Intelligent Health Management System Based on Capacitive Complementary Lubricating Oil Abrasive Sensor
The intelligent health management system based on capacitive complementary lubricating oil abrasive sensor utilizes an STM32 microcontroller for data processing, achieving real-time processing and accurate extraction of lubricating oil abrasive signals. This solves the problem of non-real-time signal processing in existing technologies and meets the online monitoring needs of mechanical equipment.
Patent Information
- Application Number
- CN202611461346.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing lubricating oil abrasive monitoring systems are prone to having weak abrasive signals drowned out by strong noise under complex working conditions. They are also difficult to eliminate circuit drift and filter background noise locally, resulting in the inability to process signals in real time and failing to meet the online monitoring needs of mechanical equipment.
An intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor is adopted. The STM32 microcontroller is used to perform data acquisition, preprocessing, adaptive signal denoising, abrasive pulse feature extraction and comprehensive health status diagnosis. Real-time processing at the edge is achieved through analog signal acquisition link, abrasive signal denoising program link and communication and control link.
It enables real-time processing of abrasive signals at the sensor edge, solves the problem of strong interference noise overwhelming the signal under complex working conditions, and realizes accurate extraction of abrasive pulse characteristics and online diagnosis and early warning of mechanical equipment.
Smart Images

Figure CN123111897A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mechanical equipment condition monitoring and fault diagnosis technology, specifically to an intelligent health management system based on a capacitive-inductive complementary lubricating oil abrasive sensor. Background Technology
[0002] Oil wear monitoring is a routine method for assessing the wear condition and providing early warning of faults in mechanical equipment. During the operation of mechanical equipment, particles generated by the wear of internal friction pairs circulate within the system along with the lubricating oil. Continuously monitoring particle information in the lubrication pipeline using oil wear sensors can reflect the internal health status of the mechanical equipment, providing fundamental data support for equipment maintenance.
[0003] Existing lubricating oil wear monitoring systems typically employ a system architecture that separates front-end data acquisition from back-end processing. In practical applications, front-end sensors are primarily responsible for picking up physical quantities in the lubricating oil pipeline and converting them into basic digital electrical signals. These raw digital signals are then transmitted directly to a remote host computer server or cloud processing center via a communication bus or wireless transmission module. Upon receiving the raw data, the host computer centrally allocates computing resources to run filtering algorithms and diagnostic models, ultimately determining the operating status of the mechanical equipment and outputting diagnostic results.
[0004] However, in actual operating conditions, the raw signals acquired by sensors contain a significant amount of background noise. Existing monitoring systems heavily rely on host computers for data processing, and a large amount of raw digital signals experience significant data transmission delays during transmission to the remote host computer. Limited by this front-end / back-end separation processing model, existing sensors cannot independently perform circuit drift elimination and background noise filtering locally. This results in signals that cannot be effectively extracted and processed in real time at the sensor edge, making it difficult to meet the real-time requirements of mechanical equipment for online status monitoring and rapid response. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides an intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor. This system solves the problems in existing technologies where weak abrasive signals are easily submerged by strong noise under complex working conditions, making extraction difficult and making it hard to accurately distinguish abrasive size and material. At the same time, existing diagnostic systems rely on host computers, resulting in insufficient real-time processing capabilities at the edge.
[0006] To achieve the above objectives, the present invention provides an intelligent health management system based on a capacitive-inductive complementary lubricating oil abrasive sensor, comprising a capacitive-inductive complementary lubricating oil abrasive sensor, an STM32 microcontroller, and an analog signal acquisition link, an abrasive signal noise reduction program link, a data storage link, and a communication and control link connected to the STM32 microcontroller, wherein the capacitive-inductive complementary lubricating oil abrasive sensor is connected to the analog signal acquisition link.
