Vibration spectrum analysis and fault early warning system of intelligent electric actuating mechanism

Abstract

The invention discloses a vibration spectrum analysis and fault early warning system for an intelligent electric actuating mechanism. The vibration spectrum analysis and fault early warning system comprises a vibration data acquisition module, a spectrum analysis module, a fault feature recognition module, an early warning generation module and a data storage module, the vibration data acquisition module acquires vibration data of the mechanical part and preprocesses the vibration data; the spectrum analysis module performs FFT conversion on the preprocessed vibration data, extracts key spectrum features and compares the key spectrum features with a reference spectrum; the fault feature recognition module stores typical frequency spectrum features of different fault types, calculates the matching degree of the key frequency spectrum features and the typical frequency spectrum features, and judges the severity of the fault; the early warning generation module generates early warning according to the fault severity and the residual life; and the data storage module stores the vibration data, the spectrogram and the early warning record. By collecting vibration data, analyzing frequency spectrum features and identifying fault types, mechanical faults are early warned in advance for 72 hours or more, and a basis is provided for preventive maintenance.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment condition monitoring technology, and in particular to a vibration spectrum analysis and fault early warning system for an intelligent electric actuator. Background Technology

[0002] Vibration is a crucial indicator of the operational status of mechanical components (motor, bearings, gearbox) in intelligent electric actuators. Mechanical faults (such as bearing wear, poor gear meshing, and motor imbalance) can lead to abnormal characteristics in the vibration spectrum. Existing vibration monitoring systems for actuators often employ simple vibration acceleration threshold alarms, which have significant drawbacks: First, they only monitor vibration amplitude and cannot identify the type of fault. For example, increased vibration acceleration may be caused by bearing wear or gear damage, making it difficult to pinpoint the root cause. Second, early warnings are delayed; threshold alarms are only triggered after the vibration amplitude exceeds a set value, by which time the fault has already progressed to a certain extent, making early warning impossible. Third, they lack spectrum analysis capabilities, failing to capture the unique frequency characteristics of faults (such as the frequency of bearing outer ring faults or gear meshing frequencies), resulting in a high rate of false alarms.

[0003] In industrial settings where equipment reliability is critical (such as petrochemicals and power), there is an urgent need for a system capable of vibration spectrum analysis, fault feature identification, and early warning to enable early detection and precise location of mechanical component failures. Summary of the Invention

[0004] To address the aforementioned problems in existing technologies, this invention provides a vibration spectrum analysis and fault early warning system for intelligent electric actuators. By collecting vibration data, analyzing spectrum characteristics, and identifying fault types, it provides early warning of mechanical faults at least 72 hours in advance, thus providing a basis for preventive maintenance.

