Intelligent building equipment fault prediction and diagnosis device based on AI

Abstract

The invention discloses an intelligent building equipment fault prediction and diagnosis device based on AI. The device comprises a data acquisition module, a data preprocessing module, a multi-dimensional feature fusion module, an AI prediction and diagnosis core module, a fault tracing module, an early warning grading module, a data storage module and a man-machine interaction module. The beneficial effects of the invention are that a complete technical system of data acquisition, preprocessing, feature fusion, prediction diagnosis and traceability early warning is constructed, fault risk prediction is realized, the functions of fault type diagnosis and fault root traceability are added, the defect that in the prior art, only prediction can be carried out, and diagnosis and traceability cannot be carried out is overcome, and the fault risk prediction and the fault root traceability are realized. The fault processing efficiency is greatly improved; according to the method, the multi-dimensional feature fusion technology is adopted, the equipment features, the equipment correlation features and the environment influence features are comprehensively considered, and compared with the mode that only the single equipment operation state sequence is analyzed in the prior art, the accuracy and robustness of the prediction diagnosis result are improved.

Description

Technical Field

[0001] This invention belongs to the field of building equipment fault prediction technology, specifically relating to an AI-based intelligent building equipment fault prediction and diagnosis device. Background Technology

[0002] With the rapid development of intelligent buildings, the types of equipment integrated within these buildings are becoming increasingly diverse, including central air conditioning systems, water supply and drainage equipment, elevator systems, security monitoring equipment, and power distribution equipment. The stable operation of these devices directly affects the normal use of the building and the user experience. However, during long-term operation, intelligent building equipment is prone to various malfunctions due to environmental factors, usage frequency, equipment aging, improper maintenance, and other factors. If these malfunctions are not detected and addressed in a timely manner, they may lead to equipment downtime, functional failure, or even safety accidents, resulting in serious economic losses. Therefore, effective fault prediction and diagnosis of intelligent building equipment, early identification of potential fault risks, and accurate location of fault causes are of significant practical importance. To address the aforementioned problems, various building equipment failure prediction schemes have been proposed in the prior art. Among them, Chinese invention patent CN118314458A discloses a method, device, equipment, and storage medium for predicting micro-failures of building equipment. The prediction method disclosed in this patent includes the following steps: collecting multimodal operating state sequences of all building equipment and performing feature analysis on the multimodal operating state sequences to obtain composite modal dynamic features; based on the composite modal dynamic features, using a hidden Markov model based on an attention mechanism to simulate the dynamic operating state under a preset future time period to obtain operating state transition trend features; establishing a probability distribution graph model based on the operating state transition trend features, and analyzing the impact of potential micro-changes based on the probability distribution graph model. The patent analyzes the transfer trends of equipment operating states, establishes a probability distribution graph model, and analyzes the influence characteristics of potential micro-changes to obtain the micro-fault risk level, thereby achieving the prediction of equipment micro-faults. This improves the accuracy and reliability of building equipment micro-fault prediction to a certain extent, helps to promptly detect and resolve equipment problems, and reduces maintenance costs and downtime. The corresponding device includes a modal feature analysis module, a transfer trend analysis module, a potential change determination module, and a micro-fault prediction module. Each module implements the corresponding function in the above steps. However, the aforementioned existing technologies still have many shortcomings: First, the patent only focuses on fault prediction and fails to achieve accurate diagnosis of faults that have already occurred and trace the causes of faults. When equipment actually fails, maintenance personnel still need to spend a lot of time troubleshooting the fault points and root causes, resulting in low fault handling efficiency. Second, the patent uses attention-based hidden Markov models and probability distribution graph models for fault prediction. The model structure is complex, has high requirements for hardware computing resources, and the model training and inference are time-consuming, making it difficult to meet the real-time requirements of fault prediction for intelligent building equipment. Third, the patent only performs simple feature analysis on the collected multimodal operating state sequences, without considering the correlation between different devices and the dynamic changes in the equipment operating environment, resulting in insufficient accuracy and robustness of the prediction results. Finally, the patent does not set up a fault warning classification mechanism and cannot provide targeted handling suggestions based on the fault risk level, which is not conducive to maintenance personnel rationally allocating resources and efficiently handling high-priority faults. Summary of the Invention

