Building equipment fault intelligent diagnosis method and system

Abstract

The invention discloses a building equipment fault intelligent diagnosis method and system, and relates to the technical field of equipment fault diagnosis. A building equipment fault intelligent diagnosis system comprises a fault component determination module and a fault comprehensive diagnosis module. According to the method, the building equipment fault diagnosis digital model is constructed through a digital twinning method, and multi-modal data processing, feature matching and model updating are combined, so that the accuracy of fault diagnosis is effectively improved, especially in monitoring and analysis of multiple parts and complex equipment; through real-time updating based on the equipment fault diagnosis digital model, equipment fault parts can be found and positioned in time, and more efficient and more accurate equipment maintenance is realized, so that the equipment downtime and the maintenance cost are reduced; and an adaptive clustering algorithm and multi-level fault diagnosis analysis are adopted, so that different equipment states and fault types can be dynamically responded, and the real-time performance and the strain capacity of fault diagnosis are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment fault diagnosis, and in particular to an intelligent diagnosis method and system for building equipment faults. Background Art

[0002] In traditional building equipment fault diagnosis, common methods mainly rely on single sensor data or manual experience and judgment. The limitations of these methods are that the data from a single sensor cannot fully reflect the multi-dimensional state of equipment operation and are prone to overlooking potential complex fault modes. In addition, the diagnostic method that relies on manual experience is highly subjective and easily affected by the operator's experience and judgment ability, resulting in reduced accuracy and efficiency of fault diagnosis. With the diversification and complexity of building equipment types, traditional diagnostic methods can no longer meet the needs of modern building systems for efficient and accurate fault diagnosis.

[0003] Therefore, it is urgent to design a method that can combine multiple data for correlation analysis and use joint hierarchical reasoning technology for comprehensive fault diagnosis. This method can improve the diagnostic accuracy, expand the decision-making dimension and enhance the adaptive ability in building equipment fault diagnosis. Summary of the invention

[0004] The present invention aims to provide a method and system for intelligent diagnosis of building equipment faults to improve the accuracy of fault diagnosis.

[0005] A method for intelligent diagnosis of building equipment faults comprises the following steps: When a building equipment fault signal is received, the current monitoring time point is recorded; the building equipment to which the building equipment fault signal belongs is identified as the faulty building equipment; wherein, for the faulty building equipment, the corresponding building equipment fault diagnosis digital model is extracted; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, …, N, N is the total number of building equipment components included in the faulty building equipment; Get the faulty building equipment including the previous M monitoring time points L of the current monitoring time point m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponds to the current monitoring time point; The first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D mThe multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; Based on the faulty building equipment monitoring data set and the building equipment data matching model, the digital model Z of the updated equipment faulty component is obtained. n and building equipment component time series monitoring data X n ; Combine all updated equipment fault parts digital model Z n The digital model of building equipment fault diagnosis of faulty building equipment is updated to obtain a new digital model of building equipment fault diagnosis; the time series monitoring data X of building equipment components is used to n Conduct fault analysis with the intelligent diagnosis model of building equipment faults to obtain comprehensive diagnosis results of building equipment faults; According to the comprehensive diagnosis results of construction equipment faults and the new digital model for construction equipment fault diagnosis, maintenance operations are performed on the faulty construction equipment.

[0006] As a preferred technical solution of the present invention, the building equipment data matching model includes a monitoring data processing layer, a component feature matching layer and an analysis result output layer; The monitoring data processing layer is used to process the building equipment monitoring data D m Multimodal monitoring data D mi Normalize the format to obtain the preprocessed multimodal monitoring data D mi '; For all pre-processed multimodal monitoring data D mi 'Perform feature matching to obtain the building equipment matching label Q mi ; At the monitoring time point L m In the example, the same building equipment is matched with label Q mi The corresponding preprocessed multimodal monitoring data D mi 'Combined to obtain construction equipment component monitoring data J mn ; Set all monitoring time points L m Building equipment component monitoring data J mn Combined to obtain the time series monitoring data X of building equipment components n ; The component feature matching layer is used to diagnose building equipment faults based on the digital model and the time series monitoring data of building equipment components X n Perform model update processing to obtain the updated digital model Z of the faulty component of the equipment n ; The analysis result output layer is used to output and update the digital model Z of the equipment fault component n .

[0007] As a preferred technical solution of the present invention, the specific steps of performing model update processing in the component feature matching layer include: The component feature matching layer includes a feature extraction layer and a model updating layer; wherein the feature extraction layer includes a text data processing unit and an image data processing unit; Extracting time series monitoring data of building equipment components X n The text data in the text data is used to obtain the time series text change data T of the building equipment components. n In the text data processing unit, the building equipment component time series text change data T n Converted into building equipment component time series text data vector T n '; Use the pre-trained dynamic time attention mechanism to analyze the time series text data vector T of building equipment parts n 'Reconstruct and obtain the time series text data feature W of building equipment components n ; Extracting time series monitoring data of building equipment components X n The image data in the image data is used to obtain the time series image change data P of the building equipment components. n In the image data processing unit, the building equipment component time series image change data P n The frequency domain features and time domain features are extracted respectively to obtain the time domain feature component P of the time series image of the building equipment components. n1 and the frequency domain feature component P of the time series image of building equipment parts n2 ; The time domain feature component P of the building equipment component time series image n1 and the frequency domain feature component P of the time series image of building equipment parts n2 The pre-trained dynamic time attention mechanism is used to combine and obtain the temporal image data features U of the building equipment components. n ; In the model update layer, based on the temporal text data features W of building equipment components n and building equipment component time series image data features U n Predictions are made to obtain the transformation factor of the digital twin model of the building equipment; the building equipment component B in the building equipment fault diagnosis digital model of the faulty building equipment is predicted according to the transformation factor of the digital twin model of the building equipment. n Update and obtain the digital model Z of the updated equipment fault component n .

