in the above embodiment and will not be repeated here.

[0098] Step S403: predicting a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold in the refrigerant circuit based on the neural network model.

[0099] Step S404: The first probability predicted by the neural network model is higher than the second probability.

[0100] Step S405: The processing unit 20 outputs the intervention signal to perform noise reduction pre-processing.

[0101] Based on the principle of the neural network model, the neural network model in the processing unit 20 is based on Figure 7 The multiple steps shown are implemented, including data collection (such as Figure 7As shown in step S10), data preprocessing (as shown in step S10), Figure 7 As shown in step S11), data segmentation (as shown in step S12), Figure 7 As shown in step S12), a neural network model is constructed (as shown in step S13). Figure 7 ), select a loss function and an optimizer (such as Figure 7 In step S14), neural network model training (as shown in Figure 7 As shown in step S15), neural network model evaluation (as shown in step S15 Figure 7 ) and neural network model deployment (as shown in step S16 in Figure 7 (as shown in step S17).

[0102] In the above process, data collection includes collecting characteristic data and target variables. In one or more embodiments of the present application, the characteristic data include operating parameters, refrigerant charge, refrigerant impurity ratio and valve component aging degree, and the target variable includes refrigerant flow sound. The content of the data collection part will be further described in detail below. Data preprocessing refers to preprocessing the collected data, including missing value processing, standardization and normalization, etc., to ensure that the data is in an appropriate range and there are no outliers. For example, in one or more embodiments of the present application, NumPy and Pandas libraries and KMeans classes can be used to establish high-performance array objects, provide flexible and efficient data indexing, data indexing and data cleaning functions, and perform data clustering, and then use MinMaxScaler for data preprocessing.

[0103] Data segmentation refers to dividing the preprocessed data set into training set, validation set and test set. The training set is used to train the neural network model, the validation set is used to adjust the hyperparameters of the neural network model, and the test set is used to evaluate the model performance. For example, 70% of the data is used as the training set and 30% of the data is used as the test set; 70% of the data can also be used as the training set, 20% of the data as the validation set, and 10% of the data as the test set.

[0104] In one or more embodiments of the present application, a convolutional neural network (CNN) is selected as the neural network structure of the neural network model. A loss function and an optimizer are selected, and both the loss function and the optimizer can select functions and optimizers in the prior art, and then the neural network model is trained, and the neural network is evaluated using mean square error or similar indicators, and finally the trained model is deployed to the processing unit 20, for example, model.h5 is generated for indoor controllers, outdoor controllers or cloud servers to call. Model.h5 can save the weights and architecture of the neural network model.

[0105] As described above, in the present application, one or more of the operating parameters and the refrigerant charge, the refrigerant impurity ratio, and the aging degree of the valve element can be selected as feature data. The following is an introduction to the data collection process. When the input data of the neural network model includes the collected operating parameters and the refrigerant charge, the neural network model is trained based on the refrigerant charge data set. The refrigerant charge data set is obtained by the following method: selecting multiple simulated air-conditioning systems, each of which has a different refrigeration capacity in the outdoor unit of the simulated air-conditioning system; configuring multiple preset refrigerant charge states and preset indoor unit combination opening and closing states, and configuring each simulated air-conditioning system to operate under different preset refrigerant charge states and different preset indoor unit combination opening and closing states; detecting the refrigerant flow sound of each simulated air-conditioning system under different preset refrigerant charge states and different preset indoor unit combination opening and closing states, and comparing the measured refrigerant flow sound with the set abnormal sound threshold, and recording the test results through binary tags; collecting the operating parameters, refrigerant charge states, indoor unit combination opening and closing states and binary tag detection results of the simulated air-conditioning system to establish a refrigerant charge data set. The preset refrigerant charging states include: standard refrigerant charging state, refrigerant under-charging state and refrigerant over-charging state, and the indoor unit combination on-off state includes: operating one indoor unit, operating some indoor units and operating all indoor units.