[0003] The purpose of this invention is to provide an AI-based intelligent building equipment fault prediction and diagnosis device, which can not only accurately predict potential fault risks of intelligent building equipment, but also quickly diagnose and trace the root cause of faults that have already occurred. It also features high real-time performance, strong accuracy, and good robustness, which can effectively improve the operation and maintenance efficiency of intelligent building equipment and reduce operation and maintenance costs.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an AI-based intelligent building equipment fault prediction and diagnosis device, comprising a data acquisition module, a data preprocessing module, a multi-dimensional feature fusion module, an AI prediction and diagnosis core module, a fault tracing module, an early warning and classification module, a data storage module, and a human-computer interaction module; The output end of the data acquisition module is connected to the input end of the data preprocessing module, and is used to collect multi-source operating data of intelligent building equipment and transmit it to the data preprocessing module; The output of the data preprocessing module is connected to the input of the multi-dimensional feature fusion module, and is used to process multi-source running data and then transmit it to the multi-dimensional feature fusion module. The output of the multi-dimensional feature fusion module is connected to the input of the AI ​​prediction and diagnosis core module. It is used to extract multi-dimensional features from multi-source running data and fuse them to obtain a fused feature vector, which is then transmitted to the AI ​​prediction and diagnosis core module. The output of the AI ​​prediction and diagnosis core module is connected to the input of the fault tracing module and the early warning and classification module, respectively. It is used to realize fault risk prediction and fault type diagnosis based on fused feature vectors, and transmit the prediction results to the early warning and classification module and the diagnosis results to the fault tracing module. The output of the fault tracing module is connected to the input of the human-computer interaction module, and is used to trace the root cause of the fault according to the fault type diagnosis result and transmit it to the human-computer interaction module. The output of the early warning classification module is connected to the input of the human-computer interaction module. It is used to classify the early warning level based on the fault risk prediction result and generate processing suggestions, which are then transmitted to the human-computer interaction module. The data storage module is bidirectionally connected to the data preprocessing module, the multi-dimensional feature fusion module, the AI ​​prediction and diagnosis core module, the fault tracing module, and the early warning and classification module, and is used to store the input data, processing data, and output results of each module. The human-computer interaction module is used to receive user operation commands and display fault prediction results, fault diagnosis results, fault tracing results, warning levels, and handling suggestions.

[0005] As a preferred embodiment of the present invention, the data acquisition module includes a sensor unit, a device interface unit, and an environmental data acquisition unit; the sensor unit is used to collect the operating status parameters of the device, including voltage, current, temperature, pressure, vibration frequency, and operating time; the device interface unit is used to establish a connection with the control module of the intelligent building equipment through a communication protocol, and to collect the device's operating log data and fault history data; the environmental data acquisition unit is used to collect the environmental parameters of the device's operating environment, including ambient temperature, humidity, air quality, and dust concentration.

[0006] As a preferred technical solution of the present invention, the data preprocessing module processes the multi-source running data as follows: first, outlier detection is performed on the collected multi-source running data to remove outlier data that deviates significantly from the normal data range; then, the moving average method is used to denoise the data to eliminate random noise in the data; finally, all the denoised data is standardized to convert data of different dimensions to the same numerical range to obtain standardized data.

[0007] As a preferred technical solution of the present invention, the multi-dimensional features extracted by the multi-dimensional feature fusion module include equipment self-features, equipment association features, and environmental impact features. The processing procedure of the multi-dimensional feature fusion module is as follows: extracting statistical features, trend features, and mutation features of equipment operating status parameters as equipment self-features; determining the strength of the association relationship and the association impact features between equipment by analyzing the correlation of operating data between different equipment; analyzing the mapping relationship between environmental parameters and equipment operating status parameters to obtain the environmental impact features of environmental factors on equipment operation; and using a weighted fusion algorithm to fuse equipment self-features, equipment association features, and environmental impact features, assigning weights according to the importance of each feature to fault prediction and diagnosis to obtain a fused feature vector.

