Fault detection method and first multi-split air conditioning system

Abstract

The embodiment of the invention provides a fault detection method and a first multi-split air conditioning system, relates to the technical field of air conditioners, and is used for expanding the application range of fault detection of the multi-split air conditioning system and realizing cross-system fault detection. The cloud server comprises the steps that a first training sample of a first multi-split air conditioning system and a second training sample of a second multi-split air conditioning system are obtained, and the first training sample comprises a training sample of the first multi-split air conditioning system in a normal operation state; the second training sample comprises a training sample of the second multi-split air conditioning system in a normal operation state and a training sample of the second multi-split air conditioning system in a fault operation state; adjusting the proportion of the minority class training samples in the second training samples; converting the feature distribution of the first training sample, wherein the converted feature distribution of the training sample is consistent with the adjusted feature distribution of the second training sample; and training a fault detection model based on the converted first training sample and the adjusted second training sample.

Description

Technical Field

[0001] The present application relates to the field of air conditioning technology, and in particular to a fault detection method and a first multi-split air conditioning system. Background Art

[0002] At present, air conditioners have become an indispensable household appliance in family life. As the utilization rate of air conditioners increases, the probability of air conditioner failure also increases.

[0003] When detecting faults in indoor air conditioner units, data-driven fault detection models are often used. However, in practice, these current data-driven fault detection models are mostly applicable only to specific types of air conditioners. Given the diversity of factors such as the number of indoor and outdoor units, and installation scenarios, data-driven fault detection methods cannot accurately detect faults in all types of air conditioners. Therefore, expanding the scope of air conditioner fault detection has become a pressing technical challenge. Summary of the Invention

[0004] The present application provides a fault detection method and a first multi-split air-conditioning system, which are used to expand the scope of application of fault detection of the multi-split air-conditioning system.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions.

[0006] In a first aspect, an embodiment of the present application provides a fault detection method, which includes: obtaining a first sample set and a second sample set; wherein the first sample set includes a plurality of first samples generated based on normal operating data of a first multi-split air-conditioning system, and the second sample set includes a plurality of second samples generated based on normal operating data and fault operating data of a second multi-split air-conditioning system, and the second samples have an operating mode label or a fault type label; when the proportion of second samples with the first label in the second sample set is less than or equal to a preset threshold, the second sample set is expanded to adjust the proportion corresponding to the first label; the first label belongs to an operating mode label or a fault type label; the first sample set is adjusted according to the feature space distribution of the second sample set; the initial model is trained according to the first sample set and the second sample set to obtain a trained fault detection model; and the fault detection model is used to perform fault detection on the first multi-split air-conditioning system.

[0007] The technical solution provided by the embodiment of the present application brings at least the following beneficial effects: since the data distribution of the operating data of the first multi-split air-conditioning system is different from the data distribution of the operating data of the second multi-split air-conditioning system, the fault detection model established based on the operating data of the second multi-split air-conditioning system is not applicable to the first multi-split air-conditioning system. However, the embodiment of the present application provides a fault detection method, which can, when establishing a fault detection model, adjust the first sample set generated based on the normal operating data and faulty operating data of the second multi-split air-conditioning system according to the feature space distribution of the second sample set generated based on the normal operating data of the first multi-split air-conditioning system, so that the feature space distribution of the first sample set is consistent with or close to the feature space distribution of the second training sample. In this way, the fault detection model trained based on the first sample set and the second sample set with consistent or close feature space distribution can perform fault detection on both the first multi-split air-conditioning system and the second multi-split air-conditioning system, thereby realizing cross-system fault detection.

[0008] Furthermore, because the faulty operating data for the second multi-split air conditioning system is difficult to obtain, it falls into the minority category. This means that the proportion of second samples with fault type labels in the second sample set is less than or equal to a preset threshold. This lack of minority samples can lead to insufficient fault classification in the fault detection model. By expanding the second sample set, the amount of samples with fault type labels can be increased, thereby improving the accuracy of the fault detection model.

[0009] In some embodiments, the historical operating data of the first air conditioner includes normal operating data and fault operating data, and the operating data of the second air conditioner includes normal operating data; the preset rule is that the difference between the normal operating data of the first air conditioner and the normal operating data of the second air conditioner is equal to the difference between the fault operating data of the first air conditioner and the fault operating data of the second air conditioner; the processor is configured to process the historical operating data of the first air conditioner and the historical operating data of the second air conditioner based on the preset rule to obtain the operating data to be tested of the second air conditioner, and is specifically configured to: based on the normal operating data of the first air conditioner and the normal operating data of the second air conditioner, calculate the difference between the average value of the normal operating data of the first air conditioner and the average value of the normal operating data of the second air conditioner; based on the difference and the normal operating data of the second air conditioner, obtain the operating data to be tested of the second air conditioner.

[0010] In some embodiments, the second sample set is expanded, including: obtaining at least one neighboring sample of the second sample with the first label; performing random linear interpolation between the second sample with the first label and each neighboring sample to generate newly added samples with the first label; and adding the newly added samples with the first label to the second sample set to adjust the proportion corresponding to the first label.

[0011] In some embodiments, the first sample set is adjusted according to the feature space distribution of the second sample set, including: an extraction step: extracting a first feature from the first sample through a feature extractor, extracting a second feature from a second sample generated based on the normal operation data of the second multi-split air-conditioning system, and extracting a third feature from the second sample generated based on the fault operation data of the second multi-split air-conditioning system; an input step: inputting the first feature and the second feature into a domain classifier to obtain a domain loss function, and inputting the third feature into a label classifier to obtain a label loss function; an updating step: updating the parameters of the feature extractor, the domain classifier and the label classifier based on the domain loss function and the label loss function through a back propagation algorithm; repeating the input step and the updating step until the domain classification loss function value of the domain classifier is greater than or equal to the first threshold, and the label loss function value of the label classifier is less than or equal to the second threshold.

