A fault diagnosis positioning technology

Abstract

The application relates to the technical field of intelligent maintenance, and discloses a fault diagnosis positioning technology, which comprises a data acquisition and alarm module, an integrated analysis and processing module, a multi-source fault positioning module and a three-dimensional fault visualization module. The data acquisition and alarm module is used for collecting equipment operation state data and generating dynamic alarm signals. The integrated analysis and processing module is used for analyzing and processing abnormal states of an identification system. The multi-source fault positioning module is used for analyzing the identified abnormal states, extracting fault features through an optimization algorithm, and positioning components that have faults or are likely to have faults in combination with a deep learning model. The three-dimensional fault visualization module is used for marking the positions and causes of fault components in a model. Through the cooperative work of the data acquisition and alarm module, the integrated analysis and processing module, the multi-source fault positioning module and the three-dimensional fault visualization module, full-process automation of fault diagnosis is realized, and the diagnosis efficiency is greatly improved. Compared with the prior art which depends on manual troubleshooting and simple alarm and is difficult to accurately position, the application effectively solves the problems of low positioning efficiency and slow response of the traditional method.

Description

Technical Field

[0001] This invention relates to the field of intelligent maintenance technology, specifically a fault diagnosis and location technology. Background Technology

[0002] Currently, fault detection mainly relies on manual troubleshooting or simple automatic alarm mechanisms. However, these traditional methods often fail to accurately pinpoint the source of the fault, resulting in low troubleshooting efficiency, long equipment repair times, and high operating costs.

[0003] In traditional methods, when equipment malfunctions, distributed sensors collect operational data, which is then transmitted to a back-end monitoring system for initial screening and aggregation. However, such methods generally lack integrated data analysis capabilities and intelligent location functions, making it difficult to meet the fault diagnosis needs of complex systems.

[0004] Existing fault detection systems are often unable to achieve efficient and accurate location when dealing with complex and ever-changing equipment faults. In particular, for deep-seated hardware and software faults, the limitations of system detection require a lot of manual intervention, which consumes a lot of time and resources to solve the problem. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a fault diagnosis and location technology that solves the problems of existing fault detection systems being unable to accurately locate the source of faults, lacking integrated data analysis capabilities and intelligent location functions, and requiring extensive manual intervention when dealing with complex and ever-changing hardware and software faults, resulting in low troubleshooting efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a fault diagnosis and location technology, comprising:

[0007] The data acquisition and alarm module is used to collect equipment operating status data and generate dynamic alarm signals;

[0008] An integrated analysis and processing module is used to build a nonlinear analysis model based on the equipment's operating status and to identify abnormal states of the system through analysis and processing.

[0009] The multi-source fault location module is used to analyze the identified abnormal states, extract fault features through optimization algorithms, and locate the components that have or may fail by combining deep learning models.

[0010] The 3D fault visualization module is used to generate a 3D model of the equipment based on the fault location results, and to mark the location and cause of the faulty component in the model.

[0011] Preferably, in the data acquisition and alarm module, the equipment operating status data includes temperature, vibration, pressure, and current, and an alarm is triggered by a dynamic threshold, wherein the dynamic threshold is calculated based on the mean and standard deviation of historical operating data.

[0012] Preferably, the integrated analysis and processing module describes the operating state of the equipment by constructing a nonlinear analysis model. The state changes of the nonlinear analysis model are described by a state transition function, and the model includes a fault disturbance term to represent the state shift caused by the fault.

[0013] Preferably, the integrated analysis and processing module analyzes the system state by constructing a positive definite function. When the derivative of the function is negative, the system is in a normal state; when the derivative is positive, the system is judged to be abnormal and enters the fault analysis stage.

[0014] Preferably, the multi-source fault location module uses an optimization algorithm to select fault features. The optimization algorithm improves fault location accuracy while minimizing the number of features through an objective function with regularization constraints.

[0015] Preferably, the multi-source fault location module constructs a graph model based on the correlation between equipment components. The graph model includes nodes and edges, where nodes represent equipment components and edges represent the dynamic coupling relationship between equipment components. The graph model is analyzed using a graph neural network.

[0016] Preferably, the graph neural network models the dynamic coupling relationship of device components through a multi-layer message passing mechanism. The node features of each layer are obtained by aggregating and linearly transforming the features of neighboring nodes, and finally output the failure probability of each device component.

[0017] Preferably, the multi-source fault location module fuses the collected multi-source data and uses a filtering algorithm to filter out random noise and outliers in the multi-source data. The filtering algorithm is dynamically adjusted based on the deviation between the observed value and the predicted state at the current moment to generate a high-precision state estimate.

[0018] Preferably, the 3D fault visualization module generates annotations of faulty components in the 3D model of the equipment based on the fault location results. The annotations highlight the location of the faulty components and provide a textual explanation of the cause of the fault.

[0019] Preferably, the three-dimensional fault visualization module generates the final model of the device by superimposing the annotation matrix with the original three-dimensional model. The annotation matrix contains index information of the faulty components and annotation style parameters, which are used to dynamically adjust the annotation style.

