Fault diagnosis method and device for air conditioner outdoor unit module and air conditioner

Abstract

The invention relates to the technical field of fault diagnosis, and discloses a fault diagnosis method and device for an air conditioner outdoor unit module and an air conditioner, and the method comprises the steps that air conditioner operation data and after-sales work order data are collected, a multi-modal training set is constructed, and the after-sales work order data comprise a maintenance timestamp, fault phenomenon description and actual replacement part information; the multi-modal training set is input into a lightweight Transform model for model training, and a fault diagnosis model is obtained; under the condition that the air conditioner triggers the fault code of the outdoor unit module, inputting the operation data before the fault into the fault diagnosis model so as to output the part fault probability; and according to the component fault probability, recommending maintenance personnel to carry corresponding spare parts for door-to-door maintenance. According to the scheme, maintenance personnel can more accurately pre-judge fault components, repeated maintenance caused by spare part missing is reduced, the one-time maintenance success rate is remarkably increased, the after-sales cost is reduced, and the user experience is improved.

Description

Technical Field

[0001] The present application relates to the technical field of air conditioner fault diagnosis, for example, to a fault diagnosis method and device for an air conditioner outdoor unit module and an air conditioner. Background Art

[0002] With the advancement of air conditioning technology, fault diagnosis methods for air conditioner outdoor unit modules have evolved from simple to complex. Early repairs relied primarily on manual labor. Relying on experience and simple testing tools, maintenance personnel would inspect each outdoor unit module individually to determine the cause of the fault. This approach was adequate in the early days of relatively simple air conditioning technology and a relatively limited number of fault types. However, with the continuous advancement of air conditioning technology, the structure and operating principles of outdoor unit modules have become increasingly complex, and fault types have become more diverse. Traditional manual repair methods have gradually exposed numerous problems and are no longer able to meet the needs of modern air conditioning fault diagnosis.

[0003] In response to issues that have arisen during the development of fault diagnosis for air conditioner outdoor unit modules, a related technology provides an air conditioner fault monitoring method based on real-time monitoring data. This method incorporates multiple monitoring modules into the air conditioner to collect various monitoring data during its operation, such as key parameters like current, voltage, and temperature, in real time. When anomalies are detected in the data, it is possible to promptly determine whether these abnormal data will cause an air conditioner failure and control whether the air conditioner should be shut down based on the severity of the failure. This solution advances the fault monitoring process to before a fault occurs. Compared to traditional manual repair methods, it improves the timeliness and accuracy of fault monitoring to a certain extent, and reduces the risk of equipment damage caused by undetected faults.

[0004] During the implementation of the embodiments of the present disclosure, it was found that at least the following problems exist in the related art: While related technologies have addressed some of the issues with early fault detection, significant shortcomings remain. Their data dimensions are relatively limited, relying primarily on real-time monitoring parameters during air conditioner operation. Furthermore, their diagnostic logic remains relatively simple, resulting in insufficient algorithm generalization. Accurate adaptation and precise diagnosis of faults in outdoor unit modules of different brands, models, and operating environments are difficult, resulting in a low accuracy rate for maintenance personnel in bringing spare parts.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to ordinary technicians in this field. Summary of the Invention

[0006] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0007] The embodiments of the present disclosure provide a fault diagnosis method and device for an air conditioner outdoor unit module, and an air conditioner, which can effectively improve the accuracy of maintenance personnel in carrying spare parts.

[0008] In some embodiments, the fault diagnosis method for the air conditioner outdoor unit module includes: collecting air conditioner operation data and after-sales work order data to construct a multimodal training set, wherein the after-sales work order data includes a maintenance timestamp, a description of the fault phenomenon, and actual replacement component information; inputting the multimodal training set into a lightweight Transformer model for model training to obtain a fault diagnosis model; when the air conditioner triggers a fault code of the outdoor unit module, inputting the operation data before the fault into the fault diagnosis model to output the component failure probability; based on the component failure probability, recommending maintenance personnel to bring corresponding spare parts for on-site repair.

[0009] In some embodiments, the fault diagnosis device for the air conditioner outdoor unit module includes: a construction module, configured to collect air conditioner operation data and after-sales work order data, and construct a multimodal training set, wherein the after-sales work order data includes a maintenance timestamp, a description of the fault phenomenon, and actual replacement component information; a training module, configured to input the multimodal training set into a lightweight Transformer model for model training to obtain a fault diagnosis model; a diagnosis module, configured to input the operation data before the fault into the fault diagnosis model when the air conditioner triggers the outdoor unit module fault code, so as to output the component failure probability; a recommendation module, configured to recommend maintenance personnel to bring corresponding spare parts for on-site repair based on the component failure probability.

[0010] In some embodiments, the fault diagnosis device for an air conditioner outdoor unit module includes: a processor and a memory storing program instructions, and the processor is configured to execute the aforementioned fault diagnosis method for an air conditioner outdoor unit module when running the program instructions.

[0011] In some embodiments, the air conditioner includes: an air conditioner outdoor unit; and the aforementioned fault diagnosis device for the air conditioner outdoor unit module, which is installed in the air conditioner outdoor unit.

[0012] The fault diagnosis method and device for an air conditioner outdoor unit module and the air conditioner provided by the embodiments of the present disclosure can achieve the following technical effects: This solution constructs a multimodal training set by fusing air conditioner operation data with after-sales work order data. It uses a lightweight Transformer model for fault diagnosis, effectively capturing deep-level features and complex dependencies within the data. Compared to traditional manual repair methods, this not only improves the efficiency and accuracy of fault diagnosis, but also significantly enhances the model's generalization capabilities through multi-source data fusion and deep learning technology. This enables maintenance personnel to more accurately predict faulty components, reduce repeated repairs due to missing spare parts, significantly improve the first-time repair success rate, reduce after-sales costs, and enhance the user experience.