[0008] As a preferred technical solution of the present invention, the intelligent diagnosis model for building equipment faults includes a time series data diagnosis layer, a comprehensive fault diagnosis layer and a diagnosis result output layer; The time series data diagnostic layer is used to monitor the time series data of building equipment components X n The improved adaptive clustering algorithm is used to perform data cutting to obtain the fault traceback point H of the building equipment components. a , a=1, 2, ..., A; based on the building equipment component failure backtracking point Ha Time series monitoring data of building equipment components X n Perform data segmentation to obtain the time slice data block G of the building equipment components a ; The comprehensive fault diagnosis layer is used to calculate the time slice data block G of the building equipment components. a and construction equipment parts B n Perform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; All construction equipment component fault diagnosis results C na Combine to obtain a building equipment fault diagnosis matrix; perform comprehensive fault diagnosis according to the building equipment fault diagnosis matrix to obtain a comprehensive diagnosis result of the building equipment fault; The diagnostic result output layer is used to output the comprehensive diagnostic results of building equipment failures.

[0009] As a preferred technical solution of the present invention, the specific steps of using the improved adaptive clustering algorithm to perform data cutting in the time series data diagnosis layer include: Construct E clusters to divide the fault center points into individual F e , e=1, 2, ..., E; each cluster fault center is divided into individual F e Contains a method for monitoring the time series data of building equipment components X n The set of backtracking points to be divided; E cluster fault centers are divided into individual F e Combine and obtain the cluster fault center point to divide the iterative population; set the maximum number of iterations; Construct different cluster fault center points to divide individual F e The specific steps include: Using K-means clustering algorithm to classify the time series monitoring data of building equipment components X n Perform initial clustering to obtain the initial cluster center set of building equipment component failures; Based on the initial cluster center set of building equipment component failures, the first half of the initial cluster center set of building equipment component failures is retained, and several backtracking points are randomly generated near the second half of the initial cluster center set of building equipment component failures to obtain the fluctuating cluster center set of building equipment component failures; the first half of the initial cluster center set of building equipment component failures and the fluctuating cluster center set of building equipment component failures are arbitrarily combined to obtain E different cluster fault center point division individuals F e ; Based on clustering fault center points, individual F e Time series monitoring data of building equipment components X nPerform simulation partitioning to obtain simulation data partition blocks; calculate the similarity between different simulation data partition blocks to obtain the corresponding average similarity; use the inverse of the average similarity as the cluster fault center point to divide the individual F e The fitness R e ; When the maximum number of iterations is reached, the cluster fault center point partition individual corresponding to the current maximum fitness is output, which is the optimal cluster fault center point partition individual; data cutting is performed based on the optimal cluster fault center point partition individual to obtain the building equipment component fault backtracking point H a , a=1, 2,…,A.

[0010] As a preferred technical solution of the present invention, the specific steps of performing fault diagnosis in the comprehensive fault diagnosis layer include: The comprehensive fault diagnosis layer includes a component fault identification layer and a comprehensive fault result analysis layer; In the component fault identification layer, according to the building equipment component time slice data block G a and construction equipment parts B n Perform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; Among them, the comprehensive fault result analysis layer includes a result matrix analysis unit and a feature analysis unit; In the result matrix analysis unit, all the building equipment component fault diagnosis results C na Combine them to obtain the building equipment fault diagnosis matrix; In the feature analysis unit, feature extraction is performed based on the building equipment fault diagnosis matrix to obtain the building equipment fault diagnosis matrix features; feature recognition is performed on the building equipment fault diagnosis matrix features to obtain the comprehensive diagnosis results of the building equipment faults.

[0011] As a preferred technical solution of the present invention, the specific steps of the model update layer in the training component feature matching layer include: Collecting several groups of digital model change prediction training samples; each group of digital model change prediction training samples contains verified model transformation factors and corresponding data features; combining several groups of digital model change prediction training samples to obtain a digital model change prediction training set; The digital model change prediction training set is used for model training to obtain an initial model update layer. The initial model update layer is model evaluated. If the initial model update layer passes the model evaluation, the initial model update layer is used as the model update layer in the component feature matching layer. Otherwise, the digital model change prediction training set is used to continue model training.