[0012] In some embodiments, adjusting the first sample set according to the feature space distribution of the second sample set includes: inputting the first sample set and the second sample set into a domain adaptation model to adjust the first sample set; wherein the domain adaptation model adopts a DANN structure.

[0013] In some embodiments, obtaining a first sample set and a second sample set includes: obtaining first experimental data of a first multi-split air-conditioning system and second experimental data of a second multi-split air-conditioning system; the first experimental data includes normal operating data of the first multi-split air-conditioning system, and the second experimental data includes normal operating data and fault operating data of the second multi-split air-conditioning system; preprocessing the first experimental data and the second experimental data to obtain processed first experimental data and second experimental data; the preprocessing includes one or more of the following: data cleaning, data reduction, data encoding, and data normalization; generating a first sample set and a second sample set based on the processed first experimental data and the second experimental data.

[0014] In some embodiments, the first experimental data includes at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the first multi-split air-conditioning system under different test environments; the second experimental data includes at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the second multi-split air-conditioning system under different test environments.

[0015] In a second aspect, an embodiment of the present application provides a first multi-split air-conditioning system, which includes: at least one indoor unit; at least one outdoor unit; a controller, configured to: obtain a first sample set and a second sample set; wherein the first sample set includes a plurality of first samples generated based on the normal operating data of the first multi-split air-conditioning system, and the second sample set includes a plurality of second samples generated based on the normal operating data and fault operating data of the second multi-split air-conditioning system, and the second samples have an operating mode label or a fault type label; when the proportion of the second samples with the first label in the second sample set is less than or equal to a preset threshold, the second sample set is expanded to adjust the proportion corresponding to the first label; the first label belongs to an operating mode label or a fault type label; the first sample set is adjusted according to the feature space distribution of the second sample set; the initial model is trained based on the first sample set and the second sample set to obtain a trained fault detection model; and the fault detection model is used to perform fault detection on the first multi-split air-conditioning system.

[0016] In a third aspect, an embodiment of the present application provides a controller comprising: one or more processors; one or more memories; wherein the one or more memories are used to store computer program codes, the computer program codes comprising computer instructions, and when the one or more processors execute the computer instructions, the controller executes any one of the fault detection methods provided in the second aspect.

[0017] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which includes computer instructions. When the computer instructions are executed on a computer, the computer executes any one of the fault detection methods provided in the second aspect.

[0018] In a fifth aspect, an embodiment of the present invention provides a computer program product, which can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement any fault detection method provided in the second aspect.

[0019] It should be noted that the above-mentioned computer instructions may be stored in whole or in part on a computer-readable storage medium. The computer-readable storage medium may be packaged together with the processor of the controller or separately from the processor of the controller, and this application does not limit this.

[0020] The beneficial effects described in the second to fifth aspects of this application can be analyzed by referring to the beneficial effects of the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

[0022] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0023] Figure 2 A schematic diagram of the hardware structure of a cloud server provided in an embodiment of the present application;

[0024] Figure 3 A schematic diagram of the composition of a multi-split air conditioning system provided in an embodiment of the present application;

[0025] Figure 4 A schematic diagram of the hardware structure of a multi-split air conditioning system provided in an embodiment of the present application;

[0026] Figure 5 A flowchart of a fault detection method provided in an embodiment of the present application;

[0027] Figure 6 A flowchart of another fault detection method provided in an embodiment of the present application;

[0028] Figure 7 A schematic diagram of the structure of a SMOTE adjustment module provided in an embodiment of the present application;

[0029] Figure 8 A flowchart of another fault detection method provided in an embodiment of the present application;

[0030] Figure 9 A schematic diagram of the structure of a DANN provided in an embodiment of the present application;

[0031] Figure 10 A flowchart of another fault detection method provided in an embodiment of the present application;

[0032] Figure 11 A flowchart of another fault detection method provided in an embodiment of the present application;

[0033] Figure 12 A schematic diagram of the hardware structure of a fault detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0035] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0036] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.

[0037] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "connected" and "connect" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections. Those skilled in the art will understand the specific meanings of the above terms in this application based on the specific circumstances. Furthermore, when describing pipelines, the terms "connected" and "connected" used in this application have the meaning of conducting electricity. The specific meanings need to be understood in the context.

[0038] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0039] In order to expand the scope of application of fault detection of multi-split air-conditioning systems, an embodiment of the present application provides a fault detection method. Since the data distribution of the operating data of the first multi-split air-conditioning system is different from the data distribution of the operating data of the second multi-split air-conditioning system, the fault detection model established based on the operating data of the second multi-split air-conditioning system is not applicable to the first multi-split air-conditioning system. However, an embodiment of the present application provides a fault detection method. When establishing a fault detection model, the first sample set generated based on the normal operating data and faulty operating data of the second multi-split air-conditioning system is adjusted according to the feature space distribution of the second sample set generated based on the normal operating data of the first multi-split air-conditioning system, so that the feature space distribution of the first sample set is consistent with or close to the feature space distribution of the second training sample. In this way, the fault detection model trained based on the first sample set and the second sample set with consistent or close feature space distribution can perform fault detection on both the first multi-split air-conditioning system and the second multi-split air-conditioning system, thereby realizing cross-system fault detection.

[0040] Furthermore, because the faulty operating data for the second multi-split air conditioning system is difficult to obtain, it falls into the minority category. This means that the proportion of second samples with fault type labels in the second sample set is less than or equal to a preset threshold. This lack of minority samples can lead to insufficient fault classification in the fault detection model. By expanding the second sample set, the amount of samples with fault type labels can be increased, thereby improving the accuracy of the fault detection model.