[0013] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] One or more embodiments are exemplarily described by corresponding drawings. These exemplary descriptions and drawings do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a scale limitation. In addition, Figure 1 is a schematic diagram of a fault diagnosis method for an air conditioner outdoor unit module provided by an embodiment of the present disclosure; Figure 2 is a schematic diagram of a method for obtaining a fault diagnosis model provided by an embodiment of the present disclosure; Figure 3 is a schematic diagram of a method for obtaining the failure probability of each component provided by an embodiment of the present disclosure; Figure 4 is a schematic diagram of another fault diagnosis method for an air conditioner outdoor unit module provided by an embodiment of the present disclosure; Figure 5 is a schematic diagram of a fault diagnosis device for an air conditioner outdoor unit module provided by an embodiment of the present disclosure; Figure 6 Schematic diagram of another fault diagnosis device for an air conditioner outdoor unit module provided by an embodiment of the present disclosure.

[0015] Reference numerals: 30: processor; 31: memory; 32: communication interface; 33: bus; 200: fault diagnosis device for air-conditioning outdoor unit module; 300: fault diagnosis device for air-conditioning outdoor unit module; 51: construction module; 52: training module; 53: diagnosis module; 54: recommendation module. DETAILED DESCRIPTION

[0016] In order to be able to understand the features and technical content of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The accompanying drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the sake of convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0017] In the description and claims of the embodiments of the present disclosure, as well as in the accompanying drawings, the terms "first," "second," and the like are used to distinguish similar items and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate to describe the embodiments of the present disclosure herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions.

[0018] Unless otherwise stated, the term "plurality" means two or more.

[0019] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0020] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0021] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0022] Combine Figure 1 As shown, optionally, an embodiment of the present disclosure discloses a fault diagnosis method for an air conditioner outdoor unit module, comprising: S11, the air conditioner collects air conditioner operation data and after-sales work order data to build a multimodal training set, where the after-sales work order data includes maintenance timestamp, fault phenomenon description and actual replacement parts information.

[0023] In S12, the air conditioner inputs the multimodal training set into the lightweight Transformer model for model training to obtain a fault diagnosis model.

[0024] S13, when the air conditioner triggers a fault code of the outdoor unit module, the air conditioner inputs the air conditioner operation data before the fault into the fault diagnosis model to output the component failure probability.

[0025] S14: Based on the probability of component failure, the air conditioner recommends that maintenance personnel bring the corresponding spare parts for on-site repair.

[0026] In this solution, the air conditioner can collect air conditioner operation data and after-sales work order data. Among them, the air conditioner operation data includes: indoor and outdoor ambient temperature, compressor frequency, coil temperature, electronic expansion valve opening, fan speed, current and voltage and other key parameters. It can be understood that the air conditioner operation data can fully reflect the status of the air conditioner outdoor unit module during operation. Specifically, the air conditioner operation data can be collected at a certain sampling frequency f so as to capture the real-time operating condition changes during the operation of the air conditioner. In one example, the setting of the sampling frequency f can be determined based on the change rate of the air conditioner operation data and the real-time requirements of fault diagnosis to ensure that key operating status changes can be captured while avoiding data redundancy and waste of computing resources.

[0027] In this solution, after-sales work order data can be obtained from the after-sales service module of the ERP (Enterprise Resource Planning) system. This work order data includes the repair order number, repair timestamp, a text description of the fault phenomenon (such as "compressor does not start," "external fan makes abnormal noise") written by the maintenance engineer on-site, and information about the actual replaced parts. This information can include the actual replaced part code, model, and specifications. The work order data can also include verification results of the device status after the repair. This way, after collecting air conditioner operating data and after-sales work order data, a multimodal training set can be constructed based on these data. This multimodal training set allows the model to simultaneously learn the air conditioner's operating status and historical maintenance experience, thereby gaining a more comprehensive understanding of fault modes and improving the accuracy and generalization of fault diagnosis.

[0028] The air conditioner first inputs the multimodal training set into a lightweight Transformer model. This model, with its unique architecture, can process complex time series data and output the probability of each component failure. These failure probabilities serve as the model's initial diagnostic results for the input data and provide a foundation for subsequent model optimization. Furthermore, the air conditioner calculates the model's loss value based on the difference between the failure probabilities output by the lightweight Transformer model and the actual fault labels. Loss is an important metric for measuring the model's prediction accuracy and is calculated using a loss function. For example, the loss function can be the cross-entropy loss function. Based on the calculated loss value, the air conditioner uses the backpropagation algorithm to update the model's parameters. This process allows the model to gradually improve its prediction accuracy by continuously adjusting the parameters. The iterative training process continues until preset conditions are met, such as when the loss value reaches a preset convergence threshold or when a set number of iterations has been reached. When these conditions are met, the training process terminates, and the resulting model is a fully trained fault diagnosis model suitable for practical fault diagnosis tasks.

[0029] In another approach, the air conditioner can also use other methods to input the multimodal training set into the lightweight Transformer model for model training. For example, a learning rate decay mechanism is introduced during the training process, and the learning rate is gradually reduced as the training progresses, so that the model can adjust the parameters more finely in the later stages of training, thereby improving the stability and accuracy of the model. In addition, an early stopping mechanism can be used to monitor the performance of the model on the validation set. If the model performance no longer improves after a certain number of iterations, the training is terminated in advance to prevent overfitting and ensure that the model has good generalization capabilities. With this solution, by inputting the multimodal training set into the lightweight Transformer model for training, the model has achieved significant improvements in fault diagnosis accuracy.

[0030] Furthermore, when the air conditioner triggers the outdoor unit module fault code (F1), the air conditioner automatically initiates the fault diagnosis process. Specifically, the air conditioner automatically extracts the air conditioner's operating data for the period t minutes before the fault occurred. This operating data includes key parameters such as indoor and outdoor ambient temperature, compressor frequency, coil temperature, electronic expansion valve opening, fan speed, and current and voltage. This operating data comprehensively reflects the operating status of the air conditioner's outdoor unit module before the fault occurred. The extracted data is then input into a pre-trained fault diagnosis model. This model, based on a lightweight Transformer architecture, can efficiently process time series data and output the probability of each component failure. The model output is a list of component failure probabilities. These probability values ​​help maintenance personnel quickly locate possible faulty components.