[0106] In one or more embodiments of the present application, the method for obtaining a refrigerant charge amount data set includes: Figure 8 Multiple steps shown.

[0107] Step S500, select N simulated air conditioning systems, each of which has an outdoor unit with different cooling capacity (4HP-32HP). The simulated air conditioning system has the same structure as the air conditioning device. In one or more embodiments of the present application, the simulated air conditioning system has multiple indoor units.

[0108] Step S501: configuring a simulated air conditioning system to operate in a first refrigerant undercharge state, where the first refrigerant undercharge state is 80% of a standard refrigerant charge.

[0109] Step S502: Turn on an indoor unit, detect the refrigerant flow sound of the simulated air-conditioning system and compare the measured refrigerant flow sound with the set abnormal sound threshold.

[0110] Step S503: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0111] Step S504: Establish a data set including operating parameters, the first refrigerant undercharge state, one indoor unit turned on and the binary tag detection result.

[0112] Step S505: Turn on some indoor units (eg, 50% of the indoor units), detect the refrigerant flow sound of the simulated air-conditioning system, and compare the measured refrigerant flow sound with the set abnormal sound threshold.

[0113] Step S506: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0114] Step S507: Establish a data set including operating parameters, first refrigerant undercharge status, 50% indoor unit opening and binary tag detection results.

[0115] Step S508: Turn on all indoor units, detect the refrigerant flow sound of the simulated air-conditioning system, and compare the measured refrigerant flow sound with the set abnormal sound threshold.

[0116] Step S509: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0117] Step S510: Establish a data set including operating parameters, a first refrigerant undercharge state, 100% indoor unit activation, and binary tag detection results.

[0118] Step S511: configuring the simulated air conditioning system to operate in a second refrigerant undercharge state, where the second refrigerant undercharge state is 90% of the standard refrigerant charge amount.

[0119] Step S512: configuring the simulated air conditioning system to operate under the conditions of turning on one indoor unit, turning on 50% of the indoor units, and turning on all the indoor units.

[0120] Step S513: Detect the refrigerant flow sound of the simulated air-conditioning system under the conditions of turning on one indoor unit, turning on 50% of the indoor units and turning on all the indoor units respectively, and compare the measured refrigerant flow sound with the set abnormal sound threshold, record the test results through binary tags, and establish a data group including operating parameters, under-charge status of the second refrigerant, turning on one indoor unit and binary tag detection results; a data group including operating parameters, under-charge status of the second refrigerant, turning on 50% of the indoor units and binary tag detection results; a data group including operating parameters, under-charge status of the second refrigerant, turning on all indoor units and binary tag detection results.

[0121] Step S514: configuring the simulated air conditioning system to operate under a standard charging state, where the standard charging state is a standard refrigerant charging amount.

[0122] Step S515: configure the simulated air conditioning system to operate under the conditions of turning on one indoor unit, turning on 50% of the indoor units, and turning on all the indoor units.

[0123] Step S516: Detect the refrigerant flow sound of the simulated air-conditioning system under the conditions of turning on one indoor unit, turning on 50% of the indoor units and turning on all the indoor units respectively, and compare the measured refrigerant flow sound with the set abnormal sound threshold, record the test results through binary tags, and establish a data group including operating parameters, standard filling status, one indoor unit turned on and binary tag detection results; a data group including operating parameters, standard filling status, 50% of the indoor units turned on and binary tag detection results; a data group including operating parameters, standard filling status, all indoor units turned on and binary tag detection results.

[0124] Step S517: configuring the simulated air conditioning system to operate in a first overcharge state and a second overcharge state, the first overcharge state being 110% of a standard refrigerant charge, and the second overcharge state being 120% of a standard refrigerant charge.

[0125] Step S518: configure the simulated air conditioning system to operate under the conditions of turning on one indoor unit, turning on 50% of the indoor units, and turning on all the indoor units.