[0008] As a preferred technical solution of the present invention, the AI ​​prediction and diagnosis core module includes a model training unit and a prediction and diagnosis unit; the model training unit is used to construct an AI hybrid model, which includes a fault prediction sub-model based on deep learning and a fault diagnosis sub-model based on machine learning; the AI ​​hybrid model is trained by historical multi-source operating data and historical fault data to obtain a pre-trained AI hybrid model; the prediction and diagnosis unit is used to input the fused feature vector into the pre-trained AI hybrid model, predict the probability of fault risk of the device in the future within a preset time period through the fault prediction sub-model, and diagnose whether the current device has a fault and the type of fault through the fault diagnosis sub-model.

[0009] As a preferred technical solution of the present invention, the fault prediction sub-model adopts an improved long short-term memory network, which optimizes the network structure by introducing a gating attention mechanism to enhance the ability to capture key time series features; the fault diagnosis sub-model adopts a gradient boosting decision tree algorithm, which improves the accuracy of fault type diagnosis by constructing multiple decision trees and performing ensemble learning.

[0010] As a preferred technical solution of the present invention, the specific implementation process of the fault tracing module is as follows: based on the fault type diagnosis result, retrieve the historical operating data and historical fault handling records of the device and related devices in the data storage module; analyze the change pattern of the device's operating status parameters before and after the fault occurs, the change of the operating status of related devices, and the change of environmental parameters; combine the structural principle of the device and the common fault cause library, and use the causal reasoning algorithm to trace the root cause of the fault, the root cause including aging of equipment components, abnormal parameter settings, influence of related devices, influence of environmental factors, or improper operation.

[0011] As a preferred technical solution of the present invention, the specific implementation process of the early warning classification module is as follows: a fault risk level classification standard is preset, and the fault risk level includes no warning, low risk warning, medium risk warning and high risk warning; according to the fault risk probability output by the fault prediction sub-model, the corresponding warning level is determined by referring to the preset risk level classification standard; corresponding handling suggestions are generated for different warning levels: no warning level requires no handling, low risk warning level suggests increasing the monitoring frequency, medium risk warning level suggests arranging planned maintenance, and high risk warning level suggests immediate shutdown for inspection and maintenance.

[0012] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a complete technical system for data acquisition, preprocessing, feature fusion, predictive diagnosis, and source tracing and early warning. It not only realizes fault risk prediction, but also adds fault type diagnosis and fault root cause tracing functions, which solves the defects of existing technologies that can only predict but not diagnose and trace the source, and greatly improves fault handling efficiency. This invention employs multi-dimensional feature fusion technology, comprehensively considering the characteristics of the equipment itself, the characteristics of equipment association, and the characteristics of environmental influence. Compared with the existing technology that only analyzes the operating state sequence of a single device, it improves the accuracy and robustness of the prediction and diagnosis results and can adapt to the complex operating environment of intelligent building equipment. The AI ​​hybrid model designed in this invention combines an improved long short-term memory network and a gradient boosting decision tree algorithm, which takes into account both the temporality of fault prediction and the classification accuracy of fault diagnosis. Moreover, the model structure is relatively simple, which reduces the hardware computing resource requirements, improves the real-time performance of prediction and diagnosis, and solves the problems of complexity and insufficient real-time performance of existing technology models. This invention includes an early warning classification module and a human-computer interaction module. It provides targeted handling suggestions based on the fault risk level and alerts maintenance personnel through audible and visual alarms. This helps maintenance personnel to allocate resources rationally, prioritize high-priority faults, and further improve the targeting and efficiency of maintenance work. Attached Figure Description