[0031] In an optimized solution, to conserve internal air conditioner storage space and improve diagnostic efficiency, the trained fault diagnosis model can be deployed to a cloud server. In this scenario, when a fault code is triggered in the outdoor unit module, the air conditioner automatically extracts operational data from the period t minutes prior to the fault and sends it to the cloud server. The inference engine on the cloud server receives this data and uses the fault diagnosis model deployed there to perform a real-time diagnosis. The diagnostic results include not only the top three components with the highest probability of failure, but also the confidence level for each component's failure probability. After manual review, this information is pushed to the after-sales service system, allowing maintenance personnel to bring the appropriate spare parts for on-site repairs based on this precise diagnostic information. This solution, by deploying the fault diagnosis model to the cloud server, not only conserves internal air conditioner storage space but also improves diagnostic accuracy and efficiency. Compared to existing technologies, this cloud-based deployment approach can more quickly process large amounts of data and provide more accurate fault diagnosis results, significantly improving repair efficiency, reducing repair time and costs, and enhancing the user experience.

[0032] Furthermore, after the air conditioner obtains the component failure probability, it can recommend that maintenance personnel bring appropriate spare parts for on-site repairs based on the component failure probability. Specifically, after the fault diagnosis model outputs the failure probability of each component, the air conditioner ranks the faulty components based on these probabilities. In one example, the fault diagnosis model outputs a list of component failure probabilities, such as a 70% probability for the compressor failure, a 20% probability for the external fan failure, and a 10% probability for the computer board failure. Based on these probabilities, the air conditioner generates a detailed repair recommendation list, recommending that maintenance personnel prioritize bringing the components with the highest failure probability. Upon receiving the repair order, maintenance personnel will prepare the appropriate spare parts based on the repair recommendation list provided by the air conditioner. For example, if the compressor has the highest probability of failure, the maintenance personnel will prioritize bringing compressor spare parts; if the external fan has the second highest probability of failure, the maintenance personnel will also bring external fan spare parts, and so on. This spare part recommendation mechanism based on failure probability ensures that maintenance personnel have the most likely replacement components upon arrival at the repair site, reducing repeated repairs due to insufficient spare parts and significantly reducing after-sales costs. At the same time, accurate spare parts recommendations enable maintenance personnel to complete repair tasks more quickly, improve the first-time repair success rate, and enhance the user experience. Furthermore, because they can quickly prepare the required spare parts based on a clear repair recommendation list, maintenance personnel can also reduce preparation time, further improving the overall efficiency of maintenance work.

[0033] In one optimized solution, the fault diagnosis model outputs both the probability of failure for each component and the corresponding confidence level. This allows the air conditioner to more accurately recommend that maintenance personnel bring the appropriate spare parts for on-site repairs based on the component failure probability and confidence level. In one example, the air conditioner ranks faulty components based on the failure probability output by the model and evaluates each component's failure prediction based on the confidence level. For example, the fault diagnosis model might output the following: a 70% probability of compressor failure with a confidence level of 0.9; a 20% probability of outdoor fan failure with a confidence level of 0.8; and a 10% probability of computer board failure with a confidence level of 0.7. In this case, the air conditioner comprehensively considers the failure probabilities and confidence levels to generate a list of recommended repairs. If a component has both a high probability of failure and a high confidence level, the air conditioner prioritizes recommending that the maintenance personnel bring the spare part. For example, if the probability of compressor failure is 70% and the confidence level is 0.9, this indicates that the fault diagnosis model's prediction of the compressor failure is highly reliable, and the air conditioner will therefore recommend that the maintenance personnel bring the compressor spare part.

[0034] In one optimized solution, the air conditioner can also set a confidence threshold, such as 0.8. Only when the fault diagnosis model's predicted confidence for a particular component failure exceeds this threshold will the air conditioner recommend that the maintenance personnel bring a spare part for that component. This further improves the reliability of maintenance recommendations and avoids spare part retrieval errors caused by inaccurate model predictions. For example, if the probability of an outdoor fan failure is 20%, but the confidence is 0.8, exceeding the set threshold, the air conditioner will still recommend that the maintenance personnel bring a spare part. Conversely, if the probability of a computer board failure is 10%, but the confidence is only 0.7, below the set threshold, the air conditioner will recommend that the maintenance personnel temporarily avoid bringing a spare part and make a decision based on on-site conditions. This solution, by combining the component failure probability and confidence output by the fault diagnosis model to make maintenance recommendations, not only improves the accuracy and reliability of maintenance recommendations but also further optimizes the allocation of maintenance resources. Compared to solutions that rely solely on failure probability, this approach, which incorporates confidence, can more accurately predict faulty components and reduce spare part retrieval errors caused by uncertainty in the fault diagnosis model's predictions. This not only improves maintenance efficiency and reduces after-sales costs, but also enhances the work efficiency of maintenance personnel and the user experience. Furthermore, by setting confidence thresholds, the air conditioner can better balance prediction accuracy and the use of maintenance resources. This ensures that, when resources are limited, maintenance personnel can bring the spare parts most likely to be replaced, further improving the first-time repair success rate.

[0035] The fault diagnosis method for the air-conditioning outdoor unit module provided by the embodiment of the present disclosure is adopted to construct a multimodal training set by fusing the air-conditioning operation data and the after-sales work order data. The use of a lightweight Transformer model for fault diagnosis can effectively capture the deep-level features and complex dependencies in the data. Compared with the traditional manual maintenance method, it not only improves the efficiency and accuracy of fault diagnosis, but also significantly enhances the generalization ability of the model through multi-source data fusion and deep learning technology. In this way, maintenance personnel can more accurately predict faulty parts, reduce repeated repairs due to missing spare parts, significantly improve the success rate of one-time repairs, reduce after-sales costs, and improve user experience.

[0036] Combine Figure 2 As shown, optionally, in S12, the air conditioner inputs the multimodal training set into the lightweight Transformer model for model training to obtain a fault diagnosis model, including: In S21, the air conditioner inputs the multimodal training set into the lightweight Transformer model to obtain the failure probability of each component output by the lightweight Transformer model.