[0041] In this application, the air conditioner performs a refrigeration cycle of the air conditioner by using a compressor, a condenser, an electronic expansion valve, an evaporator, and a four-way valve as a refrigerant circulation circuit. The refrigeration cycle includes a series of processes involving compression, condensation, expansion, and evaporation, and supplies refrigerant to the air that has been conditioned and heat exchanged.

[0042] The compressor compresses low-temperature, low-pressure refrigerant gas and discharges the compressed, high-temperature, high-pressure refrigerant gas. The discharged refrigerant gas flows into the condenser. The condenser condenses the compressed refrigerant into a liquid phase, releasing heat into the surrounding environment through the condensation process.

[0043] The electronic expansion valve throttles the high-temperature, high-pressure liquid refrigerant condensed in the condenser to a low-pressure, two-phase refrigerant. The evaporator evaporates the refrigerant throttled by the electronic expansion valve and returns the low-temperature, low-pressure refrigerant gas to the compressor. The evaporator achieves cooling by utilizing the latent heat of evaporation to exchange heat with the medium being cooled. Throughout this cycle, the air conditioner regulates the indoor temperature.

[0044] The outdoor unit of the air conditioner refers to a portion of a refrigeration cycle including a compressor and an outdoor heat exchanger, the indoor unit of the air conditioner includes an indoor heat exchanger, and an expansion valve may be provided in the indoor unit or the outdoor unit.

[0045] The indoor heat exchanger and the outdoor heat exchanger function as a condenser or an evaporator. When the indoor heat exchanger functions as a condenser, the air conditioner functions as a heater in heating mode, and when the indoor heat exchanger functions as an evaporator, the air conditioner functions as a cooler in cooling mode.

[0046] Figure 1 This is a schematic diagram of an application scenario provided by this application according to an exemplary embodiment. Figure 1 As shown, the application scenario includes multiple multi-split air-conditioning systems, such as a first multi-split air-conditioning system 101 and a second multi-split air-conditioning system 102 , and a cloud server 103 .

[0047] There is a communication connection between the cloud server 103 and multiple multi-split air-conditioning systems.

[0048] In some embodiments, a multi-split air conditioning system is a device that regulates and controls parameters such as the temperature, humidity, and flow rate of the ambient air in a building or structure.

[0049] In some embodiments, the cloud server 103 can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks, big data servers, etc. This application does not impose any special restrictions on the specific form of the cloud server 103.

[0050] In some embodiments, the multi-split air-conditioning system may send its own operating data to a cloud server so that the cloud server can detect faults of the multi-split air-conditioning system based on the operating data of the multi-split air-conditioning system.

[0051] Figure 2 This is a hardware structure diagram of a cloud server provided by this application according to an exemplary embodiment. Figure 2 As shown, the cloud server 103 includes a communicator 201 , a memory 202 and a processor 203 .

[0052] In some embodiments, the communicator 201 is used to establish a communication connection with other network entities, for example, to establish a communication connection with the first multi-split air conditioning system 101. The communicator 201 may include a radio frequency (RF) module, a cellular module, a wireless fidelity (WIFI) module, and a GPS module, etc. Taking the RF module as an example, the RF module can be used to receive and send signals, in particular, to send the received information to the processor 203 for processing; in addition, the signal generated by the processor 203 is sent out. Typically, the RF circuit may include but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc.

[0053] In some embodiments, the memory 202 can be used to store software programs and data. The processor 203 executes various functions and data processing of the cloud server 103 by running the software programs or data stored in the memory 202. The memory 202 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. The memory 202 stores an operating system that enables the processor 203 to run. In the present application, the memory 202 can store an operating system and various application programs, and may also store code for executing the fault detection method provided in the embodiments of the present application.

[0054] In some embodiments, the processor 203 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present disclosure, such as one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0055] Those skilled in the art will understand that Figure 2 The hardware structure shown in the figure does not constitute a limitation on the cloud server. The cloud server may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0056] Figure 3 This is a schematic diagram of the composition of a multi-split air conditioning system provided by the present application according to an exemplary embodiment. It should be noted that the multi-split air conditioning systems of different models involved in the embodiments of the present application are all based on Figure 1 The structural diagram of the multi-split air-conditioning system shown in FIG.

[0057] like Figure 3 As shown, the multi-split air conditioning system includes an indoor unit 301, an outdoor unit 302 and a controller 303 ( Figure 3 not shown).

[0058] In some embodiments, the number of the indoor unit 301 is at least one, and the number of the outdoor unit 302 is at least one. This application does not limit the number of the indoor unit 301 and the outdoor unit 302.

[0059] Indoor unit 301, taking indoor unit 301 as an indoor hanging unit as an example, indoor hanging units are usually installed on indoor walls, etc. For another example, indoor cabinet units are also a type of indoor unit form of indoor units.

[0060] The outdoor unit 302 is usually arranged outdoors and can be connected to a plurality of indoor units 101 for indoor heat exchange. In addition, the outdoor unit 302 is usually located outdoors on the opposite side of the indoor unit 301 across a wall.

[0061] In some embodiments, the controller 303 is a device that can generate an operation control signal based on an instruction opcode and a timing signal to instruct the multi-split air conditioning system to execute the control instruction. For example, the controller can be a central processing unit (CPU), a general-purpose processor, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. The controller can also be other devices with processing functions, such as circuits, devices, or software modules, and the embodiments of the present application do not impose any limitations on this.

[0062] In addition, both the outdoor unit 302 and the indoor unit 301 are connected to the controller 303 ( Figure 3 There is a communication connection (not shown) and related operations are performed according to the instructions of the controller 303.

[0063] In some embodiments, as Figure 4 As shown, the outdoor unit 302 includes a compressor 401, an outdoor heat exchanger 402, a liquid accumulator 403, a four-way valve 404, and a throttling device 405. The indoor unit 301 includes an indoor heat exchanger 406.