[0005] The technical solution of the present invention is as follows: A vibration spectrum analysis and fault early warning system for an intelligent electric actuator is implemented based on an STM32H743 MCU (hereinafter referred to as MCU); it includes a vibration data acquisition module, a spectrum analysis module, a fault feature identification module, an early warning generation module, and a data storage module. The vibration data acquisition module installs one triaxial vibration sensor each on the front and rear bearing housings of the motor and the gearbox housing to collect vibration acceleration data in the X, Y, and Z directions. The sampling rate is 1kHz, the sampling period is 10ms, and 1024 data points are collected to form a data frame, which is accompanied by a UTC timestamp. The vibration data acquisition module performs DC component removal, Hanning windowing, and mean filtering on the collected raw data to eliminate environmental interference and sensor noise. The spectrum analysis module performs a fast Fourier transform on the preprocessed vibration data frame, converting the time-domain data into frequency-domain data with a frequency resolution ≤1Hz. Then, it extracts the key spectral features of the frequency-domain data, as follows: (1-1) Characteristic frequencies: Motor speed frequency f1 = motor speed / 60, The failure frequency of the bearing outer ring is f2 = f1 × (1.8 × D / d - 1). The bearing inner ring failure frequency f3 = f1 × (1.8 × D / d + 1), The failure frequency of the bearing rolling elements is f4 = f1 × (D / d) × (1 - (d × cosα / D)). 2 ), Gear meshing frequency f_z = motor speed × number of gear teeth / 60, Where D is the bearing pitch circle diameter, d is the rolling element diameter, and α is the contact angle; (1-2) Spectral peaks: Identify the peak values ​​of vibration acceleration at each characteristic frequency and calculate the signal-to-noise ratio; the signal-to-noise ratio is the ratio of the peak vibration acceleration to the background noise. (1-3) Spectral harmonics: Analyze the amplitude of the second and third harmonics of the characteristic frequencies to determine the severity of the fault; The spectrum analysis module compares the spectrum data with the reference spectrum when the equipment is running normally, calculates the spectrum similarity, and determines that the spectrum is abnormal if the similarity is <0.8. The reference spectrum is calibrated before leaving the factory or collected in the early stage of normal operation of the equipment. The spectrum similarity adopts the cosine similarity algorithm, with a range of 0 to 1, where 1 indicates complete similarity. The fault feature identification module establishes a mechanical fault feature database, storing typical spectral features of different fault types. These fault types include: motor imbalance, bearing outer ring wear, bearing inner ring wear, rolling element damage, gear meshing defects, and gear tooth breakage. The fault feature identification module matches the key spectral features from the spectral analysis module with the typical spectral features in the fault database, using a weighted Euclidean distance algorithm to calculate the matching degree. A matching degree > 0.85 indicates the corresponding fault type. Simultaneously, the fault feature identification module assesses the severity of the fault based on the ratio of the characteristic frequency peak value to the baseline value, classifying it into three levels: minor, moderate, and severe. The early warning generation module predicts the remaining lifespan based on the failure development trend of the motor and bearings, and sets three levels of early warning based on the remaining lifespan and the failure severity obtained by the fault feature identification module: (2-1) When the severity of the fault is minor and the remaining life is >1000h, it is a Level 1 warning, which triggers the "potential fault warning", prompting attention to the equipment status, and no immediate maintenance is required; (2-2) When the severity of the fault is moderate and the remaining lifespan is 72~1000h, it is a level 2 warning, triggering a "fault warning". It is recommended to arrange maintenance within 1 week. (2-3) When the severity of the fault is severe and the remaining lifespan is <72h, it is a level three warning, triggering an "emergency fault warning" and prompting immediate shutdown for maintenance; The early warning generation module displays warning information in color and text on the screen of the intelligent electric actuator, showing the fault type and remaining life, and automatically generates maintenance suggestions based on the fault type. The data storage module stores raw data via eMMC, fault data via FRAM, and cloud storage via the MQTT protocol.

[0006] Furthermore, the parameters of the triaxial vibration sensor are as follows: model ICM42670P, range ±16g, frequency range 1~1000Hz, resolution 16-bit; The triaxial vibration sensor communicates with the STM32H743 MCU via an SPI interface, with a sampling rate of 1kHz. The preprocessing algorithm is integrated into the MCU firmware.

[0007] Furthermore, the Fast Fourier Transform of the spectrum analysis module is implemented through the hardware floating-point unit (FPU) of the MCU, with a 1024-point FFT taking 3ms. Spectral feature extraction is implemented through the RK3588 processor and Python scripts.

[0008] Furthermore, the fault feature identification module assesses the severity of the fault based on the ratio of the peak value of the feature frequency to the benchmark value, and the judgment conditions are as follows: A ratio of 1.2 to 1.5 indicates mild severity; a ratio of 1.5 to 2.0 indicates moderate severity; and a ratio greater than 2.0 indicates severe severity.

[0009] Furthermore, when the fault feature identification module calculates the matching degree using the weighted Euclidean distance algorithm, if the matching degree is >0.85, it is determined to be the corresponding fault type.