[0037] S22, the air conditioner calculates the model loss value according to the output failure probability of each component.

[0038] S23, the air conditioner updates the model parameters based on the model loss value until the iteration is terminated when the preset conditions are met.

[0039] S24, when the iteration is terminated, the air conditioner obtains a trained fault diagnosis model.

[0040] Among them, the preset conditions include that the model loss value meets the preset convergence conditions or the current number of iterations reaches the set number of iterations.

[0041] In this solution, the air conditioner can input the collected multimodal training set into a lightweight Transformer model. This unique architecture allows it to process complex time series data and output the failure probability of each component. These failure probabilities serve as the model's initial diagnostic results for the input data and provide a foundation for subsequent model optimization.

[0042] Furthermore, the air conditioner can calculate the model's loss value based on the difference between the failure probability of each component output by the model and the actual fault labels. Fault labels refer to real-world fault conditions that are manually annotated for supervised learning and serve as a benchmark for calculating the loss value during model training. The loss value can be calculated using a loss function. Based on the calculated model loss value, the air conditioner can update the model parameters using a backpropagation algorithm. By continuously adjusting the parameters, the fault diagnosis model can gradually improve its prediction accuracy. The iterative training process continues until preset conditions are met, such as when the loss value reaches a preset convergence threshold or when a set number of iterations has been reached. When these conditions are met, the training process terminates, and the resulting model is a fully trained fault diagnosis model, suitable for actual fault diagnosis tasks.

[0043] This approach, by feeding a multimodal training set into a lightweight Transformer model for training, can fully leverage the rich information contained in air conditioner operating data and after-sales work order data, improving the model's understanding and identification of fault modes. Compared to traditional methods, it can not only process complex time series data but also fuse multi-source data to provide more comprehensive fault diagnosis results. By accurately calculating loss values ​​and updating model parameters accordingly, the model's accuracy and generalization capabilities are significantly improved, enabling more precise prediction of faulty components and reducing false positives and missed detections. This not only improves the efficiency and accuracy of fault diagnosis, but also reduces repair time and costs, and enhances the user experience.

[0044] Understandably, for rare fault samples with a frequency of less than 5%, the air conditioner can also adopt a meta-learning optimization strategy for special processing: first, a small sample task set containing K historical cases of this fault type is constructed. Then, by calculating the meta-gradient between this task set and the base model on the current batch of data, the model parameters are adjusted in a targeted manner, with the adjustment range controlled by the meta-learning rate. This process significantly improves the ability to capture sparse fault patterns while maintaining the main structure of the base model. In this way, through the meta-learning mechanism, a high recognition accuracy can be maintained for rare fault types with insufficient data.

[0045] Combine Figure 3 As shown, optionally, in S21, the air conditioner inputs the multimodal training set into the lightweight Transformer model to obtain the failure probability of each component output by the lightweight Transformer model, including: S31, the air conditioner inputs the multimodal training set into the embedding layer of the lightweight Transformer model for feature mapping and position encoding to obtain a multidimensional time series feature vector.

[0046] S32, the air conditioner inputs the multidimensional time series feature vector into the encoder of the lightweight Transformer model for feature extraction to obtain a feature representation of the encoder output of the lightweight Transformer model.

[0047] S33, air conditioning performs global averaging compression on the feature representation output by the encoder of the lightweight Transformer model to obtain a feature vector of fixed length.

[0048] S34, the air conditioner inputs the fixed-length feature vector into the three-dimensional classification head for classification, so as to obtain the failure probability of each component output by the three-dimensional classification head.

[0049] In this solution, the air conditioner first preprocesses the collected multimodal training set. Preprocessing steps include, but are not limited to, data cleaning, normalization, and feature extraction. These steps remove noise and outliers from the data, convert data of different dimensions to the same scale, and extract key features to ensure data quality and consistency. After preprocessing, the air conditioner can input the processed data into the embedding layer of the lightweight Transformer model. The main task of the embedding layer of the lightweight Transformer model is to map the input data into a high-dimensional vector space through a linear transformation, generating an embedding vector. This process allows the data in this new space to better reflect its inherent characteristics and patterns, providing richer information for subsequent feature extraction and model training. Furthermore, to preserve the temporal order of the data, the air conditioner adds a positional encoding to the embedding vector of each time step. It should be noted that a positional encoding is a special vector that represents the position of the time step. By adding the positional encoding to the embedding vector, the model can capture the temporal order information in the time series data. Furthermore, the combination of the embedding vector and the positional encoding generates a multidimensional time series feature vector containing temporal information, which serves as input for subsequent model processing. This approach effectively converts multimodal data into multidimensional time series feature vectors suitable for model processing. This conversion not only preserves the key features of the original data but also enhances the model's understanding of time series data, thereby improving the accuracy and efficiency of fault diagnosis.

[0050] Furthermore, the air conditioner inputs the multi-dimensional time series feature vector, after preprocessing and processing by the embedding layer of the lightweight Transformer model, into the encoder of the lightweight Transformer model for feature extraction. The encoder of the lightweight Transformer model has a multi-layer structure. Each encoder layer contains a multi-head attention mechanism and a feedforward network. These two components work together to extract complex features and dependencies in the data. Specifically, the multi-head attention mechanism can simultaneously focus on different parts of the input sequence. By calculating linear transformations of the query, key, and value, it projects the input data into multiple low-dimensional subspaces and independently calculates an attention score in each subspace. These scores are then weighted and summed to produce the multi-head attention output, thereby capturing long-range dependencies in the data. This process allows the model to process information in different subspaces in parallel, improving its ability to capture complex patterns. Furthermore, the obtained attention output is passed to the feedforward network. In this embodiment, the feedforward network is a two-layer fully connected neural network with a ReLU activation function. The feedforward network performs nonlinear transformations on the attention output to further extract features. This process enhances the model's ability to represent input data, enabling it to learn more complex feature representations. To maintain model stability and training efficiency, the output of the feedforward network is residually connected to the input. This involves adding the output and input of the feedforward network, followed by layer normalization. Residual connections help address the vanishing gradient problem in deep networks, while layer normalization ensures consistent data distribution across different feature dimensions, thereby improving model training stability and convergence speed. After completing the above operations in each encoder layer, the output serves as the input for the next encoder layer. Through this layer-by-layer transfer and processing, the model is able to gradually extract more complex feature representations. As the number of layers increases, the model's understanding of the data deepens, ultimately yielding feature representations for the encoder output. These feature representations capture deep features and complex dependencies in the data, providing rich information for subsequent fault diagnosis.

