Lubricating oil quality on-line monitoring system and method

Abstract

The invention relates to the technical field of lubricating oil quality monitoring, in particular to a lubricating oil quality on-line monitoring system and method, and the system comprises a data collection module which collects lubricating oil parameters in real time through a sensor array; the communication transmission module is used for constructing a data flow network and supporting multi-level interaction; the data analysis and diagnosis module is used for realizing state evaluation and fault early warning; the user interaction and visualization module provides a human-computer interaction interface and decision support; and the intelligent decision module realizes monitoring-maintenance closed-loop control, gives an alarm in time and realizes full-link intelligent control triggered by a maintenance action. According to the method, the prediction precision and the early warning capability of the degradation trend of the oil system are effectively improved through the improved autoregression moving average model, the real-time state of the oil is accurately reflected by dynamically adjusting the alarm threshold value, and the fault positioning precision is greatly improved through the space matching algorithm.

Description

Technical Field

[0001] This invention relates to the field of lubricating oil quality monitoring technology, and in particular to an online lubricating oil quality monitoring system and method. Background Technology

[0002] Real-time monitoring of lubricating oil quality is an effective means of determining oil change intervals or predicting potential faults. Traditional offline laboratory testing of lubricating oil suffers from poor real-time performance, small cumulative data volume, and high sampling requirements. Sampling of lubricating oil in complex equipment is difficult and the real-time performance of the tests is poor. Online real-time monitoring technology for lubricating oil can monitor the performance change trend of lubricating oil and the condition of abrasive particles in real time, thereby knowing the lubrication status and wear degree of mechanical equipment, making timely pre-diagnosis of abnormal conditions of the system and equipment, reducing the downtime for equipment inspection, achieving early warning, and greatly reducing the accident rate.

[0003] Currently, existing technologies mainly focus on the research and development of online lubricating oil monitoring devices and the acquisition of detection data. However, systematic solutions have not yet been developed for efficient analysis of real-time data, the establishment of standardized parameter systems, and accurate assessment of oil condition. In particular, when monitoring data anomalies occur, existing technologies can only provide raw oil data (such as viscosity values ​​without a unified temperature reference). This fails to achieve data standardization and lacks a continuous tracking mechanism after an early warning is triggered. Furthermore, the system cannot convert viscosity values ​​acquired under different operating conditions into trend analysis based on the same temperature reference, resulting in insufficient accuracy in condition assessment. Moreover, after the early warning information is issued, there is often a lack of follow-up anomaly tracing and handling suggestions. Summary of the Invention

[0004] The purpose of this invention is to address the problems in existing online lubricating oil monitoring technologies, such as the lack of standardized data processing based on a unified temperature benchmark, insufficient accuracy in condition assessment, and the absence of a mechanism for tracing anomalies after early warning. This invention proposes an online lubricating oil quality monitoring system and method.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: An online lubricating oil quality monitoring system, the system comprising: The data acquisition module uses a sensor array to collect real-time data on the physical and chemical parameters of lubricating oil and equipment wear information. The communication transmission module constructs a data circulation network and supports multi-level interaction. The data analysis and diagnostic module enables status assessment and fault early warning; The user interaction and visualization module provides a human-computer interaction interface and decision support. The intelligent decision-making module enables closed-loop control of monitoring and maintenance, provides timely alarms, and achieves intelligent control of the entire chain for triggering maintenance actions.

[0006] Furthermore, the sensor array employs various existing detection sensors and uses a multi-parameter synchronous detection matrix to detect the moisture, viscosity, particle size, and oxidation degree parameters of the lubricating oil.

[0007] Furthermore, the communication transmission module adopts multi-protocol dynamic adaptation and protocol stack integration. The protocol conversion gateway of the communication transmission module adopts a unified data encapsulation format to convert different protocol data into a standardized format. It automatically selects the optimal protocol through the Q-Learning algorithm and updates the Q value table in real time, which can maintain low latency in multipath interference environment. The communication transmission module adopts an edge-based hierarchical division. The edge layer uses FPGA nodes to be responsible for real-time data acquisition and uploading. The local layer industrial gateway aggregates data from multiple nodes, executes local decisions, and starts a high-frequency data transmission mode when an abnormal event is triggered. The cloud layer supports big data analysis and predictive maintenance. The cloud sends machine learning model update packages to the edge nodes to realize dynamic algorithm upgrades. The communication transmission module uses lightweight encryption with AES-128-CTR mode to encrypt transmitted data. The key is dynamically negotiated through elliptic curve Diffie-Hellman (ECDH), and HMAC signature is added to verify data integrity. The receiving end verifies the validity of the tag to ensure that the data has not been tampered with.