[0064] In some embodiments, the compressor 401 is disposed between the throttling device 405 and the liquid accumulator 403 , and is configured to compress the refrigerant delivered by the liquid accumulator 403 and deliver the compressed refrigerant to the throttling device via the four-way valve 404 .

[0065] In some embodiments, one end of the outdoor heat exchanger 402 is connected to the liquid accumulator 403 via a four-way valve 404, and the other end is connected to a throttling device 405. The outdoor heat exchanger 402 has a first inlet and outlet for circulating refrigerant between the outdoor heat exchanger 402 and the suction port of the compressor 401 via the liquid accumulator 403, and a second inlet and outlet for circulating refrigerant between the outdoor heat exchanger 402 and the throttling device 405. The outdoor heat exchanger 402 exchanges heat between the heat exchanger flowing in the heat transfer pipe connected between the first inlet and outlet and the second inlet and outlet and the outdoor air. In the cooling cycle, the outdoor heat exchanger 402 operates as a condenser.

[0066] In some embodiments, one end of the liquid accumulator 403 is connected to the compressor 401, and the other end is connected to the outdoor heat exchanger 402 via a four-way valve 404. In the liquid accumulator 403, the refrigerant flowing from the outdoor heat exchanger 402 to the compressor 401 via the four-way valve 404 is separated into gas refrigerant and liquid refrigerant. Furthermore, the gas refrigerant is primarily supplied from the liquid accumulator 403 to the suction port of the compressor 401.

[0067] In some embodiments, the four ports of the four-way valve 404 are respectively connected to the compressor 401, the outdoor heat exchanger 402, the liquid reservoir 403, and the indoor heat exchanger 406. The four-way valve 404 is used to switch between cooling and heating by changing the flow direction of the refrigerant in the system pipeline.

[0068] In some embodiments, the indoor heat exchanger 406 has a third inlet and outlet for allowing liquid refrigerant to flow between the four-way valve 404 and a fourth inlet and outlet for allowing gaseous refrigerant to flow between the discharge port of the compressor 401. The indoor heat exchanger 406 exchanges heat between the refrigerant flowing in the heat transfer pipe connected between the third and fourth inlets and outlets and the indoor air.

[0069] The embodiments provided in this application are described in detail below with reference to the accompanying drawings.

[0070] like Figure 5 As shown, an embodiment of the present application provides a fault detection method. The execution subject of the method can be either a cloud server or a first multi-split air conditioning system. The present application does not limit the execution subject of the fault detection method. The method includes the following steps:

[0071] S101: Acquire a first sample set and a second sample set.

[0072] Among them, the first sample set includes multiple first samples generated based on the normal operation data of the first multi-split air-conditioning system, and the second sample set includes multiple second samples generated based on the normal operation data and fault operation data of the second multi-split air-conditioning system, and the second samples have an operation mode label or a fault type label.

[0073] Among them, the normal operating data of the first multi-split air-conditioning system is the operating data of the first multi-split air-conditioning system in normal operating state, the normal operating data of the second multi-split air-conditioning system is the operating data of the second multi-split air-conditioning system in normal operating state, and the fault operating data of the second multi-split air-conditioning system is the operating data of the second multi-split air-conditioning system in fault operating state.

[0074] The operating mode label is used to indicate the specific operating modes corresponding to the multiple second samples generated based on the normal operating data of the second multi-split air-conditioning system. During the normal operation of the second multi-split air-conditioning system, there may be different operating modes, such as cooling mode, heating mode, ventilation mode, etc. Each operating state can be better distinguished, thereby realizing the monitoring and identification of the system operating mode. The fault type label is used to indicate the specific fault type corresponding to the sample generated based on the fault operating data of the second multi-split air-conditioning system. When the second multi-split air-conditioning system fails, different fault types may be involved, such as compressor failure, sensor failure, circuit failure, etc.

[0075] By labeling each second sample with a corresponding operating mode label or fault type label, the sample data can be made more readable, which helps to perform more accurate fault monitoring, analysis and maintenance of the multi-split air-conditioning system.

[0076] In some embodiments, the cloud server can construct a first sample set based on the normal operation data in the experimental data of the first multi-split air-conditioning system, and can also construct a second sample set based on the normal operation data and fault operation data in the experimental data of the second multi-split air-conditioning system.

[0077] For example, Figure 6 As shown, step S101 can be implemented as the following steps.

[0078] Step a1: Acquire first experimental data of a first multi-split air-conditioning system and second experimental data of a second multi-split air-conditioning system.

[0079] The first experimental data includes normal operating data of the first multi-split air-conditioning system, and the second experimental data includes normal operating data and fault operating data of the second multi-split air-conditioning system.

[0080] As you can understand, the experimental data contains more categories of operating data. More categories of operating data means more diverse data, covering a wider range of scenarios. This helps the fault detection model more comprehensively learn the patterns and characteristics of multi-split air conditioning systems in different scenarios, thereby helping the fault detection model make more accurate predictions and improving the model's generalization capabilities.

[0081] In some embodiments, testers typically conduct experiments on the first multi-split air conditioning system under different test environments to obtain experimental data of the first multi-split air conditioning system under different test environments. This experimental data includes normal operating data of the first multi-split air conditioning system under different test environments. The cloud server can then obtain the experimental data of the first multi-split air conditioning system in real time and store it in a database on the cloud server for later retrieval and analysis. Thus, the first experimental data of the first multi-split air conditioning system can be obtained from the database.

[0082] Similarly, second experimental data of the second multi-split air-conditioning system can also be obtained from the database.

[0083] In some embodiments, the first experimental data may include at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the first multi-split air-conditioning system under different test environments; the second experimental data may include at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the second multi-split air-conditioning system under different test environments.