[0051] This approach, through the feature extraction process of the encoder of the lightweight Transformer model, can effectively process complex time series data and gradually extract deeper feature representations. Compared with traditional feature extraction methods, this Transformer-based architecture not only captures long-range dependencies in the data, but also enhances the model's ability to learn complex patterns through the combination of a multi-head attention mechanism and a feedforward network. In addition, residual connections and layer normalization further improve the model's training stability and convergence speed, enabling the model to more efficiently learn useful feature representations. Ultimately, these feature representations provide a solid foundation for fault diagnosis, significantly improving its accuracy and efficiency.

[0052] In this embodiment, the feature representation after processing by the encoder of the lightweight Transformer model is a high-dimensional time series feature matrix. Each row of this matrix represents a feature vector for a time step, and each column represents a feature dimension. To convert this high-dimensional time series feature matrix into a fixed-length feature vector, the air conditioner performs a global average pooling operation on the feature representation output by the encoder. Specifically, the global average pooling operation averages all time steps in each feature dimension to obtain a fixed-length feature vector. This fixed-length feature vector retains the average information of the original time series data in each feature dimension, effectively capturing the overall characteristics of the data while reducing the feature dimensionality, making subsequent processing more efficient. This solution can compress the high-dimensional time series feature matrix into a fixed-length feature vector. This process not only retains the key features of the data, but also reduces the feature dimensionality, improving the computational efficiency of the model. Compared with directly using the high-dimensional time series feature matrix, the global average pooling operation can more effectively capture the overall characteristics of the data while reducing feature redundancy, enabling the model to perform better in subsequent classification or other tasks. In addition, the global average pooling operation also has a certain ability to resist noise and can improve the robustness of the model.

[0053] As an alternative, in addition to the global average pooling operation, other alternatives can be considered to obtain fixed-length feature vectors. For example, the maximum pooling operation can be used. Specifically, the maximum pooling operation selects the maximum value along each feature dimension, thereby preserving significant features in the data. Another option is to use the adaptive pooling operation, which adaptively adjusts the size of the pooling window based on the preset target dimension to obtain a fixed-length feature vector. These alternatives can be selected based on the specific application scenario and data characteristics to further optimize the performance of the fault diagnosis model.

[0054] Furthermore, the air conditioner can input the fixed-length feature vector obtained through global average pooling into a three-dimensional classification head for classification. Here, the three-dimensional classification head is a simple fully connected layer whose primary function is to map the fixed-length feature vector into a fault probability space. Specifically, the fully connected layer uses a linear transformation to convert the input feature vector into a vector with the same number of fault categories, where each element represents the raw probability of a component failure. These raw probabilities are then processed through a Softmax activation function, which converts each raw probability to a value between 0 and 1 and ensures that the sum of all component failure probabilities is 1. Ultimately, the model outputs a probability distribution representing the probability of each component failure. For example, the probability of compressor failure is 70%, the probability of external fan failure is 20%, and the probability of computer board failure is 10%. This scheme effectively converts the feature vector into a probability distribution of component failure. The use of the Softmax activation function ensures that the output probability distribution has good mathematical properties, making the model output more intuitive and easy to interpret. Compared to directly outputting raw probabilities, the probability distribution after Softmax processing more accurately reflects the probability of each component failure, thereby improving the accuracy and reliability of fault diagnosis. In addition, this classification method has good generalization ability and can adapt to the fault diagnosis needs of air-conditioning outdoor unit modules of different brands, models and usage environments.

[0055] In one optimized solution, the air conditioner also includes a causal graph inference module. Specifically, this module performs in-depth analysis of the feature representations output by the encoder of the lightweight Transformer model. In one example, a causal directed acyclic graph (DAG) between variables is first learned from historical fault data. For example, a causal chain of "capacitor damage → current anomaly → compressor shutdown" is established. The calculated causal weight matrix (e.g., the causal strength of capacitor damage on current anomaly is 0.92) is then weighted and fused with the original Transformer attention score. A gating mechanism is used to adjust the ratio of the two. When the causal confidence exceeds a threshold (e.g., >0.8), the corresponding attention score is increased by 30% to 50%, thereby strengthening the weight of the root cause feature. In this way, in the three-dimensional classification head, the causally corrected fixed-length feature vector is concatenated with the causal weight vector. A causal constraint loss function is added to the fully connected layer to ensure that the classification results are consistent with the root cause derived from the causal graph. Compared with existing technologies that rely solely on statistical correlation for diagnosis, this solution can effectively distinguish between fundamental faults (such as capacitor damage) and derivative phenomena (such as current fluctuations) by explicitly modeling causal relationships, effectively reducing the misjudgment rate in complex concurrent fault scenarios. At the same time, the output diagnostic results are explainable (such as showing the causal chain of "capacitor damage leading to current abnormality"), allowing maintenance personnel to quickly locate the root cause of the problem rather than the superficial symptoms.

[0056] Optionally, in S31, the air conditioner inputs the multimodal training set into the embedding layer of the lightweight Transformer model for feature mapping and position encoding to obtain a multidimensional time series feature vector, including: Air conditioning preprocesses the air conditioning operation data and after-sales work order data in the multimodal training set.

[0057] The air conditioner inputs the preprocessed data into the embedding layer of the lightweight Transformer model, maps the input data to a high-dimensional vector space through linear transformation, and obtains an embedding vector.

[0058] Conditioning adds positional encoding to the embedding vector of each time step to obtain a multi-dimensional time series feature vector containing time information.