[0126] Step S519: Detect the refrigerant flow sound of the simulated air-conditioning system under the conditions of turning on one indoor unit, turning on 50% of the indoor units and turning on all the indoor units respectively, and compare the measured refrigerant flow sound with the set abnormal sound threshold, record the test results through binary tags, and establish a data group including operating parameters, a first overcharge state, one indoor unit turned on and binary tag detection results; a data group including operating parameters, a first overcharge state, 50% of the indoor units turned on and binary tag detection results; a data group including operating parameters, a first overcharge state, all indoor units turned on and binary tag detection results; a data group including operating parameters, a second overcharge state, one indoor unit turned on and binary tag detection results; a data group including operating parameters, a second overcharge state, 50% of the indoor units turned on and binary tag detection results; a data group including operating parameters, a second overcharge state, all indoor units turned on and binary tag detection results.

[0127] Step S520: Determine whether all N simulated air-conditioning systems have been traversed. If not, configure another simulated air-conditioning system to execute the above steps.

[0128] Step S521: If all N simulated air-conditioning systems have been traversed, a refrigerant charge quantity data set is established using all data sets.

[0129] Fig. 9 and Fig.10 An example of a refrigerant charge dataset.

[0130] When the input data of the neural network model includes operating parameters and refrigerant impurity ratios, the neural network model is trained based on a refrigerant impurity dataset, and the refrigerant impurity dataset is obtained by the following method: impurities of different set ratios are mixed into a set amount of refrigerant, and the refrigerant flow sound of the simulated air-conditioning system when it is running with impurities of different set ratios is detected, and the measured refrigerant flow sound is compared with the set abnormal sound threshold, and the test results are recorded through binary labels; the operating parameters of the simulated air-conditioning system, the set impurity ratio and the binary label detection results are collected to establish a refrigerant impurity dataset.

[0131] Specifically, the refrigerant impurities dataset was obtained by Fig.11 The multiple steps shown are obtained.

[0132] Step S600: mixing a set amount of impurities into a set amount of refrigerant, for example, mixing 1% of copper chips.

[0133] Step S601: Detect the refrigerant flow sound when simulating an air-conditioning system operating with impurities mixed in a set ratio.

[0134] Step S602: Compare the measured refrigerant flow sound with the set abnormal sound threshold.

[0135] Step S603: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0136] Step S604: Establish a data set including operating parameters, refrigerant impurity ratio (1%) and binary tag detection results.

[0137] Step S605: increasing the impurity ratio in the set refrigerant according to the set ratio, for example, adding 1% of copper chips so that the impurity ratio of the refrigerant reaches 2%.

[0138] Step S606: Determine whether the set upper limit of the impurity ratio has been reached. If the set upper limit of the impurity ratio has not been reached, repeat steps S601 to S605. Exemplarily, the set upper limit of the impurity ratio can be 5%, and correspondingly obtain a data group including operating parameters, refrigerant impurity ratio 2% and binary tag detection results; a data group including operating parameters, refrigerant impurity ratio 3% and binary tag detection results; a data group including operating parameters, refrigerant impurity ratio 4% and binary tag detection results; a data group including operating parameters, refrigerant impurity ratio 5% and binary tag detection results.

[0139] Step S607: Create a refrigerant impurity data set using all data sets. Fig.12 An example of a refrigerant impurity dataset.

[0140] When the input data of the neural network model includes the collected operating parameters and the aging degree of valve elements in the refrigerant circuit, the neural network model is trained based on the valve element aging degree data set, and the valve element aging degree data set is obtained by the following method: valve elements with different aging cycles are installed in the simulated air-conditioning system respectively, and the refrigerant flow sound of the simulated air-conditioning system when the valve elements with different aging cycles are installed is detected and compared with the set abnormal sound threshold, and the test results are recorded through binary labels; the operating parameters of the simulated air-conditioning system, different aging cycles and binary label detection results are collected to establish the valve element aging degree data set.

[0141] Specifically, the valve component aging degree dataset is obtained by Fig.13 The multiple steps shown are obtained.