[0084] Step a2: pre-process the first experimental data and the second experimental data to obtain processed first experimental data and second experimental data.

[0085] In some embodiments, preprocessing includes one or more of the following: data cleaning, data reduction, data encoding, and data normalization.

[0086] In some embodiments, data cleaning may be used to indicate the removal of non-stationary data from the first experimental data and the second experimental data.

[0087] For example, taking data cleaning of the first experimental data as an example, the cloud server can calculate the slope of the first experimental data in the sliding window based on the sliding window algorithm and the preset window length. If the slope is outside the preset range, the first experimental data in the sliding window will be determined as non-steady-state operation data, and the first experimental data in the sliding window will be eliminated.

[0088] In some embodiments, data reduction can be used to reduce the size of the first experimental data and the second experimental data while retaining key information and features of the first experimental data and the second experimental data.

[0089] For example, taking data reduction of the first experimental data as an example, the cloud server can map the first experimental data from a high-dimensional space to a low-dimensional space based on a dimensionality reduction method (such as a principal component analysis method) to reduce the dimension of the first experimental data, thereby reducing the complexity of the first experimental data.

[0090] In some embodiments, data encoding may be used to refer to converting the first experimental data and the second experimental data into other forms for use during transmission or storage.

[0091] For example, if the first experimental data includes wind speed data, data encoding can be used to convert the wind speed data into a digital label. For example, "low wind speed" is encoded as 0, "medium wind speed" is encoded as 1, "high wind speed" is encoded as 2, and so on.

[0092] In some embodiments, data normalization is used to convert the first experimental data and the second experimental data into a uniform scale range to eliminate dimensional differences between different variables and facilitate data analysis and processing.

[0093] In some embodiments, because the capacities of different indoor units in the first multi-split air conditioning system vary, a weighted average can be performed on the first experimental data of the first multi-split air conditioning system based on the capacities of the indoor units to better consider the performance of the entire first multi-split air conditioning system, rather than the performance of each indoor unit individually. Similarly, a weighted average can also be performed on the second experimental data of the second multi-split air conditioning system.

[0094] Step a3: Generate a first sample set and a second sample set based on the processed first experimental data and the second experimental data.

[0095] In some embodiments, the processed first experimental data can be divided into a first training sample set and a first validation sample set, with the first training sample set being referred to as the first sample set. The cloud server can also divide the processed second experimental data into a second training sample set and a second validation sample set, with the second training sample set being referred to as the second sample set. Furthermore, the first validation sample set is used to verify the accuracy of the fault detection model in step S104 described below.

[0096] S102: When the proportion of second samples with the first label in the second sample set is less than or equal to a preset threshold, expand the second sample set to adjust the proportion corresponding to the first label.

[0097] The first tag is an operation mode tag or a fault type tag.

[0098] It can be understood that if the proportion of the second samples with the first label in the second sample set is less than or equal to the preset threshold, it means that the number of the second samples with the first label is small.

[0099] In some embodiments, the proportion of the second samples with the first label in the second sample set affects the accuracy of the fault detection model in the following step S104. Therefore, the second sample set can be expanded to adjust the proportion corresponding to the first label to determine a better proportion corresponding to the first label, thereby improving the accuracy of the fault detection model in the following step S104.

[0100] In some embodiments, the second sample set may be input into a SMOTE adjustment module established based on the SMOTE technology to obtain an expanded second sample set.

[0101] Among them, such as Figure 7 As shown, the SMOTE adjustment module may include a sampling rate setting module, a K-nearest neighbor calculation module, a new sample construction module, and a data ratio adjustment module.

[0102] For example, Figure 8 As shown, in combination with the above-mentioned SMOTE adjustment module, step S102 can be implemented as the following steps.

[0103] Step b1: The K-nearest neighbor calculation module is used to obtain at least one neighbor sample of the second sample having the first label.

[0104] In some embodiments, before obtaining at least one neighboring sample of a second sample with a first label through the K-nearest neighbor calculation module, a sampling rate setting module may be used to set a sampling rate. The sampling rate refers to the ratio of the number of new samples required to generate for each second sample with a first label to the number of second samples with the first label when generating new samples.

[0105] In some embodiments, the cloud server may use a K-nearest neighbor algorithm to find at least one neighbor sample of the second sample having the first label.

[0106] For example, there may be multiple second samples with the first label. For each second sample with the first label, the cloud server may calculate the distance between the second sample with the first label and each of all other second samples with the first label. Based on the distance, K nearest neighbor samples of the second sample with the first label may be found.

[0107] Step b2: The new sample construction module is used to perform random linear interpolation processing between the second sample with the first label and each neighboring sample to generate a new sample with the first label.

[0108] In one example, a point may be randomly selected on the line connecting the second sample with the first label and each neighboring sample. This point is the newly added sample with the first label.

[0109] For example, the second sample with the first label is sample A, the neighboring sample of sample A is sample B, and a point is randomly selected on the line between sample A and sample B. This point is the newly added sample C, and sample C is a sample with the first label.

[0110] In another example, random linear interpolation processing can be performed according to the following formula (1) to obtain new minority class training samples.

[0111] x new =x+rand(0,1)*|x-xn| Formula (1)

[0112] Among them, x new is a newly added sample with the first label; x is a newly added sample with the first label in the second sample set; rand(0, 1) is used to randomly select a value between 0 and 1; xn is each neighboring sample in at least one neighboring sample of the newly added sample with the first label.

[0113] Step b3: Add the newly added samples with the first label to the second sample set to adjust the proportion corresponding to the first label.

[0114] S103: The data ratio adjustment module is used to adjust the first sample set according to the feature space distribution of the second sample set.

[0115] In some embodiments, the first sample set may be adjusted based on the feature space distribution of the second sample set so that the feature space distribution of the first sample set is consistent with or close to the feature space distribution of the second sample set.