[0059] In this solution, the air conditioner preprocesses the air conditioner operation data and after-sales work order data in the multimodal training set. This includes data cleaning to remove noise and outliers to ensure data accuracy and integrity. The data is normalized to convert data of different dimensions to the same scale for easier model processing. Feature extraction is also performed to extract key features from the raw data to enhance the model's understanding of the data. In a specific embodiment, data cleaning includes filling missing values ​​in the air conditioner operation data through interpolation or default values. Text descriptions in the after-sales work order data are removed by removing irrelevant characters and formatted text. Data normalization includes normalizing temperature data to a range of 0 to 1 and normalizing current and voltage data to the same dimension. Feature extraction also includes extracting features such as the rate of change of compressor frequency and peak module temperature to enhance the model's understanding of the data. This solution significantly improves data quality and consistency, providing a solid foundation for subsequent model training. Data cleaning ensures data accuracy and integrity and reduces the interference of noise and outliers on model training. Data normalization enables data of different dimensions to be compared and processed on the same scale, improving model training efficiency and stability. Feature extraction further enhances the model's understanding of the data, enabling it to more effectively learn key features and patterns within the data. These preprocessing steps work together to significantly improve model training and fault diagnosis accuracy, providing a crucial foundation for efficient and accurate fault diagnosis.

[0060] Furthermore, the air conditioner can input the preprocessed data into the embedding layer of the lightweight Transformer model. The main task of the embedding layer of the lightweight Transformer model is to map the raw input data into a high-dimensional vector space to extract the data's feature representation. Specifically, the embedding layer of the lightweight Transformer model converts the input m-dimensional features into n-dimensional embedding vectors through a linear transformation. This process not only transforms the data into a new space but also better reveals the data's inherent characteristics and patterns within this space. For example, through linear transformation, features that were originally difficult to distinguish in a low-dimensional space may become clearer in the high-dimensional space, providing richer information for subsequent feature extraction and model training. This approach significantly enhances the model's ability to understand the input data through the embedding layer of the lightweight Transformer model. By mapping the data into a high-dimensional vector space, the model can more effectively capture the complex patterns and feature relationships within the data. This high-dimensional representation makes the model more adept at handling complex fault diagnosis tasks, enabling more accurate fault mode identification, thereby improving the accuracy and efficiency of fault diagnosis. In addition, the linear transformation of the embedding layer of the lightweight Transformer model also provides a flexible way for the model to adjust the representation of features, so that the model can better adapt to different data characteristics and task requirements.

[0061] Furthermore, after the embedding layer of the lightweight Transformer model is processed, the air conditioner adds a position code to the embedding vector of each time step to obtain a multi-dimensional time series feature vector containing time information. The calculation method of the position code is: ;and, ; Among them, pos represents the position of the time step, i is the dimension index of the embedding vector, and d model is the dimension of the embedding vector. In this embodiment, the dimension of the embedding vector is 64. This allows the model to capture the temporal order information of the data, enabling the model to better understand the feature changes in the time dimension when processing time series data.

[0062] Optionally, in S32, the air conditioner inputs the multidimensional time series feature vector into an encoder of a lightweight Transformer model for feature extraction to obtain a feature representation of the encoder output, including: The air conditioner inputs the multi-dimensional time series feature vector into the encoder of the lightweight Transformer model. The encoder of the lightweight Transformer model includes multiple layers, and each layer of the encoder includes a multi-head attention mechanism and a feedforward network.

[0063] In each encoder layer, the air conditioner calculates the attention output using a multi-head attention mechanism and passes the attention output to the feedforward network to extract features through nonlinear transformation.

[0064] The air conditioner performs residual connection and layer normalization on the output of the feedforward network to obtain the output of the encoder of this layer.

[0065] The air conditioner takes the output of the previous layer encoder as the input of the next layer encoder, and obtains the feature representation of the encoder output by passing and processing layer by layer.

[0066] In this scheme, the air conditioner inputs the multi-dimensional time series feature vector processed by the embedding layer of the lightweight Transformer model into the encoder of the lightweight Transformer model for feature extraction. Here, the encoder of the lightweight Transformer model includes multiple layers, and each layer of the encoder contains a multi-head attention mechanism and a feedforward network. These two components work together to extract complex features and dependencies in the data. Specifically, the multi-head attention mechanism enables the model to focus on different parts of the input sequence in different subspaces, thereby more comprehensively capturing the dependencies and feature information in the data. It projects the input query (Query), key (Key), and value (Value) into multiple low-dimensional subspaces through linear transformation, calculates the attention score in each subspace, and finally obtains the output by weighted summation of these scores. This process can be expressed as:

[0067] Among them, d k is the dimension of the key vector.

[0068] Optionally, the Softmax algorithm formula is:

[0069] Among them, K is the dimension of the input variable, and each element in the input variable will be calculated according to the formula.

[0070] Furthermore, the output of the multi-head attention layer is passed to the feedforward network. In this embodiment, the feedforward network is a simple two-layer fully connected neural network that includes a ReLU activation function. The activation function can be expressed as follows:

[0071] Furthermore, the output of the feedforward network is residually connected to the input, adding the output and input of the feedforward network together. The result of the residual connection is layer-normalized to ensure that the data has a similar distribution across different feature dimensions. After layer-normalization, the output of the encoder for that layer is obtained. In this way, the air conditioner uses the output of the encoder of the previous layer as the input of the encoder of the next layer, passing and processing it layer by layer to obtain the feature representation of the encoder output. This solution, through the combination of the multi-head attention mechanism and the feedforward network, enhances the model's ability to learn complex patterns. In addition, the residual connection and layer-normalization process further improve the training stability and convergence speed of the model, allowing the model to more efficiently learn useful feature representations.

[0072] Optionally, the three-dimensional classification head includes a fully connected layer and a Softmax activation function; S34, the air conditioner inputs the fixed-length feature vector into the three-dimensional classification head for classification, to obtain the failure probability of each component output by the three-dimensional classification head, including: The air conditioner uses a fully connected layer to map the fixed-length feature vector to the fault probability space to obtain the original probability of failure of each component.

[0073] The air conditioner uses the Softmax activation function to convert the original probability into a normalized failure probability distribution to output the failure probability of each component.

