Automatic equipment fault self-detection method for industrial internet

Abstract

The invention belongs to the technical field of automation equipment management, and discloses an industrial internet automation equipment fault self-detection method comprising the following steps: S1, data acquisition; s2, data preprocessing; s3, feature extraction; s4, constructing a fault diagnosis model; s5, performing fault diagnosis; according to the method, key links of data acquisition, preprocessing, feature extraction, fault diagnosis model construction, fault diagnosis, model optimization and the like are integrated to form a set of complete automatic fault detection and diagnosis system, equipment operation state data are acquired, feature parameters capable of reflecting equipment faults are extracted after preprocessing, and the fault diagnosis model is optimized. According to the method, the fault diagnosis model is constructed, the occurrence time, the category, the position and the maintenance suggestion of the equipment fault can be accurately diagnosed, meanwhile, the diagnosis precision and efficiency are continuously improved through a continuous optimization mechanism, the operation stability and reliability of the automation equipment in the industrial internet environment are effectively improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of automated equipment management technology, and more specifically to a method for self-detection of faults in automated equipment in the industrial internet. Background Technology

[0002] In the context of the rapid development of the Industrial Internet, automated equipment has become a core element of modern industrial production systems, widely used in key sectors such as manufacturing, energy, and transportation. These devices, with their high efficiency, precision, and continuous operation, have significantly improved production efficiency, reduced labor costs, and become a vital force driving industrial transformation and upgrading. However, with the increasing complexity of equipment structures and the diversification of operating environments, the risk of failure faced by automated equipment during operation has also increased.

[0003] Traditional methods for detecting faults in automated equipment primarily rely on manual inspections and periodic maintenance. While manual inspections can detect some obvious signs of faults, limitations in personnel skills, experience, and inspection frequency often prevent them from comprehensively and promptly identifying all potential faults. This is especially true for faults that are difficult to detect initially or have subtle symptoms, making manual inspections prone to missed detections or misdiagnoses. Periodic maintenance, while capable of preventing faults to some extent, often suffers from over-maintenance or under-maintenance. Over-maintenance increases unnecessary downtime and maintenance costs, while under-maintenance can lead to sudden equipment failures during operation, causing production interruptions or even safety accidents.

[0004] Furthermore, traditional fault detection methods face the challenge of insufficient data utilization. During equipment operation, a large amount of real-time data is generated, containing crucial information about the equipment's status. However, traditional methods often simply record this data without in-depth analysis and mining. This results in a significant waste of valuable data, failing to provide effective support for fault detection.

[0005] With the rapid development of IoT, big data, and AI technologies, data-driven fault detection methods have gradually become a research hotspot. These methods collect real-time data during equipment operation and utilize advanced data analysis and machine learning algorithms to monitor and diagnose equipment status in real time. Compared with traditional fault detection methods, data-driven methods have higher accuracy and efficiency, enabling timely detection of potential faults and providing early warnings, thereby effectively preventing production interruptions and safety accidents.

[0006] However, existing data-driven fault detection methods still have some shortcomings. For example, data acquisition may be subject to various noise interferences, leading to a decline in data quality; certain key features may be overlooked during feature extraction, affecting the accuracy of diagnosis; and fault diagnosis models may lack continuous optimization mechanisms, making it difficult to adapt to changes in equipment operating conditions. Summary of the Invention

[0007] To address the aforementioned problems, this invention provides a self-detection method for automated equipment faults in the Industrial Internet, which is achieved through the following technical solutions.

[0008] A self-detection method for automated equipment faults in the Industrial Internet includes the following steps:

[0009] S1, Data Acquisition, Acquiring real-time operating status data of automated equipment during operation;

[0010] S2, data preprocessing, removes outliers and noise interference from the collected real-time running status data to improve data quality;

[0011] S3, Feature Extraction: Extracting feature parameters that can reflect the faults of automated equipment from the preprocessed real-time operating status data.

[0012] S4, Fault Diagnosis Model Construction: Based on machine learning algorithms and expert knowledge base, a fault diagnosis model is constructed, and the model is trained with historical fault data and normal data.

[0013] S5, Fault Diagnosis, takes the extracted feature parameters as the input of the model and the equipment fault diagnosis results as the output. The fault diagnosis results include the fault occurrence time, fault type, fault location, and fault repair suggestions.

[0014] S6, Fault Diagnosis Model Optimization: Based on the deviation between the diagnostic results output by the model and the actual troubleshooting results, the model parameters are continuously optimized.