[0012] An intelligent fault diagnosis system for building equipment, comprising: A fault component determination module includes a data acquisition unit and a data processing unit; The data acquisition unit is used to record the current monitoring time point when receiving a building equipment fault signal; identify the building equipment to which the building equipment fault signal belongs as the faulty building equipment; wherein, for the faulty building equipment, extract the corresponding building equipment fault diagnosis digital model; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, ..., N, N is the total number of building equipment components contained in the faulty building equipment; obtain the M monitoring time points L before the current monitoring time point of the faulty building equipment m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponding to the current monitoring time point; the first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D m The multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; The data processing unit is used to analyze the faulty building equipment monitoring data set and the building equipment data matching model to obtain an updated digital model Z of the equipment faulty component. n and building equipment component time series monitoring data X n ; A comprehensive fault diagnosis module, including a comprehensive diagnosis unit and a model updating unit; Integrated diagnostic unit for utilizing time series monitoring data from building equipment components n Conduct fault analysis with the intelligent fault diagnosis model of building equipment to obtain comprehensive fault diagnosis results of building equipment; perform maintenance operations on faulty building equipment based on the comprehensive fault diagnosis results of building equipment and the new digital fault diagnosis model of building equipment; The model updating unit is used to update the digital model Z of all the faulty parts of the equipment n The construction equipment fault diagnosis digital model of the faulty construction equipment is updated to obtain a new construction equipment fault diagnosis digital model.

[0013] The present invention has the following advantages: 1. The present invention constructs a digital model for building equipment fault diagnosis through a digital twin method, and combines multimodal data processing, feature matching and model updating to effectively improve the accuracy of fault diagnosis, especially in the monitoring and analysis of multi-component and complex equipment; through real-time updates based on the digital model of equipment fault diagnosis, it is possible to timely discover and locate equipment faulty components, achieve more efficient and accurate equipment maintenance, and thus reduce equipment downtime and maintenance costs; using adaptive clustering algorithms and multi-level fault diagnosis analysis, the model can dynamically respond to different equipment states and fault types, improve the real-time and adaptability of fault diagnosis, and ensure that the equipment can resume normal operation in the shortest time.