[0007] The STM32 microcontroller is configured to execute an intelligent diagnostic process, which includes: Data acquisition and preprocessing: The STM32 microcontroller controls the analog signal acquisition link to continuously acquire the original digital signal sequences of the inductor channel and the capacitor channel, and performs DC bias removal processing on the original digital signal sequences of the inductor channel and the capacitor channel to obtain the AC-coupled abrasive induction signal sequence. Adaptive signal denoising is performed: Based on the AC-coupled abrasive particle sensing signal sequence, the STM32 microcontroller calls the denoising algorithm in the abrasive particle signal denoising program link to separate the noisy component and perform denoising operation to obtain a clean abrasive particle signal. Abrasive Pulse Feature Extraction: Based on the pure abrasive signal, the STM32 microcontroller determines the valid abrasive events through sliding window peak detection and extracts the peak voltage and timestamp of the valid abrasive events; Perform abrasive grain size and material determination: Based on the peak voltage, the STM32 microcontroller calculates the equivalent diameter of the abrasive grain by substituting it into the mapping curve, and obtains the signal amplitude of the inductor channel and the signal amplitude of the capacitor channel. The material properties are determined based on the difference in the signal amplitude changes of the inductor channel and the capacitor channel. Perform comprehensive health status diagnosis: Based on the equivalent diameter of the abrasive grain and the timestamp, the STM32 microcontroller counts the number of abrasive grains in each size range and calculates the yield of large abrasive grains. The health status is determined based on the number of abrasive grains in each size range and the yield of large abrasive grains. When the health status is determined to be alarm state, an alarm response is triggered through the communication and control link.
[0008] Preferably, the analog signal acquisition link is connected to the STM32 microcontroller via the SPI1 interface; the abrasive signal noise reduction program link is connected to the STM32 microcontroller via the QSPI interface; the data storage link is connected to the STM32 microcontroller via the I2C interface; and the communication and control link includes a LoRa module connected to the UART interface, a TJA1050 transceiver connected to the CAN bus interface, and a human-machine interaction display module connected to the GPIO interface.
[0009] Preferably, in the data acquisition and preprocessing steps, the STM32 microcontroller enables a direct memory access channel, which directly transfers the digital signal output from the analog signal acquisition link to the STM32 microcontroller's memory to obtain the original digital signal sequences of the inductor channel and the capacitor channel; the STM32 microcontroller eliminates circuit zero drift, calculates the AC-coupled abrasive particle induction signal sequence, and stores the AC-coupled abrasive particle induction signal sequence in the data memory.
[0010] Preferably, in the adaptive signal denoising step, the STM32 microcontroller segments the AC-coupled abrasive particle induction signal sequence to obtain a segmented signal sequence; based on the segmented signal sequence, the STM32 microcontroller uses the Grey Wolf algorithm to optimize the parameters of the variational mode decomposition method and selects the optimal parameters of the variational mode decomposition method; according to the optimal parameters, the STM32 microcontroller decomposes the segmented signal sequence into multiple IMF components through the variational mode decomposition method.
[0011] Preferably, based on multiple IMF components, the STM32 microcontroller divides the multiple IMF components into clean components and noisy components by calculating the correlation coefficient between each IMF component and the original signal. For the noisy components, the STM32 microcontroller uses an improved wavelet threshold to denoise the noisy components cyclically until the correlation coefficient of the denoised noisy components meets a preset threshold. The noisy components with correlation coefficients meeting the preset threshold are then converted into clean components. Based on the clean components obtained by direct division and the clean components obtained by denoising and conversion, the STM32 microcontroller superimposes and reconstructs all clean components to obtain and output a clean abrasive grain signal.
[0012] Preferably, in the step of performing abrasive pulse feature extraction, the STM32 microcontroller performs sliding window peak detection on the clean abrasive signal in the time domain to obtain the signal amplitude in the time series; compares the signal amplitude with a set dynamic noise threshold, and determines that a valid abrasive event has occurred when the signal amplitude exceeds the dynamic noise threshold; extracts the peak voltage and pulse half-wave width when the valid abrasive event occurs, and records the timestamp corresponding to the valid abrasive event.
[0013] Preferably, in the step of determining the abrasive size and material, the mapping curve is the curve corresponding to the peak voltage and the equivalent diameter of the abrasive, established by calibrating a standard spherical particle on a calibration table; the material properties include ferromagnetic metal materials, non-ferromagnetic metal materials and non-metallic materials. The STM32 microcontroller determines the material properties of the abrasive as ferromagnetic metal materials, non-ferromagnetic metal materials and non-metallic materials based on the difference in signal amplitude changes between the inductor channel and the capacitor channel.
