Method and device for monitoring operation failure of HVAC system based on data analysis

Abstract

The present invention provides an HVAC system operation fault monitoring method and device based on data analysis, which relates to the field of data analysis technology. In the present invention, first, based on the harmonic semantic information in the equipment vibration time series data, the equipment vibration time series data is semantically enhanced to form an equipment vibration semantic vector; then, based on the surge semantic information in the equipment sound time series data, the equipment sound time series data is semantically enhanced to form an equipment sound semantic vector; thereafter, the equipment temperature time series data is semantically mined to form an equipment temperature semantic vector; further, the equipment vibration semantic vector, the equipment sound semantic vector and the equipment temperature semantic vector are fused to form an equipment operation semantic vector; finally, the operation fault data is predicted and output based on the equipment operation semantic vector. Based on the above method, the problem of relatively low reliability of operation fault monitoring in the prior art can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a method and device for monitoring operation failures of an HVAC system based on data analysis. Background Art

[0002] An HVAC system, a heating, ventilation, and air conditioning (HVAC) system, is a comprehensive facility used to regulate and control the temperature, humidity, and air quality within a building. It has numerous applications, such as intelligent low-temperature cleanroom environmental protection systems and intelligent constant-temperature biological culture cleanroom environmental protection systems. However, prior art research and development of HVAC systems primarily focuses on improving control algorithm accuracy and energy conservation. However, in these application scenarios, the HVAC system must provide continuous, stable, and effective control, making fault monitoring of the system particularly important. However, prior art primarily relies on manual periodic testing or simple empirical threshold comparisons of operating parameters across one or more dimensions. Both approaches struggle to effectively detect operational faults, making timely maintenance difficult and hindering continuous, stable, and efficient operation. Consequently, prior art suffers from relatively low reliability in operational fault monitoring. Summary of the Invention

[0003] In view of this, an object of the present invention is to provide a method and device for monitoring operation faults of an HVAC system based on data analysis, so as to improve the problem of relatively low reliability of operation fault monitoring in the prior art.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0005] A method for monitoring operational faults of an HVAC system based on data analysis, comprising:

[0006] Obtain device vibration time series data, device sound time series data, and device temperature time series data of a target device in a target HVAC system;

[0007] Based on the harmonic semantic information in the device vibration time series data, semantically enhance the device vibration time series data to form a device vibration semantic vector;

[0008] Based on the surge semantic information in the device sound time series data, semantically enhance the device sound time series data to form a device sound semantic vector;

[0009] Performing semantic mining on the device temperature time series data to form a device temperature semantic vector;

[0010] fusing the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector to form a device operation semantic vector;

[0011] The operation failure data of the target device is predicted and output based on the device operation semantic vector, wherein the operation failure data is used to characterize the probability of failure of the target device.

[0012] In some preferred embodiments, in the above-mentioned method for monitoring operation faults of an HVAC system based on data analysis, the step of fusing the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector to form the device operation semantic vector includes:

[0013] Performing a first fusion on the device temperature semantic vector and the device vibration semantic vector to form a first operation semantic vector, wherein the first fusion includes multiple fusion stages;

[0014] performing a second fusion on the device temperature semantic vector and the device sound semantic vector to form a second operation semantic vector, wherein the second fusion includes multiple fusion stages;

[0015] The first operation semantic vector and the second operation semantic vector are convolutionally fused to form a device operation semantic vector.

[0016] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of first fusing the device temperature semantic vector and the device vibration semantic vector to form a first operational semantic vector includes:

[0017] Deeply mining the vibration-temperature fusion vector of the i-th time step to form a vibration-temperature depth vector of the i-th time step, wherein, when the i-th time step is the first time step among multiple time steps, the vibration-temperature fusion vector of the i-th time step is the device vibration semantic vector;

[0018] Performing semantic fusion based on the vibration temperature depth vector of the i-th time step and the device temperature semantic vector to form a fused semantic vector of the i-th time step;

[0019] forming a mapping semantic vector according to the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step;

[0020] Upsampling the mapped semantic vector to form an upsampled semantic vector;

[0021] Linking the vibration-temperature fusion vector of the i-th time step to the upsampled semantic vector to form a vibration-temperature fusion vector of the i+1-th time step;

[0022] A first operation semantic vector is determined based on the vibration-temperature fusion vector of the last time step among the multiple time steps.

[0023] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of performing deep mining on the vibration-temperature fusion vector of the i-th time step to form the vibration-temperature depth vector of the i-th time step includes:

[0024] Downsampling the vibration temperature fusion vector of the i-th time step to form a downsampled fusion vector;

[0025] Splitting the downsampled fusion vector to form a plurality of local fusion vectors, and serializing the plurality of local fusion vectors to form a local fusion vector sequence;

[0026] The local fusion vector sequence is subjected to sequence association mining to form the vibration temperature depth vector of the i-th time step, wherein the sequence association mining includes channel splicing and channel convolution.

[0027] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of downsampling the vibration-temperature fusion vector at the i-th time step to form a downsampled fusion vector includes:

[0028] Performing a pooling operation of multiple scales on the vibration temperature fusion vector of the i-th time step to form a pooled fusion vector of multiple scales;

[0029] Performing convolution compression on the vibration temperature fusion vector of the i-th time step to form a convolution fusion vector;

[0030] Channel splicing is performed on the pooled fusion vectors of the multiple scales and the convolution fusion vector, and channel convolution is performed on the result of the channel splicing to form a downsampled fusion vector of the vibration temperature fusion vector of the i-th time step.

[0031] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of performing semantic fusion based on the vibration-temperature depth vector of the i-th time step and the equipment temperature semantic vector to form a fused semantic vector of the i-th time step includes:

[0032] Performing semantic space conversion on the device temperature semantic vector to form a device temperature conversion vector, and performing gated adjustment on the device temperature conversion vector and the vibration-temperature fusion vector of the i-th time step to form a device temperature adjustment vector, wherein a gating parameter corresponding to the gated adjustment is obtained based on the mapping of the vibration-temperature fusion vector of the i-th time step;

[0033] Sum or average the device temperature adjustment vector and the vibration temperature depth vector to form a preliminary fusion semantic vector;

[0034] The preliminary fused semantic vector is subjected to deep convolution and nonlinear activation to form the fused semantic vector of the i-th time step.