[0008] Furthermore, the data analysis and diagnosis module also includes: The data preprocessing unit performs denoising and feature enhancement on the raw sensor data. It uses a hybrid denoising algorithm, combining LOWESS local weighted regression and wavelet threshold denoising, to retain transient abrasive pulses while suppressing electromagnetic interference. The multi-parameter feature fusion unit integrates multi-dimensional data such as moisture, viscosity, particle size, and temperature to construct a joint feature vector. The weights are dynamically adjusted according to the signal characteristics and noise level of the sensor to ensure the accuracy and reliability of the fusion results.

[0009] Furthermore, the user interaction and visualization module includes: The integrated view unit, relying on multimodal data presentation, human-machine collaborative decision-making and interactive intelligent optimization, realizes data transparency, real-time decision-making and humanized operation. It integrates key parameters such as moisture content, viscosity and particle size to realize data correlation across time and space dimensions, analyzes and establishes dynamic weighted integrated view, and visualizes the health score. The mobile positioning unit enables 3D labeling of fault points and maintenance guidance, using a spatial matching algorithm to accurately overlay virtual fault markers and wear indicators onto the real device.

[0010] in, Represents the rigid body transformation matrix of the equipment CAD model. This indicates the position of the sensor in the model coordinate system. This represents the actual coordinates captured by the AR camera.

[0011] Furthermore, the intelligent decision-making module includes: The digital twin simulation unit performs virtual verification before triggering actual maintenance actions, based on dynamic early warning of oil deterioration and prediction of remaining service life, using an improved autoregressive moving average model.

[0012] in, This represents the state variable of the oil at the current moment. These are autoregressive coefficients, representing the influence of past states on the current state. The white noise term represents random error. The moving average coefficient represents the influence of past errors on the current state. To accelerate the weighting coefficients of degradation terms, This indicates the accelerated degradation term when nearing failure. The graded early warning unit triggers an alarm when the viscosity error of the ferromagnetic abrasive reaches a certain value. It also initiates an emergency alarm when the dielectric constant deviates from the baseline or the ARMA prediction is abnormal. When multiple parameters are abnormal, it stops the alarm and performs a self-test. The dynamic threshold adjustment unit and algorithm adaptively adjust the alarm threshold based on the oil deterioration rate, avoiding false alarms or missed alarms caused by fixed thresholds.

[0013] in, This indicates the alarm threshold that has been dynamically adjusted over time. Indicates the basic threshold. The viscosity degradation rate is predicted based on the ARMA model. This represents the near-failure acceleration factor, reflecting the urgency with which the oil approaches failure. It is calculated using a fuzzy evaluation model. , These are all empirical coefficients, and the threshold value is automatically adjusted when the oil deteriorates rapidly; The multi-actuator collaborative control unit optimizes resource allocation for oil purification, additive injection, and oil change operations. Based on the Q-Learning reinforcement learning model, it links the oil purification equipment, additive injection pump, and oil change valve to coordinate scheduling in response to emergencies. The maintenance effectiveness evaluation unit, based on the improved TOPSIS multi-objective evaluation, comprehensively considers the performance and weight allocation of multiple indicators, uses positive and negative ideal solutions as reference standards, and derives a comprehensive score, providing a scientific basis for the comprehensive evaluation of the evaluation object.