[0014] 2. The present invention can accurately divide the time-series monitoring data of building equipment components into reasonable data blocks through an improved adaptive clustering algorithm. This data cutting method can more accurately identify the changing trend of equipment failures, especially through the reasonable setting of backtracking points, to ensure the accurate identification of equipment failures; in the clustering process, after the initial fault clustering center is generated by the K-means clustering algorithm, it is optimized in combination with the fluctuating clustering center and the randomly generated backtracking points to ensure that the clustering center can adapt to the fluctuations of the equipment operating status. This adaptive adjustment improves the adaptability of the clustering process to different failure modes. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a schematic diagram of the structure of an intelligent diagnosis system for building equipment faults used in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0017] Embodiment 1, a method for intelligent diagnosis of building equipment faults, comprising the following steps: When a building equipment fault signal is received, the current monitoring time point is recorded; the building equipment to which the building equipment fault signal belongs is identified as the faulty building equipment; wherein, for the faulty building equipment, the corresponding building equipment fault diagnosis digital model is extracted; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, …, N, N is the total number of building equipment components included in the faulty building equipment; The digital twin method is used to construct a digital model for building equipment fault diagnosis, which can fully reflect the status and performance of building equipment and its components. The digital model of each building equipment component ensures accurate monitoring of all aspects of the equipment, which helps to detect potential faults in a timely manner. The equipment is decomposed into multiple components, and an independent fault diagnosis model is established for each component, making fault analysis more detailed and accurate. This can identify the fault of a specific component, not just the fault of the entire equipment, and improve diagnostic accuracy. Using digital twin technology, a virtual model is created for each building equipment, which enhances the digital management of the equipment. This method provides a strong data support and analysis basis for equipment health monitoring, fault prediction and maintenance decision-making. Get the faulty building equipment including the previous M monitoring time points L of the current monitoring time point m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponds to the current monitoring time point; Generally, data is collected from sensors at fixed time intervals. The monitoring time point usually corresponds to the timestamp of data collection, that is, the moment when the sensor records the data. When a building equipment failure is detected, the time when the failure signal is generated is the current monitoring time point. The size of M can be set to the monitoring data in the past few minutes or hours according to the data collection frequency and the response time of the equipment. The first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D m The multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; In building equipment monitoring, common sensor types include but are not limited to the following, each of which can collect multiple physical quantities and environmental parameters to facilitate comprehensive monitoring of the equipment status, such as temperature sensors, pressure sensors, vibration sensors, current and voltage sensors, etc. When a fault signal is received, the faulty equipment can be quickly identified and further analyzed through real-time monitoring data to ensure that the fault is located in time, buy time for subsequent processing, and reduce equipment downtime; by collecting the first M pieces of building equipment monitoring data and combining them into a faulty equipment monitoring data set, more comprehensive equipment operation status information can be provided. This fusion of multiple historical data improves the accuracy of fault diagnosis and avoids the risk of misjudgment by relying solely on a single data point; by using the monitoring data of I different sensors, the operation of the equipment can be comprehensively evaluated from multiple dimensions. Diversified data sources help to more accurately identify the type of fault and improve the reliability of diagnosis; Based on the faulty building equipment monitoring data set and the building equipment data matching model, the digital model Z of the updated equipment faulty component is obtained. n and building equipment component time series monitoring data X n ; The building equipment data matching model effectively improves the accuracy and real-time performance of intelligent diagnosis of building equipment faults and optimizes equipment maintenance capabilities through multimodal data processing, feature matching and model updating; The building equipment data matching model includes a monitoring data processing layer, a component feature matching layer, and an analysis result output layer; The monitoring data processing layer is used to process the building equipment monitoring data D m Multimodal monitoring data D mi Normalize the format to obtain the preprocessed multimodal monitoring data D mi '; For all pre-processed multimodal monitoring data D mi 'Perform feature matching to obtain the building equipment matching label Q mi ; At the monitoring time point L m In the example, the same building equipment is matched with label Q mi The corresponding preprocessed multimodal monitoring data D mi 'Combined to obtain construction equipment component monitoring data J mn ; Set all monitoring time points L m Building equipment component monitoring data J mn Combined to obtain the time series monitoring data X of building equipment components n ; The component feature matching layer is used to diagnose building equipment faults based on the digital model and the time series monitoring data of building equipment components X n Perform model update processing to obtain the updated digital model Z of the faulty component of the equipment n ; The analysis result output layer is used to output and update the digital model Z of the equipment fault component n ; By combining the building equipment data matching model with multimodal data processing, feature matching and model updating, the accuracy and real-time performance of fault diagnosis can be improved. The digital model of the equipment fault component can be updated in time to more accurately reflect the actual status of the equipment, thereby improving the accuracy of diagnosis. By continuously updating the digital model and time-series monitoring data of building equipment, the monitoring of the health status of the equipment can be enhanced, potential problems can be discovered in advance, and optimized equipment maintenance plans can be formulated, which reduces the occurrence of sudden failures and improves the efficiency of equipment maintenance. Through feature matching and model updating, a variety of monitoring data can be effectively matched with the faulty equipment model to help generate a digital model consistent with the current equipment status. This dynamic update method can reflect the real-time changes of equipment failures and improve the adaptability and flexibility of the diagnosis model. The specific steps of model updating in the component feature matching layer include: The component feature matching layer includes a feature extraction layer and a model updating layer; wherein the feature extraction layer includes a text data processing unit and an image data processing unit; Extracting time series monitoring data of building equipment components X n The text data in the text data is used to obtain the time series text change data T of the building equipment components. n In the text data processing unit, the building equipment component time series text change data T n Converted into building equipment component time series text data vector T n '; Use the pre-trained dynamic time attention mechanism to analyze the time series text data vector T of building equipment parts n 'Reconstruct and obtain the time series text data feature W of building equipment components n ; Extracting time series monitoring data of building equipment components X n The image data in the image data is used to obtain the time series image change data P of the building equipment components. n In the image data processing unit, the building equipment component time series image change data P n The frequency domain features and time domain features are extracted respectively to obtain the time domain feature component P of the time series image of the building equipment components. n1 and the frequency domain feature component P of the time series image of building equipment parts n2 ; The time domain feature component P of the building equipment component time series image n1 and the frequency domain feature component P of the time series image of building equipment parts n2 The pre-trained dynamic time attention mechanism is used to combine and obtain the temporal image data features U of the building equipment components. n ; The training process of the pre-trained dynamic temporal attention mechanism usually includes the following steps: First, the model is pre-trained using a large amount of historical data to capture the temporal dependencies and feature changes in the time series data; during the training process, the model automatically learns the weight distribution between different time steps through the attention mechanism, focusing on the moments or features that are most important to the current task and ignoring irrelevant or noisy data; the dynamic temporal attention mechanism not only considers the global time series information during training, but also adaptively adjusts the importance of each time point, so that the model can freely weight when processing different types of time series data, thereby improving the prediction accuracy and flexibility of the model; finally, the trained model can accurately reconstruct the change pattern of the time series data and effectively support the analysis and prediction of real-time data; In the model update layer, based on the temporal text data features W of building equipment components n and building equipment component time series image data features U n Predictions are made to obtain the transformation factor of the digital twin model of the building equipment; the building equipment component B in the building equipment fault diagnosis digital model of the faulty building equipment is predicted according to the transformation factor of the digital twin model of the building equipment. n Update and obtain the digital model Z of the updated equipment fault component n ; Through the text data processing unit and the image data processing unit in the feature extraction layer, rich text and image features can be extracted from the time-series monitoring data of building equipment components. The text data is reconstructed through the dynamic time attention mechanism, and the image data is extracted through the frequency domain and time domain feature extraction methods, ensuring the accurate expression of multimodal data, providing a solid foundation for subsequent diagnosis; the combination of text and image data is fused using the dynamic time attention mechanism, which can more comprehensively capture the multi-dimensional information of equipment failures. The text data provides the changing trend of the equipment operation status, and the image data reveals the physical state of the equipment components through the time domain and frequency domain features. The combination of the two significantly improves the comprehensiveness and accuracy of fault diagnosis; by using the time-series text data features and the time-series image data features for prediction in the model update layer, the transformation factors of the digital twin model of building equipment can be generated in real time, thereby dynamically updating the equipment fault diagnosis model. This dynamic update method can quickly adapt to changes in equipment status and improve the real-time and accuracy of fault diagnosis; combining multiple data features and adaptive update mechanisms, the equipment fault diagnosis model can accurately reflect the current health status of the equipment, and the updated digital model is more in line with the actual equipment situation, thereby improving the accuracy and reliability of diagnosis; The specific steps of the model update layer in the training component feature matching layer include: Collecting several groups of digital model change prediction training samples; each group of digital model change prediction training samples contains verified model transformation factors and corresponding data features; combining several groups