[0014] When the inductor channel signal exhibits a large positive inductor pulse and the capacitor channel signal exhibits a large positive capacitor pulse, the abrasive material is determined to be a ferromagnetic metal material. When the inductor channel signal shows a small negative inductor pulse and the capacitor channel signal shows a large positive capacitor pulse, the abrasive material is determined to be a non-ferromagnetic metal material. When the inductor channel signal has almost no inductive signal and the capacitor channel signal shows a weak capacitor pulse, the abrasive material is determined to be a non-metallic material.
[0015] Preferably, in the step of performing comprehensive health status diagnosis, the STM32 microcontroller uses the number of abrasive grains in each size range and the yield of large abrasive grains as status feature data, and compares the status feature data with the system's preset three-level alarm threshold. When the equivalent diameter of the abrasive particles detected per unit time is less than 50μm and the yield of large abrasive particles is lower than the preset threshold, the current health status is determined to be normal. When the number of abrasive particles with a size in the range of 50μm to 100μm increases monotonically over multiple consecutive evaluation cycles, the current health status is determined to be a warning state. When large abrasive particles exceeding 100μm in size are found, or when the ratio of large abrasive particle yields in two adjacent evaluation cycles is calculated and the ratio is greater than or equal to the preset exponential surge coefficient, the current health status is determined to be an alarm state.
[0016] Preferably, when the current health status is determined to be an alarm state, the STM32 microcontroller generates a specific fault code and sends the specific fault code through the CAN bus in the communication and control link; at the same time, the STM32 microcontroller controls the external LED indicator to flash or controls the buzzer to sound an alarm.
[0017] This invention provides an intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor. It offers the following advantages: 1. This invention connects the analog signal acquisition link and the abrasive signal noise reduction program link to an STM32 microcontroller, and executes DC bias removal and adaptive noise reduction algorithms sequentially within the microcontroller. This overcomes the shortcomings of existing systems that rely on a host computer, resulting in large data transmission delays. It enables the original digital signals of the inductor and capacitor channels to be directly eliminated at the sensor end for circuit zero drift elimination and background noise filtering, thereby extracting pure abrasive signals. This achieves real-time processing of abrasive signals at the sensor edge, meeting the needs of online monitoring of mechanical equipment.
[0018] 2. This invention processes AC-coupled abrasive particle induction signal sequences by invoking a joint denoising mechanism combining variational mode decomposition and improved wavelet thresholding. After optimizing parameters using the Grey Wolf algorithm, multimodal components are decomposed, and the noisy components are subjected to cyclic wavelet denoising and signal reconstruction to obtain pure abrasive particle signals. Then, the peak voltage of effective abrasive particle events is extracted using sliding window technology. This solves the problem that strong interference noise overwhelms the original pulse signal under complex working conditions, making feature extraction impossible, and achieves accurate extraction of abrasive particle pulse features in lubricating oil.
[0019] 3. This invention calculates the equivalent diameter of abrasive particles by substituting the extracted peak voltage into the mapping curve, and determines whether the abrasive particles are ferromagnetic, non-ferromagnetic, or non-metallic materials by combining the difference in signal amplitude changes between the inductive and capacitive channels. It also compares the number of abrasive particles in each size range with the yield of large abrasive particles with a preset three-level alarm threshold. When the health status is determined to be alarm state, it directly controls the external indicator light to flash or the buzzer to sound an alarm. This solves the problems that single-channel sensors cannot distinguish abrasive particle materials and the lag in traditional equipment status diagnosis, and realizes online diagnosis and early warning of mechanical equipment wear status. Attached Figure Description
[0020] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the adaptive algorithm noise reduction process of the present invention. Detailed Implementation
[0021] 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.