[0116] It can be understood that the feature distribution of the first training sample is converted to be consistent or close to the feature distribution of the second training sample. In this way, after the fault detection model is trained based on the second training sample whose feature distribution is consistent or close to the first training sample, fault detection can be performed on both the first multi-split air-conditioning system and the second multi-split air-conditioning system, thereby realizing cross-system fault detection.

[0117] In some embodiments, the cloud server may input the first sample set and the second sample set into a domain adaptation model to adjust the first sample set.

[0118] Among them, the domain adaptation model adopts the adversarial neural network (DANN) structure.

[0119] like Figure 9 As shown in Figure 1, the basic structure of a DANN consists of a feature extractor, a label classifier, and a domain classifier. DANN employs adversarial learning, pitting the label classifier and domain classifier against each other during training to achieve a balance between label classification loss and domain classification loss. This optimizes the feature extractor so that the feature spatial distribution of the features extracted by the feature extractor from the first sample in the first sample set is consistent with or close to the feature spatial distribution of the features extracted from the second sample in the second sample set. Consequently, a fault detection model trained on the second sample, whose feature spatial distribution is consistent with or close to the first sample, can detect faults in both the first and second multi-split air conditioning systems, thus achieving cross-system fault detection.

[0120] For example, Figure 10 As shown, step S103 can be implemented as the following steps.

[0121] Step c1: Using a feature extractor, extract a first feature from the first sample, extract a second feature from a second sample generated based on normal operation data of the second multi-split air-conditioning system, and extract a third feature from a second sample generated based on fault operation data of the second multi-split air-conditioning system.

[0122] In some embodiments, the feature extractor is a convolutional neural network composed of multiple one-dimensional convolutional layers and one-dimensional maximum pooling layers stacked together. The input data is convolved in the convolutional layer and dimensionality reduced in the maximum pooling layer to extract effective features.

[0123] In some embodiments, in the feature extractor, nonlinear mapping and feature extraction can be performed on the input data to better capture and represent the complexity of the input data. This process can be expressed as the following formula (2).

[0124] G f =sigm(Wx+b) Formula (2)

[0125] Step c2: input the first feature into a domain classifier to obtain a domain loss function, and input the second feature into a label classifier to obtain a label loss function.

[0126] The domain classifier is used to distinguish the source of the input data, that is, to distinguish whether the input data is data from the first multi-split air conditioning system or the second multi-split air conditioning system. The label classifier is used to distinguish the label type of the input data, including the operating mode label or the fault type label.

[0127] In some embodiments, the label classifier is a fully connected neural network composed of multiple fully connected layers stacked together, and the data is mapped into an output vector with a length equal to the number of categories through the fully connected neural network; the domain classifier is a fully connected neural network composed of a gradient reversal layer and multiple fully connected layers stacked together, and during feature extraction, the data is mapped into an output value with a length of 1 through the fully connected neural network.

[0128] In some embodiments, the first feature is input into a domain classifier to obtain a first classification probability, and a domain loss function may be determined based on the first classification probability.

[0129] In some embodiments, in the domain classifier, a nonlinear transformation can be performed on the input features to better distinguish whether the input features come from the first multi-split air-conditioning system or the second multi-split air-conditioning system in the feature distribution space. This process can be expressed by the following formula (3).

[0130] G d =sigm(u T G f (x)+z) Formula (3)

[0131] Similarly, the second feature is input into the label classifier to obtain a second classification probability. Based on the second label classification probability, the label loss function can be determined.

[0132] In some embodiments, the label classifier performs a nonlinear transformation on the input features, allowing the label classifier to better adapt to various complex data distributions and patterns, thereby more effectively distinguishing the corresponding label types of the input features. This process can be expressed as the following formula (4).

[0133] G y =softmax(VG f (x)+c) Formula (4)

[0134] Step c3: Update the parameters of the feature extractor, domain classifier, and label classifier based on the domain loss function and the label loss function through the back propagation algorithm.

[0135] In some embodiments, during backpropagation, the gradient of the label classifier is first calculated using the label loss function. The gradient is then propagated to the feature extractor according to the chain rule to update the parameters of the feature extractor. Next, the gradient of the domain classifier is calculated using the domain loss function. The gradient is also propagated to the feature extractor to update the parameters of the feature extractor.

[0136] During the domain adaptation process, the feature extractor is continuously optimized so that the feature space distribution of the features extracted by the feature extractor from the first sample in the first sample set is consistent with or close to the feature space distribution of the features extracted from the second sample in the second sample set.

[0137] Step c4: Repeat steps c2 and c3 until the convergence condition is reached.

[0138] In some embodiments, the convergence condition may include any of the following:

[0139] The domain classification loss function value of the domain classifier is greater than or equal to a first threshold, and the label loss function value of the label classifier is less than or equal to a second threshold; the number of iterations reaches a preset number.

[0140] It can be understood that, on the one hand, the domain classification loss function value of the domain classifier is greater than or equal to the first threshold, indicating that the domain classifier cannot distinguish whether the features of the sample are from the first multi-split air-conditioning system or from the second multi-split air-conditioning system. This means that the fault detection model established based on the adjusted first sample set and second sample set can detect both the first multi-split air-conditioning system and the second multi-split air-conditioning system.

[0141] On the other hand, the loss function value of the label classifier is less than or equal to the second threshold, indicating that the label classifier has high label classification accuracy for the sample, which helps to improve the accuracy of the fault detection model.

[0142] S104: Train the initial model according to the first sample set and the second sample set to obtain a trained fault detection model.

[0143] In some embodiments, the first sample set and the second sample set are divided into a training set and a test set. An initial model is selected, and training and parameter optimization are performed on the initial model to obtain a trained fault detection model.