[0014] A method for online monitoring of lubricating oil quality, applied to the aforementioned online monitoring system for lubricating oil quality, includes the following steps: Step S1: Monitor the operating parameters of the lubricating oil system in real time using a sensor array; Step S2: The collected parameters are transmitted to the data analysis and diagnosis module through the communication module for preliminary processing; In step S3, the data analysis and diagnosis module uses multimodal dynamic fusion to integrate multi-dimensional data such as moisture, viscosity, particle size, and temperature to construct a joint feature vector. The weights are dynamically adjusted according to the signal characteristics and noise level of the sensors to ensure the accuracy and reliability of the results. Step S4: The user interaction and visualization module uses view fusion and spatial matching algorithms to visualize the health score and overlay virtual fault indicators onto the real device. Step S5: Based on the diagnostic results of the data analysis and diagnosis module, a virtual verification is performed before the intelligent decision-making module triggers the actual maintenance action. The alarm threshold is adaptively adjusted based on the dynamic threshold adjustment algorithm to improve the response speed and data processing accuracy. Step S6: Optimize the multi-actuator collaborative control unit to promptly respond to maintenance after an alarm is issued; Step S7 evaluates the maintenance effect, considering the performance and weighting of multiple indicators to obtain a comprehensive score, providing a scientific basis for the comprehensive evaluation of the evaluation object, ensuring that the maintenance effect reaches the expected level in all aspects, thereby improving the reliability and service life of the equipment.

[0015] The beneficial effects of the technical solution provided by this invention include at least the following: 1. This invention effectively improves the prediction accuracy and early warning capability of oil system degradation trends by introducing an improved autoregressive moving average model, and introduces an accelerated degradation term. The model can more accurately capture the degradation trend of the oil system, especially when it is close to failure, thereby improving the prediction accuracy. Based on the improved ARMA model, the algorithm can dynamically predict the remaining life of the oil and issue an early warning signal before failure, realizing dynamic early warning. This helps the staff to take timely measures to avoid equipment failure and downtime losses. The adjustment coefficient λ and acceleration coefficient k in the algorithm can be flexibly adjusted according to different oil systems and application scenarios, thereby improving the model's versatility and adaptability.

[0016] 2. This invention accurately reflects the real-time status of the oil by dynamically adjusting the alarm threshold, making it more accurate and timely, reducing false alarms or missed alarms caused by fixed thresholds, and improving the accuracy of early warning. The algorithm can adaptively adjust the alarm threshold according to different oil deterioration conditions, improving the algorithm's adaptability and robustness. Through timely and accurate early warning, this algorithm can reduce unnecessary maintenance operations, extend equipment life, and reduce maintenance costs. When the oil deteriorates rapidly or is close to failure, the algorithm can quickly tighten the alarm threshold, improving the sensitivity and timeliness of early warning, thereby ensuring the safe operation of the equipment.

[0017] 3. This invention greatly improves the accuracy of fault location through spatial matching algorithms and combines AR technology, enabling maintenance personnel to perform virtual annotation and maintenance guidance on actual equipment, reducing the time and error of manual location. This improves the automation and intelligence level of fault location, reduces the reliance on human experience, and brings significant improvements and enhancements to the field of industrial maintenance.

[0018] 4. This invention utilizes an improved TOPSIS multi-objective evaluation method, comprehensively considering the performance and weight allocation of multiple indicators, and employing positive and negative ideal solutions as reference standards to calculate the relative closeness. The performance of multiple indicators can be comprehensively considered to obtain a comprehensive score, which facilitates the ranking and comparison of the evaluated objects and provides a scientific basis for the comprehensive evaluation of the evaluated objects. Attached Figure Description

[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0020] Figure 1 This is a schematic diagram of the system structure provided in an embodiment of the present invention; Figure 2 This is a flowchart of a method provided in an embodiment of the present invention. Detailed Implementation

[0021] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an online lubricating oil quality monitoring system and method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0024] The following description, in conjunction with the accompanying drawings, details the specific scheme of the online lubricating oil quality monitoring system and method provided by the present invention.

[0025] Example Please see Figure 1 The diagram illustrates a system configuration of an online lubricating oil quality monitoring system according to an embodiment of the present invention. The system includes: The data acquisition module uses a sensor array to collect real-time data on the physical and chemical parameters of lubricating oil and equipment wear information. The sensor array is based on quantum sensing, biomimetic fluid and dynamic topology reconstruction technology, and constructs a three-in-one architecture of "multimodal perception-intelligent compensation-self-evolution". The sensor array uses a variety of existing detection sensors and adopts a multi-parameter synchronous detection matrix to break through the limitations of traditional sensors with single parameters and low sensitivity.