[0142] Step S700: In an initial state, a valve element of a first aging cycle is installed in a simulated air conditioning system. The service life of the valve element of the first aging cycle may be 0.

[0143] Step S701: Detect the refrigerant flow sound of the simulated air-conditioning system and compare the measured refrigerant flow sound with a set abnormal sound threshold.

[0144] Step S702: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0145] Step S703: establishing a data set including operating parameters, a first aging cycle and binary tag detection results.

[0146] Step S704: placing the valve element in an aging test environment until the timing period ends. The timing period is generated according to the design service life of the valve element, for example, the timing period may be 10% of the design service life of the valve element, the temperature, humidity and salinity of the aging test environment are higher than the temperature, humidity and salinity of the air conditioning device use environment, and the temperature, humidity and salinity of the air conditioning device use environment may be set according to the refrigeration equipment standard, for example, according to the equipment environment requirements provided in the technical specifications of the air conditioning device.

[0147] Step S705: Install the valve element that has undergone the aging test in a simulated air-conditioning system, detect the refrigerant flow sound of the simulated air-conditioning system, and compare the measured refrigerant flow sound with the set abnormal sound threshold.

[0148] Step S706: If the measured refrigerant flow sound is higher than the set abnormal sound threshold, the test result is recorded through label 1; if the measured refrigerant flow sound is lower than the set abnormal sound threshold, the test result is recorded through label 0.

[0149] Step S707: Establish a data set including operation parameters, update aging cycle and binary tag detection results. The update aging cycle is the sum of the timing cycle and the update aging cycle of the previous cycle.

[0150] Step S708: Determine whether the updated aging cycle has reached the set aging cycle upper limit. If it has not reached the set aging cycle upper limit, the valve element is placed in the aging test environment again until the timing cycle ends, and the above steps S701 to S707 are repeated. The set aging cycle upper limit can be 20% of the design service life.

[0151] Step S709: If the set aging cycle upper limit is reached, all data groups are used to establish a valve element aging degree data set. Fig.14 This is an example of a valve component aging dataset.

[0152] When the input data of the neural network model includes the collected operating parameters, refrigerant charge amount, refrigerant impurity ratio and aging degree of valve components in the refrigerant circuit, the neural network model is trained based on the refrigerant charge amount data set, the refrigerant impurity data set and the valve component aging degree data set.

[0153] Since abnormal noise caused by refrigerant flow is not a common problem and only occurs in certain specific situations, the neural network model trained based on the refrigerant charge data set, refrigerant impurity data set, and valve element aging degree data set may have data imbalance problems, that is, the number of samples labeled 1 and 0 is unbalanced, and the number of samples of the latter will be greater than the number of samples of the former. The neural network model tends to better predict the category with a larger number, which leads to bias in the neural network model when predicting and reduces the accuracy of the prediction of abnormal noise. To solve this problem, the category imbalance can be improved by calculating the category weight when training the model, that is, different category weights are given to the two categories labeled 1 and 0, and the category weight is inversely proportional to the number of category samples. In this way, the neural network model is trained by increasing the attention of the minority category, so that the neural network model adapts to the imbalanced category distribution of the label 1 and the label 0, and improves the prediction performance of the minority category.

[0154] Finally, the neural network model will output the prediction results and express the results in the form of probability values, as shown in the following table.

[0155] Tag classification Label 0 (second probability) Label 1 (first probability) Probability value 0.99873957 0.42907459

[0156] When the first probability is higher than the second probability, the processing unit 20 outputs an intervention signal to perform noise reduction preprocessing. The intervention signal can be sent to the control terminal in the form of an electrical signal, or it can be sent to the cloud server to prompt the manufacturer to perform advance processing, such as refrigerant filling, valve inspection, or adding sound insulation materials in advance, etc., to ensure that users are not polluted by noise caused by abnormal refrigerant sounds.