[0144] Optionally, the ratio of the number of samples in the training set to the number of samples in the test set may be 9: 1. The initial model may be a support vector machine (SVM) model.

[0145] S105: Perform fault detection on the first multi-split air-conditioning system using a fault detection model.

[0146] In some embodiments, after obtaining the trained fault detection model, real-time operating data of the first multi-split air-conditioning system can be obtained, and the real-time operating data can be input into the fault detection model to obtain the fault detection result of the first multi-split air-conditioning system.

[0147] In some embodiments, after obtaining the trained fault detection model, the real-time operation data of the first multi-split air-conditioning system can be used as experimental data of the first multi-split air-conditioning system, and the above steps can be re-executed to update the fault detection model.

[0148] Exemplarily, the fault detection model has a target domain input channel and a source domain input channel. The target domain input channel is used to input real-time operating data of the first multi-split air conditioning system, while the source domain input channel is used to input other labeled operating data. The real-time operating data of the first multi-split air conditioning system can be input into the fault detection model through the target domain input channel, and the above steps can be repeated to update the fault detection model.

[0149] It is understandable that other labeled operation data contains multiple types of operation data, such as various types of fault data. When the real-time operation data in the target domain is insufficient or incomplete, transfer learning can help the fault detection model learn a wider range of fault modes and characteristics, thereby improving the fault detection model's ability to detect target domain faults, as well as improving the model's performance and generalization capabilities.

[0150] based on Figure 5 In the embodiment shown, the embodiment of the present application provides a fault detection method. Since the data distribution of the operating data of the first multi-split air-conditioning system is different from the data distribution of the operating data of the second multi-split air-conditioning system, the fault detection model established based on the operating data of the second multi-split air-conditioning system is not applicable to the first multi-split air-conditioning system. However, the embodiment of the present application provides a fault detection method. When establishing a fault detection model, the first sample set generated based on the normal operating data and faulty operating data of the second multi-split air-conditioning system is adjusted according to the feature space distribution of the second sample set generated based on the normal operating data and faulty operating data of the second multi-split air-conditioning system, so that the feature space distribution of the first sample set is consistent with or close to the feature space distribution of the second training sample. In this way, the fault detection model trained based on the first sample set and the second sample set with consistent or close feature space distribution can perform fault detection on both the first multi-split air-conditioning system and the second multi-split air-conditioning system, thereby realizing cross-system fault detection.

[0151] Furthermore, because the faulty operating data for the second multi-split air conditioning system is difficult to obtain, it falls into the minority category. This means that the proportion of second samples with fault type labels in the second sample set is less than or equal to a preset threshold. This lack of minority samples can lead to insufficient fault classification in the fault detection model. By expanding the second sample set, the amount of samples with fault type labels can be increased, thereby improving the accuracy of the fault detection model.

[0152] The following is an illustrative description of the complete process of the fault detection method provided in the embodiment of the present application.

[0153] like Figure 11 As shown, data preprocessing is performed on the experimental data of the first multi-split air-conditioning system (such as the normal operating data of the first multi-split air-conditioning system) and the experimental data of the second multi-split air-conditioning system (such as the normal operating data and the fault operating data of the second multi-split air-conditioning system) to obtain preprocessed data, and the processed data is divided into a training sample set (i.e., a first sample set and a second sample set) and a verification sample set.

[0154] Next, the second sample set is input into the SMOTE adjustment module, and the second sample set is expanded to adjust the proportion of samples corresponding to the first label, thereby enriching the second sample set.

[0155] Then, the first sample set and the expanded second sample set are input into the DANN module to adjust the first sample set so that the feature space distribution of the first sample set is consistent with or close to the feature space distribution of the expanded second sample set.

[0156] Finally, after training the initial model based on the first sample set and the second sample set whose feature space distribution is consistent or close to each other, and obtaining the trained fault detection model, the real-time operation data of the first multi-split air-conditioning system can be used as the experimental data of the first multi-split air-conditioning system, and the above steps can be re-executed to update the fault detection model, and fault detection can be performed on the first multi-split air-conditioning system.

[0157] Those skilled in the art will appreciate that in one or more of the above examples, the functions described herein can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0158] In the embodiments of the present invention, electronic products, etc., can be divided into functional modules according to the above-mentioned method examples. For example, each functional module can be divided according to each function, or two or more functions can be integrated into a processing module. The above-mentioned integrated modules can be implemented in the form of hardware or software functional modules. It should be noted that the division of modules in the embodiments of the present invention is schematic and is only a logical functional division. In actual implementation, other division methods may be used.

[0159] In the case of dividing each functional module into corresponding functional modules, Figure 12 A schematic diagram of the structure of a fault detection device provided in an embodiment of the present application is shown in FIG. Figure 12 As shown, the fault detection device 1200 may include: an acquisition module 1201 , an expansion module 1202 , an adjustment module 1203 , a training module 1204 and a detection module 1205 .

[0160] In some embodiments, the acquisition module 1201 is configured to acquire a first sample set and a second sample set.

[0161] In some embodiments, the expansion module 1202 is configured to expand the second sample set to adjust the proportion corresponding to the first label when the proportion of second samples with the first label in the second sample set is less than or equal to a preset threshold.

[0162] In some embodiments, the adjustment module 1203 is configured to adjust the first sample set according to the feature space distribution of the second sample set.

[0163] In some embodiments, the training module 1204 is configured to train the initial model based on the first sample set and the second sample set to obtain a trained fault detection model.

[0164] In some embodiments, the detection module 1205 is configured to perform fault detection on the first multi-split air conditioning system using a fault detection model.

[0165] In the case of adopting an integrated unit, the fault detection device 1200 may further include: a storage module and a communication module.

[0166] In some embodiments, the storage module is used to store program codes and data of the fault detection device. In some embodiments, the communication module can be a transceiver, a transceiver circuit or a communication interface.