[0035] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of forming a mapping semantic vector based on the mapping parameter distribution at the i-th time step and the fused semantic vector at the i-th time step includes:

[0036] Determining a mapping parameter distribution for the i-th time step, wherein, when the i-th time step is the first time step among the multiple time steps, each parameter in the mapping parameter distribution for the i-th time step is 1; and when the i-th time step is a time step other than the first time step among the multiple time steps, the mapping parameter distribution for the i-th time step is determined based on the mapping parameter distribution for the previous time step;

[0037] A bitwise multiplication operation is performed on the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step to form a mapping semantic vector.

[0038] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of semantically enhancing the equipment vibration time series data based on the harmonic semantic information in the equipment vibration time series data to form an equipment vibration semantic vector includes:

[0039] Performing a time-frequency transformation on the device vibration time series data to form a device vibration spectrum diagram, and determining the harmonic frequency corresponding to the device vibration time series data based on the device vibration spectrum diagram;

[0040] Based on the harmonic frequency, filtering the device vibration time series data to form device vibration harmonic data, and performing time-frequency transformation on the device vibration harmonic data to form a harmonic vibration spectrum diagram;

[0041] Perform convolution mining on the device vibration spectrum graph and the harmonic vibration spectrum graph respectively to form a device vibration spectrum vector and a harmonic vibration spectrum vector;

[0042] Based on the harmonic vibration spectrum vector, the device vibration spectrum vector is mined for significant features to form a device vibration semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic vibration spectrum vector.

[0043] In some preferred embodiments, in the above-mentioned method for monitoring operational faults of an HVAC system based on data analysis, the step of semantically enhancing the equipment sound time series data based on surge semantic information in the equipment sound time series data to form an equipment sound semantic vector includes:

[0044] Filtering the device sound time series data based on the surge frequency to form device sound filter data, performing time-frequency transformation on the device sound filter data to form a harmonic sound spectrogram, and performing time-frequency transformation on the device sound time series data to form a device sound spectrogram;

[0045] Performing convolution mining on the harmonic sound spectrum graph and the device sound spectrum graph respectively to form a harmonic sound spectrum vector and a device sound spectrum vector;

[0046] Based on the harmonic sound spectrum vector, the device sound spectrum vector is mined for significant features to form a device sound semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic sound spectrum vector.

[0047] An embodiment of the present invention further provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the above-mentioned method for monitoring operation faults of an HVAC system based on data analysis.

[0048] The data analysis-based HVAC system operational fault monitoring method and device provided by the present invention first performs semantic enhancement on the equipment vibration time series data based on the harmonic semantic information in the data to form an equipment vibration semantic vector. Then, based on the surge semantic information in the equipment sound time series data, the equipment sound time series data is semantically enhanced to form an equipment sound semantic vector. Then, semantic mining is performed on the equipment temperature time series data to form an equipment temperature semantic vector. Furthermore, the equipment vibration semantic vector, the equipment sound semantic vector, and the equipment temperature semantic vector are fused to form an equipment operation semantic vector. Finally, operational fault data is predicted and output based on the equipment operation semantic vector. Based on the above method, in the first aspect, the semantic fusion of equipment time series data in the three dimensions of vibration, sound, and temperature enables the generated equipment operation semantic vector to comprehensively represent the operating process of the target equipment. Secondly, the semantic information in the vibration dimension is enhanced based on harmonic semantic information. Harmonics are highly representative of internal imbalances within equipment (such as compressors). Therefore, the equipment vibration semantic vector can not only represent the global semantic information of vibration but also provide important characterization of internal anomalies. Thirdly, because the semantic information of the sound dimension is semantically enhanced based on surge semantic information, which is a phenomenon of periodic fluctuations in pressure and flow caused by dynamic blockage of the gas flow path during equipment (compressor) operation, surge plays an important role in the semantic representation of some abnormal conditions. Therefore, the device sound semantic vector can be used to represent the global semantic information of the sound and also to provide important characterization of some equipment anomalies. Based on this, it is possible to fully explore the potential semantic information of the target equipment during operation and focus on important semantic information, thereby improving the reliability of fault prediction and further improving the relatively low reliability of operational fault monitoring existing in existing technologies.

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a structural block diagram of an electronic device provided by an embodiment of the present invention.

[0051] Figure 2 A schematic flow chart of the modules included in the device for monitoring operation faults of an HVAC system based on data analysis provided by an embodiment of the present invention.

[0052] Figure 3 A schematic diagram of the steps of a method for monitoring operational faults of an HVAC system based on data analysis provided by an embodiment of the present invention.

[0053] Figure 4 A schematic diagram of semantic enhancement provided by an embodiment of the present invention.

[0054] Figure 5 A schematic diagram of the first fusion provided in an embodiment of the present invention.

[0055] Figure 6 A schematic diagram of sequence association mining provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0057] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0058] like Figure 1 As shown, an embodiment of the present invention provides an electronic device, wherein the electronic device may include a memory and a processor.

[0059] Specifically, the memory and processor are directly or indirectly electrically connected to each other to enable data transmission or interaction. For example, the electrical connection may be achieved via one or more communication buses or signal lines. The memory may store at least one software function module (computer program) in the form of software or firmware. The processor may be configured to execute the executable computer program stored in the memory, thereby implementing the data analysis-based HVAC system operational fault monitoring method provided in an embodiment of the present invention (as described below). For details, please refer to the relevant description below.