[0010] Furthermore, the workflow is as follows: S1. Data acquisition: The triaxial vibration sensor acquires one set of triaxial vibration data every 10ms, which is then preprocessed to form a 1024-point data frame. S2, spectrum analysis, namely: performing a fast Fourier transform on the data frame of S1 to extract the motor speed frequency and the bearing outer ring fault frequency; S3, Feature Matching: Detect the peak value and signal-to-noise ratio in the spectrum of S2, match them with typical spectral features in the fault database, calculate the matching degree, and determine the severity of the fault. S4, Early Warning Generation: Based on the fault development trend, the remaining lifespan is predicted, and an early warning is triggered according to the remaining lifespan and the fault severity of S3. The local display screen shows text information, and a text message is pushed to the maintenance personnel remotely. S5. Data storage: Vibration data, spectrum analysis results, and early warning records are stored in eMMC and the cloud.

[0011] The beneficial technical effects of this invention are as follows: Multi-directional vibration acquisition and spectrum analysis accurately capture fault characteristic frequencies, avoiding the limitations of single-directional monitoring; Fault feature matching and severity assessment enable accurate fault type localization and reduce the false judgment rate; Provide early warnings at least 72 hours in advance to allow sufficient time for preventative maintenance and reduce unplanned downtime; Complete data storage and export facilitate fault tracing and maintenance effectiveness verification. Attached Figure Description

[0012] Figure 1 This is a structural block diagram of the vibration spectrum analysis and fault early warning system of the present invention. Detailed Implementation

[0013] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0014] like Figure 1 As shown, the embodiment includes a vibration data acquisition module, a spectrum analysis module, a fault feature identification module, an early warning generation module, and a data storage module. These modules work together as follows: I. Vibration Data Acquisition Module 1. Sensor deployment: Install one triaxial vibration sensor (model ICM42670P, range ±16g, frequency range 1~1000Hz, resolution 16-bit) in each of the front and rear bearing housings of the motor and the gearbox housing to collect vibration acceleration data in the X, Y, and Z directions, covering the main vibration directions of the mechanical components; 2. Acquisition parameters: Sampling rate 1kHz (to meet the frequency analysis requirements within 1000Hz), sampling period 10ms, each data frame consists of 1024 data points, with an accompanying UTC timestamp (accurate to milliseconds). 3. Data preprocessing: The collected raw data is processed by removing DC components, adding Hanning windows (to reduce spectral leakage), and mean filtering (window size 5) to eliminate environmental interference and sensor noise. The preprocessing time is ≤1ms.

[0015] II. Spectrum Analysis Module 1. FFT Transformation: Perform Fast Fourier Transform (FFT) on the preprocessed vibration data frame to convert the time domain data into the frequency domain data. The frequency resolution is ≤1Hz (1024-point FFT, sampling rate 1kHz), and the transformation time is ≤5ms. 2. Spectral Feature Extraction: Extracting key features from frequency domain data, including: (1) Characteristic frequencies: Calculate the motor speed frequency (f0 = motor speed / 60), bearing fault characteristic frequencies (outer ring fault frequency f0 × (1.8 × D / d - 1), inner ring fault frequency f0 × (1.8 × D / d + 1), rolling element fault frequency f0 × (D / d) × (1 - (d × cosα / D)). 2 ), where D is the bearing pitch circle diameter, d is the rolling element diameter, α is the contact angle, and the gear meshing frequency (f_z = motor speed × number of gear teeth / 60); (2) Spectral peak: Identify the peak value of vibration acceleration at each characteristic frequency and calculate the ratio of the peak value to the background noise (signal-to-noise ratio). (3) Spectrum harmonics: Analyze the amplitude of the second and third harmonics of the characteristic frequency to determine the severity of the fault (the higher the harmonic amplitude, the more severe the fault). 3. Spectrum Comparison: Compare the real-time spectrum with the reference spectrum during normal operation of the equipment (calibrated before leaving the factory or collected at the beginning of normal operation of the equipment), and calculate the spectrum similarity (using the cosine similarity algorithm, ranging from 0 to 1, where 1 indicates complete consistency). If the similarity is <0.8, it is judged as spectrum abnormality.