[0020] This invention provides a fault diagnosis and location technology. It has the following beneficial effects:

[0021] 1. This invention utilizes a data acquisition and alarm module, an integrated analysis and processing module, a multi-source fault location module, and a 3D fault visualization module. Through the collaborative work of these modules, the entire process from data acquisition to fault location and visualization is automated, significantly improving the efficiency of fault diagnosis. Compared to existing technologies that rely on manual investigation and simple alarm methods, making it difficult to accurately locate the source of the fault, this invention effectively solves the shortcomings of traditional methods, such as low fault location efficiency and long response time.

[0022] 2. This invention employs a combination of a dynamic threshold judgment mechanism and deep learning algorithms to intelligently analyze and process multi-source data. By dynamically adjusting thresholds to adapt to changes in equipment operating status, it further improves the sensitivity and accuracy of fault detection. Compared to existing technologies that lack integrated data analysis capabilities and rely on fixed threshold judgments leading to frequent false alarms, this invention can more efficiently adapt to the changing operating environments of complex equipment.

[0023] 3. This invention dynamically assesses equipment status and identifies abnormal system states by constructing a nonlinear analysis model and positive definite functions. The multi-level modeling method employed not only captures the complex relationships between equipment state variables but also accurately identifies potential fault risks. Compared to existing technologies that rely heavily on manual intervention to analyze complex hardware and software faults, this invention reduces human involvement and achieves precise analysis and early warning of complex faults.

[0024] 4. This invention utilizes a 3D fault visualization module to intuitively present fault location results through highlighted annotations, dynamic adjustments, and textual descriptions. By overlaying an annotation matrix to generate an intuitive 3D model, it helps maintenance personnel quickly locate faulty components, shortening troubleshooting time. Compared to existing technologies that rely solely on data reports or static graphics, this invention overcomes the shortcomings of existing methods, such as unintuitive information presentation and difficulty in effectively guiding maintenance. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the system architecture of the present invention; Detailed Implementation

[0026] The technical solutions in 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.

[0027] To better understand the present invention, the above content will be described in detail below with reference to specific embodiments.

[0028] Please see the appendix Figure 1 Example 1: This embodiment of the invention provides a fault diagnosis and location technology, including: a data acquisition and alarm module for acquiring equipment operating status data and generating dynamic alarm signals; an integrated analysis and processing module for constructing a nonlinear analysis model based on the equipment operating status and identifying abnormal states of the system through analysis and processing; a multi-source fault location module for analyzing the identified abnormal states, extracting fault features through optimization algorithms, and locating components that have or may fail by combining deep learning models; and a three-dimensional fault visualization module for generating a three-dimensional model of the equipment based on the fault location results and marking the location and cause of the faulty components in the model.

[0029] In this embodiment, the data acquisition and alarm module collects multi-dimensional status data of the equipment in real time by deploying various types of sensors (such as temperature, vibration, pressure, and current sensors) at key parts of the equipment. The module has a built-in dynamic threshold judgment mechanism that dynamically calculates alarm thresholds by analyzing the mean and standard deviation of historical operating data. When the operating status of the equipment deviates from the normal range, the module will automatically trigger an alarm and record abnormal data, providing basic support for subsequent analysis and localization.

[0030] The integrated analysis and processing module constructs a nonlinear analysis model based on the equipment's operational data to describe the dynamic coupling relationships between various components. It models the dynamic changes in equipment operation through state transition functions, introduces fault disturbance terms to identify key state changes that may lead to anomalies, and further analyzes system stability by constructing positive definite functions to accurately determine whether the equipment's operating state is normal. When system instability or abnormal offset is detected, the module transmits the anomaly information to the subsequent positioning module.

[0031] The multi-source fault location module utilizes optimization algorithms to perform feature selection on the collected multi-source data, eliminating redundant information and extracting features highly correlated with the fault. Then, it constructs the dynamic relationships between equipment components using a graph model, where nodes represent equipment components and edges represent dynamic coupling relationships. Combining deep learning technology with the multi-layer message passing mechanism of graph neural networks, the module can effectively analyze the interaction characteristics between components, ultimately locating specific components that may have failed or have already failed.

[0032] The 3D fault visualization module highlights faulty components in the device's 3D model based on the location results. By overlaying a label matrix, it dynamically displays the location, status, and possible causes of the faulty components. The highlighting method is intuitive and easy to understand, and combined with text descriptions, it helps maintenance personnel quickly understand equipment problems, accelerates the troubleshooting and handling process, and improves the overall system maintenance efficiency.

[0033] In the data acquisition and alarm module, the equipment operating status data includes temperature, vibration, pressure, and current. The system uses dynamic thresholds to determine whether to trigger an alarm. These dynamic thresholds are calculated based on the mean and standard deviation of historical operating data.

[0034] In this embodiment, alarm signals are generated by collecting equipment operating status data and combining it with a dynamic threshold judgment mechanism. This provides data support for subsequent integrated analysis and fault location, ensuring that the entire system can promptly identify potential problems and take further action.