[0015] As a further embodiment of the present invention, in step S1, real-time operating status data is collected through a deployed sensor network, which includes, but is not limited to, vibration sensors, temperature sensors, pressure sensors, and current sensors.

[0016] As a further aspect of the present invention, in step S2, data preprocessing specifically includes:

[0017] Data cleaning: Check for missing or outlier values ​​in the real-time running status data. For missing values, fill them with the mean, median, or interpolation. For outliers, judge and process them according to the data distribution characteristics and business logic, and delete, correct, or retain and mark them.

[0018] Data denoising: Filtering algorithms are used to denoise the data. Wavelet transform filtering algorithm is used for vibration data, and moving average filtering algorithm is used for temperature, pressure and current data.

[0019] Data normalization: Normalize the cleaned and denoised data to map it to a uniform range in order to eliminate the influence of different units on data analysis.

[0020] As a further embodiment of the present invention, in step S3, the characteristic parameters include, but are not limited to, abnormal power supply characteristics, abnormal gas supply characteristics, sensor position offset characteristics, abnormal sensor sensitivity characteristics, relay grounding adhesion characteristics, abnormal pressure regulation characteristics of flow limiter and pressure reducing valve, electrical circuit open circuit characteristics, air pipe damage characteristics, and abnormal solenoid valve reversing characteristics.

[0021] As a further embodiment of the present invention, in step S4, the prepared historical fault data and normal data are divided into a training set and a test set in a 3:1 ratio. The selected machine learning algorithm is trained using the training set, and the trained model is evaluated using the test set.

[0022] As a further embodiment of the present invention, in step S4, a fault diagnosis model is constructed based on a hybrid model of deep learning convolutional neural network and long short-term memory network.

[0023] As a further aspect of the present invention, the specific method for optimizing the fault diagnosis model in step S6 is as follows:

[0024] Deviation assessment: In practical applications, the diagnostic results output by the model are compared with the fault results obtained through manual troubleshooting, and the deviation between the two is calculated.

[0025] Parameter adjustment: Adjust the parameters of the fault diagnosis model based on the results of the deviation assessment;

[0026] Continuous iteration: Establish a continuous optimization mechanism for the fault diagnosis model, regularly collect new historical fault data and normal data, and retrain and evaluate the model.

[0027] The beneficial effects of the present invention are as follows:

[0028] 1. By deploying a sensor network to collect real-time equipment operating status data, and after data cleaning, noise reduction and normalization, outliers and noise interference in the data are effectively removed, improving data quality and providing a reliable data foundation for subsequent feature extraction and model training;

[0029] 2. Extract multi-dimensional feature parameters that can reflect equipment faults from the preprocessed data, such as power supply abnormality features, gas source abnormality features, sensor position offset features, etc., to ensure the comprehensiveness and pertinence of feature extraction and improve the accuracy of fault diagnosis.

[0030] 3. Based on machine learning algorithms and expert knowledge base, a fault diagnosis model is built. Through training with historical fault data and normal data, it can accurately diagnose the occurrence time, type, location and maintenance suggestions of equipment faults, thereby improving the efficiency and reliability of fault diagnosis.

[0031] 4. Establish a continuous optimization mechanism for the fault diagnosis model. Through deviation assessment, parameter adjustment and continuous iteration, new data is collected regularly to retrain and evaluate the model, ensuring that the model can be continuously optimized as the equipment operating status changes, thereby improving the model's adaptability and diagnostic accuracy.

[0032] 5. This invention enables self-detection of faults in automated equipment, reducing the need for manual inspections and periodic maintenance, and lowering maintenance costs. Simultaneously, by promptly identifying potential faults, it prevents production interruptions and safety accidents caused by escalating faults, further reducing operational risks for enterprises.

[0033] 6. By monitoring and diagnosing equipment faults in real time, this invention ensures that equipment operates in optimal condition, improving production efficiency. Simultaneously, accurate fault diagnosis and repair recommendations help to quickly restore equipment operation, reducing downtime and enhancing overall production efficiency. Attached Figure Description

[0034] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 The flowchart of an automated equipment fault self-detection method for the Industrial Internet described in this invention. Detailed Implementation

[0036] 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.

[0037] like Figure 1As shown, an automated equipment fault self-detection method for the Industrial Internet includes the following steps:

[0038] S1, Data Acquisition, Acquiring real-time operating status data of automated equipment during operation;

[0039] S2, data preprocessing, removes outliers and noise interference from the collected real-time running status data to improve data quality;

[0040] S3, Feature Extraction: Extracting feature parameters that can reflect the faults of automated equipment from the preprocessed real-time operating status data.