[0026] The moisture sensor combines resistive and capacitive sensors to measure changes in the dielectric constant and resistivity of the oil to detect its moisture content. Multiple moisture sensors are arranged at the bottom of the oil tank to monitor the accumulation of free water and emulsified water. The viscosity sensor combines ultrasonic and vibration sensors. Based on the correlation between the propagation speed and viscosity in the oil, as well as the damping effect of the constant vibration element in the oil, the viscosity is calculated. Multiple viscosity sensors are arranged at the main oil inlet position, which can monitor lubrication performance in real time at any time.

[0027] The particle sensor combines laser scattering sensor and inductive sensor. It can use the scattering characteristics of laser to count particle size and quantity, and can also distinguish between ferromagnetic and non-ferromagnetic particles based on the electromagnetic induction signal of metal wear particles. Multiple particle sensors are arranged near mechanical drive components to facilitate the rapid capture of early wear signals.

[0028] Simultaneous detection of multiple key parameters enables comprehensive assessment of oil oxidation, contamination, and wear status. After processing, multi-sensor data combined with algorithm calculations can achieve early fault warnings, avoiding sudden equipment downtime. Distributed layout covering key monitoring points improves data representativeness. The multi-parameter sensor matrix enables refined and intelligent management of lubricating oil status, significantly improving equipment reliability and reducing operation and maintenance costs.

[0029] The communication transmission module constructs a data circulation network and supports multi-level interaction. The communication transmission module adopts multi-protocol dynamic adaptation and protocol stack integration to support heterogeneous interaction between sensors, PLCs, and cloud platforms. The protocol conversion gateway of the communication transmission module adopts a unified data encapsulation format to convert different protocol data into a standardized JSON format. The fields include timestamps, sensor IDs, and parameter key-value pairs. Based on link quality and device type, the optimal protocol is automatically selected through the Q-Learning algorithm.

[0030] The communication transmission module adopts an edge-layered structure. The edge layer uses FPGA nodes to handle real-time data acquisition and preprocessing, uploading basic parameters such as viscosity and temperature at a frequency of 1Hz. The local layer industrial gateway aggregates data from multiple nodes, executes local decisions, and activates a 10Hz high-frequency data transmission mode when an abnormal event is triggered. The cloud layer uses a cloud platform based on Kafka+Spark to realize big data analysis and predictive maintenance. The cloud distributes machine learning model update packages to the edge nodes to achieve dynamic algorithm upgrades.

[0031] To prevent data leakage, the communication transmission module strengthens the security of communication transmission. Lightweight encryption uses AES-128-CTR mode to encrypt transmitted data. The key is dynamically negotiated through elliptic curve Diffie-Hellman and an HMAC signature is added to verify data integrity. The receiving end verifies the validity of the tag to ensure that the data has not been tampered with.

[0032] The data analysis and diagnostic module enables status assessment and fault early warning; The data analysis and diagnostics module includes: The data preprocessing unit performs denoising and feature enhancement on the raw sensor data on moisture viscosity, dielectric constant, and particle size. It utilizes a hybrid denoising algorithm, combining Lowes local weighted regression and wavelet thresholding to preserve transient abrasive pulses while suppressing electromagnetic interference.

[0033] in This refers to the processed signal, that is, the signal after noise reduction and feature enhancement. This represents the raw signal, i.e., the unprocessed sensor data. This indicates the size of the dynamic window, which is dynamically adjusted according to the signal gradient. LOWESS is an abbreviation for Local Weighted Regression Smoothing Algorithm, used to smooth the original signal to reduce noise and retain the main features of the signal. WaveletShrink represents the wavelet thresholding denoising algorithm, which further removes high-frequency noise from the signal.

[0034] The multi-parameter feature fusion unit integrates multi-dimensional data such as moisture, viscosity, pH, and temperature to construct a joint feature vector.

[0035] in This represents the output parameter value of the i-th sensor at time t. The weight represents the weight of the i-th sensor at time t, and the weights represent the weights of the outputs from multiple sensors fused together using a weighted summation method. This indicates that different importance is assigned to the outputs of different sensors during the fusion process, and the weights represent this. Dynamic adjustments are made based on the sensor's signal characteristics and noise levels to ensure the accuracy and reliability of the fusion results.