[0016] III. Fault Feature Identification Module 1. Fault Database: Establish a mechanical fault characteristic database to store typical spectral characteristics of different fault types (motor imbalance, bearing outer ring wear, bearing inner ring wear, rolling element damage, gear meshing problems, gear tooth breakage), for example: (1) Wear of the outer ring of the bearing: A significant peak appears at the failure frequency of the outer ring, the signal-to-noise ratio is >5, and the amplitude of the second harmonic is >30% of the amplitude of the fundamental wave; (2) Poor gear meshing: The peak value at the meshing frequency increases, and the amplitude of the third harmonic is greater than 50% of the amplitude of the fundamental frequency; 2. Feature matching: Match the real-time spectrum features with the features in the fault database, and use the weighted Euclidean distance algorithm to calculate the matching degree (range 0~1, 1 indicates a perfect match). If the matching degree is >0.85, it is determined to be the corresponding fault type. 3. Severity assessment of faults: The severity of faults is assessed based on the ratio of the peak value of the characteristic frequency to the baseline value, and is divided into three levels: slight (ratio 1.2~1.5), moderate (1.5~2.0), and severe (>2.0), which provides a basis for the warning level.

[0017] IV. Early Warning Generation Module 1. Early Warning Level Classification: Based on the severity of the fault and the predicted remaining lifespan (based on the fault development trend), a three-level early warning system is set: (1) Level 1 warning (minor fault): The remaining lifespan is >1000h, triggering the "potential fault warning", prompting attention to the equipment status, and no immediate maintenance is required; (2) Level 2 warning (moderate fault): The remaining lifespan is 72~1000h, triggering the "fault warning". It is recommended to arrange maintenance within 1 week; (3) Level 3 warning (serious fault): The remaining lifespan is <72h, triggering the "emergency fault warning", prompting immediate shutdown for maintenance; 2. Warning Output: The local IPS display screen provides warning information in color (green / yellow / red) and text, showing the fault type (e.g., "bearing outer ring wear") and remaining life (e.g., "800h"); the remote DCS / cloud platform synchronously receives the warning signal and pushes it to maintenance personnel via SMS / email; 3. Maintenance suggestion generation: Automatically generate maintenance suggestions based on the fault type, such as "Bearing outer ring wear, it is recommended to replace SKF 6205 bearing, required tools: bearing puller, torque wrench".

[0018] V. Data Storage Module 1. Storage content: Store raw vibration data (7 days of retention, rolling over), spectrum data (30 days of retention), fault characteristic data (1 year of retention), and early warning records (≥10 years of retention). 2. Storage Media: Raw data and spectrum data are stored in eMMC (32GB), while fault characteristics and early warning records are stored in FRAM (1Mbit) and the cloud platform to ensure that data is not lost in the event of power failure; 3. Data Export: Supports data export (CSV / PDF format) via local infrared maintenance port or remote communication interface. Exported files include vibration time-domain curves, spectrum diagrams, and fault analysis reports, facilitating fault tracing.

[0019] The hardware configuration of this embodiment is as follows: 1. Vibration data acquisition module: Triaxial vibration sensor ICM42670P, which communicates with STM32H743 MCU via SPI interface, with a sampling rate of 1kHz. The preprocessing algorithm is integrated into the MCU firmware. 2. Spectrum Analysis Module: The FFT transformation is implemented through the hardware floating-point unit (FPU) of the MCU. The 1024-point FFT takes 3ms. The spectrum feature extraction is implemented through Python scripts (edge ​​layer RK3588 processor). 3. Fault Feature Identification Module: The fault database is stored in FRAM, and the matching algorithm is implemented in C language. The matching degree calculation time is ≤10ms. 4. Warning generation module: The warning logic is integrated into the MCU. The warning signal controls the LED indicator through GPIO output. Remote warning is sent via SMS through the 4G module. 5. Data storage module: eMMC (KLM8G1GEME) stores raw data, FRAM (FM25W256) stores fault data, and cloud storage is implemented through the MQTT protocol.