[0074] In this solution, the air conditioner can input a fixed-length feature vector, obtained through global average pooling, into a three-dimensional classification head for classification. Here, the three-dimensional classification head is a simple fully connected layer whose primary function is to map the fixed-length feature vector into a fault probability space. Specifically, the fully connected layer uses a linear transformation to convert the input feature vector into a vector with the same number of fault categories, where each element represents the raw probability of a component failure. These raw probabilities are then processed through a Softmax activation function, which converts each raw probability to a value between 0 and 1 and ensures that the sum of all component failure probabilities is 1. Ultimately, the model outputs a probability distribution representing the probability of each component failure. For example, the probability of a compressor failure is 70%, the probability of an outdoor fan failure is 20%, and the probability of a computer board failure is 10%. This solution effectively converts the feature vector into a probability distribution of component failures. The use of the Softmax activation function ensures that the output probability distribution has good mathematical properties, making the model output more intuitive and easy to interpret. Compared to directly outputting raw probabilities, the probability distribution after Softmax processing more accurately reflects the probability of each component failure, thereby improving the accuracy and reliability of fault diagnosis. In addition, this classification method has good generalization ability and can adapt to the fault diagnosis needs of air-conditioning outdoor unit modules of different brands, models and usage environments.

[0075] Combine Figure 4 As shown, optionally, after recommending that a maintenance person bring corresponding spare parts for on-site repair, the method further includes: S41: When receiving maintenance feedback information, the air conditioner continuously updates the multimodal training set according to the maintenance feedback information.

[0076] S42, the air conditioner re-inputs the updated multimodal training set into the lightweight Transformer model for model training to improve the accuracy of the model in fault diagnosis.

[0077] In this solution, to further enhance the model's accuracy and adaptability, after recommending a maintenance technician with the appropriate spare parts for on-site repair, the air conditioner continuously optimizes the model based on repair feedback. Specifically, after the maintenance technician completes the repair and provides feedback, the air conditioner receives feedback containing the actual faulty component, repair measures, and repair results. This information is used to update the multimodal training set, adding the new repair case to the existing training data. Furthermore, the air conditioner re-inputs the updated multimodal training set into the lightweight Transformer model for model training. In this way, the model learns the latest failure modes and repair experience, continuously improving the accuracy of fault diagnosis. By continuously updating the training dataset and model parameters, this solution gradually improves the model's ability to recognize complex fault patterns, reduces false positives and missed detections, and ultimately improves repair efficiency, reduces after-sales costs, and enhances the user experience. Furthermore, this approach reduces reliance on large amounts of historical data, enabling the model to more quickly adapt to new equipment and environmental conditions, maintaining the reliability of its diagnostic capabilities.

[0078] Combine Figure 5 As shown, optionally, an embodiment of the present disclosure provides a fault diagnosis device 200 for an air conditioner outdoor unit module, comprising a construction module 51, a training module 52, a diagnosis module 53, and a recommendation module 54. The construction module 51 is configured to collect air conditioner operation data and after-sales work order data to construct a multimodal training set, wherein the after-sales work order data includes a maintenance timestamp, a description of the fault phenomenon, and information about actual replacement parts; the training module 52 is configured to input the multimodal training set into a lightweight Transformer model for model training to obtain a fault diagnosis model; the diagnosis module 53 is configured to input the operation data before the fault into the fault diagnosis model to output the probability of component failure when the air conditioner triggers a fault code of the outdoor unit module; the recommendation module 54 is configured to recommend maintenance personnel to carry corresponding spare parts for on-site repair based on the probability of component failure.

[0079] The fault diagnosis device 200 for an air conditioner outdoor unit module provided by the disclosed embodiments not only improves the efficiency and accuracy of fault diagnosis but also significantly enhances the model's generalization capabilities through multi-source data fusion and deep learning technology. This allows maintenance personnel to more accurately predict faulty components, reduce repeated repairs due to missing spare parts, significantly increase the first-time repair success rate, reduce after-sales costs, and enhance the user experience.

[0080] Combine Figure 6 As shown, an embodiment of the present disclosure provides a device 300 for an air conditioner outdoor unit module, comprising a processor 30 and a memory 31. Optionally, the device 300 may also include a communication interface 32 and a bus 33. The processor 30, the communication interface 32, and the memory 31 may communicate with each other via the bus 33. The communication interface 32 may be used for information transmission. The processor 30 may invoke logic instructions stored in the memory 31 to execute the fault diagnosis method for the air conditioner outdoor unit module of the above embodiment.

[0081] In addition, the logic instructions in the memory 31 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0082] Memory 31, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of the present disclosure. Processor 30 executes the program instructions / modules stored in memory 31 to perform functional applications and data processing, thereby implementing the fault diagnosis method for the air conditioner outdoor unit module in the above-mentioned embodiment.

[0083] The memory 31 may include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the terminal device. Furthermore, the memory 31 may include high-speed random access memory and non-volatile memory.

[0084] The embodiment of the present disclosure provides an air conditioner, comprising: an air conditioner outdoor unit, and the above-mentioned fault diagnosis device 200 (300) for the air conditioner outdoor unit module. The fault diagnosis device 200 (300) for the air conditioner outdoor unit module is installed in the air conditioner outdoor unit. The installation relationship described here is not limited to placement inside the air conditioner outdoor unit, but also includes installation connections with other components of the air conditioner, including but not limited to physical connections, electrical connections, or signal transmission connections. It can be understood by those skilled in the art that the fault diagnosis device 200 (300) for the air conditioner outdoor unit module can be adapted to a feasible air conditioner outdoor unit, thereby realizing other feasible embodiments.

[0085] An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned fault diagnosis method for an air conditioner outdoor unit module.

[0086] The technical solutions of the embodiments of the present disclosure may be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present disclosure. The aforementioned storage medium may be a non-transitory storage medium, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, among other media capable of storing program code.