[0036] in, This represents the mean value of the sliding window for the i-th sensor, reflecting the stability of the sensor output. Let represent the sliding window standard deviation of the i-th sensor, and let represent the standard deviation of the measurements of the i-th sensor over the past 60 seconds. The numerator is an exponential function. Calculate the difference between the measured value of the i-th sensor and its sliding window mean, taking into account the sensor's noise level. The smaller the difference, the larger the value of the exponential function, indicating that the sensor's measurement is more reliable. The denominator... This involves summing the exponential values ​​of all sensors to normalize the weight values. The sum of the weight values ​​of all sensors is 1, ensuring a reasonable allocation of weights.

[0037] Weight The calculation is based on the difference between the sensor output and its sliding window mean, taking into account the influence of the standard deviation. This method can dynamically reflect the stability and noise level of the sensor output, thereby allocating weights more reasonably. When lubricating oil contamination causes a sudden change in the viscosity-dielectric constant correlation, the weight of abnormal sensors is automatically reduced. Dynamic weight adjustment can adjust the fusion result in real time according to the signal characteristics and noise level of the sensor, improving the robustness and accuracy of the system. Weighted fusion can effectively reduce the influence of noise on the signal and improve the reliability and accuracy of the signal.

[0038] The user interaction and visualization module provides a human-computer interaction interface and decision support. The user interaction and visualization module includes: The fusion view unit, based on the needs of multimodal data presentation, human-machine collaborative decision-making, and interactive intelligent optimization, achieves innovative design with data transparency, real-time decision-making, and user-friendly operation. It integrates key parameters such as viscosity, dielectric constant, granularity, and temperature, supports data association across time and spatial dimensions, and is based on a dynamic weighted fusion view.

[0039] This represents the dynamic weighted fusion view data at time t, indicating the overall state. This represents the weight of the i-th sensor at time t. Let be the parameter value of the i-th sensor at time t. The health score of the i-th sensor at time t is dynamically displayed in the form of a heatmap, showing the comprehensive status of multiple parameters. The color mapping function converts health scores into colors for visualization, dynamically displaying the comprehensive status of multiple parameters in the form of a heatmap. When particle size exceeds the standard, the red area expands, and when viscosity is abnormal, a flashing warning appears. In addition to the dynamic weighted fusion view, the algorithm also incorporates a spatiotemporal correlation map. By binding sensor locations to the device's 3D model and using a particle system to simulate the abrasive flow path, the algorithm can display the changing trends of parameters in time and space, enhancing the spatial understanding of the data.

[0040] The mobile positioning unit, combined with AR glasses or a mobile device, enables 3D marking of fault points and repair guidance. Utilizing a spatial matching algorithm, virtual fault markers and mechanical wear indicators are accurately overlaid onto the real device. The core algorithm is as follows:

[0041] in, Represents the rigid body transformation matrix of the equipment CAD model. This indicates the position of the sensor in the model coordinate system. This represents the true coordinates captured by the AR camera. The algorithm finds the optimal alignment between the model and the real world by minimizing the Euclidean distance between the sensor position and the real position, and uses the rigid body transformation matrix T_model to represent the sensor position. Transform from the model coordinate system to the actual coordinate system, and then calculate the transformed position and the actual position. The differences between them are addressed by continuously adjusting the rigid body transformation matrix. The rotation and translation parameters in the model minimize the difference between the transformed position and the actual position, thereby achieving accurate alignment between the model and the real world.

[0042] The spatial matching algorithm greatly improves the accuracy of fault location, and combined with AR technology, it enables maintenance personnel to make virtual annotations and provide maintenance guidance on actual equipment, reducing the time and error of manual location. This improves the automation and intelligence of fault location, reduces the reliance on human experience, and brings significant improvements and enhancements to the field of industrial maintenance.

[0043] The prediction visualization is achieved by combining the grey prediction GM(1,1) model with an LSTM neural network. A force-directed graph layout is used, where node size represents the failure probability and edge weight represents the causal relationship strength. Interactive parameters are tuned, allowing users to drag parameter thresholds. The impact of different strategies is simulated in real time, mainly based on a sensitivity analysis algorithm.

[0044] in, This represents the change in relative lifetime. When the user adjusts the model parameters, the system's predicted remaining lifetime will change. θ represents a function value of the autoregressive moving average model, a commonly used time series forecasting model that captures autocorrelation and moving average properties in data. θ represents the model's parameters, which can be adjusted by the user through drag-and-drop interface. Represents the partial derivatives of the model with respect to the parameters. This indicates the amount of change in the parameter θ when the user adjusts the parameter by dragging and dropping the interface.