[0013] Figure 1 This is a block diagram of the AI-based intelligent building equipment fault prediction and diagnosis device of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Please see Figure 1 The present invention provides an AI-based intelligent building equipment fault prediction and diagnosis device, including a data acquisition module, a data preprocessing module, a multi-dimensional feature fusion module, an AI prediction and diagnosis core module, a fault tracing module, an early warning and classification module, a data storage module, and a human-computer interaction module. The output of the data acquisition module is connected to the input of the data preprocessing module, and is used to collect multi-source operating data of intelligent building equipment and transmit it to the data preprocessing module; The output of the data preprocessing module is connected to the input of the multi-dimensional feature fusion module, and is used to process multi-source running data and then transmit it to the multi-dimensional feature fusion module; The output of the multi-dimensional feature fusion module is connected to the input of the AI ​​prediction and diagnosis core module. It is used to extract multi-dimensional features from multi-source running data and fuse them to obtain a fused feature vector, which is then transmitted to the AI ​​prediction and diagnosis core module. The output of the AI ​​prediction and diagnosis core module is connected to the input of the fault tracing module and the early warning and classification module, respectively. It is used to realize fault risk prediction and fault type diagnosis based on fused feature vectors, and transmit the prediction results to the early warning and classification module and the diagnosis results to the fault tracing module. The output of the fault tracing module is connected to the input of the human-machine interaction module, and is used to trace the root cause of the fault based on the fault type diagnosis results and transmit them to the human-machine interaction module. The output of the early warning classification module is connected to the input of the human-machine interaction module. It is used to classify the early warning level based on the fault risk prediction results and generate handling suggestions, which are then transmitted to the human-machine interaction module. The data storage module is bidirectionally connected to the data preprocessing module, the multi-dimensional feature fusion module, the AI ​​prediction and diagnosis core module, the fault tracing module, and the early warning and classification module, and is used to store the input data, processing data, and output results of each module. The human-computer interaction module is used to receive user operation commands and display fault prediction results, fault diagnosis results, fault tracing results, warning levels, and handling suggestions.

[0016] In this embodiment, the data acquisition module includes a sensor unit, a device interface unit, and an environmental data acquisition unit. The sensor unit is used to collect the operating status parameters of the device, including voltage, current, temperature, pressure, vibration frequency, and operating time. The device interface unit is used to establish a connection with the control module of the intelligent building device through a communication protocol and collect the device's operating log data and fault history data. The environmental data acquisition unit is used to collect the environmental parameters of the device's operating environment, including ambient temperature, humidity, air quality, and dust concentration.

[0017] In this embodiment, the data preprocessing module processes the multi-source running data as follows: First, outlier detection is performed on the collected multi-source running data to remove outlier data that deviates significantly from the normal data range; then, the moving average method is used to denoise the data to eliminate random noise in the data; finally, all the denoised data is standardized to convert data of different dimensions to the same numerical range to obtain standardized data.

[0018] In this embodiment, the multi-dimensional features extracted by the multi-dimensional feature fusion module include equipment self-features, equipment association features, and environmental impact features. The processing procedure of the multi-dimensional feature fusion module is as follows: extracting statistical features, trend features, and mutation features of equipment operating status parameters as equipment self-features; determining the strength of the association relationship and the association impact features between equipment by analyzing the correlation of operating data between different devices; analyzing the mapping relationship between environmental parameters and equipment operating status parameters to obtain the environmental impact features of environmental factors on equipment operation; and using a weighted fusion algorithm to fuse equipment self-features, equipment association features, and environmental impact features, assigning weights according to the importance of each feature to fault prediction and diagnosis to obtain a fused feature vector.

[0019] In this embodiment, the AI ​​prediction and diagnosis core module includes a model training unit and a prediction and diagnosis unit. The model training unit is used to construct an AI hybrid model, which includes a fault prediction sub-model based on deep learning and a fault diagnosis sub-model based on machine learning. The AI ​​hybrid model is trained using historical multi-source operating data and historical fault data to obtain a pre-trained AI hybrid model. The prediction and diagnosis unit is used to input the fused feature vector into the pre-trained AI hybrid model, predict the probability of fault risk of the device within a preset time period through the fault prediction sub-model, and diagnose whether the device currently has a fault and the type of fault through the fault diagnosis sub-model. The fault prediction sub-model adopts an improved long short-term memory network, and optimizes the network structure by introducing a gating attention mechanism to enhance the ability to capture key time series features. The fault diagnosis sub-model adopts a gradient boosting decision tree algorithm, and improves the accuracy of fault type diagnosis by constructing multiple decision trees and performing ensemble learning.

