Method and system for detecting cleanliness of lubricating oil based on intelligent sensing element

Abstract

The application discloses a lubricating oil cleanliness detection method and system based on intelligent sensing elements, and relates to the technical field of lubricating oil detection. The method comprises the following steps: constructing a detection branch in a lubricating oil circulation loop; extracting a lubricating oil sample in the lubricating oil circulation loop through the detection branch to flow through a high-frequency impedance sensing area, an optical sensing area and a micro-pressure difference sensing area in sequence, taking an optical pulse event as a main trigger event, and establishing an associated time window for the main trigger event; inquiring whether there are associated signal events in the high-frequency impedance sensing area and the micro-pressure difference sensing area; based on the inquiry result, performing multi-sensor feature fusion decision, completing effective solid particle counting, and generating a cleanliness detection result. The technical problems that the lubricating oil cleanliness detection precision is low and the solid particles in the lubricating oil cannot be effectively identified in the prior art are solved, the accurate identification of the solid particles is achieved through multi-sensing feature fusion, and the technical effect of improving the lubricating oil cleanliness detection precision is achieved.

Description

Technical Field

[0001] This invention relates to the field of lubricating oil testing technology, and more specifically to a method and system for testing the cleanliness of lubricating oil based on intelligent sensing elements. Background Technology

[0002] Lubricating oil cleanliness is a crucial indicator of the health of a mechanical equipment's lubrication system, directly impacting wear, service life, and reliability. Current lubricating oil cleanliness testing primarily relies on offline sampling analysis or single-sensor signal detection methods, such as optical particle counting, impedance analysis, or pressure difference methods. However, offline sampling suffers from long cycles and poor real-time performance, making it difficult to reflect the lubricating oil's operating status promptly. Single-sensor methods are susceptible to interference from factors like air bubbles, oil temperature fluctuations, and changes in impurity morphology, leading to insufficient particle identification accuracy and high false and false detection rates. Furthermore, different types of solid particles (such as metal shavings, carbon powder, and impurity particles) differ in their optical and electrical characteristics, making accurate differentiation difficult with single-channel detection methods. Summary of the Invention

[0003] This application provides a method and system for detecting the cleanliness of lubricating oil based on intelligent sensing elements, which solves the technical problems of low accuracy in detecting the cleanliness of lubricating oil and the inability to effectively identify solid particles in lubricating oil in the prior art.

[0004] The first aspect of this application provides a method for detecting the cleanliness of lubricating oil based on a smart sensing element, the method comprising:

[0005] A detection branch is constructed in the lubricating oil circulation loop, comprising a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area. A lubricating oil sample is extracted from the lubricating oil circulation loop through the detection branch and flows sequentially through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area. An optical pulse event detected by the optical sensing area is used as the primary triggering event, and an associated time window is established for this primary triggering event. For the primary triggering event, within the associated time window, it is queried whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area. Based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate a cleanliness detection result.

[0006] A second aspect of this application provides a lubricating oil cleanliness detection system based on intelligent sensing elements, the system comprising:

[0007] The detection branch construction module constructs a detection branch in the lubricating oil circulation loop, which includes a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area. The sample detection module extracts a lubricating oil sample from the lubricating oil circulation loop through the detection branch, allowing it to flow sequentially through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area. An optical pulse event detected by the optical sensing area is used as the main triggering event, and an associated time window is established for this main triggering event. The associated signal query module queries whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area within the associated time window for the main triggering event. The feature fusion module performs multi-sensor feature fusion decision-making based on the query results, completes the effective solid particle count, and generates a cleanliness detection result.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, a detection branch is constructed in the lubricating oil circulation loop, comprising a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area. Then, a lubricating oil sample is extracted from the circulation loop via the detection branch and flows sequentially through these three areas. An optical pulse event detected by the optical sensing area is used as the primary trigger event, and a correlation time window is established for this event. Further, for the primary trigger event, within the correlation time window, it is checked whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area. Finally, based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate a cleanliness detection result. This solves the technical problems of low accuracy in lubricating oil cleanliness detection and the inability to effectively identify solid particles in lubricating oil in existing technologies. It achieves the technical effect of accurately identifying solid particles through multi-sensor feature fusion, thereby improving the accuracy of lubricating oil cleanliness detection. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the 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.

[0011] Figure 1 A schematic flowchart of a lubricating oil cleanliness detection method based on intelligent sensing elements provided in an embodiment of this application;

[0012] Figure 2 This is a schematic diagram of the structure of a lubricating oil cleanliness detection system based on intelligent sensing elements provided in an embodiment of this application.