[0035] The data acquisition and alarm module relies on a network of multiple distributed sensors deployed at critical locations within the equipment to collect various operational status data. This data includes, but is not limited to, key parameters such as temperature, vibration, pressure, and current. These parameters are acquired in real-time by the sensors, forming a complete multidimensional dataset that describes the current operating status of the equipment. In one possible implementation, these sensors are positioned in critical components or areas prone to failure, such as motor bearings, pipe interfaces, or core components of the control unit.

[0036] Specifically, the collected data is processed through a built-in dynamic threshold determination mechanism. The dynamic threshold is set based on the device's historical operating data, combined with differences in the device's normal operating conditions and workload, and is calculated using statistical methods. The formula for calculating the dynamic threshold is as follows:

[0037] in, Indicates the first The dynamic threshold of each sensor, Indicates sensor Historical average, Indicates its historical standard deviation, This is a sensitivity factor, adjusted according to the specific operating requirements and application scenarios of the equipment. For example, for equipment components operating in high-load environments, The value of can be larger to improve the system's sensitivity.

[0038] In another implementation, the dynamic threshold can be adjusted based on the operating conditions of different devices. For example, for components that operate in a constant state for extended periods, the dynamic threshold may be set more strictly, while for components that frequently start and stop or experience large load variations, the dynamic threshold calculation may be appropriately relaxed to avoid frequent false alarms. Alternatively, the adjustment of the dynamic threshold can also incorporate external environmental parameters, such as the temperature and humidity of the device's operating environment, to further optimize the alarm mechanism.

[0039] In this embodiment, when the collected data value exceeds the dynamic threshold, the system generates an alarm signal and records the alarm information in the fault data log, serving as an important input data source for the subsequent integration and analysis processing module. Specifically, the alarm signal contains various information, such as the time of alarm triggering, the identification of the relevant sensor, the operating parameters that triggered the alarm, and the specific magnitude of exceeding the threshold.

[0040] In some embodiments, to improve the accuracy and real-time performance of data acquisition, the data acquisition and alarm module can also employ multi-channel data fusion technology. In a specific implementation, multi-channel data is processed through timestamp synchronization and signal strength weighted fusion to eliminate abnormal fluctuations caused by signal delay or interference. As a possible implementation, the fused data can effectively reflect the device's operating status, providing a more accurate basis for dynamic threshold calculation and alarm judgment.

[0041] In another possible implementation, the data acquisition and alarm module can use different types of sensors for different equipment component types and operating scenarios. For example, for high-speed rotating equipment, high-precision vibration sensors can be used first; for large electrical equipment, the density of current and temperature sensors can be increased to enhance the comprehensiveness and reliability of data acquisition.

[0042] In summary, the data acquisition and alarm module undertakes the tasks of basic data input and preliminary anomaly detection in the entire fault diagnosis and location technology. Its dynamic threshold judgment mechanism can quickly detect abnormalities in the equipment's operating status and provide efficient and reliable data support for subsequent analysis, processing, and fault location.

[0043] The integrated analysis and processing module describes the equipment's operating state by constructing a nonlinear analysis model. The state changes in the nonlinear analysis model are described by a state transition function, and the model includes a fault disturbance term to represent the state shift caused by a fault. The integrated analysis and processing module analyzes the system state by constructing a positive definite function. When the derivative of this function is negative, the system is in a normal state; when the derivative is positive, the system is judged to be abnormal and enters the fault analysis stage.

[0044] In this embodiment, the integrated analysis and processing module receives multidimensional data from the data acquisition and alarm module, analyzes the dynamic characteristics of the equipment's operating status, and provides processed abnormal data to the multi-source fault location module. The integrated analysis and processing module describes the equipment's operating status by constructing a nonlinear analysis model, evaluates the system's stability and abnormal status using positive definite functions, and outputs possible fault characteristics or offset parameters.

[0045] In this embodiment, a nonlinear analysis model is used to characterize the operating state of the equipment and its changing relationships. This model describes the dynamic behavior of the system through a state transition function, specifically as follows: in, The derivative of the device state variable with respect to time represents the rate of change of the state. This is the current state vector of the device, which contains multiple state variables, such as temperature and vibration frequency. External input signals represent the control variables or external excitations of the equipment; It is a nonlinear state transition function, representing the dynamic changes of the system under normal conditions; This is the fault disturbance term, which reflects the abnormal state shift of the system caused by faults.

[0046] In some embodiments, the state transition function The specific form can be designed according to the operating principle and dynamic characteristics of the equipment. For example, for a vibration system, the state transition function can adopt a second-order nonlinear vibration equation; for a multivariable control system, the function form may involve nonlinear expressions of coupling relationships.

[0047] Specifically, the fault disturbance term It is an important component of nonlinear analysis models, and can be estimated through statistical analysis of historical equipment data or real-time monitoring. In one possible implementation, It can indicate the magnitude of the impact of certain specific failure modes, such as the performance degradation caused by the aging of a certain component.