[0060] In addition, the above software function modules can be based on data analysis of the HVAC system operation fault monitoring device including Figure 2 Each module shown, wherein each module specifically includes:

[0061] A time series data acquisition module, used to acquire device vibration time series data, device sound time series data, and device temperature time series data of a target device in a target HVAC system;

[0062] a harmonic semantic enhancement module, configured to perform semantic enhancement on the device vibration time series data based on the harmonic semantic information in the device vibration time series data to form a device vibration semantic vector;

[0063] a surge semantic enhancement module, configured to perform semantic enhancement on the device sound time series data based on surge semantic information in the device sound time series data to form a device sound semantic vector;

[0064] A semantic mining module, configured to perform semantic mining on the device temperature time series data to form a device temperature semantic vector;

[0065] a semantic fusion module, configured to fuse the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector to form a device operation semantic vector;

[0066] A fault prediction module is used to predict and output operation fault data of the target device based on the device operation semantic vector, wherein the operation fault data is used to characterize the probability of failure of the target device.

[0067] Optionally, the memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0068] Optionally, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0069] and, Figure 1 The structure shown is only for illustration, and the electronic device may also include Figure 1 More or fewer components than shown, or with Figure 1The different configurations shown may, for example, include a communication unit for exchanging information with other devices (various sensors, etc.).

[0070] In an alternative example, the electronic device may be a server with data processing capabilities.

[0071] Combine Figure 3 The embodiment of the present invention further provides a method for monitoring the operation failure of an HVAC system based on data analysis, which can be applied to the above-mentioned electronic device. The method steps defined in the process related to the method for monitoring the operation failure of an HVAC system based on data analysis can be implemented by the electronic device. Figure 3 The specific process shown is explained in detail.

[0072] Step S110 , obtaining device vibration time series data, device sound time series data, and device temperature time series data of a target device in a target HVAC system.

[0073] In an embodiment of the present invention, the electronic device can obtain device vibration time series data, device sound time series data, and device temperature time series data of a target device in a target HVAC system. For example, the target device may be a compressor (responsible for driving the circulation of refrigerant, ensuring that heat can be effectively transferred and released, playing a key role in the air conditioning system, and being a core component for implementing the refrigerant cycle). Thus, data can be collected during the operation of the compressor using a vibration sensor, sound sensor, and temperature sensor provided on the compressor, thereby generating corresponding device vibration time series data, device sound time series data, and device temperature time series data, respectively.

[0074] Step S120 : Based on the harmonic semantic information in the device vibration time series data, semantic enhancement is performed on the device vibration time series data to form a device vibration semantic vector.

[0075] In an embodiment of the present invention, after acquiring the device vibration time series data, the electronic device can perform semantic enhancement on the device vibration time series data based on the harmonic semantic information in the device vibration time series data to form a device vibration semantic vector. In other words, the global semantic information of the device vibration can be semantically enhanced based on the harmonic semantic information, so that the device vibration semantic vector can not only represent the global semantic information of the device vibration, but also achieve important representation of the corresponding harmonic semantic information, taking into account both the comprehensiveness and accuracy of the semantic representation.

[0076] Step S130 : Based on the surge semantic information in the device sound time series data, semantic enhancement is performed on the device sound time series data to form a device sound semantic vector.

[0077] In an embodiment of the present invention, after acquiring the device sound time-series data, the electronic device can perform semantic enhancement on the device sound time-series data based on the surge semantic information in the device sound time-series data to form a device sound semantic vector. In other words, the global semantic information of the device sound can be semantically enhanced based on the surge semantic information, so that the device sound semantic vector can not only represent the global semantic information of the device sound, but also achieve important representation of the corresponding surge semantic information, thus ensuring both comprehensiveness and accuracy of the semantic representation.

[0078] Step S140 , performing semantic mining on the device temperature time series data to form a device temperature semantic vector.

[0079] In an embodiment of the present invention, after acquiring the device temperature time series data, the electronic device may perform semantic mining on the device temperature time series data to form a device temperature semantic vector, that is, directly mining the global semantic information of the device temperature time series data.

[0080] Step S150 , fusing the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector to form a device operation semantic vector.

[0081] In an embodiment of the present invention, after forming the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector, the electronic device may fuse the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector to form a device operation semantic vector. In other words, the three dimensions of semantic information may be fused to obtain multi-dimensional semantic information about the operation process.

[0082] Step S160: outputting the operation fault data of the target device based on the prediction of the device operation semantic vector.

[0083] In an embodiment of the present invention, after forming the device operation semantic vector, the electronic device can predict and output the operation fault data of the target device based on the device operation semantic vector. The operation fault data is used to characterize the probability of failure of the target device, such as 0-1. It should be noted that in other embodiments, the device operation semantic vector can also be adjusted in combination with the operation data of the related devices of the target device, and then the prediction output is made based on the adjusted semantic vector. For example, when the target device is a compressor, the related devices can be condensers and evaporators, etc., and the data of the three dimensions mentioned above can also be collected and semantically fused for each device to form a corresponding device operation semantic vector. Then, based on mechanisms such as cross-attention, the device operation semantic vector corresponding to each related device and the device operation semantic vector corresponding to the target device can be fused, and then the prediction output is made based on the fused vector. It should be noted that the relevance of the related devices lies in:

[0084] Condenser blockage or poor heat dissipation will cause the refrigerant to be unable to release heat effectively, causing the compressor to overheat. Overheating of the compressor may cause damage to internal components such as bearings or motors;

[0085] If the evaporator is clogged or cannot dissipate heat effectively, it will cause abnormal refrigerant pressure and too low compressor suction pressure, which will affect the efficiency of the compressor and may shorten the life of the compressor.

[0086] Based on the above method, in the first aspect, the semantic fusion of device time series data across the three dimensions of vibration, sound, and temperature enables the resulting device operation semantic vector to comprehensively characterize the target device's operating process. Secondly, the semantic information in the vibration dimension is enhanced based on harmonic semantic information. Harmonics are highly representative of internal imbalances within devices (such as compressors). Therefore, the device vibration semantic vector can not only represent the global semantic information of vibration, but also provide important characterizations of internal anomalies. Thirdly, the semantic information in the sound dimension is enhanced based on surge semantic information. Surge is a periodic fluctuation in pressure and flow caused by dynamic blockage of the gas flow path during device (compressor) operation. Therefore, surge plays an important role in the semantic characterization of some anomalies. Therefore, the device sound semantic vector can also represent the global semantic information of sound, and also provide important characterizations of some device anomalies. Based on this, we can achieve comprehensive mining of potential semantic information during the operation of the target equipment and focus on important semantic information, thereby improving the reliability of fault prediction and further improving the relatively low reliability of operation fault monitoring in the existing technology.