[0013] Figure labeling: Detection branch construction module 11, sample detection module 12, correlation signal query module 13, feature fusion module 14. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structure, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1, as Figure 1 As shown, this application provides a method for detecting the cleanliness of lubricating oil based on intelligent sensing elements, wherein the method includes:

[0016] A detection branch is constructed in the lubricating oil circulation loop, the detection branch including a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area.

[0017] A detection branch is set up in the lubricating oil circulation loop. The detection branch is connected to the main lubricating oil pipeline via a tee connector, forming a controllable bypass structure for extracting a portion of the lubricating oil sample from the main loop for online testing. A solenoid valve and a flow stabilizing device are installed at the inlet end of the detection branch to control the stability of the extraction flow rate and to link with the flow feedback signal of the main loop to achieve adaptive flow rate adjustment.

[0018] The detection branch is sequentially arranged with a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area. Specifically: the high-frequency impedance sensing area uses a pair of parallel plate electrodes embedded in the flow channel wall, connected to a high-frequency excitation power supply and a signal acquisition module, to detect changes in the dielectric constant caused by particles, bubbles, and contaminants in the oil; the optical sensing area consists of a aligned light source and a photodetector, arranged on opposite sides of the flow channel. The flow channel cross-section is designed as a flat rectangle to form an optical sheath flow focusing, so that particles passing through the beam generate a recognizable optical pulse signal; the micro-differential pressure sensing area consists of a capillary section with a gradually decreasing cross-sectional area that remains constant. The inlet and outlet are respectively connected to high-sensitivity micro-differential pressure sensors to sense transient pressure fluctuations caused by particles passing through. All the sensing areas are connected by short-distance buffer flow channels. The interior of these channels is smooth and without sharp angles to reduce secondary turbulence.

[0019] Furthermore, the high-frequency impedance sensing region is composed of a pair of parallel plate electrodes embedded in the channel wall of the detection branch; the optical sensing region is composed of a collimated light source and a photodetector pair located on both sides of the channel, and the cross-section of the channel in the optical sensing region is constructed as a flat rectangle to form an optical sheath flow focusing effect; the micro-differential pressure sensing region is composed of a capillary segment whose cross-sectional area gradually decreases and then remains constant, with a micro-differential pressure sensor connected to the inlet end and the outlet end respectively; the high-frequency impedance sensing region and the optical sensing region, as well as the optical sensing region and the micro-differential pressure sensing region, are all connected by a buffer channel.

[0020] The high-frequency impedance sensing area consists of a pair of parallel plate electrodes, which are embedded in the inner surface of the flow channel wall of the detection branch. The electrode material is preferably a high-temperature resistant and corrosion-resistant conductive alloy. The electrodes are connected to the signal acquisition module through a high-frequency excitation source, with the excitation frequency set in the range of 100kHz to 10MHz, for detecting changes in the dielectric constant of the lubricating oil and transient impedance fluctuations caused by the passage of solid particles.

[0021] The optical sensing region is located downstream of the high-frequency impedance sensing region and consists of collimated light sources and photodetector pairs located on opposite sides of the flow channel. The collimated light source is preferably a semiconductor laser with a wavelength in the range of 630nm to 850nm. The emitted light is collimated by a collimating lens to form a narrow beam that passes through the central region of the flow channel. The photodetector is a high-sensitivity PIN photodiode used to receive the optical pulse signal generated by particles blocking light. The cross-section of the flow channel within the optical sensing region is processed into a flat rectangular structure, with a preferred aspect ratio of 3:1 to 5:1, to create an optical sheath focusing effect in the fluid, causing solid particles to concentrate at the center of the optical path, thereby obtaining a clear optical pulse signal.

[0022] The micro-differential pressure sensing area is located downstream of the optical sensing area and consists of a capillary segment whose cross-sectional area gradually decreases and then remains constant. The inner diameter of the capillary is preferably 0.5 mm to 1.5 mm, and the ratio of the length of the contracted section to the length of the constant diameter section is 1:(3~5), which is used to enhance the local fluid pressure difference response when particles pass through. A micro-differential pressure sensor is connected to the inlet and outlet ends of the capillary, respectively. The sensor sensitivity is not less than 10 Pa, and the sampling frequency is in the range of 1 kHz to 10 kHz, which is used to capture transient pressure difference waveforms.

[0023] The high-frequency impedance sensing area and the optical sensing area, as well as the optical sensing area and the micro-differential pressure sensing area, are all connected by a buffer channel. The length of the buffer channel is preferably 10 to 20 times the channel diameter. The inner wall is polished to reduce the risk of turbulence and particle deposition, thereby ensuring that the oil sample maintains a stable laminar flow state in each sensing area.

[0024] The detection branch extracts a lubricating oil sample from the lubricating oil circulation loop and sequentially flows through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area. The optical pulse event detected by the optical sensing area is used as the main triggering event, and an associated time window is established for the main triggering event.