of digital model change prediction training samples to obtain a digital model change prediction training set; The digital model change prediction training set is used to perform model training to obtain an initial model update layer; the initial model update layer is model evaluated, and if the initial model update layer passes the model evaluation, the initial model update layer is used as the model update layer in the component feature matching layer; otherwise, the digital model change prediction training set is used to continue model training; By collecting digital model change prediction training samples and conducting training, it can be ensured that the model update layer can accurately predict the changes in the equipment digital model based on the existing model transformation factors and data features. This training based on sample data improves the model's adaptability and predictive ability to equipment status changes. By conducting model evaluation on the initial model update layer, it can be ensured that only verified and excellent models are adopted, thereby improving the effectiveness and stability of the model. If the initial model fails the evaluation, it can still be optimized, ensuring continuous improvement and high-quality output during the training process. Combine all updated equipment fault parts digital model Z n The digital model of building equipment fault diagnosis of faulty building equipment is updated to obtain a new digital model of building equipment fault diagnosis; the time series monitoring data X of building equipment components is used to n Conduct fault analysis with the intelligent diagnosis model of building equipment faults to obtain comprehensive diagnosis results of building equipment faults; According to the comprehensive diagnosis results of building equipment faults and the new digital model for building equipment fault diagnosis, maintenance operations are performed on faulty building equipment. The intelligent diagnosis model for building equipment faults uses adaptive clustering algorithms and multi-level fault diagnosis analysis to achieve accurate identification and comprehensive diagnosis of building equipment faults, thereby improving the efficiency and accuracy of equipment fault detection. The intelligent diagnosis model of building equipment faults includes a time series data diagnosis layer, a comprehensive fault diagnosis layer, and a diagnosis result output layer; The time series data diagnostic layer is used to monitor the time series data of building equipment components X n The improved adaptive clustering algorithm is used to perform data cutting to obtain the fault traceback point H of the building equipment components. a , a=1, 2, ..., A; based on the building equipment component failure backtracking point H a Time series monitoring data of building equipment components X n Perform data segmentation to obtain the time slice data block G of the building equipment components a ; The comprehensive fault diagnosis layer is used to calculate the time slice data block G of the building equipment components. a and construction equipment parts Bn Perform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; All construction equipment component fault diagnosis results C na Combine to obtain a building equipment fault diagnosis matrix; perform comprehensive fault diagnosis according to the building equipment fault diagnosis matrix to obtain a comprehensive diagnosis result of the building equipment fault; The diagnostic result output layer is used to output the comprehensive diagnostic results of building equipment failures; By updating the digital model of equipment failure components and combining time series monitoring data with the improved adaptive clustering algorithm, the failure mode of building equipment can be identified more accurately, which enables fault diagnosis to be predicted and analyzed based on the actual operation of the equipment, thereby improving the accuracy and efficiency of fault detection; the adaptive clustering algorithm in the time series data diagnosis layer can dynamically trace back faults based on the real-time monitoring data of the equipment and automatically identify the time point of potential equipment failures; through this dynamic adaptation process, diagnosis can update fault analysis in real time and provide continuously optimized equipment status assessment; the comprehensive fault diagnosis layer generates a building equipment fault diagnosis matrix by summarizing the fault diagnosis results of each component, and conducts detailed fault analysis from multiple levels. This multi-level diagnostic method helps to fully identify the overall problems of the equipment and avoid the risk of a single component failure going unnoticed; based on the analysis of time series data and component failure traceback points, the failure trend of the equipment can be identified in advance, providing earlier warnings for equipment maintenance. This fault prediction capability provides equipment managers with enough time to perform preventive maintenance and avoid losses caused by sudden failures; The specific steps of using the improved adaptive clustering algorithm to perform data cutting in the time series data diagnosis layer include: Construct E clusters to divide the fault center points into individual F e , e=1, 2, ..., E; each cluster fault center is divided into individual F e Contains a method for monitoring the time series data of building equipment components X n The set of backtracking points to be divided; E cluster fault centers are divided into individual F e Combine and obtain the cluster fault center point to divide the iterative population; set the maximum number of iterations; the maximum number of iterations is set by professional technicians according to actual conditions; Construct different cluster fault center points to divide individual F e The specific steps include: Using K-means clustering algorithm to classify the time series monitoring data of building equipment components X n Perform initial clustering to obtain the initial cluster center set of building equipment component failures; Based on the initial cluster center set of building equipment component failures, the first half of the initial cluster center set of building equipment component failures is retained, and several backtracking points are randomly generated near the second half of the initial cluster center set of building equipment component failures to obtain the fluctuating cluster center set of building equipment component failures; the first half of the initial cluster center set of building equipment component failures and the fluctuating cluster center set of building equipment component failures are arbitrarily combined to obtain E different cluster fault center point division individuals F e ; Based on clustering fault center points, individual F e Time series monitoring data of building equipment components X n Perform simulation partitioning to obtain simulation data partition blocks; calculate the similarity between different simulation data partition blocks to obtain the corresponding average similarity; use the inverse of the average similarity as the cluster fault center point to divide the individual F e The fitness R e ; When the maximum number of iterations is reached, the cluster fault center point partition individual corresponding to the current maximum fitness is output, which is the optimal cluster fault center point partition individual; data cutting is performed based on the optimal cluster fault center point partition individual to obtain the building equipment component fault backtracking point H a , a=1,2,…,A; Through the improved adaptive clustering algorithm, the time series monitoring data of building equipment components can be accurately divided into reasonable data blocks. This data cutting method can more finely identify the changing trend of equipment failures, especially through the reasonable setting of backtracking points, which ensures the accurate identification of equipment failures. In the clustering process, after the initial fault clustering center is generated by the K-means clustering algorithm, it is optimized by combining the fluctuating clustering center and the randomly generated backtracking points to ensure that the clustering center can adapt to the fluctuation of the equipment operation status. This adaptive adjustment improves the adaptability of the clustering process to different fault modes. By setting the maximum number of iterations and optimizing according to the fitness of the clustering fault center point, it is ensured that the clustering process gradually converges to the optimal solution in multiple iterations. This iterative optimization process makes the final clustering result more accurate and effectively avoids the situation of local optimal solution. By accurately dividing the data blocks and calculating their similarity, it can provide clearer and structured data for subsequent fault analysis. These data blocks help improve the fault diagnosis model's understanding and processing capabilities of equipment component time series data. The specific steps of fault diagnosis in the comprehensive fault diagnosis layer include: The comprehensive fault diagnosis layer includes a component fault identification layer and a comprehensive fault result analysis layer; In the component fault identification layer, according to the building equipment component time slice data block G a and construction equipment parts B nPerform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; Among them, the comprehensive fault result analysis layer includes a result matrix analysis unit and a feature analysis unit; In the result matrix analysis unit, all the building equipment component fault diagnosis results C na Combine them to obtain the building equipment fault diagnosis matrix; In the feature analysis unit, feature extraction is performed based on the building equipment fault diagnosis matrix to obtain the building equipment fault diagnosis matrix features; feature recognition is performed on the building equipment fault diagnosis matrix features to obtain the comprehensive diagnosis results of the building equipment faults; The function of the result matrix analysis unit is to combine the fault diagnosis results of all building equipment components to form a fault diagnosis matrix. First, the fault diagnosis results of each component are collected. For example, the fault diagnosis result of each component is calculated based on historical data, sensor data and other information. The fault diagnosis results of all components are combined in chronological order and component number to form a diagnosis matrix. A machine learning model can be used to analyze the matrix and train a classifier or regression model to generate a fault diagnosis matrix under future input data. The function of the feature analysis unit is to extract features based on the fault diagnosis matrix and then perform fault identification. The feature extraction process includes identifying key patterns or relationships that affect the equipment status from the original matrix, and then analyzing these features to obtain comprehensive fault diagnosis results. In this step, data preprocessing and feature extraction techniques, such as principal component analysis, convolutional neural networks or autoencoders, can be used to convert the fault diagnosis matrix into a more concise and representative feature vector. The extracted features are input into a deep neural network classifier for training to learn how to identify and predict equipment faults based on different feature patterns. Through the component fault identification layer, the time slice data blocks and component information of the building equipment components are used to perform independent fault diagnosis, which can accurately identify the fault type and fault status of each equipment component. This provides detailed component-level information for the overall equipment fault diagnosis and ensures the accuracy of the diagnosis results. In the comprehensive fault result analysis layer, by combining all component fault diagnosis results into a fault diagnosis matrix, the overall status of the building equipment can be comprehensively analyzed. This global perspective helps to reveal the correlation and mutual influence between multiple component failures, thereby improving the comprehensiveness and accuracy of fault diagnosis. The feature analysis unit can effectively screen out the most diagnostically valuable information by extracting key features from the fault diagnosis matrix and performing feature recognition. This process helps to improve the intelligence level of the fault diagnosis model, enabling it to automatically extract useful diagnostic features from complex diagnostic data. Through the analysis and feature recognition of the fault diagnosis matrix, reliable comprehensive fault diagnosis results can be obtained in a short time. This efficient diagnostic process not only improves the speed of fault identification, but also improves the reliability of fault prediction, avoiding missed diagnosis or misdiagnosis.