[0020] The workflow of the example is as follows: S1. Data acquisition: The vibration sensor acquires a set of triaxial vibration data every 10ms, which is then preprocessed to form a 1024-point data frame. S2. Spectrum Analysis: Perform FFT transformation on the data frame to extract the motor speed frequency (f0=25Hz, motor speed 1500rpm) and the bearing outer ring fault frequency (f=25×(1.8×50 / 15 -1)=125Hz). S3. Feature matching: A peak (acceleration 1.2g) appears at 125Hz in the real-time spectrum, with a signal-to-noise ratio of 6. The feature matching degree with "bearing outer ring wear" in the fault database is 0.92, which is judged as a moderate fault. S4. Warning Generation: Based on the fault development trend, the remaining life is predicted to be 800 hours, triggering a level 2 warning. The local display shows "Bearing outer ring wear warning, remaining life 800 hours", and a text message is sent remotely to the maintenance personnel. S5. Data storage: Vibration data, spectrum diagrams, and early warning records are stored to eMMC and the cloud, supporting subsequent export and analysis.

[0021] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, and for those of ordinary skill in the art, various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. Therefore, the present invention is not limited to the specific details without departing from the general concept defined by the claims and their equivalents.

Claims

1. A vibration spectrum analysis and fault early warning system for an intelligent electric actuator, characterized in that: Implemented based on STM32H743 MCU, hereinafter referred to as MCU; including vibration data acquisition module, spectrum analysis module, fault feature identification module, early warning generation module and data storage module; The vibration data acquisition module installs one triaxial vibration sensor each on the front and rear bearing housings of the motor and the gearbox housing to collect vibration acceleration data in the X, Y, and Z directions. The sampling rate is 1kHz, the sampling period is 10ms, and 1024 data points are collected to form a data frame, which is accompanied by a UTC timestamp. The vibration data acquisition module performs DC component removal, Hanning windowing, and mean filtering on the collected raw data to eliminate environmental interference and sensor noise. The spectrum analysis module performs a fast Fourier transform on the preprocessed vibration data frame, converting the time-domain data into frequency-domain data with a frequency resolution ≤1Hz. Then, it extracts the key spectral features of the frequency-domain data, as follows: (1-1) Characteristic frequencies: Motor speed frequency f1 = motor speed / 60, The failure frequency of the bearing outer ring is f2 = f1 × (1.8 × D / d - 1). The bearing inner ring failure frequency f3 = f1 × (1.8 × D / d + 1), The failure frequency of the bearing rolling elements is f4 = f1 × (D / d) × (1 - (d × cosα / D)). 2 ), Gear meshing frequency f_z = motor speed × number of gear teeth / 60, Where D is the bearing pitch circle diameter, d is the rolling element diameter, and α is the contact angle; (1-2) Spectral peaks: Identify the peak values ​​of vibration acceleration at each characteristic frequency and calculate the signal-to-noise ratio; the signal-to-noise ratio is the ratio of the peak vibration acceleration to the background noise. (1-3) Spectral harmonics: Analyze the amplitude of the second and third harmonics of the characteristic frequencies to determine the severity of the fault; The spectrum analysis module compares the spectrum data with the reference spectrum when the equipment is running normally, calculates the spectrum similarity, and determines that the spectrum is abnormal if the similarity is <0.