[0025] When a lubricating oil sample is extracted from the lubricating oil circulation loop via the detection branch, the inlet of the detection branch is connected to the main circulation pipeline, and the outlet returns to the main loop through a reflux valve, forming a closed bypass channel. The lubricating oil sample flowing through the detection branch maintains a constant flow velocity under the action of a flow stabilization device, preferably within the range of 0.05~0.3 m / s, to ensure accurate calculation of the transmission delay between the various sensing zones. The lubricating oil sample flows sequentially through the high-frequency impedance sensing zone, the optical sensing zone, and the micro-differential pressure sensing zone. During this process, the optical sensing zone monitors in real time the blocking effect of passing particles or bubbles on the light beam. When the photodetector output signal shows a momentary drop or rise pulse with an amplitude exceeding a preset threshold, it is defined as an optical pulse event. This optical pulse event is set as the main trigger event to initiate time synchronization and data correlation analysis of multiple sensing signals.

[0026] Furthermore, establish associated time windows for each main triggering event, including:

[0027] Acquire real-time flow velocity data of the detection branch; calculate a first transmission delay based on the real-time flow velocity data and the flow path distance from the high-frequency impedance sensing area to the optical sensing area; calculate a second transmission delay based on the real-time flow velocity data and the flow path distance from the optical sensing area to the micro-differential pressure sensing area; establish a forward correlation time window by tracing back the first transmission delay based on the occurrence time of the main triggering event, for correlation of signal events in the high-frequency impedance sensing area; establish a backward correlation time window by extending the second transmission delay backward based on the occurrence time of the main triggering event, for correlation of signal events in the micro-differential pressure sensing area.

[0028] Specifically, real-time flow velocity data within the detection branch is acquired, and this flow velocity data is collected by a miniature flow sensor installed at the inlet or steady-flow section of the detection branch. After signal filtering and averaging, the flow velocity data serves as a reference for the flow velocity of the oil sample between the various sensing zones. Based on the structural parameters of the detection branch, the flow path distance from the high-frequency impedance sensing zone to the optical sensing zone is pre-recorded. And the flow channel distance from the optical sensing area to the micro-pressure difference sensing area. Based on the real-time flow velocity v, the transmission delay time of each segment is calculated, and the following results are obtained: First transmission delay This describes the time required for particles in an oil sample to flow from the high-frequency impedance sensing region to the optical sensing region; the second propagation delay. This is used to describe the time it takes for particles to flow from the optical sensing area to the micro-pressure sensing area.

[0029] When the optical sensing area detects a main trigger event, the system uses the timestamp of the event as a reference to trace back a time period and establish a forward correlation time window to retrieve the impedance signal data of the high-frequency impedance sensing area during that time period; at the same time, it extends the time period backward and establishes a backward correlation time window to retrieve the differential pressure signal data of the micro differential pressure sensing area during that time period.

[0030] For the main triggering event, within the associated time window, query whether there are any associated signal events in the high-frequency impedance sensing area and the micro differential pressure sensing area.

[0031] Furthermore, for the main triggering event, within the associated time window, querying whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area includes:

[0032] Within the associated time window, the raw impedance signal data collected by the high-frequency impedance sensor in the high-frequency impedance sensing area is extracted; the raw impedance signal data is processed to identify abrupt pulses in signal amplitude and phase angle; the phase angle of the abrupt pulse is compared with a preset phase threshold, wherein the phase threshold is set based on the difference in dielectric constants of bubbles, oil, and water; if an abrupt pulse with a phase angle lower than the preset phase threshold is identified, it is determined that there is an associated bubble characteristic signal event in the high-frequency impedance sensing area.

[0033] Within the forward correlation time window, the raw impedance signal data for the corresponding time period is extracted from the high-frequency impedance sensor in the high-frequency impedance sensing area. Digital filtering and baseline normalization are performed on the raw impedance signal data. The filtering uses a finite impulse response (FIR) algorithm to suppress high-frequency noise, and a moving average (MA) algorithm is used to remove low-frequency drift caused by fluid disturbances. The processed impedance signal contains real-time amplitude and phase angle information. The system performs point-by-point derivative analysis on the signal to extract abrupt amplitude changes and corresponding phase angle changes. When the amplitude difference between consecutive sampling points exceeds a preset amplitude threshold (e.g., set to 5 times the steady-state root mean square error) and the phase angle change rate is higher than a preset change rate threshold, the point is determined to be an abrupt pulse event. Subsequently, the phase angle of the abrupt pulse is compared with a preset phase threshold, which can be determined through pre-experimental calibration as approximately 80% to 85% of the oil baseline phase angle. If the detected abrupt pulse phase angle is lower than the preset phase threshold, the system determines the high-frequency impedance event as a bubble characteristic signal event and records the event timestamp as a bubble signal associated with the current main triggering event.