[0018] Example 2, a building equipment fault intelligent diagnosis system, see Figure 1 As shown, including: A fault component determination module includes a data acquisition unit and a data processing unit; The data acquisition unit is used to record the current monitoring time point when receiving a building equipment fault signal; identify the building equipment to which the building equipment fault signal belongs as the faulty building equipment; wherein, for the faulty building equipment, extract the corresponding building equipment fault diagnosis digital model; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, ..., N, N is the total number of building equipment components contained in the faulty building equipment; obtain the M monitoring time points L before the current monitoring time point of the faulty building equipment m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponding to the current monitoring time point; the first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D m The multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; The data processing unit is used to analyze the faulty building equipment monitoring data set and the building equipment data matching model to obtain an updated digital model Z of the equipment faulty component. n and building equipment component time series monitoring data X n ; A comprehensive fault diagnosis module, including a comprehensive diagnosis unit and a model updating unit; Integrated diagnostic unit for utilizing time series monitoring data from building equipment components n Conduct fault analysis with the intelligent fault diagnosis model of building equipment to obtain comprehensive fault diagnosis results of building equipment; perform maintenance operations on faulty building equipment based on the comprehensive fault diagnosis results of building equipment and the new digital fault diagnosis model of building equipment; The model updating unit is used to update the digital model Z of all the faulty parts of the equipment n The construction equipment fault diagnosis digital model of the faulty construction equipment is updated to obtain a new construction equipment fault diagnosis digital model.