[0048] The integrated analysis and processing module further evaluates the system state stability using positive definite functions. Generally, positive definite functions are constructed as energy functions of device state variables to measure system stability. A typical form of a positive definite function is:

[0049] in, It is a positive definite function used to describe the energy of the device in its current state; It is a symmetric positive definite matrix, whose elements are used to weight the contributions of different state variables. This represents the transpose of the matrix.

[0050] In one possible implementation, a positive definite function

[0051] The derivative of is used to determine the trend of changes in the system state, and its calculation formula is: in, For time The derivative of is used to describe the rate of change of the system's state energy; State vector The gradient; State vector Regarding time The derivative;

[0052] Specifically, when When the system is in a stable state; when When the system is deemed unstable, the fault analysis phase begins.

[0053] As an alternative, the positive definite function can be adjusted according to the dynamic characteristics of different devices. For example, for devices with high dynamic loads, additional damping terms can be introduced to enhance sensitivity to vibration-related faults. In another possible implementation, the construction of the positive definite function can also be combined with the calculation of the multivariate covariance matrix to improve the ability to describe the coupled variables;

[0054] In some embodiments, the integrated analysis and processing module can also generate a continuous set of system stability assessment data by calculating the value and derivative of the positive definite function in real time. This data can be used to monitor the changing trends of the system's operating state and predict possible instability regions or abnormal states in advance. For example, in high-temperature operating scenarios, when the system state variable (such as temperature) gradually approaches the boundary value of the positive definite function, the module can issue an early warning signal, providing a reference for the multi-source fault location module regarding the probability of a fault.

[0055] As an extended implementation approach, the integrated analysis and processing module can improve the adaptability to system operating conditions by introducing external environmental parameters (such as ambient temperature and humidity) to correct the nonlinear analysis model. For example, in a high-humidity environment, the module can adjust the weights of certain sensitive variables in the state transition function to enhance the accuracy of fault analysis.

[0056] Therefore, the integrated analysis and processing module, through the construction of a nonlinear analysis model and the use of positive definite functions, can effectively identify changes in the system's operating state and abnormal regions, and provide reliable input data for the multi-source fault location module. Its design not only improves the accuracy of system operating state assessment but also provides strong technical support for the early detection and location of equipment faults.

[0057] The multi-source fault location module uses an optimization algorithm to select fault features. The optimization algorithm improves fault location accuracy by minimizing the number of features through an objective function with regularization constraints.

[0058] In this embodiment, the multi-source fault location module receives abnormal state data output by the integration and analysis processing module. It then uses an optimization algorithm to select fault features, minimizing the number of features while improving location accuracy, thereby more effectively locating components in the system that may have failed or have already failed. The optimization algorithm achieves a trade-off in fault feature selection through an objective function, while combining regularization constraints to ensure the robustness and performance of the model.

[0059] The multi-source fault localization module first preprocesses the input data, including data normalization, noise removal, and correlation analysis. Specifically, the normalization step unifies the numerical range of each fault feature to the same scale, thus avoiding imbalances caused by features with larger numerical ranges affecting the objective function of the optimization algorithm. In one possible implementation, the normalization formula is as follows:

[0060] in, Represents the normalized eigenvalues. Represents the original feature values. and These represent the mean and standard deviation of the feature, respectively.

[0061] The optimization algorithm uses an objective function to select fault features. The objective function not only considers fault location accuracy but also penalizes redundant features through a regularization term to reduce model complexity and improve the robustness of fault location. The expression for the objective function is as follows: in, This represents the objective function value of the optimization algorithm; Represents a set of features; This indicates the fault location accuracy based on the current feature set; The weight factor represents the strength of regularization and is used to control model complexity.

[0062] In one possible implementation, the optimization algorithm can employ Bayesian optimization, searching for combinations of different feature subsets to find the optimal subset of fault features under specific conditions. In some embodiments, the optimization process dynamically adjusts the regularization strength to adapt to different data scales and feature dimensions. For example, when the number of features is large, the regularization strength can be appropriately increased. The value of is adjusted to significantly reduce the impact of redundant features.

[0063] Specifically, when the dimensionality of fault features is high, the multi-source fault location module eliminates redundant features through correlation analysis. Alternatively, correlation analysis can be calculated based on mutual information or Pearson correlation coefficients between features.

[0064] For the mutual information method, the formula is as follows: in, Representation of features and target Mutual information; express and Joint probability distribution;

[0065] and They represent and The marginal probability distribution.

[0066] In another possible implementation, the multi-source fault localization module can further enhance its fault feature extraction capabilities by incorporating deep learning techniques. Specifically, a graph neural network (GNN) can be used to model the dynamic coupling relationships between equipment components. Nodes represent equipment components, edges represent the relationships between components, and node features are updated and aggregated through a multi-layer message passing mechanism. For example, for a certain mechanical equipment, node features could be the vibration amplitude of a component, and edge weights could be the coupling stiffness between components. Through multiple iterations, the graph neural network can effectively extract the dynamic characteristics between components and output the fault probability distribution of each component.

[0067] Generally, combining optimization algorithms with deep learning techniques can further improve the overall performance of multi-source fault location modules. As an extended application, the fault feature extraction results of the module can also be input into other analysis tools, such as predictive maintenance models or real-time alarm systems, providing richer data support for the intelligent management of equipment operation.