[0045] This method allows users to simulate the effects of different strategies in real time and optimize parameter settings. Through sensitivity analysis and partial derivative calculation, the impact of parameter changes on prediction results can be evaluated more accurately, thereby improving prediction accuracy. Users can adjust parameters in real time by dragging and dropping the interface and immediately see the impact of parameter changes on prediction results. This helps users optimize parameter settings and improve system reliability and efficiency. The method simulates and displays the effects of different strategies in real time, allowing users to intuitively see the impact of parameter adjustments on prediction results, thereby enhancing user experience and satisfaction. Sensitivity analysis and real-time parameter adjustment provide a dynamic and accurate system optimization method, which helps improve prediction accuracy, optimize parameter settings, and enhance user experience.

[0046] The intelligent decision-making module enables closed-loop control of monitoring and maintenance, provides timely alarms, and achieves intelligent control of the entire chain for triggering maintenance actions; The digital twin simulation unit performs virtual verification before triggering actual maintenance actions, based on dynamic early warning of oil deterioration and prediction of remaining service life, using an improved autoregressive moving average model.

[0047] in, This represents the state variable of the oil at the current moment. These are autoregressive coefficients, representing the influence of past states on the current state. The white noise term represents random error. The moving average coefficient represents the influence of past errors on the current state. Weighting coefficients for accelerating degradation terms:

[0048] This represents the accelerated degradation term nearing failure, where k is the acceleration factor. The historical average failure time is used to activate the acceleration term when the dielectric constant of the oil changes abruptly, thereby improving the accuracy of remaining lifetime (RUL) prediction. By combining an improved autoregressive moving average model with an accelerated degradation term, the prediction accuracy and early warning capability of oil system degradation trends are effectively improved, which has significant practical application value.

[0049] The graded early warning unit triggers an alarm when the viscosity error of the ferromagnetic abrasive reaches a certain value, or when the dielectric constant deviates from the baseline or the ARMA prediction is abnormal; in the case of multiple parameter joint abnormalities, the unit will shut down and perform a self-check. The dynamic threshold adjustment unit and algorithm adaptively adjust the alarm threshold based on the oil deterioration rate, avoiding false alarms or missed alarms caused by fixed thresholds.

[0050] in, This indicates the alarm threshold that has been dynamically adjusted over time. Indicates the basic threshold. The viscosity degradation rate is predicted based on the ARMA model. This represents the near-failure acceleration factor, reflecting the urgency with which the oil approaches failure. It is calculated using a fuzzy evaluation model. , These are all empirical coefficients. When the oil deteriorates rapidly, the threshold is automatically adjusted. For example, if the moisture content suddenly increases, the sensitivity of the moisture warning is improved.

[0051] Dynamically adjusting alarm thresholds can more accurately reflect the real-time status of the oil, reduce false alarms or missed alarms caused by fixed thresholds, and improve the accuracy of early warnings. The algorithm can adaptively adjust the alarm thresholds according to different oil deterioration conditions, improving the algorithm's adaptability and robustness. Through timely and accurate early warnings, the algorithm can reduce unnecessary maintenance operations, extend equipment lifespan, and reduce maintenance costs. When the oil deteriorates rapidly or is close to failure, the algorithm can quickly tighten the alarm thresholds, improving the sensitivity and timeliness of early warnings, thereby ensuring the safe operation of the equipment and enhancing safety.

[0052] The multi-actuator collaborative control unit optimizes resource allocation for operations such as oil purification, additive injection, and oil change. Based on a Q-Learning reinforcement learning model, it coordinates the oil purification equipment, additive injection pump, and oil change valve to handle emergencies through collaborative scheduling.

[0053] in, This represents the dielectric constant deviation, viscosity, particle size, water content, and temperature. Indicates the action response, where This indicates that the vacuum oil purification process has been initiated. This indicates the injection of antioxidant additives. This indicates that the oil change valve is opened, which triggers the coordinated operation of the oil purification equipment, additive injection pump, and oil change valve to respond to emergencies. The maintenance effectiveness evaluation unit, based on the improved TOPSIS multi-objective evaluation, comprehensively considers the performance and weight allocation of multiple indicators, uses positive and negative ideal solutions as reference standards, and derives a comprehensive score, providing a scientific basis for the comprehensive evaluation of the evaluation objects. The relative proximity of the i-th evaluation object... Calculation:

[0054] The closer the value is to 1, the closer the evaluated object is to the ideal solution, and the better the overall evaluation. This represents the distance to the positive ideal solution of the i-th evaluation object. The positive ideal solution is the ideal state in which all indicators reach their optimal values. The smaller the distance, the closer the evaluation object is to the positive ideal solution. This represents the distance to the negative ideal solution of the i-th evaluation object. The negative ideal solution is the ideal state where all indicators reach their worst values. The larger the distance, the further the evaluation object is from the negative ideal solution, and the better the overall evaluation. The calculation is as follows:

[0055]

[0056] This represents the weight of the i-th indicator (1). The weight reflects the importance of each indicator in the comprehensive evaluation; the larger the weight, the greater the influence of that indicator on the comprehensive evaluation. This represents the value of the i-th evaluation object on the j-th index. This value is based on actual observed data and is used to calculate the distance to the positive and negative ideal solutions. Let represent the positive ideal value of the j-th indicator, which is the optimal value that the indicator can achieve among all evaluation objects. This represents the negative ideal value of the j-th indicator, which is the worst possible value that the indicator can reach among all evaluation objects. m is the number of indicators, representing the total number of evaluation indicators involved in the evaluation object.

[0057] Among the evaluation indicators, For viscosity recovery rate, the target is ≥90%. For dielectric constant deviation, the target is ≤0.3. For execution time, with a target time of ≤30 minutes, the relative proximity is calculated. The performance of multiple indicators can be comprehensively considered to obtain a comprehensive score, which makes it easier to rank and compare the evaluation objects.

[0058] In this way, an online monitoring system and method for lubricating oil quality can be realized.

[0059] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An online monitoring system for lubricating oil quality, characterized in that, The system includes: The data acquisition module uses a sensor array to collect real-time data on the physical and chemical parameters of lubricating oil and equipment wear information. The communication transmission module constructs a data circulation network and supports multi-level interaction. The data analysis and diagnostic module enables status assessment and fault early warning; The user interaction and visualization module provides a human-computer interaction interface and decision support. The intelligent decision-making module enables closed-loop control of monitoring and maintenance, provides timely alarms, and achieves intelligent control of the entire chain for triggering maintenance actions.

2. The online lubricating oil quality monitoring system according to claim 1, characterized in that, The sensor array employs various existing detection sensors and uses a multi-parameter synchronous detection matrix to detect parameters such as moisture, viscosity, particle size, and oxidation degree of lubricating oil.

3. The online lubricating oil quality monitoring system according to claim 1, characterized in that: The communication transmission module adopts multi-protocol dynamic adaptation and protocol stack integration. The protocol conversion gateway of the communication transmission module adopts a unified data encapsulation format to convert different protocol data into a standardized format. It automatically selects the optimal protocol through the Q-Learning algorithm and updates the Q value table in real time, which can maintain low latency in multipath interference environment. The communication transmission module adopts an edge-based hierarchical division. The edge layer uses FPGA nodes to be responsible for real-time data acquisition and uploading. The local layer industrial gateway aggregates data from multiple nodes, executes local decisions, and starts a high-frequency data transmission mode when an abnormal event is triggered. The cloud layer supports big data analysis and predictive maintenance. The cloud sends machine learning model update packages to the edge nodes to realize dynamic algorithm upgrades. The communication transmission module uses lightweight encryption with AES-128-CTR mode to encrypt transmitted data. The key is dynamically negotiated through elliptic curve Diffie-Hellman (ECDH), and HMAC signature is added to verify data integrity. The receiving end verifies the validity of the tag to ensure that the data has not been tampered with.

4. The online lubricating oil quality monitoring system according to claim 1, characterized in that, The data analysis and diagnosis module also includes: The data preprocessing unit performs denoising and feature enhancement on the raw sensor data. It uses a hybrid denoising algorithm, combining LOWESS local weighted regression and wavelet threshold denoising, to retain transient abrasive pulses while suppressing electromagnetic interference. The multi-parameter feature fusion unit integrates multi-dimensional data such as moisture, viscosity, particle size, and temperature to construct a joint feature vector. The weights are dynamically adjusted according to the signal characteristics and noise level of the sensor to ensure the accuracy and reliability of the fusion results.