8. The reference spectrum is calibrated before leaving the factory or collected in the early stage of normal operation of the equipment. The spectrum similarity adopts the cosine similarity algorithm, with a range of 0 to 1, where 1 indicates complete similarity. The fault feature identification module establishes a mechanical fault feature database, storing typical spectral features of different fault types. These fault types include: motor imbalance, bearing outer ring wear, bearing inner ring wear, rolling element damage, gear meshing defects, and gear tooth breakage. The fault feature identification module matches the key spectral features from the spectral analysis module with the typical spectral features in the fault database, using a weighted Euclidean distance algorithm to calculate the matching degree. A matching degree > 0.85 indicates the corresponding fault type. Simultaneously, the fault feature identification module assesses the severity of the fault based on the ratio of the characteristic frequency peak value to the baseline value, classifying it into three levels: minor, moderate, and severe. The early warning generation module predicts the remaining lifespan based on the failure development trend of the motor and bearings, and sets three levels of early warning based on the remaining lifespan and the failure severity obtained by the fault feature identification module: (2-1) When the severity of the fault is minor and the remaining life is >1000h, it is a Level 1 warning, triggering the "potential fault warning", prompting attention to the equipment status, and no immediate maintenance is required; (2-2) When the severity of the fault is moderate and the remaining lifespan is 72~1000h, it is a level 2 warning, triggering a "fault warning". It is recommended to arrange maintenance within 1 week. (2-3) When the severity of the fault is severe and the remaining life is <72h, it is a level three warning, triggering an "emergency fault warning", prompting an immediate shutdown for maintenance; The early warning generation module displays warning information in color and text on the screen of the intelligent electric actuator, showing the fault type and remaining life, and automatically generates maintenance suggestions based on the fault type. The data storage module stores raw data via eMMC, fault data via FRAM, and cloud storage via the MQTT protocol.

2. The vibration spectrum analysis and fault early warning system for an intelligent electric actuator according to claim 1, characterized in that: The parameters of the triaxial vibration sensor are as follows: model ICM42670P, range ±16g, frequency range 1~1000Hz, resolution 16-bit. The triaxial vibration sensor communicates with the STM32H743 MCU via an SPI interface, with a sampling rate of 1kHz. The preprocessing algorithm is integrated into the MCU firmware.

3. The vibration spectrum analysis and fault early warning system for an intelligent electric actuator according to claim 2, characterized in that: The Fast Fourier Transform (FFT) of the spectrum analysis module is implemented through the hardware floating-point unit (FPU) of the MCU. The 1024-point FFT takes 3ms. The spectrum feature extraction is implemented through the RK3588 processor and Python script.

4. The vibration spectrum analysis and fault early warning system for an intelligent electric actuator according to claim 1, characterized in that: The fault feature identification module assesses the severity of the fault based on the ratio of the peak value of the feature frequency to the baseline value, and the judgment conditions are as follows: A ratio of 1.2 to 1.5 indicates mild severity; a ratio of 1.5 to 2.0 indicates moderate severity; and a ratio greater than 2.0 indicates severe severity.

5. The vibration spectrum analysis and fault early warning system for an intelligent electric actuator according to claim 1, characterized in that: When the fault feature identification module uses the weighted Euclidean distance algorithm to calculate the matching degree, if the matching degree is >0.85, it is determined to be the corresponding fault type.

6. The vibration spectrum analysis and fault early warning system for an intelligent electric actuator according to claim 1, characterized in that, The workflow is as follows: S1. Data acquisition: The triaxial vibration sensor acquires one set of triaxial vibration data every 10ms, which is then preprocessed to form a 1024-point data frame. S2, spectrum analysis, namely: performing a fast Fourier transform on the data frame of S1 to extract the motor speed frequency and the bearing outer ring fault frequency; S3, Feature Matching: Detect the peak value and signal-to-noise ratio in the spectrum of S2, match them with typical spectral features in the fault database, calculate the matching degree, and determine the severity of the fault. S4, Early Warning Generation: Based on the fault development trend, the remaining lifespan is predicted, and an early warning is triggered according to the remaining lifespan and the fault severity of S3. The local display screen shows text information, and a text message is pushed to the maintenance personnel remotely. S5. Data storage: Vibration data, spectrum analysis results, and early warning records are stored in eMMC and the cloud.

Images