[0068] Therefore, the multi-source fault location module can not only effectively extract features highly correlated with the fault, but also reasonably constrain the number of features, providing accurate and efficient input data for the subsequent 3D fault visualization module. It plays a crucial role in feature selection and data optimization throughout the entire fault diagnosis process, significantly improving the accuracy and efficiency of fault diagnosis.

[0069] The multi-source fault location module constructs a graph model based on the relationship between equipment components. The graph model includes nodes and edges, where nodes represent equipment components and edges represent the dynamic coupling relationship between equipment components. The graph model is analyzed through a graph neural network.

[0070] Graph neural networks model the dynamic coupling relationships of device components through a multi-layer message passing mechanism. The node features of each layer are obtained by aggregating and linearly transforming the features of neighboring nodes, and finally output the failure probability of each device component.

[0071] In this embodiment, the integrated analysis and processing module provides dynamic characteristic data of the equipment status, while the multi-source fault location module analyzes the complex dynamic relationships between equipment components by further modeling these data. A graph model is constructed based on the relationships between equipment components to reveal hidden fault propagation paths and coupling characteristics between components within the equipment structure. Analysis of the graph model allows for accurate identification of components that may fail in the system, further improving the accuracy of fault location. Specifically, the graph model consists of nodes and edges. Nodes represent the various components of the equipment, while edges describe the dynamic coupling relationships between these components. Dynamic coupling relationships can be defined based on physical connections, signal transmission paths, or functional dependencies. For example, in some mechanical equipment, nodes can represent rotating shafts, bearings, or gears, while edges represent the transmission connections between components. In one possible implementation, the edge weights can also reflect the coupling strength between equipment components, such as transmission efficiency or the proportion of vibration energy transfer.

[0072] Specifically, graph models are analyzed using graph neural networks. Graph neural networks are deep learning models capable of directly processing graph-structured data. They extract complex relationships between equipment components through multi-layered message passing and aggregation of node features. In one possible implementation, node features can include equipment operating parameters such as vibration frequency, temperature changes, and pressure values, while edge features can describe the signal transmission strength or coupling stiffness between components.

[0073] Generally, graph neural networks (GNNs) employ a multi-layer message-passing mechanism to model the dynamic coupling relationships between device components. In each layer, node features are updated and computed by aggregating the features of their neighboring nodes. The update formula for node features is as follows: in, For nodes In the Layer feature representation; For nodes In the Layer feature representation; For nodes The set of neighboring nodes; This is an aggregation function used to aggregate features of neighboring nodes, such as summation or averaging. For the first Layer weight matrix; For the first Layer bias vector; For activation functions;

[0074] In one possible implementation, the aggregation function can choose between weighted averaging or max pooling to adapt to different data distribution characteristics. For example, in a vibration signal propagation scenario, weighted averaging can better reflect the actual energy distribution between nodes.

[0075] In some embodiments, the final output of the graph neural network is a feature representation of each node, which can be further converted into a failure probability distribution of the nodes. Specifically, by applying a classifier to the output node features, the probability of failure for each component can be predicted. For example, for a given node... The probability of failure can be expressed as: in, For nodes The probability of failure; For nodes Features in the last layer of a graph neural network; and Let's define the weights and biases of the classifier.

[0076] Alternatively, the classifier can employ a simple linear layer or introduce more nonlinear structures to improve the accuracy of fault classification. In some embodiments, historical data of nodes can also be incorporated as additional feature inputs to enhance the ability to identify long-term fault patterns.

[0077] In another possible implementation, the edge features of a graph neural network can also be updated dynamically. The update mechanism for edge features is similar to that for node features, capturing changes in the relationships between nodes through feature aggregation from neighboring nodes. This mechanism is particularly suitable for fault propagation scenarios, such as the cascading vibration effects caused by bearing failure in a mechanical system.

[0078] Generally, combining graphical models and graphical neural networks can effectively improve the accuracy of multi-source fault location. As an extension technique, the fault location results can also be compared and verified with the data output by the integrated analysis and processing module, thereby improving the overall reliability of the system. Thus, the multi-source fault location module can not only effectively identify potential faults in complex equipment, but also provide high-quality input data for the subsequent 3D fault visualization module.

[0079] The multi-source fault location module fuses the collected multi-source data and uses a filtering algorithm to filter out random noise and outliers in the multi-source data. The filtering algorithm is dynamically adjusted based on the deviation between the observed value and the predicted state at the current moment to generate a state estimate.

[0080] In this embodiment, the multi-source fault location module utilizes multi-source data provided by the data acquisition and alarm module, combined with the results of the integrated analysis and processing module, to further analyze and process the equipment status. Generally, multi-source data includes observations from multiple sensors, which may exhibit data deviations and outliers due to environmental noise, sensor malfunctions, or signal interference. To ensure data accuracy and reliability, the multi-source fault location module introduces a filtering algorithm to effectively filter out random noise and outliers in the data through dynamic adjustment, ultimately generating a high-precision status estimate, providing reliable data support for subsequent fault diagnosis.