[0034] Furthermore, for the main triggering event, within the associated time window, querying whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area includes:

[0035] Within the associated time window, the raw differential pressure signal data collected by the micro differential pressure sensor is extracted; based on the real-time flow velocity data of the detection branch, the raw differential pressure signal is standardized and compensated to eliminate the influence of flow velocity fluctuations on the pressure baseline, and the attenuation coefficient of the standardized differential pressure signal is calculated as a compressibility feature value; the compressibility feature value is compared with a preset feature threshold, and if the compressibility feature value is lower than the preset feature threshold, it is determined that there is an associated compressible feature signal event in the micro differential pressure sensing area.

[0036] Within the backward correlation time window, the raw differential pressure signal data for the corresponding time period is extracted from the sensor data of the micro differential pressure sensing area. The raw differential pressure signal data is then standardized with flow rate compensation to eliminate the pressure baseline offset caused by instantaneous flow fluctuations. Specifically, the differential pressure signal amplitude is adjusted according to the ratio of real-time flow rate to steady-state average flow rate to decouple the signal from flow rate changes. Next, the moving average algorithm is used to remove the slowly changing pressure baseline trend to obtain a stable standardized differential pressure signal.

[0037] After obtaining the standardized differential pressure signal, the signal change process is analyzed, and the pressure fluctuation recovery rate is calculated. This recovery rate is then used as a characteristic parameter reflecting the compressibility of the fluid, namely the compressibility characteristic value. The lower the compressibility characteristic value, the slower the fluid recovers after local disturbance, indicating a buffering effect caused by compressible materials. Conversely, a higher compressibility characteristic value indicates that the disturbance was caused by the passage of solid particles, and the oil recovers rapidly.

[0038] The system compares the calculated compressibility feature value with a preset feature threshold. When the compressibility feature value is lower than the feature threshold, it is determined that there is a compressible signal event related to the main trigger event time in the micro differential pressure sensing area.

[0039] Based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate cleanliness detection results.

[0040] Furthermore, based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate cleanliness detection results, including:

[0041] Based on the query results, determine whether the main triggering event is simultaneously associated with the bubble characteristic signal event of the high-frequency impedance sensing area and the compressible material characteristic signal event of the micro-differential pressure sensing area; if so, determine that the main triggering event is bubble interference and remove it; if not, determine that the main triggering event is effective solid particles and count them to generate the cleanliness detection result.

[0042] For each primary triggering event, the system acquires three signal identification markers within its associated time window: presence of bubble characteristic signal event, presence of compressible material characteristic signal event, and optical pulse characteristic. Based on timestamps, the system performs matching analysis on these three types of signals. When the timestamp of the primary triggering event overlaps with the time point marked as a bubble characteristic signal event in the high-frequency impedance sensing area, and simultaneously has a temporal proximity (time difference less than the fluid transport delay) with the compressible material characteristic signal event in the micro-differential pressure sensing area, the system considers the optical signal to be caused by bubbles or compressible impurities. In this case, the primary triggering event is marked as a "bubble interference event" and removed from the particle count dataset to prevent misjudgments of cleanliness caused by false particle counts.

[0043] If the detection results show that the main triggering event is not simultaneously associated with the aforementioned bubble characteristic signal event or compressible material characteristic signal event, the system assumes that the optical pulse originates from the passage of actual solid particles. In this case, the optical signal waveform of the event is quantitatively analyzed, and the equivalent particle size is calculated based on the pulse amplitude and width, and recorded as a valid particle event. The system accumulates and statistically analyzes all valid particle events within a preset detection cycle to obtain the total number of valid particles. Combining this with standard lubricant cleanliness evaluation criteria (such as ISO 4406 level or NAS 1638 standard), the system generates corresponding cleanliness level results based on the particle quantity distribution within different particle size ranges.

[0044] Furthermore, after performing multi-sensor feature fusion decision-making, it also includes:

[0045] Based on the query results, a primary triggering event associated with either the bubble characteristic signal event in the high-frequency impedance sensing area or the compressible material characteristic signal event in the micro-pressure difference sensing area is identified and marked as an ambiguous event. The optical pulse amplitude, high-frequency impedance phase angle change, and micro-pressure difference compressibility feature value corresponding to the ambiguous event are extracted to form a multi-dimensional feature vector. The multi-dimensional feature vector is input into a pre-trained lightweight machine learning classification model. Based on the classification results, the ambiguous event is determined to be either a solid particle or a bubble. Based on the determination results, counting or elimination operations are performed.