[0022] See attached document Figure 1 , Figure 1This is an overall architecture diagram of an intelligent health management system based on a capacitive-inductance complementary lubricating oil abrasive sensor according to an embodiment of the present invention. The intelligent health management system based on the capacitive-inductance complementary lubricating oil abrasive sensor uses an STM32 microcontroller as its core processing unit and achieves physical mapping and data interaction with peripheral functional links through various bus interfaces. In the specific hardware layout, the analog signal acquisition link is connected to the capacitive-inductance complementary lubricating oil abrasive sensor at its front end. The sensor detects abrasive particles in the lubricating oil using a dual-modal method based on inductive and capacitive methods, outputting the original inductive and capacitive channel signals. The STM32 microcontroller is connected to the analog signal acquisition link via an SPI1 interface. The analog signal acquisition link integrates start-up conversion, clock, data input, data output, and backup channels, responsible for the synchronous digital acquisition of the front-end capacitive dual-channel physical signals. The STM32 microcontroller is connected to the abrasive signal noise reduction program link via a QSPI interface, loading the clock, abrasive data, noise reduction algorithm, and backup data within the abrasive signal noise reduction program link. The data storage link is connected to the STM32 microcontroller via an I2C interface, performing clock synchronization and abrasive data recording. On the communication and control link, the STM32 microcontroller is configured with three independent physical interfaces: the UART interface is connected to the LoRa module to perform receiving diagnostic frames, data return, soft reset, and interrupt response; the CAN bus interface is connected to the TJA1050 transceiver, utilizing the internal conversion, demodulation, differential, reference level provision, and anti-reflection function modules of the TJA1050 transceiver to ensure the stability of data transmission over long distances; and the GPIO interface is directly connected to the human-machine interface display module to drive the buzzer and LED_R / G / B three-color indicator lights to perform local status indication.
[0023] Based on the aforementioned hardware architecture, the STM32 microcontroller has an internally integrated intelligent diagnostic process. Each data processing step is interconnected, with the results of the preceding steps serving as input parameters for the subsequent steps. The overall intelligent diagnostic process encompasses the following steps.
[0024] Step S1 involves data acquisition and preprocessing. The STM32 microcontroller controls the SPI1 interface to enable the direct memory access channel, continuously acquiring the raw digital signal sequences from the inductor and capacitor channels. A DC bias removal operation is performed on the raw digital signal sequences to eliminate circuit zero drift, and the AC-coupled abrasive particle induction signal sequence is calculated. This AC-coupled abrasive particle induction signal sequence serves as the reference data for subsequent processing and is output to the next step.
[0025] After receiving the AC-coupled abrasive particle sensing signal sequence output in step S1, the adaptive noise reduction algorithm in step S2 is executed. The STM32 microcontroller calls the noise reduction algorithm in the abrasive particle signal noise reduction program chain, using a joint noise reduction mechanism combining variational mode decomposition and improved wavelet thresholding. Through iterative optimization, the input AC-coupled abrasive particle sensing signal sequence is processed to separate the noisy components. Noise reduction operations are performed on the separated noisy components, and the processed components are superimposed with the clean components from the separation process to reconstruct a clean abrasive particle signal with background noise filtered out.
[0026] Based on the clean abrasive grain signal output in step S2, abrasive grain feature extraction is performed in step S3. The STM32 microcontroller applies a sliding window to the clean abrasive grain signal in the time domain and monitors the signal amplitude within the sliding window. When the monitored signal amplitude exceeds the set dynamic noise threshold, the STM32 microcontroller determines that a valid abrasive grain event has occurred, extracts the peak voltage and pulse half-wave width feature parameters for the valid abrasive grain event, and records the timestamp. After the feature parameters are extracted, they are passed to the next determination module.
[0027] Using the peak voltage and pulse half-wave width characteristic parameters obtained in step S3, the abrasive grain size and material determination in step S4 are performed. The STM32 microcontroller retrieves a pre-established mapping curve based on a standard spherical particle on a calibration platform, substitutes the extracted peak voltage into the mapping curve, and calculates the equivalent diameter of the abrasive grain. The STM32 microcontroller extracts the signal amplitudes of the inductor and capacitor channels, calculates the difference in signal amplitude changes between the inductor and capacitor channels, and determines the material properties of the abrasive grain as ferromagnetic metal, non-ferromagnetic metal, or non-metallic material based on the difference in signal amplitude changes.