[0081] The core of multi-source data fusion lies in the design and application of filtering algorithms. Filtering algorithms dynamically adjust the data by combining the current observations with the system's predicted state. Specifically, the filtering algorithm employs a Kalman filter-based implementation, and its state estimate is updated using the following formula: in, Indicates the current time State estimation; This represents the state estimate predicted at the previous moment; This represents the observation value at the current moment; This represents the observation matrix, used to map state variables to the observation space; This represents the filter gain, used to dynamically adjust the impact of observations on the estimation.

[0082] In general, filter gain

[0083] The calculation formula is: in, The state prediction covariance matrix is ​​the covariance matrix of the state estimation error obtained in the prediction step, describing the magnitude of the prediction error. Its dimension is the same as the state vector, reflecting the uncertainty of the state prediction. Represents the observation matrix The transpose of is used to reflect the observations in the state variable space. This represents the observation noise covariance matrix, used to describe the distribution characteristics of random noise during the observation process, and is usually determined by the sensor's measurement noise. Its dimension is the same as the observed value. This represents the inverse of the weighted sum of the covariance matrices. This term includes the prediction covariance (i.e., the uncertainty of state prediction) and the observation noise covariance (i.e., the uncertainty of measurement error). The weighting ratio is obtained through the inversion operation and is used to dynamically adjust the influence of observations on state estimation.

[0084] In one possible implementation, the filtering algorithm adjusts the covariance matrix... and The value can be adjusted to adapt to different data noise levels and equipment operating environments. For example, when the sensor signal has high noise, the value can be appropriately increased. The value of this can be reduced, thereby decreasing the impact of observations on state estimation; and in the case of low noise, it can be reduced The value is used to enhance the weight of the observation.

[0085] Specifically, filtering algorithms can not only eliminate random noise, but also detect and remove outliers. For example, in a high-load mechanical system, if the output value of a sensor suddenly exceeds the normal range, the filtering algorithm can identify the abnormal signal based on the deviation between the predicted value and the observed value of the current state, and dynamically adjust the state estimate to avoid the impact of outliers on system analysis.

[0086] In some embodiments, to improve the efficiency of the filtering algorithm, multi-source data fusion can be implemented using distributed computing. Specifically, multi-source data can be divided into different groups according to their source and type, each group is processed independently using a filtering algorithm, and the results of the group processing are finally aggregated to generate an overall high-precision state estimate. In another possible implementation, the filtering algorithm can also incorporate an adaptive parameter adjustment strategy, dynamically optimizing the filtering parameters based on real-time acquired data to further improve the robustness of data processing.

[0087] Alternatively, filtering algorithms can be combined with deep learning models to preprocess anomalous patterns in multi-source data. For example, by training on historical data, deep learning models can identify potential anomalous signal features and use these features to guide the parameter adjustment of the filtering algorithm. This combination can further improve the accuracy and reliability of multi-source data fusion.

[0088] In another possible implementation, the filtering algorithm can also be applied to dynamic weight allocation to quantify the reliability of different data sources. For example, higher weights can be assigned to certain critical sensors, thus increasing their impact on the state estimation results; while for noisy data sources, their weights can be reduced to decrease interference with the overall estimation.

[0089] Through the aforementioned technologies, the multi-source fault location module can efficiently fuse multi-source data, filtering out noise and outliers, thereby providing strong support for accurate estimation of equipment operating status. The flexibility and dynamic adjustment capability of the filtering algorithm ensure that the module can adapt to complex and ever-changing operating environments, providing accurate data input for subsequent fault location and 3D visualization.

[0090] The 3D fault visualization module generates annotations of faulty components in the 3D model of the equipment based on the fault location results. The annotations highlight the location of the faulty components and provide a textual explanation of the cause of the fault.

[0091] The 3D fault visualization module generates the final model of the device by overlaying the annotation matrix with the original 3D model. The annotation matrix contains index information of the faulty components and annotation style parameters, which are used to dynamically adjust the annotation style.

[0092] In this embodiment, the 3D fault visualization module utilizes the fault location results provided by the multi-source fault location module to process and annotate the 3D model of the equipment. Generally, this module displays the diagnostic results by generating annotations of faulty components in the 3D model, intuitively presenting fault information to maintenance personnel. The annotations include the location of the faulty component and related fault causes, and are visually enhanced through highlighting, color changes, or animation effects to help maintenance personnel quickly identify and address problematic components.

[0093] Specifically, it processes data based on the original equipment's 3D model. The 3D model is typically generated from the equipment's CAD data, representing the various components and their geometric relationships. Specifically, this module first uses the component index information provided in the fault location results to locate the corresponding component in the 3D model and generate an annotation matrix. The annotation matrix structure includes the faulty component's index information and annotation style parameters, used to define the specific annotation method, such as highlight color, transparency variation, or flashing frequency.