[0046] Specifically, when a primary triggering event is temporally correlated only with any one type of characteristic signal event in the high-frequency impedance sensing area or the micro-differential pressure sensing area, and does not simultaneously meet the dual-area matching condition, the primary triggering event is marked as an ambiguous event. The system extracts multi-dimensional feature parameters corresponding to the ambiguous event from the multi-sensor signal database, including: optical pulse amplitude, high-frequency impedance phase angle change, and micro-differential pressure compressibility feature value. The above three feature quantities are used to form a feature vector and input into a pre-trained lightweight machine learning classification model. The lightweight classification model can use a support vector machine or a lightweight convolutional neural network. The training samples of the model come from a particle-bubble mixed fluid sample dataset calibrated in a laboratory environment, where each sample contains synchronously acquired optical, impedance, and differential pressure signals and their real labels (solid particles or bubbles). During the training phase, the model optimizes the parameters by minimizing the cross-entropy loss function, enabling it to learn the discrimination boundaries of various signal distributions in the feature space.

[0047] The system inputs the feature vector of the real-time fuzzy event into the model and outputs the category prediction result. If the classification result is solid particles, the system includes the event in the valid particle statistics; if the classification result is bubbles or compressible impurities, they are discarded.

[0048] Furthermore, generating cleanliness test results also includes:

[0049] Within a preset single detection cycle, the proportion of events marked as ambiguous to the total number of triggered events is defined as the ambiguity rate. If the ambiguity rate is lower than a first preset threshold, the cleanliness detection result is directly output with a high reliability label. If the ambiguity rate is higher than the first preset threshold but lower than a second preset threshold, the cleanliness detection result is output with a suggested review label. If the ambiguity rate is higher than the second preset threshold, the detection result is determined to be unreliable, and a sensor abnormality or oil condition abnormality alarm is generated.

[0050] In generating cleanliness test results, in addition to outputting the effective solid particle count, the system also quantitatively evaluates the reliability of the test results based on the statistical results of fuzzy events. Specifically, within each preset detection cycle, the total number of optical main triggering events is counted, as well as the number of events marked as fuzzy events after multi-sensor fusion and machine learning; the proportion of fuzzy events is calculated and defined as the fuzziness rate. The fuzziness rate reflects the proportion of uncertain samples during the detection process and can be used to measure the state stability of the detection system and the complexity of the oil flow state.

[0051] The system classifies reliability based on the ambiguity rate and preset thresholds: when the ambiguity rate is below the first preset threshold, it indicates a stable detection process and high signal consistency; the system directly outputs the cleanliness detection result with a high reliability label. When the ambiguity rate is between the first and second preset thresholds, it indicates slight fluctuations in the oil condition or signal; the system outputs the cleanliness detection result with a suggested review label to prompt operators to retest in subsequent cycles. When the ambiguity rate is above the second preset threshold, it indicates a significant anomaly in the current detection environment; the system determines the detection result is unreliable and automatically generates an abnormal alarm signal. Specifically, when multiple sensor signal waveforms simultaneously exhibit baseline drift or a sudden increase in noise, the system generates a sensor abnormality alarm; when the frequency of optical trigger events changes drastically and the micro-differential pressure signal deviates from its steady-state trend, the system generates an oil condition abnormality alarm.

[0052] Furthermore, it also includes self-diagnostic and calibration steps:

[0053] Standard calibration particles of known size and material are periodically injected into the detection branch; signals generated when the standard calibration particles flow through the high-frequency impedance sensing area, optical sensing area, and micro-differential pressure sensing area are simultaneously collected; the particle size detected by the optical sensing area is compared with the standard value; if there is a deviation, the optical size calculation model is automatically calibrated; bubble identification verification is completed by verifying the correct response of the high-frequency impedance sensing area and the micro-differential pressure sensing area to the standard calibration particles.

[0054] The self-diagnosis and calibration process is triggered periodically, with the cycle automatically set based on system uptime or cumulative testing counts, such as once every 1000 testing cycles or every 24 hours. During the calibration cycle, the system injects pre-made standard calibration particle samples into the testing branch through a built-in calibration module or an external calibration unit. These particles have known particle size and material and are suspended in a base lubricating oil medium to simulate the actual testing environment.

[0055] As the standard calibration particle passes through the detection branch, the output signals from the high-frequency impedance sensing area, optical sensing area, and micro-differential pressure sensing area are acquired. The particle pulse signal detected by the optical sensing area is used to calculate the equivalent optical size of the particle. The system compares this calculated value with the actual size of the standard particle. When there is a systematic deviation in the comparison result (e.g., the deviation exceeds 5%), the system automatically corrects the calibration coefficients of the optical size calculation model, including the beam width compensation parameter, the transmitted light intensity linearization coefficient, and the detection gain parameter, to restore the accuracy of the size measurement.