[0028] After extracting the abrasive equivalent diameter and material property data output in step S4, the comprehensive health status diagnosis in step S5 is performed. The STM32 microcontroller uses timestamps to count the number of abrasive particles in each size range per unit time and compares the statistical results with the system's preset three-level alarm thresholds. When the equivalent diameter of the abrasive particles is less than 50μm and the yield of large abrasive particles is lower than the preset threshold, it is determined to be in a normal state; when the number of abrasive particles in the 50 to 100μm range continues to increase, it is determined to be in a warning state; when large abrasive particles exceeding 100μm appear or the yield of large abrasive particles increases exponentially, it is determined to be in an alarm state. When an alarm state is triggered, the STM32 microcontroller generates a specific fault code, calls the CAN bus to send the specific fault code, and simultaneously drives the GPIO interface to control the external LED indicator to flash or trigger the buzzer alarm.
[0029] In the intelligent health management system based on a capacitive complementary lubricating oil wear sensor, step S1 is performed for data acquisition and preprocessing. Step S1 is further divided into steps S101 and S102.
[0030] Step S101: The digital signal sequences of the inductor and capacitor channels are continuously acquired via direct memory access (DMI). The STM32 microcontroller sends a start conversion command to the analog signal acquisition link via the SPI1 interface. Synchronized with a clock signal, the analog signal acquisition link interacts with the STM32 microcontroller through data input and output channels. The STM32 microcontroller internally enables the DMI, which directly transfers the digital signals output from the analog signal acquisition link to the STM32 microcontroller's memory, acquiring the original digital signal sequences of the inductor and capacitor channels.
[0031] Step S102: Based on the original digital signal sequences of the inductor channel and capacitor channel obtained in step S101, the STM32 microcontroller performs DC bias removal processing on the original digital signal sequences of the inductor channel and capacitor channel to eliminate circuit zero drift, thereby obtaining the AC-coupled abrasive particle induction signal sequence x. ind (n) and x cap (n). The STM32 microcontroller converts the AC-coupled abrasive particle sensing signal sequence x ind (n) and x cap (n) is stored in the data memory as the input basis for the subsequent execution step S2.
[0032] See attached document Figure 2 , Figure 2 This is a flowchart of an adaptive algorithm for noise reduction according to an embodiment of the present invention. In an intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor, step S2 is performed for adaptive signal noise reduction. The AC-coupled abrasive sensing signal sequence obtained in step S102 is uniformly represented as the preprocessed signal x(n) in step S2. Step S2 is further subdivided into steps S201 to S205.
[0033] In step S201, the STM32 microcontroller receives the preprocessed signal x(n) output in step S102. The STM32 microcontroller segments the preprocessed signal x(n) to obtain a segmented signal sequence.
[0034] In step S202, based on the segmented signal sequence obtained in step S201, the STM32 microcontroller uses the Grey Wolf algorithm to optimize the parameters of the variational mode decomposition method. The STM32 microcontroller selects the optimal parameters K and α for the variational mode decomposition method through the parameter optimization process. Parameters K and α are the operating parameters required by the variational mode decomposition method.
[0035] In step S203, after obtaining the optimal parameters K and α, the STM32 microcontroller decomposes the segmented signal sequence into multiple IMF components based on the optimal parameters K and α using the variational mode decomposition method.
[0036] In step S204, based on the multiple IMF components output in step S203, the STM32 microcontroller calculates the correlation coefficient between each IMF component and the original signal to separate the multiple IMF components into clean components and noisy components. For the noisy components, the STM32 microcontroller uses an improved wavelet threshold to denoise them. The STM32 microcontroller determines whether the correlation coefficient of the denoised noisy components meets a preset threshold. If the correlation coefficient does not meet the preset threshold, the STM32 microcontroller repeatedly uses the improved wavelet threshold to denoise the noisy components until the correlation coefficient of the denoised noisy components meets the preset threshold. The STM32 microcontroller then converts the noisy components with correlation coefficients meeting the preset threshold into clean components.
[0037] In step S205, based on the pure components obtained directly from step S204 and the pure components obtained from the noise reduction conversion in step S204, the STM32 microcontroller superimposes and reconstructs all pure components to obtain and output the pure abrasive particle signal s(n). The STM32 microcontroller uses the pure abrasive particle signal s(n) as the input basis for the subsequent execution of step S3.