[0094] In one possible implementation, the label matrix can be constructed using the following formula:

[0095] in, This represents the final generated 3D model; A three-dimensional model representing the original equipment; Indicates the brightness intensity factor; This represents the annotation matrix, which contains the index information and style parameters of the faulty components;

[0096] High brightness intensity factor The value is dynamically adjusted based on the severity of the fault. For example, for critical components or high-priority faults, the value can be adjusted accordingly. Set a larger value to enhance the visual effect; while for low-priority faults, a smaller value can be used to reduce visual interference.

[0097] Specifically, annotation style parameters can include not only highlight colors but also transparency and animation effects. For example, in some embodiments, when a serious fault is detected in a component of the device, the component will be highlighted in red in the 3D model with a flashing effect to alert maintenance personnel. In another possible implementation, different transparency settings can be used to display the internal structure of the component, helping maintenance personnel to more intuitively understand the location of the fault and its relationship with surrounding components.

[0098] As an alternative, the 3D fault visualization module can also provide text descriptions related to the annotations. These descriptions typically include the name of the faulty component, the cause of the fault, and recommended corrective actions. For example, in a 3D model of electrical equipment, when an overload problem is detected on a circuit board, the text description might read, "Component: Circuit Board A, Cause of Fault: Overload, Recommended Action: Check circuit connections and replace fuse."

[0099] In another possible implementation, the annotation matrix can also be dynamically linked to the device's operational status data, enabling real-time visualization updates. For example, when the status parameters of a component change, the annotation style in the 3D model will also be updated synchronously. Specifically, the color intensity or flashing frequency of the annotations can be adjusted by monitoring real-time changes in parameter values. For instance, when the temperature of a component gradually rises and approaches a warning value, its color can gradually transition from green to red, indicating an increasing risk of failure.

[0100] In some embodiments, the 3D fault visualization module can also support interactive functions, such as allowing maintenance personnel to click on highlighted faulty components to view their detailed information or historical data. As an extended function, interactive operations can also trigger linkages with other systems, such as sending maintenance tasks to a work order management system or requesting more detailed diagnostic reports.

[0101] Through the methods described above, the 3D fault visualization module can intuitively display fault information and provide maintenance personnel with accurate fault location and cause analysis. The design of the annotation matrix and its dynamic adjustment function ensure the model's adaptability and practicality in complex operating environments, providing strong support for the rapid troubleshooting and repair of equipment faults. Example

[0102] In this embodiment, by enhancing data fusion and analysis capabilities, an artificial intelligence engine combined with a big data platform is used to construct a real-time data stream intelligent parsing and pattern mining method. This method is used to identify potential equipment risks in advance and proactively intervene to prevent escalation of faults. The integrated analysis and processing module is deeply integrated with an enterprise-level big data warehouse in this embodiment. Deep learning models are used to extract features and learn association rules from massive amounts of monitoring data, ultimately achieving the prediction and proactive intervention of potential faults.

[0103] Typically, equipment generates a large amount of monitoring data during operation, including status parameters such as temperature, pressure, vibration, and power consumption. This data is collected by various sensors and uploaded to the integration and analysis processing module via the data acquisition and alarm module. In this embodiment, the integration and analysis processing module connects to an enterprise-level big data warehouse to capture this monitoring data in real time and performs intelligent analysis using deep learning algorithms.

[0104] Specifically, the integrated analysis and processing module includes the following key steps:

[0105] Real-time data stream access and processing;

[0106] Feature extraction and pattern mining in deep neural networks;

[0107] Prediction and early warning of potential failures.

[0108] Real-time data stream access and processing

[0109] In this embodiment, the integrated analysis and processing module captures real-time data streams by connecting to an enterprise-level big data warehouse. Data processing involves the following steps:

[0110] 1. Data stream access:

[0111] Data is uploaded from multiple sensor sources to an enterprise-level big data warehouse in the form of message queues. As one possible implementation, the message queues utilize high-throughput message middleware such as Kafka to deliver high-frequency monitoring data, ensuring real-time performance. The data captured in real time includes device status parameters, historical operating records, and external environmental conditions.

[0112] 2. Data preprocessing:

[0113] To improve the efficiency of data analysis, raw monitoring data needs to be standardized and cleaned. Data cleaning includes removing missing values, eliminating outlier data points, and standardizing data formats.

[0114] Feature extraction and pattern mining of deep neural networks

[0115] The integrated analysis and processing module utilizes deep neural networks to extract features and mine patterns from the processed monitoring data. The deep neural network structure includes an input layer, multiple hidden layers, and an output layer, used to extract features and output prediction results. .

[0116] 1. Data Input:

[0117] in, Indicates time The monitoring data at any given time includes various status parameters.

[0118] 2. Feature extraction:

[0119] Deep neural networks extract features from data through multiple hidden layers. Alternatively, the hidden layers can employ convolutional neural networks (CNNs) to extract local features from the data, or long short-term memory networks (LSTMs) to capture the dynamic changes in time series data. The feature update formula for the hidden layers is: in, For the first Feature representation of hidden layers; and These are the weight matrix and the bias vector, respectively; This is an activation function, such as the ReLU function.