[0056] Simultaneously, the signal responses of the high-frequency impedance sensing region and the micro-differential pressure sensing region are cross-validated. For each standard particle passing event, the system checks whether there are corresponding amplitude abrupt changes and reasonable phase responses in the impedance signal, confirming that they are consistent with the typical electrical characteristics of solid particles. If the high-frequency impedance signal shows a bubble-like low phase response or the micro-differential pressure signal shows a low attenuation characteristic, it is determined that there is abnormal drift or sensitivity attenuation in the sensing channel, and the system automatically performs bubble identification verification and sensing channel health detection. If the abnormality persists, a sensor status abnormality prompt is triggered, suggesting manual maintenance or automatic recalibration.

[0057] In summary, the embodiments of this application have at least the following technical effects:

[0058] First, a detection branch is constructed in the lubricating oil circulation loop, comprising a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area. Then, a lubricating oil sample is extracted from the circulation loop via the detection branch and flows sequentially through these three areas. An optical pulse event detected by the optical sensing area is used as the primary trigger event, and a correlation time window is established for this event. Further, for the primary trigger event, within the correlation time window, it is checked whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area. Finally, based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate a cleanliness detection result. This solves the technical problems of low accuracy in lubricating oil cleanliness detection and the inability to effectively identify solid particles in lubricating oil in existing technologies. It achieves the technical effect of accurately identifying solid particles through multi-sensor feature fusion, thereby improving the accuracy of lubricating oil cleanliness detection.

[0059] Example 2, based on the same inventive concept as the lubricating oil cleanliness detection method based on intelligent sensing elements in the foregoing examples, such as... Figure 2 As shown, this application provides a lubricating oil cleanliness detection system based on intelligent sensing elements, wherein the system includes:

[0060] Detection branch construction module 11: Constructs a detection branch in the lubricating oil circulation loop, the detection branch including a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area; Sample detection module 12: Extracts lubricating oil samples from the lubricating oil circulation loop through the detection branch, which sequentially flow through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area, using the optical pulse event detected by the optical sensing area as the main triggering event, and establishing an associated time window for the main triggering event; Associated signal query module 13: For the main triggering event, within the associated time window, queries whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area; Feature fusion module 14: Based on the query results, performs multi-sensor feature fusion decision-making, completes the effective solid particle count, and generates cleanliness detection results.

[0061] Furthermore, the sample detection module 12 is used to perform the following method:

[0062] Acquire real-time flow velocity data of the detection branch; calculate a first transmission delay based on the real-time flow velocity data and the flow path distance from the high-frequency impedance sensing area to the optical sensing area; calculate a second transmission delay based on the real-time flow velocity data and the flow path distance from the optical sensing area to the micro-differential pressure sensing area; establish a forward correlation time window by tracing back the first transmission delay based on the occurrence time of the main triggering event, for correlation of signal events in the high-frequency impedance sensing area; establish a backward correlation time window by extending the second transmission delay backward based on the occurrence time of the main triggering event, for correlation of signal events in the micro-differential pressure sensing area.

[0063] Furthermore, the associated signal query module 13 is used to perform the following method:

[0064] Within the associated time window, the raw impedance signal data collected by the high-frequency impedance sensor in the high-frequency impedance sensing area is extracted; the raw impedance signal data is processed to identify abrupt pulses in signal amplitude and phase angle; the phase angle of the abrupt pulse is compared with a preset phase threshold, wherein the phase threshold is set based on the difference in dielectric constants of bubbles, oil, and water; if an abrupt pulse with a phase angle lower than the preset phase threshold is identified, it is determined that there is an associated bubble characteristic signal event in the high-frequency impedance sensing area.

[0065] Furthermore, the associated signal query module 13 is used to perform the following method:

[0066] Within the associated time window, the raw differential pressure signal data collected by the micro differential pressure sensor is extracted; based on the real-time flow velocity data of the detection branch, the raw differential pressure signal is standardized and compensated to eliminate the influence of flow velocity fluctuations on the pressure baseline, and the attenuation coefficient of the standardized differential pressure signal is calculated as a compressibility feature value; the compressibility feature value is compared with a preset feature threshold, and if the compressibility feature value is lower than the preset feature threshold, it is determined that there is an associated compressible feature signal event in the micro differential pressure sensing area.

[0067] Furthermore, the feature fusion module 14 is used to perform the following method:

[0068] Based on the query results, determine whether the main triggering event is simultaneously associated with the bubble characteristic signal event of the high-frequency impedance sensing area and the compressible material characteristic signal event of the micro-differential pressure sensing area; if so, determine that the main triggering event is bubble interference and remove it; if not, determine that the main triggering event is effective solid particles and count them to generate the cleanliness detection result.