[0087] The above description and the accompanying drawings sufficiently illustrate the embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless expressly required, individual components and functions are optional, and the order of operations may vary. Portions and features of some embodiments may be included in or replace portions and features of other embodiments. Moreover, the terms used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, the singular forms "a", "an" and "the" are intended to also include the plural forms unless the context clearly indicates otherwise. Similarly, the term "and / or" as used in this application means any and all possible combinations of one or more of the associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be referred to the description of the method part.

[0088] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0089] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices and equipment) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units may be merely a logical functional division. In actual implementation, other divisions may be used, such as combining or integrating multiple units or components into another system, or omitting or disabling some features. Furthermore, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interface, or indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to implement the embodiments according to actual needs. Furthermore, the functional units in the embodiments disclosed herein may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit.

[0090] The flowcharts and block diagrams in the accompanying drawings show the possible implementation architectures, functions and operations of the systems, methods and computer program products according to the embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical functions. In some alternative implementations, the functions marked in the box 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, or they can sometimes be executed in the opposite order, which can depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified function or action, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A fault diagnosis method for an air conditioner outdoor unit module, characterized in that: include: Collect air conditioner operation data and after-sales work order data to build a multimodal training set. The after-sales work order data includes maintenance timestamps, fault phenomenon descriptions, and actual replacement parts information. Inputting the multimodal training set into a lightweight Transformer model for model training to obtain a fault diagnosis model; In the event that the air conditioner triggers a fault code for the outdoor unit module, the air conditioner operation data before the fault is input into the fault diagnosis model to output the component failure probability; Based on the probability of component failure, it is recommended that maintenance personnel bring the corresponding spare parts for on-site repair.

2. The method according to claim 1, characterized in that The multimodal training set is input into the lightweight Transformer model for model training to obtain a fault diagnosis model, including: Inputting the multimodal training set into a lightweight Transformer model to obtain the failure probability of each component output by the lightweight Transformer model; Calculating a model loss value based on the output failure probability of each component; Update the model parameters based on the model loss value until the iteration is terminated when the preset conditions are met; When the iteration is terminated, the trained fault diagnosis model is obtained; Among them, the preset conditions include that the model loss value meets the preset convergence condition or the current number of iterations reaches the set number of iterations.

3. The method according to claim 2, characterized in that Inputting the multimodal training set into the lightweight Transformer model to obtain the failure probability of each component output by the lightweight Transformer model includes: Inputting the multimodal training set into the embedding layer of the lightweight Transformer model for feature mapping and position encoding to obtain a multidimensional time series feature vector; Inputting the multidimensional time series feature vector into an encoder of a lightweight Transformer model for feature extraction to obtain a feature representation of an encoder output of the lightweight Transformer model; Performing global average pooling compression on the feature representation output by the encoder of the lightweight Transformer model to obtain a feature vector of fixed length; The fixed-length feature vector is input into a three-dimensional classification head for classification, so as to obtain the failure probability of each component output by the three-dimensional classification head.

4. The method according to claim 3, characterized in that The multimodal training set is input into the embedding layer of the lightweight Transformer model for feature mapping and position encoding to obtain a multidimensional time series feature vector, including: Preprocessing the air conditioning operation data and the after-sales work order data in the multimodal training set; The preprocessed data is input into the embedding layer of the lightweight Transformer model, and the input data is mapped to a high-dimensional vector space through linear transformation to obtain an embedding vector; Position encoding is added to the embedding vector of each time step to obtain a multi-dimensional time series feature vector containing time information.

5. The method according to claim 3, characterized in that Inputting the multidimensional time series feature vector into an encoder of a lightweight Transformer model for feature extraction to obtain a feature representation output by the encoder of the lightweight Transformer model includes: Inputting the multidimensional time series feature vector into an encoder of a lightweight Transformer model, wherein the encoder of the lightweight Transformer model includes multiple layers, and each layer of the encoder includes a multi-head attention mechanism and a feedforward network; In each encoder layer, the multi-head attention mechanism is used to calculate the attention output, and the attention output is passed to the feedforward network to extract features through nonlinear transformation; Performing residual connection and layer normalization processing on the output of the feedforward network to obtain the output of the layer encoder; The output of the previous layer encoder is used as the input of the next layer encoder, so as to obtain the feature representation of the encoder output of the lightweight Transformer model through layer-by-layer transmission and processing.

6. The method according to claim 3, characterized in that The three-dimensional classification head includes a fully connected layer and a Softmax activation function; inputting the fixed-length feature vector into the three-dimensional classification head for classification to obtain the failure probability of each component output by the three-dimensional classification head includes: Mapping the fixed-length feature vector to a fault probability space using the fully connected layer to obtain the original probability of each component failure; The Softmax activation function is used to convert the original probability into a normalized failure probability distribution to output the failure probability of each component.

7. The method according to claim 1, characterized in that After the recommended maintenance personnel bring the corresponding spare parts for on-site maintenance, the method further includes: When maintenance feedback information is received, continuously updating the multimodal training set according to the maintenance feedback information; The updated multimodal training set is re-input into the lightweight Transformer model for model training to improve the accuracy of the model in fault diagnosis.

8. A fault diagnosis device for an air conditioner outdoor unit module, characterized in that: include: A construction module is configured to collect air conditioner operation data and after-sales work order data to construct a multimodal training set, wherein the after-sales work order data includes a maintenance timestamp, a description of the fault phenomenon, and information about actual replaced parts; A training module is configured to input the multimodal training set into a lightweight Transformer model for model training to obtain a fault diagnosis model; a diagnostic module configured to input pre-fault operating data into the fault diagnosis model to output a component failure probability when the air conditioner triggers a fault code for the outdoor unit module; The recommendation module is configured to recommend maintenance personnel to carry corresponding spare parts for on-site repair based on the probability of component failure.

9. A fault diagnosis device for an air conditioner outdoor unit module, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the fault diagnosis method for an air conditioner outdoor unit module according to any one of claims 1 to 7 when running the program instructions.

10. An air conditioner, characterized in that: include: Air conditioner outdoor unit; The fault diagnosis device for an air conditioner outdoor unit module according to claim 8 or 9 is installed in the air conditioner outdoor unit.

Images