5. The online lubricating oil quality monitoring system according to claim 1, characterized in that, The user interaction and visualization module includes: The integrated view unit, relying on multimodal data presentation, human-machine collaborative decision-making and interactive intelligent optimization, realizes data transparency, real-time decision-making and humanized operation. It integrates key parameters such as moisture content, viscosity and particle size to realize data correlation across time and space dimensions, analyzes and establishes dynamic weighted integrated view, and visualizes the health score. The mobile positioning unit enables 3D labeling of fault points and maintenance guidance. Utilizing a spatial matching algorithm, it accurately overlays virtual fault markers and wear indicators onto the actual device, as shown in the following example: in, Represents the rigid body transformation matrix of the equipment CAD model. This indicates the position of the sensor in the model coordinate system. This represents the actual coordinates captured by the AR camera.

6. The method for online monitoring of lubricating oil quality according to claim 1, characterized in that, The intelligent decision-making module includes: The digital twin simulation unit performs virtual verification before triggering actual maintenance actions. Based on dynamic early warning of oil deterioration and prediction of remaining service life, an improved autoregressive moving average model is used, as shown in the following equation: in, This represents the state variable of the oil at the current moment. These are autoregressive coefficients, representing the influence of past states on the current state. The white noise term represents random error. The moving average coefficient represents the influence of past errors on the current state. To accelerate the weighting coefficients of degradation terms, This indicates an accelerated degradation term that occurs near the point of failure. The graded early warning unit triggers an alarm when the viscosity error of the ferromagnetic abrasive reaches a certain value, activates an emergency alarm when the dielectric constant deviates from the baseline or when the ARMA model prediction is abnormal, and shuts down and performs a self-check when multiple parameters are abnormal. The dynamic threshold adjustment unit and algorithm adaptively adjust the alarm threshold based on the oil deterioration rate, avoiding false alarms or missed alarms caused by fixed thresholds. in, This indicates the alarm threshold that has been dynamically adjusted over time. Indicates the basic threshold. The viscosity degradation rate is predicted based on the ARMA model. This represents the near-failure acceleration factor, reflecting the urgency with which the oil approaches failure. It is calculated using a fuzzy evaluation model. , These are all empirical coefficients, and the threshold value is automatically adjusted when the oil deteriorates rapidly; The multi-actuator collaborative control unit optimizes resource allocation for oil purification, additive injection, and oil change operations. Based on the Q-Learning reinforcement learning model, it links the oil purification equipment, additive injection pump, and oil change valve to coordinate scheduling in response to emergencies. The maintenance effectiveness evaluation unit, based on the improved TOPSIS multi-objective evaluation, comprehensively considers the performance and weight allocation of multiple indicators, uses positive and negative ideal solutions as reference standards, and derives a comprehensive score, providing a scientific basis for the comprehensive evaluation of the evaluation object.

7. A method for online monitoring of lubricating oil quality, applied to the online monitoring system for lubricating oil quality according to any one of claims 1-6, characterized in that, Includes the following steps: Step S1: Monitor the operating parameters of the lubricating oil system in real time using a sensor array; Step S2: The collected parameters are transmitted to the data analysis and diagnosis module through the communication module for preliminary processing; In step S3, the data analysis and diagnosis module uses multimodal dynamic fusion to integrate multi-dimensional data such as moisture, viscosity, particle size, and temperature to construct a joint feature vector. The weights are dynamically adjusted according to the signal characteristics and noise level of the sensors to ensure the accuracy and reliability of the results. Step S4: The user interaction and visualization module uses view fusion and spatial matching algorithms to visualize the health score and overlay virtual fault indicators onto the real device. Step S5: Based on the diagnostic results of the data analysis and diagnosis module, a virtual verification is performed before the intelligent decision-making module triggers the actual maintenance action. The alarm threshold is adaptively adjusted based on the dynamic threshold adjustment algorithm to improve the response speed and data processing accuracy. Step S6: Optimize the multi-actuator collaborative control unit to promptly respond to maintenance after an alarm is issued; Step S7 evaluates the maintenance effect, considering the performance and weighting of multiple indicators to obtain a comprehensive score, providing a scientific basis for the comprehensive evaluation of the evaluation object, ensuring that the maintenance effect reaches the expected level in all aspects, thereby improving the reliability and service life of the equipment.

Images