[0019] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A method for intelligent diagnosis of building equipment faults, characterized in that: The following steps are involved: When a building equipment failure signal is received, the current monitoring time point is recorded; Identify the building equipment to which the building equipment fault signal belongs as the faulty building equipment; wherein, for the faulty building equipment, extract the corresponding building equipment fault diagnosis digital model; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, …, N, N is the total number of building equipment components included in the faulty building equipment; Get the faulty building equipment including the previous M monitoring time points L of the current monitoring time point m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponds to the current monitoring time point; The first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D m The multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; Based on the faulty building equipment monitoring data set and the building equipment data matching model, the digital model Z of the updated equipment faulty component is obtained. n and building equipment component time series monitoring data X n ; Combine all updated equipment fault parts digital model Z n The digital model of building equipment fault diagnosis of faulty building equipment is updated to obtain a new digital model of building equipment fault diagnosis; the time series monitoring data X of building equipment components is used to n Conduct fault analysis with the intelligent diagnosis model of building equipment faults to obtain comprehensive diagnosis results of building equipment faults; According to the comprehensive diagnosis results of construction equipment faults and the new digital model for construction equipment fault diagnosis, maintenance operations are performed on the faulty construction equipment.

2. A construction equipment fault intelligent diagnosis method according to claim 1, characterized in that: The building equipment data matching model includes a monitoring data processing layer, a component feature matching layer, and an analysis result output layer; The monitoring data processing layer is used to process the building equipment monitoring data D m Multimodal monitoring data D mi Normalize the format to obtain the preprocessed multimodal monitoring data D mi '; For all pre-processed multimodal monitoring data D mi 'Perform feature matching to obtain the building equipment matching label Q mi ; At the monitoring time point L m In the example, the same building equipment is matched with label Q mi The corresponding preprocessed multimodal monitoring data D mi 'Combined to obtain construction equipment component monitoring data J mn ; Set all monitoring time points L m Building equipment component monitoring data J mn Combined to obtain the time series monitoring data X of building equipment components n ; The component feature matching layer is used to diagnose building equipment faults based on the digital model and the time series monitoring data of building equipment components X n Perform model update processing to obtain the updated digital model Z of the faulty component of the equipment n ; The analysis result output layer is used to output and update the digital model Z of the equipment fault component n .

3. A construction equipment fault intelligent diagnosis method according to claim 2, characterized in that: The specific steps of model updating in the component feature matching layer include: The component feature matching layer includes a feature extraction layer and a model updating layer; wherein the feature extraction layer includes a text data processing unit and an image data processing unit; Extracting time series monitoring data of building equipment components X n The text data in the text data is used to obtain the time series text change data T of the building equipment components. n In the text data processing unit, the building equipment component time series text change data T n Converted into building equipment component time series text data vector T n '; Use the pre-trained dynamic time attention mechanism to analyze the time series text data vector T of building equipment parts n 'Reconstruct and obtain the time series text data feature W of building equipment components n ; Extracting time series monitoring data of building equipment components X n The image data in the image data is used to obtain the time series image change data P of the building equipment components. n In the image data processing unit, the building equipment component time series image change data P n The frequency domain features and time domain features are extracted respectively to obtain the time domain feature component P of the time series image of the building equipment components. n1 and the frequency domain feature component P of the time series image of building equipment parts n2 ; The time domain feature component P of the building equipment component time series image n1 and the frequency domain feature component P of the time series image of building equipment parts n2 The pre-trained dynamic time attention mechanism is used to combine and obtain the temporal image data features U of the building equipment components. n ; In the model update layer, based on the temporal text data features W of building equipment components n and building equipment component time series image data features U n Predictions are made to obtain the transformation factor of the digital twin model of the building equipment; the building equipment component B in the building equipment fault diagnosis digital model of the faulty building equipment is predicted according to the transformation factor of the digital twin model of the building equipment. n Update and obtain the digital model Z of the updated equipment fault component n .

4. A construction equipment fault intelligent diagnosis method according to claim 3, characterized in that: The intelligent diagnosis model of building equipment faults includes a time series data diagnosis layer, a comprehensive fault diagnosis layer, and a diagnosis result output layer; The time series data diagnostic layer is used to monitor the time series data of building equipment components X n The improved adaptive clustering algorithm is used to perform data cutting to obtain the fault traceback point H of the building equipment components. a , a=1,2,…,A; Based on the fault traceback point H of building equipment components a Time series monitoring data of building equipment components X n Perform data segmentation to obtain the time slice data block G of the building equipment components a ; The comprehensive fault diagnosis layer is used to calculate the time slice data block G of the building equipment components. a and construction equipment parts B n Perform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; All construction equipment component fault diagnosis results C na Combine them to obtain the building equipment fault diagnosis matrix; Perform comprehensive fault diagnosis according to the building equipment fault diagnosis matrix to obtain comprehensive diagnosis results of building equipment faults; The diagnostic result output layer is used to output the comprehensive diagnostic results of building equipment failures.