[0041] S4, Fault Diagnosis Model Construction: Based on machine learning algorithms and expert knowledge base, a fault diagnosis model is constructed, and the model is trained with historical fault data and normal data.

[0042] S5, Fault Diagnosis, takes the extracted feature parameters as the input of the model and the equipment fault diagnosis results as the output. The fault diagnosis results include the fault occurrence time, fault type, fault location, and fault repair suggestions.

[0043] S6, Fault Diagnosis Model Optimization: Based on the deviation between the diagnostic results output by the model and the actual troubleshooting results, the model parameters are continuously optimized.

[0044] By integrating key processes such as data acquisition, preprocessing, feature extraction, fault diagnosis model construction, fault diagnosis, and model optimization, a complete automated fault detection and diagnosis system has been formed.

[0045] Its core lies in using sensor networks to collect equipment operating status data in real time, and after cleaning, noise reduction and normalization, extracting feature parameters that can reflect equipment faults, and building a fault diagnosis model based on machine learning algorithms and expert knowledge base.

[0046] This model, trained using historical fault and normal data, can accurately diagnose the time, type, location, and maintenance recommendations of equipment faults. At the same time, it continuously improves diagnostic accuracy and efficiency through a continuous optimization mechanism.

[0047] This method effectively improves the stability and reliability of automated equipment operation in the industrial internet environment and reduces maintenance costs.

[0048] Preferably, in step S1, real-time operating status data is collected through a deployed sensor network, which includes, but is not limited to, vibration sensors, temperature sensors, pressure sensors, and current sensors.

[0049] Step S1 is further refined, explicitly stating that real-time operational status data is collected through a deployed sensor network, and listing the types of components in the sensor network, including vibration sensors, temperature sensors, pressure sensors, and current sensors.

[0050] This requirement emphasizes the core role of sensor networks in data acquisition, ensuring the diversity and real-time nature of data sources, and providing a rich data foundation for subsequent data processing and analysis.

[0051] Preferably, in step S2, data preprocessing specifically includes:

[0052] Data cleaning: Check for missing or outlier values ​​in the real-time running status data. For missing values, fill them with the mean, median, or interpolation. For outliers, judge and process them according to the data distribution characteristics and business logic, and delete, correct, or retain and mark them.

[0053] Data denoising: Filtering algorithms are used to denoise the data. Wavelet transform filtering algorithm is used for vibration data, and moving average filtering algorithm is used for temperature, pressure and current data.

[0054] Data normalization: Normalize the cleaned and denoised data to map it to a uniform range in order to eliminate the influence of different units on data analysis.

[0055] Data cleaning is used to handle missing and outlier values, ensuring data integrity and accuracy; data denoising uses filtering algorithms to eliminate noise interference in the data, improving data quality; data normalization maps data to a unified range, eliminating the influence of different units on data analysis.

[0056] This requirement ensures the purity and consistency of the preprocessed data, providing reliable data support for subsequent feature extraction and model training.

[0057] Preferably, in step S3, the characteristic parameters include, but are not limited to, abnormal power supply characteristics, abnormal gas supply characteristics, sensor position offset characteristics, abnormal sensor sensitivity characteristics, relay grounding adhesion characteristics, abnormal pressure regulation characteristics of flow limiter and pressure reducing valve, electrical circuit open circuit characteristics, air pipe damage characteristics, and abnormal solenoid valve reversing characteristics.

[0058] The document lists characteristic parameters that can reflect faults in automated equipment, such as abnormal power supply characteristics, abnormal air supply characteristics, and sensor position offset characteristics.

[0059] The selection of these characteristic parameters is based on practical experience in equipment operation and expert knowledge, and can comprehensively and accurately reflect the fault status of the equipment.

[0060] This requirement ensures the relevance and effectiveness of feature extraction, providing crucial input for the subsequent construction of fault diagnosis models.

[0061] Preferably, in step S4, the prepared historical fault data and normal data are divided into a training set and a test set in a 3:1 ratio. The selected machine learning algorithm is trained using the training set, and the trained model is evaluated using the test set.

[0062] By rationally dividing the dataset, using the training set to train the machine learning algorithm, and using the test set to evaluate the trained model, the generalization ability and diagnostic accuracy of the model are ensured.

[0063] This requirement provides a scientific method and process for constructing fault diagnosis models, improving the reliability and practicality of the models.

[0064] Preferably, in step S4, a fault diagnosis model is constructed based on a hybrid model of deep learning convolutional neural network and long short-term memory network.

[0065] This hybrid model combines the advantages of convolutional neural networks in feature extraction with the capabilities of long short-term memory networks in time series data processing, enabling it to better handle complex data patterns in fault diagnosis of automated equipment.