[0038] In the intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor, after signal noise reduction in step S2, step S3 is performed to extract abrasive pulse features. Step S3 is further subdivided into steps S301 to S303.
[0039] In step S301, based on the pure abrasive particle signal s(n) output in step S205, the STM32 microcontroller performs sliding window peak detection on the pure abrasive particle signal s(n) in the time domain. The STM32 microcontroller obtains the signal amplitude in the time series through the sliding window.
[0040] In step S302, the STM32 microcontroller compares the signal amplitude acquired within the sliding window with a set dynamic noise threshold. When the signal amplitude exceeds the dynamic noise threshold, the STM32 microcontroller determines that a valid abrasive event has occurred.
[0041] In step S303, for the determined valid abrasive event, the STM32 microcontroller extracts the peak voltage Vp and pulse half-wave width W at the time the valid abrasive event occurs, and records the timestamp corresponding to the valid abrasive event. The STM32 microcontroller stores the extracted peak voltage Vp, pulse half-wave width W, and timestamp, and outputs the peak voltage Vp, pulse half-wave width W, and timestamp as abrasive pulse characteristic parameters to step S4.
[0042] After obtaining the abrasive pulse characteristic parameters, step S4 is executed to determine the abrasive size and material. Step S4 is further divided into steps S401 to S403.
[0043] Step S401: Calibrate using standard spherical particles on a calibration stage to establish a mapping curve between peak voltage Vp and the equivalent diameter of the abrasive particles.
[0044] In step S402, during real-time diagnosis, based on the peak voltage Vp output in step S303, the STM32 microcontroller substitutes the peak voltage Vp into the formula to calculate the equivalent diameter of the abrasive grain.
[0045] In step S403, the STM32 microcontroller acquires the amplitude of the inductor channel signal and the amplitude of the capacitor channel signal, and determines the material properties of the abrasive grain based on the difference in amplitude changes between the two channels. The material properties include ferromagnetic metal materials, non-ferromagnetic metal materials, and non-metallic materials. Specifically, when the inductor channel signal exhibits a large positive inductance pulse and the capacitor channel signal exhibits a large positive capacitance pulse, the abrasive grain material is determined to be a ferromagnetic metal material; when the inductor channel signal exhibits a small negative inductance pulse and the capacitor channel signal exhibits a large positive capacitance pulse, the abrasive grain material is determined to be a non-ferromagnetic metal material; when the inductor channel signal has almost no inductance signal and the capacitor channel signal exhibits a weak capacitance pulse, the abrasive grain material is determined to be a non-metallic material. The STM32 microcontroller stores the calculated equivalent diameter and material properties of the abrasive grain, and outputs these parameters to step S5 as the data basis for step S5.
[0046] In the intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor, based on the equivalent diameter and material properties of the abrasive particles output in step S4, step S5 is executed to perform a comprehensive health status diagnosis. Step S5 is further subdivided into steps S501 to S503.
[0047] In step S501, the STM32 microcontroller receives the equivalent diameter of the abrasive grains output in step S4 and the timestamp recorded in step S303. Based on the equivalent diameter of the abrasive grains and the timestamp, the STM32 microcontroller counts the number of abrasive grains in each size range per unit time.
[0048] The STM32 microcontroller extracts the total number of abrasive grains with an equivalent diameter greater than or equal to 50 μm within a set time window and calculates the large abrasive grain yield using a preset formula. The formula for calculating the large abrasive grain yield is as follows:
[0049] In the formula, For high abrasive yield; The total number of abrasive grains with an equivalent diameter greater than or equal to 50 μm, obtained within a set time window; To set the duration of the time window, the STM32 microcontroller calculates the generation rate of large abrasive grains per unit time using the large abrasive grain yield formula, and uses this generation rate as the large abrasive grain yield.
[0050] The STM32 microcontroller stores the number of abrasive grains in each size range and the yield of large abrasive grains as state characteristic data per unit time.
[0051] In step S502, the STM32 microcontroller retrieves the system's preset three-level alarm thresholds. The STM32 microcontroller compares the status characteristic data output in step S501 with the three-level alarm thresholds to determine the device's health status. The health statuses corresponding to the three-level alarm thresholds include normal state, warning state, and alarm state.