[0157] In one or more embodiments of the present application, Fig.15 As shown, the neural network model performs multiple prediction tasks, specifically, including Fig.15 Multiple steps shown.

[0158] Step S801: Collecting operating parameters of the air conditioning device. The operating parameters include one or more of the set temperature, return air temperature, supply air temperature, set wind speed, gas pipe temperature, liquid pipe temperature, indoor throttling element opening, outdoor temperature, compressor operating frequency, and outdoor throttling element opening.

[0159] Step S802: Obtain the refrigerant charge amount, the refrigerant impurity ratio, and the aging degree of valve components in the refrigerant circuit.

[0160] Step S803: predicting a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit based on the neural network model, and predicting a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold.

[0161] Step S804: Based on the neural network model, the cause of the refrigerant flow sound is inferred to be the refrigerant charge amount, the refrigerant impurity ratio or the aging degree of the valve element. The prediction of multiple tasks can be achieved by using different activation functions in the output layer. For example, the sigmoid activation function is used to process the binary label, and the linear activation function is used to process the output of the cause of the refrigerant flow sound.

[0162] Step S805: The neural network model infers that the first probability is higher than the second probability.

[0163] Step S806: Perform noise reduction preprocessing according to the estimated cause.

[0164] By expanding the output layer of the neural network model, it is not only possible to predict the presence or absence of audible sounds, but also to provide information about the specific causes that may have caused the audible sounds.

[0165] In the description of the above embodiments, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0166] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. Air conditioning device, include: a refrigerant circuit connecting the compressor, the outdoor heat exchanger, the throttling element, and the indoor heat exchanger so that the refrigerant circulates therein; It is characterized by further comprising: A processing unit is configured to collect operating parameters of the air conditioning device; based on a neural network model, predict a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold, wherein the input data of the neural network model includes the collected operating parameters and also includes one or more of the refrigerant charge amount, the refrigerant impurity ratio and the degree of aging of the valve element arranged in the refrigerant circuit; when the first probability is higher than the second probability, an intervention signal is output to perform noise reduction preprocessing.

2. The air conditioning device according to claim 1, It is characterized in that The input data of the neural network model includes the collected operating parameters and refrigerant charge amount; The neural network model is trained based on a refrigerant charge amount data set, and the refrigerant charge amount data set is obtained by the following method: Select multiple simulated air conditioning systems, each of which has an outdoor unit with a different cooling capacity; Configure multiple preset refrigerant charging states and preset indoor unit combination on-off states, and configure each simulated air-conditioning system to operate under different preset refrigerant charging states and different preset indoor unit combination on-off states; Detect the refrigerant flow sound of each simulated air-conditioning system under different preset refrigerant charging states and different preset indoor unit combination opening and closing states, compare the measured refrigerant flow sound with the set abnormal sound threshold, and record the test results through binary tags; The operating parameters, refrigerant charging status, indoor unit combination opening and closing status and binary label detection results of the simulated air-conditioning system are collected to establish a refrigerant charging amount dataset.

3. The air conditioning device according to claim 2, It is characterized in that The preset refrigerant charging state includes: a standard refrigerant charging state, a refrigerant under-charging state and a refrigerant over-charging state; the indoor unit combination opening and closing state includes: operating one indoor unit, operating some indoor units and operating all indoor units.

4. The air conditioning device according to claim 1, It is characterized in that The input data of the neural network model includes the collected operating parameters and refrigerant impurity ratio; The neural network model is trained based on a refrigerant impurity dataset, and the refrigerant impurity dataset is obtained by the following method: Mix impurities of different set proportions into a set amount of refrigerant; Detect the refrigerant flow sound when the simulated air-conditioning system is running with impurities mixed in different set proportions, compare the measured refrigerant flow sound with the set abnormal sound threshold, and record the test results through binary tags; The operating parameters of the simulated air-conditioning system, the set impurity ratios and the binary label detection results were collected to establish a refrigerant impurity dataset.