[0120] 3. Pattern mining:

[0121] High-dimensional features extracted by deep neural networks are input into an association rule mining module to identify relationships between key features. For example, a trained model can be used to identify the correlation between certain features (such as hard drive read / write latency and temperature rise) and specific failure events. The mathematical expression of association rules is as follows: in, The confidence level of the association rule; Indicates simultaneous occurrence and The proportion of data; Indicates an event The proportion of occurrence.

[0122] Prediction and early warning of potential failures

[0123] In this embodiment, the analysis results of a deep learning model are used to predict potential faults and proactively intervene.

[0124] 1. Fault prediction:

[0125] When the characteristic pattern matches historical failure patterns with a high degree of similarity, the system generates a fault warning signal. For example, if server hard drive read / write latency and temperature rise are identified as critical patterns before a system crash, the system can issue an alert several hours in advance. The output formula for the prediction result is: in, This is the result vector of fault prediction; This is the output of the last hidden layer; and These are the weights and biases for the output layer.

[0126] 2. Proactive intervention:

[0127] When the system detects a potential fault, it can trigger automated actions to prevent escalation. For example, if a hard drive is detected to be at risk of failure, the system can automatically migrate data to a spare hard drive and notify the administrator to replace the device.

[0128] In some embodiments, the integrated analysis and processing module can also incorporate real-time external environmental data (such as temperature, humidity, and power fluctuations) to correct the prediction results. For example, in high-temperature operating environments, the normal read / write latency range of the hard drive may need to be dynamically adjusted to avoid misjudgments.

[0129] In another possible implementation, the system can continuously optimize the deep learning model through an adaptive training mechanism. For example, the system can use newly collected monitoring data to update the model weights online, thereby improving the accuracy of fault prediction.

[0130] In this embodiment, the integrated analysis and processing module can process massive amounts of data in real time, automatically identify potential equipment failure modes, and take early intervention measures to avoid system downtime and equipment damage. Simultaneously, by combining deep neural networks and association rule mining techniques, it achieves comprehensive analysis and accurate prediction of equipment status, providing strong technical support for intelligent operation and maintenance.

[0131] Although embodiments of the invention have been shown and described, 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. A fault diagnosis and location technology, characterized in that, include: The data acquisition and alarm module is used to collect equipment operating status data and generate dynamic alarm signals; An integrated analysis and processing module is used to build a nonlinear analysis model based on the equipment's operating status and to identify abnormal states of the system through analysis and processing. The multi-source fault location module is used to analyze the identified abnormal states, extract fault features through optimization algorithms, and locate the components that have or may fail by combining deep learning models. The 3D fault visualization module is used to generate a 3D model of the equipment based on the fault location results, and to mark the location and cause of the faulty component in the model; The integrated analysis and processing module describes the operating state of the equipment by constructing a nonlinear analysis model. The state changes of the nonlinear analysis model are described by a state transition function, and the model includes a fault disturbance term to represent the state shift caused by the fault. The integrated analysis and processing module analyzes the system state by constructing a positive definite function. When the derivative of the function is negative, the system is in a normal state. When the derivative is positive, the system is judged to be abnormal and enters the fault analysis stage. The multi-source fault location module uses an optimization algorithm to select fault features. The optimization algorithm improves fault location accuracy while minimizing the number of features through an objective function with regularization constraints. The expression for the objective function is: ,in, This represents the objective function value of the optimization algorithm. Represents the feature set, This indicates the fault location accuracy based on the current feature set. The weight factor represents the strength of regularization and is used to control model complexity.

2. The fault diagnosis and location technology according to claim 1, characterized in that, In the data acquisition and alarm module, the equipment operating status data includes temperature, vibration, pressure, and current, and an alarm is triggered by a dynamic threshold, which is calculated based on the mean and standard deviation of historical operating data.

3. The fault diagnosis and location technology according to claim 1, characterized in that, The multi-source fault location module constructs a graph model based on the correlation between equipment components. The graph model includes nodes and edges, where nodes represent equipment components and edges represent the dynamic coupling relationship between equipment components. The graph model is analyzed using a graph neural network.

4. The fault diagnosis and location technology according to claim 3, characterized in that, The graph neural network models the dynamic coupling relationship of device components through a multi-layer message passing mechanism. The node features of each layer are obtained by aggregating and linearly transforming the features of neighboring nodes, and finally outputs the failure probability of each device component.

5. The fault diagnosis and location technology according to claim 1, characterized in that, The multi-source fault location module fuses the collected multi-source data and uses a filtering algorithm to filter out random noise and outliers in the multi-source data. The filtering algorithm is dynamically adjusted based on the deviation between the observed value and the predicted state at the current moment to generate a high-precision state estimate.

6. The fault diagnosis and location technology according to claim 1, characterized in that, The 3D fault visualization module generates annotations of faulty components in the 3D model of the equipment based on the fault location results. The annotations highlight the location of the faulty components and provide a textual explanation of the cause of the fault.

7. The fault diagnosis and location technology according to claim 6, characterized in that, The three-dimensional fault visualization module generates the final model of the device by superimposing the annotation matrix with the original three-dimensional model. The annotation matrix contains index information of the faulty components and annotation style parameters, which are used to dynamically adjust the annotation style.

Images