[0069] Furthermore, the feature fusion module 14 is used to perform the following method:

[0070] Based on the query results, a primary triggering event associated with either the bubble characteristic signal event in the high-frequency impedance sensing area or the compressible material characteristic signal event in the micro-pressure difference sensing area is identified and marked as an ambiguous event. The optical pulse amplitude, high-frequency impedance phase angle change, and micro-pressure difference compressibility feature value corresponding to the ambiguous event are extracted to form a multi-dimensional feature vector. The multi-dimensional feature vector is input into a pre-trained lightweight machine learning classification model. Based on the classification results, the ambiguous event is determined to be either a solid particle or a bubble. Based on the determination results, counting or elimination operations are performed.

[0071] Furthermore, the detection branch construction module 11 is used to perform the following method:

[0072] The high-frequency impedance sensing region consists of a pair of parallel plate electrodes embedded in the channel wall of the detection branch; the optical sensing region consists of a collimated light source and a photodetector pair located on both sides of the channel, and the cross-section of the channel in the optical sensing region is constructed as a flat rectangle to form an optical sheath flow focusing effect; the micro-differential pressure sensing region consists of a capillary segment whose cross-sectional area gradually decreases and then remains constant, with a micro-differential pressure sensor connected to the inlet end and the outlet end respectively; the high-frequency impedance sensing region and the optical sensing region, as well as the optical sensing region and the micro-differential pressure sensing region, are all connected by a buffer channel.

[0073] Furthermore, the feature fusion module 14 is used to perform the following method:

[0074] Within a preset single detection cycle, the proportion of events marked as ambiguous to the total number of triggered events is defined as the ambiguity rate. If the ambiguity rate is lower than a first preset threshold, the cleanliness detection result is directly output with a high reliability label. If the ambiguity rate is higher than the first preset threshold but lower than a second preset threshold, the cleanliness detection result is output with a suggested review label. If the ambiguity rate is higher than the second preset threshold, the detection result is determined to be unreliable, and a sensor abnormality or oil condition abnormality alarm is generated.

[0075] Furthermore, the feature fusion module 14 is used to perform the following method:

[0076] Standard calibration particles of known size and material are periodically injected into the detection branch; signals generated when the standard calibration particles flow through the high-frequency impedance sensing area, optical sensing area, and micro-differential pressure sensing area are simultaneously collected; the particle size detected by the optical sensing area is compared with the standard value; if there is a deviation, the optical size calculation model is automatically calibrated; bubble identification verification is completed by verifying the correct response of the high-frequency impedance sensing area and the micro-differential pressure sensing area to the standard calibration particles.

[0077] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for detecting the cleanliness of lubricating oil based on intelligent sensing elements, characterized in that, The method includes: A detection branch is constructed in the lubricating oil circulation loop, the detection branch including a high-frequency impedance sensing area, an optical sensing area and a micro-differential pressure sensing area; The detection branch extracts a lubricating oil sample from the lubricating oil circulation loop and flows it sequentially through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area. The optical pulse event detected by the optical sensing area is used as the main triggering event, and an associated time window is established for the main triggering event. For the main triggering event, within the associated time window, query whether there are associated signal events between the high-frequency impedance sensing area and the micro differential pressure sensing area; Based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate cleanliness detection results; This includes establishing an associated time window for each main triggering event, including: Obtain the real-time flow velocity data of the detection branch; Based on the real-time flow velocity data and the flow channel distance from the high-frequency impedance sensing area to the optical sensing area, the first transmission delay is calculated. Based on the real-time flow velocity data and the flow channel distance from the optical sensing area to the micro differential pressure sensing area, the second transmission delay is calculated. Based on the occurrence time of the main triggering event, the first transmission delay is traced back to establish a forward correlation time window for correlation of signal events in the high-frequency impedance sensing area. Based on the occurrence time of the main triggering event, the second transmission delay is extended backward to establish a backward correlation time window, which is used to correlate the signal events of the micro differential pressure sensing area. Based on the query results, multi-sensor feature fusion decision-making is performed to complete the effective solid particle count and generate cleanliness detection results, including: Based on the query results, determine whether the main triggering event is simultaneously associated with the bubble characteristic signal event of the high-frequency impedance sensing area and the compressible material characteristic signal event of the micro-differential pressure sensing area. If so, determine that the main triggering event is bubble interference and remove it; If not, determine that the main triggering event is a valid solid particle and count it to generate the cleanliness detection result; This process, following multi-sensor feature fusion decision-making, also includes: Based on the query results, identify the main triggering event associated with one of the bubble characteristic signal events in the high-frequency impedance sensing area and the compressible material characteristic signal events in the micro-differential pressure sensing area, and mark it as an fuzzy event. The optical pulse amplitude, high-frequency impedance phase angle change, and micro-pressure difference compressibility feature value corresponding to the blurred event are extracted to form a multi-dimensional feature vector; The multidimensional feature vector is input into a pre-trained lightweight machine learning classification model. Based on the classification results, the ambiguous event is determined to be either a solid particle or a bubble. Based on the determination results, a counting or elimination operation is performed.