[0052] When the STM32 microcontroller determines that the equivalent diameter of all detected abrasive particles is less than 50μm and the yield of large abrasive particles is lower than a preset threshold, the STM32 microcontroller determines that the current health status is normal.
[0053] The STM32 microcontroller establishes a continuous evaluation cycle based on the time axis. When the number of abrasive particles with a size in the range of 50μm to 100μm is detected to increase monotonically over multiple consecutive evaluation cycles, the STM32 microcontroller determines the current health status as a warning state.
[0054] When the STM32 microcontroller detects large abrasive particles with a size exceeding 100μm, or calculates the ratio of large abrasive particle yields in two adjacent evaluation cycles, and this ratio is greater than or equal to a preset exponential surge coefficient, the STM32 microcontroller determines the current health status as an alarm state.
[0055] In this embodiment, the thresholds of 50μm and 100μm are set for a specific equipment lubrication system; in actual engineering applications, those skilled in the art can make adaptive adjustments to the above thresholds according to the equipment type, operating conditions and historical wear data.
[0056] In step S503, when the STM32 microcontroller determines the health status to be in an alarm state, it generates a specific fault code. The STM32 microcontroller sends this fault code via the CAN bus in the communication and control link. Simultaneously, the STM32 microcontroller controls an external LED indicator to flash or activates a buzzer to sound an alarm.
[0057] For the implementation of the underlying communication protocol of the CAN bus and the underlying hardware driver circuit of the external LED indicator and buzzer, those skilled in the art can use conventional embedded hardware development schemes for configuration. The implementation of the CAN bus communication protocol and the peripheral driver circuit is a well-known technology in this field, and will not be described in detail here.
[0058] 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. An intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor, characterized in that, include: A capacitive-inductive complementary abrasive sensor for lubricating oil is provided, connected to an analog signal acquisition link. An STM32 microcontroller is also provided, along with the analog signal acquisition link, abrasive signal noise reduction program link, data storage link, and communication and control link connected to the STM32 microcontroller. The STM32 microcontroller is configured to execute an intelligent diagnostic process, which includes: Data acquisition and preprocessing: The STM32 microcontroller controls the analog signal acquisition link to continuously acquire the original digital signal sequence of the inductor channel and the original digital signal sequence of the capacitor channel, and performs DC bias removal processing on the original digital signal sequence of the inductor channel and the original digital signal sequence of the capacitor channel to obtain the AC coupled abrasive induction signal sequence. Adaptive signal denoising is performed: Based on the AC-coupled abrasive particle sensing signal sequence, the STM32 microcontroller calls the denoising algorithm in the abrasive particle signal denoising program link to separate the noisy component and perform denoising operation to obtain a clean abrasive particle signal; Abrasive Pulse Feature Extraction: Based on the pure abrasive signal, the STM32 microcontroller determines valid abrasive events through sliding window peak detection, and extracts the peak voltage and timestamp of the valid abrasive events; Perform abrasive grain size and material determination: Based on the peak voltage, the STM32 microcontroller substitutes the mapping curve to calculate the equivalent diameter of the abrasive grain; obtain the inductor channel signal amplitude and the capacitor channel signal amplitude, and determine the material properties based on the difference in the signal amplitude changes of the inductor channel and the capacitor channel; Perform comprehensive health status diagnosis: Based on the equivalent diameter of the abrasive grains and the timestamp, the STM32 microcontroller counts the number of abrasive grains in each size range and calculates the yield of large abrasive grains. Based on the number of abrasive grains in each size range and the yield of large abrasive grains, the health status is determined. When the health status is determined to be alarm state, an alarm response is triggered through the communication and control link.
2. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, The analog signal acquisition link is connected to the STM32 microcontroller via the SPI1 interface; the abrasive signal noise reduction program link is connected to the STM32 microcontroller via the QSPI interface; the data storage link is connected to the STM32 microcontroller via the I2C interface; the communication and control link includes a LoRa module connected to the UART interface, a TJA1050 transceiver connected to the CAN bus interface, and a human-machine interaction display module connected to the GPIO interface.
3. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the data acquisition and preprocessing steps, the STM32 microcontroller enables a direct memory access channel (DMI). This DMI directly transfers the digital signal output from the analog signal acquisition link to the STM32 microcontroller's memory, acquiring the original digital signal sequences of the inductor channel and the capacitor channel. The STM32 microcontroller eliminates circuit zero drift, calculates the AC-coupled abrasive particle induction signal sequence, and stores the AC-coupled abrasive particle induction signal sequence in the data memory.
4. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the step of performing adaptive signal denoising, the STM32 microcontroller segments the AC-coupled abrasive particle sensing signal sequence to obtain a segmented signal sequence; based on the segmented signal sequence, the STM32 microcontroller uses the Grey Wolf algorithm to optimize the parameters of the variational mode decomposition method and selects the optimal parameters of the variational mode decomposition method; according to the optimal parameters, the STM32 microcontroller decomposes the segmented signal sequence into multiple IMF components using the variational mode decomposition method.
5. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 4, characterized in that, Based on the multiple IMF components, the STM32 microcontroller calculates the correlation coefficient between each IMF component and the original signal; For the noisy component, the STM32 microcontroller uses an improved wavelet threshold to denoise the noisy component repeatedly until the correlation coefficient of the denoised noisy component meets a preset threshold. The noisy component with the correlation coefficient meeting the preset threshold is then converted into a clean component. Based on the pure components obtained by direct partitioning and the pure components obtained by noise reduction conversion, the STM32 microcontroller superimposes and reconstructs all pure components to obtain and output the pure abrasive grain signal.
6. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the step of performing abrasive pulse feature extraction, the STM32 microcontroller performs sliding window peak detection on the pure abrasive signal in the time domain to obtain the signal amplitude in the time series; the signal amplitude is compared with a set dynamic noise threshold, and when the signal amplitude exceeds the dynamic noise threshold, it is determined that the effective abrasive event has occurred; Extract the peak voltage and pulse half-wave width at the time of the effective abrasive event, and record the timestamp corresponding to the effective abrasive event.
7. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the step of determining the abrasive grain size and material, the mapping curve is a mapping curve between the peak voltage and the equivalent diameter of the abrasive grain established by calibration on a calibration table using standard spherical particles.
8. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the step of determining the size and material of the abrasive grains, the material properties include ferromagnetic metal materials, non-ferromagnetic metal materials, and non-metallic materials. The STM32 microcontroller determines the material properties of the abrasive grains by combining the signal amplitude changes of the inductor channel and the capacitor channel. When the inductor channel signal is a large positive inductor pulse and the capacitor channel signal is a large positive capacitor pulse, the abrasive material is determined to be a ferromagnetic metal material. When the inductor channel signal shows a small negative inductor pulse and the capacitor channel signal shows a large positive capacitor pulse, the abrasive material is determined to be a non-ferromagnetic metal material. When the inductor channel signal shows almost no inductive signal and the capacitor channel signal shows a weak capacitor pulse, the abrasive material is determined to be a non-metallic material.
9. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 1, characterized in that, In the step of performing comprehensive health status diagnosis, the STM32 microcontroller uses the number of abrasive grains in each size range and the yield of large abrasive grains as status feature data, and compares the status feature data with the system's preset three-level alarm threshold. When the equivalent diameter of the abrasive particles detected per unit time is less than 50μm and the yield of large abrasive particles is lower than a preset threshold, the current health status is determined to be normal. When the number of abrasive particles with a size in the range of 50μm to 100μm increases monotonically over multiple consecutive evaluation cycles, the current health status is determined to be a warning state. When large abrasive particles exceeding 100 μm in size are found, or when the ratio of the yield of such large abrasive particles in two adjacent evaluation cycles is greater than or equal to a preset exponential surge coefficient, the current health status is determined to be an alarm state.
10. The intelligent health management system based on a capacitive complementary lubricating oil abrasive sensor according to claim 9, characterized in that, When the current health status is determined to be the alarm state, the STM32 microcontroller generates a specific fault code and sends the specific fault code through the CAN bus in the communication and control link; at the same time, the STM32 microcontroller controls the external LED indicator to flash or controls the buzzer to sound an alarm.