[0087] In the first part, it should be noted that for step S110, the specific method of obtaining the device vibration time series data, device sound time series data and device temperature time series data of the target device in the target HVAC system is not limited and can be selected accordingly according to actual needs.

[0088] For example, in an alternative embodiment, the electronic device can communicate with the corresponding sensor in real time to obtain the corresponding time series data, and then perform subsequent processing to achieve timely fault prediction. For another example, in another alternative embodiment, the electronic device can also obtain the corresponding time series data from a database, and then perform subsequent processing, such as periodically predicting faults. In this way, after each sensor generates the corresponding time series data, it can first be stored in the corresponding database.

[0089] In the second part, it should be noted that for step S120, the specific method of semantically enhancing the device vibration time series data is not limited and can be selected according to actual needs.

[0090] For example, in an alternative embodiment, time domain semantic mining can be performed on the device vibration time series data and the harmonic components in the device vibration time series data respectively (for example, this can be achieved through LSTM (long short-term memory network) and GRU (gated recurrent unit)) to obtain two corresponding semantic vectors. Then, the two semantic vectors can be spliced, summed, averaged or attention-processed, etc. to obtain a device vibration semantic vector.

[0091] For example, in another alternative embodiment, in order to ensure that the harmonic semantic information can be effectively integrated during the semantic enhancement process, thereby improving the semantic representation ability of the formed device vibration semantic vector, the above-mentioned step S120 may further include the following executable content (combined with Figure 4 shown):

[0092] In the first step, a time-frequency transformation (such as Fourier transform) can be performed on the device vibration time series data to form a device vibration spectrum diagram, and based on the device vibration spectrum diagram, the harmonic frequencies corresponding to the device vibration time series data are determined. For example, the fundamental frequency can be determined first, and then, based on n times the fundamental frequency, each harmonic frequency can be obtained, such as 2 times the fundamental frequency, 3 times the fundamental frequency, etc.;

[0093] In a second step, the device vibration time series data may be filtered based on the harmonic frequency (for example, a corresponding filter may be designed based on the harmonic frequency to filter out information of other frequencies) to form device vibration harmonic data, and the device vibration harmonic data may be subjected to a time-frequency transform (such as a Fourier transform) to form a harmonic vibration spectrum diagram;

[0094] In the third step, convolution mining can be performed on the device vibration spectrum graph and the harmonic vibration spectrum graph respectively to form a device vibration spectrum vector and a harmonic vibration spectrum vector;

[0095] In the fourth step, significant feature mining can be performed on the device vibration spectrum vector based on the harmonic vibration spectrum vector to form a device vibration semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic vibration spectrum vector. For example, the harmonic vibration spectrum vector can be linearly transformed (such as through a fully connected network layer), and then the result of the linear transformation is nonlinearly activated to obtain the gating parameters. The gating parameters and the device vibration spectrum vector can then be bitwise multiplied, and the result of the bitwise multiplication operation and the device vibration spectrum vector can be averaged or added to obtain the device vibration semantic vector. In addition, when performing cross-attention processing, the result of the cross-attention processing can be averaged or added to the device vibration spectrum vector to obtain the device vibration semantic vector. In addition, in the cross-attention processing, the query vector corresponds to the harmonic vibration spectrum vector, and the key vector and value vector correspond to the device vibration spectrum vector (the specific processing process can refer to the relevant existing technology). In addition, the gated adjustment processing process can refer to the following formula:

[0096] Y=sigmiod(Ax1+b)⊙x2+x2;

[0097] Where Y is the device vibration semantic vector, sigmiod is the function corresponding to nonlinear activation (other functions can also be used as required), A and b are the weight matrix and bias vector of linear transformation, x1 is the harmonic vibration spectrum vector, x2 is the device vibration spectrum vector, and ⊙ represents bitwise multiplication.

[0098] In the third part, it should be noted that for step S130, the specific method of semantically enhancing the device sound timing data is not limited and can be selected according to actual needs.

[0099] For example, in an alternative embodiment, time domain semantic mining can be performed on the device sound timing data and the surge component in the device sound timing data respectively (for example, this can be achieved through LSTM (long short-term memory network) and GRU (gated recurrent unit)) to obtain two corresponding semantic vectors. Then, the two semantic vectors can be spliced, summed, averaged or attention-processed to obtain the device sound semantic vector.

[0100] For example, in another alternative embodiment, to ensure effective integration of surge semantic information during semantic enhancement, thereby improving the semantic representation capability of the generated device sound semantic vector, the aforementioned step S130 may include the following executable contents:

[0101] In the first step, the device sound time series data may be filtered based on a surge frequency (the surge frequency may be configured based on a typical surge frequency of the target device, for example, when the target device is a compressor, the surge frequency is typically between tens of hertz and hundreds of hertz) to form device sound filtered data, and a time-frequency transform (such as a Fourier transform) may be performed on the device sound filtered data to form a harmonic sound spectrum graph, and a time-frequency transform (such as a Fourier transform) may be performed on the device sound time series data to form a device sound spectrum graph;

[0102] In the second step, convolution mining can be performed on the harmonic sound spectrum graph and the device sound spectrum graph respectively to form a harmonic sound spectrum vector and a device sound spectrum vector;

[0103] In the third step, based on the harmonic sound spectrum vector, the device sound spectrum vector can be mined for significant features to form a device sound semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic sound spectrum vector, as described above.

[0104] In the fourth part, it should be noted that for step S140 , the specific method of performing semantic mining on the device temperature time series data is not limited and can be selected according to actual needs.