5. A construction equipment fault intelligent diagnosis method according to claim 4, characterized in that: The specific steps of using the improved adaptive clustering algorithm to perform data cutting in the time series data diagnosis layer include: Construct E clusters to divide the fault center points into individual F e , e=1, 2, ..., E; each cluster fault center is divided into individual F e Contains a method for monitoring the time series data of building equipment components X n The set of backtracking points to be divided; E cluster fault centers are divided into individual F e Combine and obtain the cluster fault center points to divide the iterative population; set the maximum number of iterations; Construct different cluster fault center points to divide individual F e The specific steps include: Using K-means clustering algorithm to classify the time series monitoring data of building equipment components X n Perform initial clustering to obtain the initial cluster center set of building equipment component failures; Based on the initial cluster center set of building equipment component failures, the first half of the initial cluster center set of building equipment component failures is retained, and several backtracking points are randomly generated near the second half of the initial cluster center set of building equipment component failures to obtain the fluctuation cluster center set of building equipment component failures; the first half of the initial cluster center set of building equipment component failures and the fluctuation cluster center set of building equipment component failures are arbitrarily combined to obtain E different cluster fault center point division individuals F e ; Based on clustering fault center points, individual F e Time series monitoring data of building equipment components X n Perform simulation partitioning to obtain simulation data partition blocks; calculate the similarity between different simulation data partition blocks to obtain the corresponding average similarity; use the inverse of the average similarity as the cluster fault center point to divide the individual F e The fitness R e ; When the maximum number of iterations is reached, the cluster fault center point partition individual corresponding to the current maximum fitness is output, which is the optimal cluster fault center point partition individual; data cutting is performed based on the optimal cluster fault center point partition individual to obtain the building equipment component fault backtracking point H a , a=1, 2,…,A.

6. A construction equipment fault intelligent diagnosis method according to claim 5, characterized in that: The specific steps of fault diagnosis in the comprehensive fault diagnosis layer include: The comprehensive fault diagnosis layer includes a component fault identification layer and a comprehensive fault result analysis layer; In the component fault identification layer, according to the building equipment component time slice data block G a and construction equipment parts B n Perform fault diagnosis and obtain the fault diagnosis result C of the building equipment components na ; Among them, the comprehensive fault result analysis layer includes a result matrix analysis unit and a feature analysis unit; In the result matrix analysis unit, all the building equipment component fault diagnosis results C na Combine them to obtain the building equipment fault diagnosis matrix; In the feature analysis unit, feature extraction is performed based on the building equipment fault diagnosis matrix to obtain the building equipment fault diagnosis matrix features; feature recognition is performed on the building equipment fault diagnosis matrix features to obtain the comprehensive diagnosis results of the building equipment faults.

7. A construction equipment fault intelligent diagnosis method according to claim 6, characterized in that: The specific steps of the model update layer in the training component feature matching layer include: Collecting several groups of digital model change prediction training samples; each group of digital model change prediction training samples contains verified model transformation factors and corresponding data features; combining several groups of digital model change prediction training samples to obtain a digital model change prediction training set; The digital model change prediction training set is used for model training to obtain an initial model update layer. The initial model update layer is model evaluated. If the initial model update layer passes the model evaluation, the initial model update layer is used as the model update layer in the component feature matching layer. Otherwise, the digital model change prediction training set is used to continue model training.

8. An intelligent fault diagnosis system for building equipment, characterized in that: The system applies a construction equipment fault intelligent diagnosis method as described in any one of claims 1 to 7, comprising: A fault component determination module includes a data acquisition unit and a data processing unit; The data acquisition unit is used to record the current monitoring time point when receiving a building equipment fault signal; identify the building equipment to which the building equipment fault signal belongs as the faulty building equipment; wherein, for the faulty building equipment, extract the corresponding building equipment fault diagnosis digital model; the building equipment fault diagnosis digital model is constructed by the digital twin method; the building equipment fault diagnosis digital model includes N building equipment components B n , n=1, 2, ..., N, N is the total number of building equipment components contained in the faulty building equipment; obtain the M monitoring time points L before the current monitoring time point of the faulty building equipment m Construction equipment monitoring data D m , m=1, 2, ..., M; monitoring time point L M Corresponding to the current monitoring time point; the first M pieces of building equipment monitoring data D m Combined, we get the faulty building equipment monitoring data set; among them, D m ={D mi |i=1,2,…,I};D mi Represents building equipment monitoring data D m The multimodal monitoring data comes from different sensors, and I represents the total number of different sensors; The data processing unit is used to analyze the faulty building equipment monitoring data set and the building equipment data matching model to obtain an updated digital model Z of the equipment faulty component. n and building equipment component time series monitoring data X n ; A comprehensive fault diagnosis module, including a comprehensive diagnosis unit and a model updating unit; Integrated diagnostic unit for utilizing time series monitoring data from building equipment components n Conduct fault analysis with the intelligent fault diagnosis model of building equipment to obtain comprehensive fault diagnosis results of building equipment; perform maintenance operations on faulty building equipment based on the comprehensive fault diagnosis results of building equipment and the new digital fault diagnosis model of building equipment; The model updating unit is used to update the digital model Z of all the faulty parts of the equipment n The construction equipment fault diagnosis digital model of the faulty construction equipment is updated to obtain a new construction equipment fault diagnosis digital model.

Images