[0020] In this embodiment, the specific implementation process of the fault tracing module is as follows: based on the fault type diagnosis result, retrieve the historical operating data and historical fault handling records of the device and related devices in the data storage module; analyze the change pattern of the device's operating status parameters before and after the fault occurs, the change of the operating status of related devices, and the change of environmental parameters; combine the structural principle of the device and the common fault cause library, and use the causal reasoning algorithm to trace the root cause of the fault. The root cause includes aging of equipment components, abnormal parameter settings, influence of related devices, influence of environmental factors, or improper operation.

[0021] In this embodiment, the specific implementation process of the early warning classification module is as follows: a fault risk level classification standard is preset, which includes no warning, low risk warning, medium risk warning and high risk warning; based on the fault risk probability output by the fault prediction sub-model, the corresponding early warning level is determined by referring to the preset risk level classification standard; corresponding handling suggestions are generated for different early warning levels: no warning level requires no handling, low risk warning level suggests increasing the monitoring frequency, medium risk warning level suggests arranging planned maintenance, and high risk warning level suggests immediate shutdown for inspection and maintenance.

[0022] In this embodiment, the data storage module adopts a distributed storage architecture, including a real-time database and a historical database. The real-time database is used to store real-time collected multi-source operating data, preprocessed data, and real-time prediction and diagnosis results. The historical database is used to store historical multi-source operating data, historical fault data, historical prediction and diagnosis results, fault tracing results, and early warning processing records. The data storage module supports fast data reading, writing, querying, and backup functions.

[0023] In this embodiment, the human-computer interaction module includes a display unit, an input unit, and an alarm unit. The display unit is used to visually display the equipment operating status, fault prediction results, fault diagnosis results, fault tracing results, warning levels, and handling suggestions. The input unit is used to receive operation instructions input by the user, including parameter settings, data queries, and model update instructions. The alarm unit is used to remind maintenance personnel through sound and light alarms when the warning level is a medium-risk warning or a high-risk warning.

[0024] Although embodiments of the invention have been shown and described (see the detailed description above), it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An AI-based intelligent building equipment fault prediction and diagnosis device, characterized in that: It includes a data acquisition module, a data preprocessing module, a multi-dimensional feature fusion module, an AI prediction and diagnosis core module, a fault tracing module, an early warning and classification module, a data storage module, and a human-computer interaction module; The output end of the data acquisition module is connected to the input end of the data preprocessing module, and is used to collect multi-source operating data of intelligent building equipment and transmit it to the data preprocessing module; The output of the data preprocessing module is connected to the input of the multi-dimensional feature fusion module, and is used to process multi-source running data and then transmit it to the multi-dimensional feature fusion module. The output of the multi-dimensional feature fusion module is connected to the input of the AI ​​prediction and diagnosis core module. It is used to extract multi-dimensional features from multi-source running data and fuse them to obtain a fused feature vector, which is then transmitted to the AI ​​prediction and diagnosis core module. The output of the AI ​​prediction and diagnosis core module is connected to the input of the fault tracing module and the early warning and classification module, respectively. It is used to realize fault risk prediction and fault type diagnosis based on fused feature vectors, and transmit the prediction results to the early warning and classification module and the diagnosis results to the fault tracing module. The output of the fault tracing module is connected to the input of the human-computer interaction module, and is used to trace the root cause of the fault according to the fault type diagnosis result and transmit it to the human-computer interaction module. The output of the early warning classification module is connected to the input of the human-computer interaction module. It is used to classify the early warning level based on the fault risk prediction result and generate processing suggestions, which are then transmitted to the human-computer interaction module. The data storage module is bidirectionally connected to the data preprocessing module, the multi-dimensional feature fusion module, the AI ​​prediction and diagnosis core module, the fault tracing module, and the early warning and classification module, and is used to store the input data, processing data, and output results of each module. The human-computer interaction module is used to receive user operation commands and display fault prediction results, fault diagnosis results, fault tracing results, warning levels, and handling suggestions.

2. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The data acquisition module includes a sensor unit, a device interface unit, and an environmental data acquisition unit. The sensor unit is used to collect the operating status parameters of the device, including voltage, current, temperature, pressure, vibration frequency, and operating time. The device interface unit is used to establish a connection with the control module of the intelligent building equipment through a communication protocol and collect the device's operating log data and fault history data. The environmental data acquisition unit is used to collect the environmental parameters of the device's operating environment, including ambient temperature, humidity, air quality, and dust concentration.

3. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The data preprocessing module processes multi-source operational data in the following ways: First, outlier detection is performed on the collected multi-source operational data to remove outlier data that deviates significantly from the normal data range; then, the moving average method is used to denoise the data and eliminate random noise in the data; finally, all the denoised data is standardized to convert data of different dimensions to the same numerical range to obtain standardized data.

4. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The multi-dimensional feature fusion module extracts multi-dimensional features including equipment self-features, equipment association features, and environmental impact features. The processing procedure of the multi-dimensional feature fusion module is as follows: extracting statistical features, trend features, and abrupt change features of equipment operating status parameters as equipment self-features; determining the strength of the association relationship and the association impact features between different equipment by analyzing the correlation of operating data; analyzing the mapping relationship between environmental parameters and equipment operating status parameters to obtain the environmental impact features of environmental factors on equipment operation; and using a weighted fusion algorithm to fuse equipment self-features, equipment association features, and environmental impact features, assigning weights according to the importance of each feature to fault prediction and diagnosis to obtain a fused feature vector.

5. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The AI ​​prediction and diagnosis core module includes a model training unit and a prediction and diagnosis unit. The model training unit is used to construct an AI hybrid model, which includes a fault prediction sub-model based on deep learning and a fault diagnosis sub-model based on machine learning. The AI ​​hybrid model is trained using historical multi-source operating data and historical fault data to obtain a pre-trained AI hybrid model. The prediction and diagnosis unit is used to input the fused feature vector into the pre-trained AI hybrid model, predict the probability of fault risk of the equipment within a preset time period through the fault prediction sub-model, and diagnose whether the current equipment has a fault and the type of fault through the fault diagnosis sub-model.

6. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 5, characterized in that: The fault prediction sub-model adopts an improved long short-term memory network, which optimizes the network structure by introducing a gating attention mechanism to enhance the ability to capture key time series features; the fault diagnosis sub-model adopts a gradient boosting decision tree algorithm, which improves the accuracy of fault type diagnosis by constructing multiple decision trees and performing ensemble learning.

7. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The specific implementation process of the fault tracing module is as follows: Based on the fault type diagnosis result, retrieve the historical operating data and historical fault handling records of the device and related devices from the data storage module; analyze the change pattern of the device's operating status parameters before and after the fault, the change of the operating status of related devices, and the change of environmental parameters; combine the structural principle of the device and the common fault cause library, and use the causal reasoning algorithm to trace the root cause of the fault. The root cause includes aging of equipment components, abnormal parameter settings, influence of related devices, influence of environmental factors, or improper operation.

8. The AI-based intelligent building equipment fault prediction and diagnosis device according to claim 1, characterized in that: The specific implementation process of the early warning classification module is as follows: a fault risk level classification standard is preset, and the fault risk level includes no early warning, low risk early warning, medium risk early warning and high risk early warning; Based on the failure risk probability output by the failure prediction sub-model, the corresponding early warning level is determined by comparing it with the preset risk level classification standard. Corresponding handling suggestions are generated for different warning levels. No action is required if there is no warning level. For low-risk warning levels, it is recommended to increase the monitoring frequency. For medium-risk warning levels, it is recommended to arrange planned maintenance. For high-risk warning levels, it is recommended to immediately shut down the system for inspection and maintenance.

Images