[0105] For example, in an alternative embodiment, the device temperature time series data can be subjected to time domain semantic mining (for example, this can be achieved through LSTM (long short-term memory network) and GRU (gated recurrent unit)) to obtain the corresponding device temperature semantic vector, or the results of the semantic mining can be subjected to self-attention processing to achieve semantic enhancement, that is, through internal association mining, the mining of significant features can be achieved, thereby forming a device temperature semantic vector.

[0106] For example, in another alternative embodiment, the device temperature time series data can be converted into a time-frequency data to form a corresponding device temperature spectrum diagram. Then, the device temperature spectrum diagram can be convolutionally mined to form a corresponding device temperature semantic vector. Alternatively, the result of the convolution mining can be self-attention processed to achieve semantic enhancement, that is, through internal association mining, the mining of significant features can be achieved, thereby forming a device temperature semantic vector.

[0107] In the fifth part, it should be noted that for step S150, the specific method of fusing the device vibration semantic vector, the device sound semantic vector and the device temperature semantic vector is not limited and can be selected accordingly according to actual needs.

[0108] For example, in an alternative embodiment, the device vibration semantic vector, the device sound semantic vector, and the device temperature semantic vector can be spliced together, and then convolution mining can be performed on the spliced result to form a device operation semantic vector.

[0109] For another example, in another alternative embodiment, in order to improve the accuracy of fusion, the above-mentioned step S150 may further include step S151, step S152 and step S153, and the specific content of each step is described as follows.

[0110] Step S151 : performing a first fusion on the device temperature semantic vector and the device vibration semantic vector to form a first operation semantic vector.

[0111] In an embodiment of the present invention, the device temperature semantic vector and the device vibration semantic vector may be first fused to form a first operation semantic vector, wherein the first fusion includes multiple fusion stages, thereby achieving full and effective fusion.

[0112] Step S152: performing a second fusion on the device temperature semantic vector and the device sound semantic vector to form a second operation semantic vector.

[0113] In an embodiment of the present invention, a second fusion can be performed on the device temperature semantic vector and the device sound semantic vector to form a second operation semantic vector. The second fusion includes multiple fusion stages, thereby achieving full and effective fusion. It should be noted that the specific fusion method for the second fusion can be the same as that for the first fusion.

[0114] Step S153: Convolutionally fuse the first operation semantic vector and the second operation semantic vector to form a device operation semantic vector.

[0115] In an embodiment of the present invention, after forming the first operation semantic vector and the second operation semantic vector, the first operation semantic vector and the second operation semantic vector can be convolutionally fused (i.e., first spliced and then convolutionally mined) to form a device operation semantic vector.

[0116] That is, in the present invention, in one aspect, semantic enhancement is first performed internally within the vibration dimension based on harmonic semantic information. Then, external semantic enhancement is performed on the semantic information in the temperature dimension based on semantic information, thereby obtaining a first operational semantic vector. In another aspect, semantic enhancement is first performed internally within the sound dimension based on surge semantic information. Then, external semantic enhancement is performed on the semantic information in the temperature dimension based on semantic information, thereby obtaining a second operational semantic vector. Thirdly, the first and second operational semantic vectors are further fused to achieve high-precision and comprehensive capture of global semantic information.

[0117] Alternatively, in the above step S151, the specific manner of first fusing the device temperature semantic vector and the device vibration semantic vector is not limited. For example, in an alternative embodiment, in order to achieve full fusion of semantic information of two dimensions in multiple stages, the above step S151 may further include steps S151a, S151b, S151c, S151d, S151e, and S151f. The specific contents of each step are as follows (combined with Figure 5 shown).

[0118] Step S151a: Deeply mine the vibration-temperature fusion vector of the i-th time step to form a vibration-temperature depth vector of the i-th time step.

[0119] In an embodiment of the present invention, the vibration-temperature fusion vector of the i-th time step can be deeply mined to form a vibration-temperature depth vector of the i-th time step. When the i-th time step is the first time step among multiple time steps, the vibration-temperature fusion vector of the i-th time step is the device vibration semantic vector. In other words, the device vibration semantic vector can be deeply mined first to form the vibration-temperature depth vector of the first time step. Furthermore, i is an integer greater than or equal to 1.

[0120] Step S151b: performing semantic fusion based on the vibration temperature depth vector of the i-th time step and the device temperature semantic vector to form a fused semantic vector of the i-th time step.

[0121] In an embodiment of the present invention, after forming the vibration-temperature depth vector of the i-th time step, semantic fusion can be performed based on the vibration-temperature depth vector of the i-th time step and the device temperature semantic vector to form a fused semantic vector of the i-th time step. In this way, the fused semantic vector of the i-th time step can synchronously represent the semantic information of the two dimensions of temperature and vibration. In this way, the fusion of the semantic information of the temperature dimension can be achieved at each time step.

[0122] Step S151c: forming a mapping semantic vector according to the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step.

[0123] In this embodiment of the present invention, a mapping semantic vector is formed based on the mapping parameter distribution at the i-th time step and the fused semantic vector at the i-th time step. In other words, the fused semantic vector at the corresponding time step can be further adjusted based on the mapping parameter distribution to enhance the representational capabilities of the formed mapping semantic vector.

[0124] Step S151d: upsample the mapped semantic vector to form an upsampled semantic vector.

[0125] In an embodiment of the present invention, taking into account that in the above-mentioned deep mining process, semantic vector compression is generally achieved, the mapping semantic vector can be upsampled (such as interpolation, deconvolution, depooling, etc.) to form an upsampled semantic vector, so that the size of the semantic vector is restored to the original size, that is, the size of the device vibration semantic vector.

[0126] Step S151e: linking the vibration-temperature fusion vector of the i-th time step to the upsampled semantic vector to form a vibration-temperature fusion vector of the i+1-th time step.

[0127] In an embodiment of the present invention, after the upsampled semantic vector is formed, the vibration-temperature fusion vector of the i-th time step can be linked to the upsampled semantic vector to form the vibration-temperature fusion vector of the i+1-th time step. For example, the vibration-temperature fusion vector of the i-th time step and the upsampled semantic vector can be added or averaged to form the vibration-temperature fusion vector of the i+1-th time step, such as forming the vibration-temperature fusion vector of the second time step, the vibration-temperature fusion vector of the third time step, and the vibration-temperature fusion vector of the fourth time step in sequence.