[0167] An embodiment of the present invention further provides a computer-readable storage medium, which includes computer-executable instructions. When the computer-executable instructions are executed on a computer, the computer executes the method provided in the above embodiment.

[0168] An embodiment of the present invention further provides a computer program product, which can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement the method provided in the above embodiment.

[0169] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0170] The above is only a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or replacements within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A fault detection method, characterized in that: The method comprises: Obtaining a first sample set and a second sample set; wherein the first sample set includes a plurality of first samples generated based on normal operating data of the first multi-split air conditioning system, and the second sample set includes a plurality of second samples generated based on normal operating data and fault operating data of the second multi-split air conditioning system, wherein the second samples have operating mode labels or fault type labels; When a proportion of second samples with a first label in the second sample set is less than or equal to a preset threshold, expanding the second sample set to adjust the proportion corresponding to the first label; the first label belongs to the operating mode label or the fault type label; adjusting the first sample set according to the feature space distribution of the second sample set; Training an initial model based on the first sample set and the second sample set to obtain a trained fault detection model; The fault detection model is used to perform fault detection on the first multi-split air conditioning system.

2. The method according to claim 1, characterized in that The expanding the second sample set includes: Obtain at least one neighboring sample of the second sample having the first label; Performing random linear interpolation processing between the second sample with the first label and each of the neighboring samples to generate newly added samples with the first label; The newly added samples with the first label are added to the second sample set to adjust the proportion corresponding to the first label.

3. The method according to claim 1, characterized in that The adjusting the first sample set according to the feature space distribution of the second sample set includes: Extraction step: extracting, by a feature extractor, a first feature from the first sample, extracting a second feature from a second sample generated based on normal operation data of the second multi-split air-conditioning system, and extracting a third feature from a second sample generated based on fault operation data of the second multi-split air-conditioning system; Input step: inputting the first feature and the second feature into a domain classifier to obtain a domain loss function, and inputting the third feature into a label classifier to obtain a label loss function; Updating step: updating the parameters of the feature extractor, the domain classifier, and the label classifier based on the domain loss function and the label loss function through a back propagation algorithm; The input step and the update step are repeatedly performed until a domain classification loss function value of the domain classifier is greater than or equal to a first threshold, and a label loss function value of the label classifier is less than or equal to a second threshold.

4. The method according to claim 1, wherein The adjusting the first sample set according to the feature space distribution of the second sample set includes: The first sample set and the second sample set are input into a domain adaptation model to adjust the first sample set; wherein the domain adaptation model adopts a DANN structure.

5. The method according to claim 1, wherein The obtaining of the first sample set and the second sample set includes: Acquire first experimental data of the first multi-split air conditioning system and second experimental data of the second multi-split air conditioning system; the first experimental data includes normal operation data of the first multi-split air conditioning system, and the second experimental data includes normal operation data and fault operation data of the second multi-split air conditioning system; Preprocessing the first experimental data and the second experimental data to obtain processed first experimental data and second experimental data; the preprocessing includes one or more of the following: data cleaning, data reduction, data encoding, and data normalization; The first sample set and the second sample set are generated based on the processed first experimental data and the second experimental data.

6. The method according to claim 5, characterized in that The first experimental data includes at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the first multi-split air-conditioning system under different test environments; the second experimental data includes at least one of the following: the operating conditions, load rate, set temperature, and air volume level of the second multi-split air-conditioning system under different test environments.

7. A first multi-split air conditioning system, characterized in that: include: At least one indoor unit; At least one outdoor unit; The controller is configured as: Obtaining a first sample set and a second sample set; wherein the first sample set includes a plurality of first samples generated based on normal operating data of the first multi-split air conditioning system, and the second sample set includes a plurality of second samples generated based on normal operating data and fault operating data of the second multi-split air conditioning system, wherein the second samples have operating mode labels or fault type labels; When a proportion of second samples with a first label in the second sample set is less than or equal to a preset threshold, expanding the second sample set to adjust the proportion corresponding to the first label; the first label belongs to the operating mode label or the fault type label; adjusting the first sample set according to the feature space distribution of the second sample set; Training an initial model based on the first sample set and the second sample set to obtain a trained fault detection model; The fault detection model is used to perform fault detection on the first multi-split air conditioning system.

8. The first multi-split air conditioning system according to claim 7, characterized in that: The controller is configured to perform expansion processing on the second sample set, including: Obtain at least one neighboring sample of the second sample having the first label; Performing random linear interpolation processing between the second sample with the first label and each of the neighboring samples to generate newly added samples with the first label; The newly added samples with the first label are added to the second sample set to adjust the proportion corresponding to the first label.

9. The first multi-split air conditioning system according to claim 7, characterized in that: The controller is configured to adjust the first sample set according to the feature space distribution of the second sample set, including: Extraction step: extracting, by a feature extractor, a first feature from the first sample, extracting a second feature from a second sample generated based on normal operation data of the second multi-split air-conditioning system, and extracting a third feature from a second sample generated based on fault operation data of the second multi-split air-conditioning system; Input step: inputting the first feature and the second feature into a domain classifier to obtain a domain loss function, and inputting the third feature into a label classifier to obtain a label loss function; Updating step: updating the parameters of the feature extractor, the domain classifier, and the label classifier based on the domain loss function and the label loss function through a back propagation algorithm; The input step and the update step are repeatedly performed until the domain classification loss function value of the domain classifier is maximized and the label loss function value of the label classifier is minimized.

10. The first multi-split air conditioning system according to claim 7, characterized in that: The controller is configured to adjust the first sample set according to the feature space distribution of the second sample set, including: The first sample set and the second sample set are input into a domain adaptation model to adjust the first sample set; wherein the domain adaptation model adopts a DANN structure.

Images