[0066] This requirement improves the fault diagnosis model's ability to adapt to complex data and its diagnostic accuracy.

[0067] Preferably, in step S6, the specific method for optimizing the fault diagnosis model is as follows:

[0068] Deviation assessment: In practical applications, the diagnostic results output by the model are compared with the fault results obtained through manual troubleshooting, and the deviation between the two is calculated.

[0069] Parameter adjustment: Adjust the parameters of the fault diagnosis model based on the results of the deviation assessment;

[0070] Continuous iteration: Establish a continuous optimization mechanism for the fault diagnosis model, regularly collect new historical fault data and normal data, and retrain and evaluate the model.

[0071] Deviation assessment calculates the deviation by comparing the model output with the actual investigation results; parameter adjustment adjusts the model parameters based on the deviation assessment results; continuous iteration establishes a continuous model optimization mechanism, regularly collecting new data to retrain and evaluate the model.

[0072] This requirement ensures that the fault diagnosis model can be continuously optimized and its diagnostic performance improved as new data accumulates.

[0073] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for self-detection of faults in automated equipment in the Industrial Internet, characterized in that, Includes the following steps: S1, Data Acquisition, Acquiring real-time operating status data of automated equipment during operation; S2, data preprocessing, removes outliers and noise interference from the collected real-time running status data to improve data quality; S3, Feature Extraction: Extracting feature parameters that can reflect the faults of automated equipment from the preprocessed real-time operating status data. S4, Fault Diagnosis Model Construction: Based on machine learning algorithms and expert knowledge base, a fault diagnosis model is constructed, and the model is trained with historical fault data and normal data. S5, Fault Diagnosis, takes the extracted feature parameters as the input of the model and the equipment fault diagnosis results as the output. The fault diagnosis results include the fault occurrence time, fault type, fault location, and fault repair suggestions. S6, Fault Diagnosis Model Optimization: Based on the deviation between the diagnostic results output by the model and the actual troubleshooting results, the model parameters are continuously optimized.

2. The self-detection method for automated equipment faults in the Industrial Internet according to claim 1, characterized in that, In step S1, real-time operating status data is collected through a deployed sensor network, which includes, but is not limited to, vibration sensors, temperature sensors, pressure sensors, and current sensors.

3. The self-detection method for automated equipment faults in the Industrial Internet according to claim 2, characterized in that, In step S2, data preprocessing specifically includes: Data cleaning: Check for missing or outlier values ​​in the real-time running status data. For missing values, fill them with the mean, median, or interpolation. For outliers, judge and process them according to the data distribution characteristics and business logic, and delete, correct, or retain and mark them. Data denoising: Filtering algorithms are used to denoise the data. Wavelet transform filtering algorithm is used for vibration data, and moving average filtering algorithm is used for temperature, pressure and current data. Data normalization: Normalize the cleaned and denoised data to map it to a uniform range in order to eliminate the influence of different units on data analysis.

4. The self-detection method for automated equipment faults in the Industrial Internet according to claim 1, characterized in that, In step S3, the characteristic parameters include, but are not limited to, abnormal power supply characteristics, abnormal gas supply characteristics, abnormal sensor position offset characteristics, abnormal sensor sensitivity characteristics, abnormal relay grounding characteristics, abnormal pressure regulation characteristics of flow limiter and pressure reducing valve, electrical circuit open circuit characteristics, air pipe damage characteristics, and abnormal solenoid valve reversing characteristics.

5. The self-detection method for automated equipment faults in the Industrial Internet according to claim 1, characterized in that, In step S4, the prepared historical fault data and normal data are divided into a training set and a test set in a 3:1 ratio. The selected machine learning algorithm is trained using the training set, and the trained model is evaluated using the test set.

6. The self-detection method for automated equipment faults in the Industrial Internet according to claim 1, characterized in that, In step S4, a fault diagnosis model is constructed based on a hybrid model of deep learning convolutional neural network and long short-term memory network.

7. The self-detection method for automated equipment faults in the Industrial Internet according to claim 1, characterized in that, In step S6, the specific method for optimizing the fault diagnosis model is as follows: Deviation assessment: In practical applications, the diagnostic results output by the model are compared with the fault results obtained through manual troubleshooting, and the deviation between the two is calculated. Parameter adjustment: Adjust the parameters of the fault diagnosis model based on the results of the deviation assessment; Continuous iteration: Establish a continuous optimization mechanism for the fault diagnosis model, regularly collect new historical fault data and normal data, and retrain and evaluate the model.

Images