[0128] Step S151f: determining a first operation semantic vector based on the vibration-temperature fusion vector of the last time step among the multiple time steps.

[0129] In an embodiment of the present invention, after completing the fusion of multiple stages, the first operation semantic vector can be determined based on the vibration-temperature fusion vector of the last time step among the multiple time steps. For example, the vibration-temperature fusion vector of the last time step can be used as the first operation semantic vector, or other mapping processing can be performed.

[0130] Optionally, in the above-mentioned step S151a, the specific method of deep mining the vibration-temperature fusion vector of the i-th time step is not limited. For example, in an alternative embodiment, in order to realize the mining of deep and advanced latent semantic information through deep mining, the above-mentioned step S151a may further include step a1, step a2 and step a3, and the specific content of each step is described as follows.

[0131] Step a1: down-sample the vibration-temperature fusion vector of the i-th time step to form a down-sampled fusion vector.

[0132] In an embodiment of the present invention, the vibration-temperature fusion vector of the i-th time step can be downsampled to form a downsampled fusion vector. In this way, downsampling can reduce the amount of data, reduce the computational complexity, improve the processing efficiency, and make subsequent deep mining more efficient.

[0133] Step a2: dividing the downsampled fusion vector into multiple local fusion vectors, and serializing the multiple local fusion vectors to form a local fusion vector sequence.

[0134] In an embodiment of the present invention, after forming the downsampled fusion vector, the downsampled fusion vector can be segmented to form multiple local fusion vectors (for example, a sliding window segmentation can be performed, and the step size and window size of the sliding window can be configured according to actual needs, or can be determined during the learning process of the corresponding samples and labels), and the multiple local fusion vectors are serialized to form a sequence of local fusion vectors. In this way, segmenting the downsampled vector into multiple local fusion vectors helps capture local features in the data and improves sensitivity to local changes. In addition, serializing the local vectors to form an ordered data sequence facilitates sequence analysis and helps identify and mine patterns and associations in the sequence.

[0135] Step a3: performing sequence association mining on the local fusion vector sequence to form the vibration temperature depth vector of the i-th time step.

[0136] In an embodiment of the present invention, after forming the local fusion vector sequence, the local fusion vector sequence can be subjected to sequence association mining to form the vibration temperature depth vector of the i-th time step. The sequence association mining includes channel splicing and channel convolution. In this way, sequence association mining is implemented through channel splicing and channel convolution, which can further extract the deep features of the data, enhance the feature expression ability, and form a reliable vibration temperature depth vector. It should be noted that, in combination with channel splicing and channel convolution, Figure 6, the size of each local fusion vector can be n*m, and the local fusion vector is a. In this way, the size of the spliced vector formed by channel splicing can be a*n*m. Then, a vibration temperature depth vector can be formed by performing channel convolution. The size of the vibration temperature depth vector can be n*m, wherein the size of the convolution kernel of the channel convolution is a*1*1. In addition, in some other embodiments, after performing channel convolution, the result of the channel convolution can be further processed, such as normalization processing, self-attention processing, residual linking with the result of channel convolution, normalization processing, etc., so as to form the vibration temperature depth vector of the i-th time step.

[0137] Optionally, in the above step a1, the specific manner of downsampling the vibration-temperature fusion vector at the i-th time step is not limited. For example, in an alternative embodiment, in order to extract rich semantic information during the compression process, the above step a1 may further include the following practicable contents:

[0138] In the first step, the vibration-temperature fusion vector of the i-th time step can be pooled at multiple scales to form pooled fusion vectors of multiple scales. In this way, local semantic information of different scales can be captured. For example, the sizes of the pooled fusion vectors of multiple scales can be 1 / 2, 1 / 4, 1 / 8, etc. of the vibration-temperature fusion vector of the i-th time step, respectively.

[0139] In the second step, the vibration temperature fusion vector of the i-th time step may be convolved and compressed to form a convolution fusion vector. For example, convolution may be performed using a 3*3 convolution kernel with a step size of 2 and without edge padding.

[0140] In the third step, channel splicing can be performed on the pooled fusion vectors of the multiple scales and the convolution fusion vectors (it should be noted that since the sizes of the vectors may be different, they can be upsampled to the largest size first and then channel splicing can be performed), and channel convolution can be performed on the result of channel splicing (such as the convolution kernel is b*1*1, b is the cumulative number of the pooled fusion vectors of the multiple scales and the convolution fusion vectors) to form a downsampled fusion vector of the vibration temperature fusion vector of the i-th time step. In this way, the vector's ability to express nonlinear semantic information is increased through the operations of channel splicing and channel convolution.

[0141] Optionally, in the above step S151b, the specific manner of performing semantic fusion based on the vibration temperature depth vector at the i-th time step and the device temperature semantic vector is not limited. In order to achieve reliable fusion of semantic information of different temperatures, the above step S151b may further include the following executable content:

[0142] In the first step, the device temperature semantic vector can be converted into a semantic space to form a device temperature conversion vector, and the device temperature conversion vector and the vibration temperature fusion vector of the i-th time step can be gated to form a device temperature adjustment vector, wherein the gating parameters corresponding to the gating adjustment are obtained based on the vibration temperature fusion vector mapping of the i-th time step (for the specific processing process, please refer to the relevant description in the previous text). In addition, since temperature and vibration actually belong to two different dimensions and are in different semantic spaces, in order to achieve semantic fusion, that is, to ensure the reliability of gated adjustment, it is necessary to first perform semantic space conversion on the device temperature semantic vector, so as to convert it into a semantic space with a similar vibration dimension. Specifically, The device temperature semantic vector is transformed at least once (when multiple transformations are performed, the object of the first transformation is the device temperature semantic vector, and the objects of subsequent transformations are the output of the previous transformation) to form a device temperature conversion vector. Specifically, this can be achieved through the corresponding transformation matrix and bias parameters. That is, the device temperature semantic vector and the transformation matrix are matrix multiplied, and then the result of the matrix multiplication and the bias parameters are added (bit by bit) to achieve the conversion. In this way, the device temperature semantic vector is converted to a space more suitable for fusion, which not only improves the fusion efficiency but also enhances the expressiveness of the features. This conversion involves nonlinear transformation, which enables better capture of the potential semantic relationship between device temperature and vibration.

