Multi-source data fusion analysis method for lubricating oil production
By collecting and analyzing multi-source data from the lubricant production process, identifying equipment anomalies and adjusting parameters, the problem of false alarms and missed detections in lubricant production is solved, highly sensitive detection of oil flow and contamination is achieved, and system stability and equipment reliability are improved.
Patent Information
- Application Number
- CN202510849095.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-24
- Publication Date
- 2025-10-14
AI Technical Summary
Existing lubricant production detection systems are prone to false alarms or missed detections when faced with fluctuations in production process parameters. They find it difficult to accurately identify oil flow anomalies caused by dynamic changes in equipment, and to quickly locate the source of disturbances in multi-device coupled systems. Furthermore, they are unable to detect oil contamination with high sensitivity and ensure the continuity of the production process.
By collecting the agitator speed signal and pipeline pressure distribution data, analyzing the pipeline pressure changes when the agitator speed is abnormal, and combining the oil viscosity and flow rate differences to calculate the shear stress distribution index, the oil particle suspension state and the degree of pollution diffusion are detected, the pollution deposition area is identified, and the equipment chain abnormality distribution map is generated to adjust the agitator speed and pipeline flow distribution.
It achieves comprehensive monitoring of the lubricant production process, timely discovers pollution deposition problems and diagnoses equipment abnormalities, and improves system operation stability and equipment reliability.
Smart Images

Figure CN120771768A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of information technology, and in particular to a lubricating oil production multi-source data fusion analysis method. BACKGROUND
[0002] As an important part of the petrochemical industry, the product quality of lubricating oil production directly affects the operation reliability and service life of mechanical equipment, and plays a key role in the modern industrial system. With the increasing demand for lubricating oil quality, abnormal detection and quality control in the production process become increasingly important. Existing detection systems often use fixed threshold monitoring, which is prone to false positives or missed detections when facing fluctuations in production process parameters. In the production process of lubricating oil, changes in the viscosity of raw materials directly affect the operating state of the mixer, causing abnormal speed pulses. This mixing abnormality changes the flow field distribution in the reactor, causing changes in the flow state of the oil in the pipeline, and further exacerbating turbulent flow. The instability of turbulent flow in the pipeline produces pressure transient fluctuations, which can cause sealing failure and leakage when transmitted to the pump valve system. In the mode of multi-variety and small-batch production, frequent process switching causes the oil to bear uneven shear stress, which easily forms local pollution and spreads in the system, and eventually may trigger a chain reaction of abnormality between devices. How to accurately identify this oil flow state abnormality caused by dynamic changes in equipment, quickly locate the disturbance source in a complex multi-device coupled system, while taking into account the high sensitivity detection of oil pollution and the continuity of the production process, has become a key problem to be solved in the field of lubricating oil production abnormality detection. SUMMARY
[0003] The present application provides a lubricating oil production multi-source data fusion analysis method, mainly comprising:
[0004] Collect the mixer speed signal and obtain the pressure distribution data of the corresponding pipeline. If the mixer speed fluctuation amplitude exceeds the preset threshold range, it is marked as a mixing speed abnormality;
[0005] Analyze the change in pipeline pressure distribution when the mixing speed is abnormal, combine the monitored oil flow viscosity change and oil circulation flow rate difference, and fuse to calculate the shear stress distribution index in the pipeline;
[0006] If the shear stress distribution index is abnormal, detect the oil particle suspension state and the pollution particle deposition rate. If the suspension state and the deposition rate exceed the corresponding preset threshold, determine the oil pollution diffusion degree by the suspension state and the deposition rate;
[0007] According to the detected wear degree and oil temperature fluctuation, the oil pollution diffusion path is analyzed, the pollution deposition area is determined based on the oil pollution diffusion path and diffusion degree, the concentration and particle size distribution of abnormal particles in the area are detected and recognized, and the pollution deposition feature vector is obtained by vectorization;
[0008] The pollution distribution mode and deposition trend are obtained by statistical analysis of the pollution deposition feature vector, the interference effect of the deposition feature vector on vibration transmission is analyzed in combination with the equipment vibration transmission path, and the vibration abnormal signal time sequence correlation caused by pollution deposition from the agitator to the pump valve is obtained;
[0009] According to the abnormal signal time sequence correlation, the abnormal equipment is recognized, the operation parameters of the abnormal equipment are collected, and the equipment operation state coupling is obtained by synchronous analysis, the equipment fault chain reaction characteristics are analyzed, and the equipment chain abnormal distribution map is generated according to the chain reaction characteristics;
[0010] The node vibration frequency and abnormal signal time sequence correlation in the equipment chain abnormal distribution map are recognized, the abnormal probability of each node is calculated, if the abnormal probability of a node exceeds a preset threshold and is located at the starting position of the equipment vibration transmission path, the disturbance source is judged, and the speed of the agitator and the pipeline flow distribution in the production process are adjusted in combination with historical oil circulation flow rate difference and pollution particle deposition rate data.