2. The lubricating oil cleanliness detection method based on intelligent sensing elements as described in claim 1, characterized in that, For the main triggering event, within the associated time window, query whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area, including: Within the associated time window, the raw impedance signal data collected by the high-frequency impedance sensor in the high-frequency impedance sensing area is extracted; The original impedance signal data is processed to identify abrupt pulses in signal amplitude and phase angle; The phase angle of the abrupt pulse is compared with a preset phase threshold, wherein the phase threshold is set based on the difference in dielectric constants of bubbles, oil and water; If a sudden pulse with a phase angle lower than the preset phase threshold is detected, it is determined that there is an associated bubble characteristic signal event in the high-frequency impedance sensing area.

3. The lubricating oil cleanliness detection method based on intelligent sensing elements as described in claim 1, characterized in that, For the main triggering event, within the associated time window, query whether there are associated signal events in the high-frequency impedance sensing area and the micro-differential pressure sensing area, including: Within the associated time window, the raw differential pressure signal data collected by the micro differential pressure sensor is extracted; Based on the real-time flow velocity data of the detection branch, the original differential pressure signal is standardized and compensated to eliminate the influence of flow velocity fluctuations on the pressure baseline. The attenuation coefficient of the standardized differential pressure signal is calculated as a compressibility characteristic value. The compressibility feature value is compared with a preset feature threshold. If the compressibility feature value is lower than the preset feature threshold, it is determined that there is an associated compressible feature signal event in the micro-pressure differential sensing area.

4. The lubricating oil cleanliness detection method based on intelligent sensing elements as described in claim 1, characterized in that, The high-frequency impedance sensing region is composed of a pair of parallel plate electrodes embedded in the channel wall of the detection branch; the optical sensing region is composed of collimated light sources and photodetector pairs located on both sides of the channel, and the cross-section of the channel in the optical sensing region is constructed as a flat rectangle to form an optical sheath focusing effect. The micro differential pressure sensing area is composed of a capillary segment whose cross-sectional area gradually decreases and then remains constant, with a micro differential pressure sensor connected to the inlet end and the outlet end respectively. The high-frequency impedance sensing area and the optical sensing area, as well as the optical sensing area and the micro-differential pressure sensing area, are all connected by a buffer channel.

5. The lubricating oil cleanliness detection method based on intelligent sensing elements as described in claim 1, characterized in that, The generation of cleanliness test results also includes: The fuzziness rate is defined as the proportion of the number of events marked as fuzzy events to the total number of triggered events within a preset single detection period. If the blur rate is lower than the first preset threshold, the cleanliness detection result is directly output and a high confidence mark is added. If the blur rate is higher than the first preset threshold but lower than the second preset threshold, the cleanliness detection result is output, and a suggested review label is added. If the ambiguity rate is higher than the second preset threshold, the detection result is determined to be unreliable, and an alarm for sensor abnormality or oil condition abnormality is generated.

6. The lubricating oil cleanliness detection method based on intelligent sensing elements as described in claim 1, characterized in that, It also includes self-diagnostic and calibration steps: Periodically inject standard calibration particles of known size and material into the detection branch; The system simultaneously collects signals generated when the standard calibration particles flow through the high-frequency impedance sensing area, optical sensing area, and micro-differential pressure sensing area. The particle size detected by the optical sensing area is compared with the standard value. If there is a deviation, the optical size calculation model is automatically calibrated. The bubble identification verification was completed by verifying the correct response of the high-frequency impedance sensing area and the micro-differential pressure sensing area to the standard calibration particles.

7. A lubricating oil cleanliness detection system based on intelligent sensing elements, characterized in that, For implementing the lubricating oil cleanliness detection method based on intelligent sensing elements according to any one of claims 1-6, the system comprises: Detection branch construction module: Construct a detection branch in the lubricating oil circulation loop, wherein the detection branch includes a high-frequency impedance sensing area, an optical sensing area, and a micro-differential pressure sensing area; Sample detection module: The detection branch extracts a lubricating oil sample from the lubricating oil circulation loop and flows it sequentially through the high-frequency impedance sensing area, the optical sensing area, and the micro-differential pressure sensing area. The optical pulse event detected by the optical sensing area is used as the main triggering event, and an associated time window is established for the main triggering event. Correlation signal query module: For the main triggering event, within the correlation time window, query whether there are any related signal events between the high-frequency impedance sensing area and the micro differential pressure sensing area; Feature fusion module: Based on the query results, it performs multi-sensor feature fusion decision-making, completes the effective solid particle count, and generates cleanliness detection results.

Images