[0143] In the second step, the device temperature adjustment vector and the vibration temperature depth vector may be summed or averaged to form a preliminary fusion semantic vector;

[0144] In the third step, the preliminary fused semantic vector can be subjected to deep convolution and nonlinear activation (i.e., deep convolution is performed first, and then, nonlinear activation is performed on the result of the deep convolution) to form the fused semantic vector of the i-th time step, wherein the deep convolution can be implemented by cascading multiple convolution layers, i.e., the input of the first convolution layer is the preliminary fused semantic vector, the input of the second and subsequent convolution layers is the output of the previous convolution layer, and the output of the last convolution layer can be used as the fused semantic vector of the i-th time step after nonlinear activation.

[0145] Optionally, in the above step S151c, the specific manner of the mapping parameter distribution at the i-th time step and the fused semantic vector at the i-th time step is not limited. For example, in an alternative embodiment, in order to capture important semantic information by performing corresponding mapping, the above step S151c may further include the following executable content:

[0146] In the first step, a mapping parameter distribution of the i-th time step can be determined, wherein, when the i-th time step is the first time step among the multiple time steps, each parameter in the mapping parameter distribution of the i-th time step is 1 (in this way, uniform mapping can be achieved, thereby ensuring that all fused semantic vectors are treated equally in the initial stage); when the i-th time step is a time step other than the first time step among the multiple time steps, the mapping parameter distribution of the i-th time step is determined based on the mapping parameter distribution of the previous time step. In this way, this dynamic adjustment enables the semantic vector of the current time step to be gradually optimized based on the previous state information;

[0147] In a second step, a bitwise multiplication operation may be performed on the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step to form a mapping semantic vector.

[0148] Wherein, it can be selected that, in the above steps, the specific method of determining the mapping parameter distribution of the current time step based on the mapping parameter distribution of the previous time step can be:

[0149] First, the difference (absolute difference) of the vibration temperature fusion vector of the previous time step (i.e., the vibration temperature fusion vector of the i-1th time step, i is greater than or equal to 3) and the vibration temperature fusion vector of the previous two time steps (i.e., the vibration temperature fusion vector of the i-2th time step) can be calculated to obtain the corresponding difference vector. Then, the difference vector can be nonlinearly activated (mapped to the interval 0-1) to obtain an activation difference vector. Then, the activation difference vector can be downsampled to obtain a corresponding downsampled difference vector (the size can be the same as the size of the mapping parameter distribution). Then, the downsampled difference vector and the mapping parameter distribution of the previous time step can be bitwise multiplied to form the mapping parameter distribution of the current time step.

[0150] In the sixth part, it should be noted that for step S160, the specific method of outputting the operation fault data of the target device based on the device operation semantic vector prediction is not limited and can be selected accordingly according to actual needs.

[0151] For example, in an alternative embodiment, the device operation semantic vector can be fully connected to form a corresponding fully connected vector, the size of which can be 1*1. Then, the fully connected vector can be identity mapped or linearly mapped to obtain a probability value of 0-1, which is used as operation fault data.

[0152] In summary, the present invention provides an operation fault monitoring method and device for an HVAC system based on data analysis. First, based on the harmonic semantic information in the equipment vibration time series data, the equipment vibration time series data is semantically enhanced to form an equipment vibration semantic vector; then, based on the surge semantic information in the equipment sound time series data, the equipment sound time series data is semantically enhanced to form an equipment sound semantic vector; thereafter, the equipment temperature time series data is semantically mined to form an equipment temperature semantic vector; further, the equipment vibration semantic vector, the equipment sound semantic vector and the equipment temperature semantic vector are fused to form an equipment operation semantic vector; finally, the operation fault data is predicted and output based on the equipment operation semantic vector. Based on the above method, in the first aspect, since the equipment time series data in the three dimensions of vibration, sound and temperature are semantically fused, the formed equipment operation semantic vector can comprehensively characterize the operation process of the target equipment. Secondly, because the semantic information of the vibration dimension is enhanced based on harmonic semantic information, and harmonics are highly representative of imbalances within equipment (such as compressors), the device vibration semantic vector can be used to represent not only the global semantic information of vibration but also important internal anomalies. Thirdly, because the semantic information of the sound dimension is enhanced based on surge semantic information, which is a periodic fluctuation in pressure and flow caused by dynamic blockage of the gas flow path during equipment (compressor) operation, it is important for the semantic representation of certain anomalies. Therefore, the device sound semantic vector can be used to represent not only the global semantic information of sound but also important equipment anomalies. This allows for comprehensive mining of latent semantic information during the operation of the target equipment and a focused focus on important semantic information, thereby improving the reliability of fault prediction and addressing the relatively low reliability of operational fault monitoring in existing technologies.

[0153] In the several embodiments provided in the embodiments of the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of the devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or can be implemented using a combination of dedicated hardware and computer instructions.

[0154] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0155] If the functions are implemented in the form of software modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, electronic device, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks. It should be noted that, in this document, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or device. Without further constraints, an element defined by the phrase "comprises a..." does not preclude the existence of additional identical elements in the process, method, article or apparatus that includes the element.