5. The air conditioning device according to claim 4, It is characterized in that The refrigerant impurity dataset was obtained by the following method: Mixing a set proportion of impurities into a set amount of refrigerant; Detect the refrigerant flow sound when the simulated air conditioning system is running with impurities mixed in a set proportion; The measured refrigerant flow sound is compared with the set abnormal sound threshold, the test results are recorded through the binary tag, and a data set including operating parameters, refrigerant impurity ratio and binary tag detection results is established; Increase the proportion of impurities in the set refrigerant according to the set proportion; It is estimated whether the set upper limit of the impurity ratio is reached. If the set upper limit of the impurity ratio is not reached, the refrigerant flow sound of the simulated air-conditioning system when running with the set impurity ratio increased is tested again, and the measured refrigerant flow sound is compared with the set abnormal sound threshold. The test results are recorded through binary tags, and a data group including operating parameters, increased refrigerant impurity ratio and binary tag test results is established until the set upper limit of the impurity ratio is reached, and the refrigerant impurity data set is established using all data groups.

6. The air conditioning device according to claim 1, It is characterized in that The input data of the neural network model includes the collected operating parameters and the aging degree of valve components in the refrigerant circuit; The neural network model is trained based on a valve component aging degree data set, and the valve component aging degree data set is obtained by the following method: installing valve elements with different aging cycles in the simulated air-conditioning system respectively, detecting the refrigerant flow sound of the simulated air-conditioning system when the valve elements with different aging cycles are installed, comparing the measured refrigerant flow sound with a set abnormal sound threshold, and recording the test results through a binary tag; The operating parameters of the simulated air-conditioning system, different aging cycles and binary label detection results are collected to establish a valve component aging degree dataset.

7. The air conditioning device according to claim 6, It is characterized in that In the initial state, the valve element has a first aging cycle; After collecting the operating parameters of the simulated air-conditioning system, the first aging cycle and the binary tag detection results, the valve element is placed in the aging test environment for multiple times until a timing cycle ends. After each timing cycle ends, the valve element is set in the simulated air-conditioning system, the operating parameters of the simulated air-conditioning system are collected again, and the aging cycle and the binary tag detection results are updated, wherein the updated aging cycle is the sum of the timing cycle and the updated aging cycle of the previous cycle; the timing cycle is generated according to the designed service life of the valve element; the temperature, humidity and salinity of the aging test environment are higher than the temperature, humidity and salinity of the air conditioning device use environment.

8. The air conditioning device according to any one of claims 1 to 7, It is characterized in that The operating parameters include one or more of a set temperature, a return air temperature, a supply air temperature, a set wind speed, a gas pipe temperature, a liquid pipe temperature, an indoor throttling element opening, an outdoor temperature, a compressor operating frequency, and an outdoor throttling element opening.

9. Air conditioning device, include: a refrigerant circuit connecting the compressor, the outdoor heat exchanger, the throttling element, and the indoor heat exchanger so that the refrigerant circulates therein; It is characterized by further comprising: A processing unit is configured to collect operating parameters of the air conditioning device; based on a neural network model, predict a first probability of generating a refrigerant flow sound higher than a set abnormal sound threshold in the refrigerant circuit and a second probability of not generating a refrigerant flow sound higher than the set abnormal sound threshold, wherein input data of the neural network model include the collected operating parameters, refrigerant charge, refrigerant impurity ratio and aging degree of valve elements arranged in the refrigerant circuit; based on the neural network model, infer that the cause of generating the refrigerant flow sound is the refrigerant charge, the refrigerant impurity ratio, or the aging degree of the valve elements; when the first probability is higher than the second probability, output an intervention signal to perform noise reduction preprocessing.

10. The air conditioning device according to claim 9, It is characterized in that The operating parameters include one or more of a set temperature, a return air temperature, a supply air temperature, a set wind speed, a gas pipe temperature, a liquid pipe temperature, an indoor throttling element opening, an outdoor temperature, a compressor operating frequency, and an outdoor throttling element opening.

Images