[0156] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for monitoring operational faults of an HVAC system based on data analysis, characterized in that: include: Obtain device vibration time series data, device sound time series data, and device temperature time series data of a target device in a target HVAC system; Based on the harmonic semantic information in the device vibration time series data, semantically enhance the device vibration time series data to form a device vibration semantic vector; Based on the surge semantic information in the device sound time series data, semantically enhance the device sound time series data to form a device sound semantic vector; Performing semantic mining on the device temperature time series data to form a device temperature semantic vector; Deeply mine the vibration-temperature fusion vector of the i-th time step to form the vibration-temperature depth vector of the i-th time step, wherein, when the i-th time step is the first time step among multiple time steps, the vibration-temperature fusion vector of the i-th time step is the device vibration semantic vector; semantic fusion is performed based on the vibration-temperature depth vector of the i-th time step and the device temperature semantic vector to form the fusion semantic vector of the i-th time step; a mapping semantic vector is formed based on the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step; upsample the mapping semantic vector to form an upsampled semantic vector; link the vibration-temperature fusion vector of the i-th time step to the upsampled semantic vector to form the vibration-temperature fusion vector of the i+1-th time step; and a first operation semantic vector is determined based on the vibration-temperature fusion vector of the last time step among the multiple time steps; performing a second fusion on the device temperature semantic vector and the device sound semantic vector to form a second operation semantic vector, wherein the second fusion includes multiple fusion stages; Performing convolution fusion on the first operation semantic vector and the second operation semantic vector to form a device operation semantic vector; The operation failure data of the target device is predicted and output based on the device operation semantic vector, wherein the operation failure data is used to characterize the probability of failure of the target device.

2. The method for monitoring operational faults of an HVAC system based on data analysis according to claim 1, wherein: The step of performing deep mining on the vibration-temperature fusion vector of the i-th time step to form the vibration-temperature depth vector of the i-th time step includes: Downsampling the vibration temperature fusion vector of the i-th time step to form a downsampled fusion vector; Splitting the downsampled fusion vector to form a plurality of local fusion vectors, and serializing the plurality of local fusion vectors to form a local fusion vector sequence; The local fusion vector sequence is subjected to sequence association mining to form the vibration temperature depth vector of the i-th time step, wherein the sequence association mining includes channel splicing and channel convolution.

3. The method for monitoring operational faults of an HVAC system based on data analysis according to claim 2, wherein: The step of downsampling the vibration temperature fusion vector of the i-th time step to form a downsampled fusion vector includes: Performing a pooling operation of multiple scales on the vibration temperature fusion vector of the i-th time step to form a pooled fusion vector of multiple scales; Performing convolution compression on the vibration temperature fusion vector of the i-th time step to form a convolution fusion vector; Channel splicing is performed on the pooled fusion vectors of the multiple scales and the convolution fusion vector, and channel convolution is performed on the result of the channel splicing to form a downsampled fusion vector of the vibration temperature fusion vector of the i-th time step.

4. The method for monitoring operational faults of an HVAC system based on data analysis according to claim 1, wherein: The step of performing semantic fusion based on the vibration temperature depth vector of the i-th time step and the device temperature semantic vector to form a fused semantic vector of the i-th time step includes: Performing semantic space conversion on the device temperature semantic vector to form a device temperature conversion vector, and performing gated adjustment on the device temperature conversion vector and the vibration-temperature fusion vector of the i-th time step to form a device temperature adjustment vector, wherein a gating parameter corresponding to the gated adjustment is obtained based on the mapping of the vibration-temperature fusion vector of the i-th time step; Sum or average the device temperature adjustment vector and the vibration temperature depth vector to form a preliminary fusion semantic vector; The preliminary fused semantic vector is subjected to deep convolution and nonlinear activation to form the fused semantic vector of the i-th time step.

5. The method for monitoring operational faults of an HVAC system based on data analysis according to claim 1, wherein: The step of forming a mapping semantic vector based on the mapping parameter distribution of the i-th time step and the fused semantic vector of the i-th time step includes: Determining a mapping parameter distribution for the i-th time step, wherein, when the i-th time step is the first time step among the multiple time steps, each parameter in the mapping parameter distribution for the i-th time step is 1; and when the i-th time step is a time step other than the first time step among the multiple time steps, the mapping parameter distribution for the i-th time step is determined based on the mapping parameter distribution for the previous time step; A bitwise multiplication operation is performed on the mapping parameter distribution of the i-th time step and the fusion semantic vector of the i-th time step to form a mapping semantic vector.

6. The method for monitoring operational faults of an HVAC system based on data analysis according to any one of claims 1 to 5, wherein: The step of performing semantic enhancement on the device vibration time series data based on the harmonic semantic information in the device vibration time series data to form a device vibration semantic vector includes: Performing a time-frequency transformation on the device vibration time series data to form a device vibration spectrum diagram, and determining the harmonic frequency corresponding to the device vibration time series data based on the device vibration spectrum diagram; Based on the harmonic frequency, filtering the device vibration time series data to form device vibration harmonic data, and performing time-frequency transformation on the device vibration harmonic data to form a harmonic vibration spectrum diagram; Perform convolution mining on the device vibration spectrum graph and the harmonic vibration spectrum graph respectively to form a device vibration spectrum vector and a harmonic vibration spectrum vector; Based on the harmonic vibration spectrum vector, the device vibration spectrum vector is mined for significant features to form a device vibration semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic vibration spectrum vector.

7. The method for monitoring operational faults of an HVAC system based on data analysis according to any one of claims 1 to 5, wherein: The step of semantically enhancing the device sound time series data based on the surge semantic information in the device sound time series data to form a device sound semantic vector includes: Filtering the device sound time series data based on the surge frequency to form device sound filter data, performing time-frequency transformation on the device sound filter data to form a harmonic sound spectrogram, and performing time-frequency transformation on the device sound time series data to form a device sound spectrogram; Performing convolution mining on the harmonic sound spectrum graph and the device sound spectrum graph respectively to form a harmonic sound spectrum vector and a device sound spectrum vector; Based on the harmonic sound spectrum vector, the device sound spectrum vector is mined for significant features to form a device sound semantic vector, wherein the significant feature mining includes at least one of gated adjustment and cross-attention processing, and the gating parameters corresponding to the gated adjustment are obtained based on the mapping of the harmonic sound spectrum vector.

8. An electronic device, characterized in that: The system comprises a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the method for monitoring operation faults of an HVAC system based on data analysis according to any one of claims 1 to 7.

Images