[0011] The technical scheme provided by the embodiment of the present application can include the following beneficial effects:
[0012] The application discloses a kind of lubricating oil production multi-source data fusion analysis method, by collecting agitator speed signal and pipeline pressure distribution data, the change of pipeline pressure when stirring speed is abnormal is analyzed, shear stress distribution index is calculated in combination with oil viscosity and flow rate difference.When index is abnormal, the state of oil particle suspension and deposition rate are detected, the pollution diffusion degree and path are determined, and the feature vector of pollution deposition area is identified.Further analyze the interference of pollution deposition to equipment vibration transmission, identify abnormal equipment and analyze fault chain reaction, generate equipment chain abnormal distribution map.Finally, according to the abnormal probability of node, the disturbance source is judged, and the speed of the agitator and the pipeline flow distribution in the production process are adjusted.The present application realizes the comprehensive monitoring of oil circulation system, can find pollution deposition problem in time and diagnose equipment anomaly, effectively improve system operation stability and equipment reliability. BRIEF DESCRIPTION OF DRAWINGS
[0013] Fig. 1 It is a flow chart of the lubricating oil production multi-source data fusion analysis method of the present application.
[0014] Fig. 2 It is a schematic diagram of the lubricating oil production multi-source data fusion analysis method of the present application. DETAILED DESCRIPTION
[0015] In order for those skilled in the technical field to better understand the technical solutions in the specification, the technical solutions in the specification will be clearly and completely described below in combination with the drawings in the embodiments of the specification. Obviously, the described embodiments are only part of the embodiments of the specification, not all. Based on the embodiments in the specification, all other embodiments obtained by those of ordinary skill in the art without creative labor should belong to the scope of protection of the specification.
[0016] As Figs. 1-2 The lubricating oil production multi-source data fusion analysis method of the embodiment can specifically include:
[0017] Step S101, collect the stirring speed signal, and obtain the pressure distribution data of the corresponding pipeline. If the stirring speed fluctuation amplitude exceeds the preset threshold range, it is marked as stirring speed abnormality.
[0018] The analog voltage signal output by the stirring speed sensor is collected, and the digitized speed data is obtained through the AD converter. At the same time, the pressure values of the multi-point pressure sensor in the pipeline are read, and the time sequence data correlation matrix of the speed and pressure distribution is established. The speed fluctuation variance in the continuous time window is calculated from the speed data. According to the comparison between the speed fluctuation variance and the preset fluctuation threshold, if the speed fluctuation variance exceeds the preset threshold range, the abnormal marking state is triggered, the pressure distribution value corresponding to the abnormal marking state is obtained, and the pressure distribution value is used to form a pressure feature data group for abnormal pressure state recognition. The pressure feature data group is processed by moving average filtering to remove random noise interference in pressure measurement, and the smoothed pressure change data sequence is obtained. The pressure difference value between adjacent measuring points is extracted from the pressure change data sequence as the pressure gradient value. According to the pressure gradient value, the pipeline resistance state evaluation relationship is established. If the pressure gradient value deviates from the preset normal operation interval, it is judged that there is flow resistance abnormality in the pipeline, and the resistance abnormality level identifier and the corresponding stirring speed adjustment parameter are output.
[0019] For example, during the stirring speed signal collection process, the speed sensor usually outputs a 0-10V analog voltage signal, and the voltage value is linearly corresponding to the actual stirring speed.
[0020] Exemplarily, the sensor outputs 3V voltage when the stirrer operates at 300 rpm, and outputs 6V voltage at 600 rpm. After the AD converter digitizes the analog signal, the accurate sequence of rotational speed values can be obtained. The synchronous pressure data acquisition involves pressure sensors at different positions in the pipeline, which are distributed at key positions such as the inlet, middle section and outlet of the pipeline. By establishing a time series data correlation matrix, the time correspondence between the rotational speed change and the pressure change at each measuring point can be captured, and this correlation reflects the direct influence of stirring intensity on the fluid state in the pipeline.
[0021] In a possible implementation, the rotational speed fluctuation variance calculation adopts a sliding time window method, and 30 seconds is set as a calculation window, and a numerical value is updated every 5 seconds. When the stirrer operates normally, the rotational speed fluctuation variance is generally maintained within a small range, and device aging or mechanical failure can cause unstable rotational speed, and the fluctuation variance significantly increases. The determination of the preset threshold range is based on statistical analysis of historical operation data, and the fluctuation variance generally does not exceed 3% of the rated rotational speed under normal working conditions. After the abnormality marking state is triggered, the system immediately records the values of each pressure measuring point at the corresponding time, to form a pressure feature data set. The technical effect of this method lies in accurately positioning the time node at which the abnormality occurs, and retaining the pressure distribution characteristics under the abnormal state, to provide a reliable data basis for subsequent analysis.
[0022] Specifically, the pressure feature data set can effectively eliminate the interference of measurement noise after moving average filtering processing. The moving average filtering replaces the current sampling value with the arithmetic average value of a plurality of consecutive sampling points, and the filtering window length is determined according to the pressure signal characteristics.
[0023] For example, the original pressure data contains high-frequency noise generated due to pipeline vibration, and after 5-point moving average filtering, the pressure change curve becomes smooth, and the real pressure change trend is revealed. In the pressure gradient value extraction process, the pressure difference between adjacent measuring points can reflect the resistance distribution condition of the fluid in the pipeline, and the greater the pressure difference, the higher the resistance of the pipeline.
[0024] It should be noted that the pipeline resistance state evaluation relationship is established on the basis of the basic principles of fluid mechanics, and the pressure gradient should be regularly distributed under normal flow state. When partial blockage occurs in the pipeline, the pressure upstream of the blockage position will significantly increase, and the pressure downstream will relatively decrease, forming an abnormal pressure gradient distribution. The preset normal operation interval is obtained by statistical analysis of a large number of normal working condition data, and the pressure gradient mean value is usually taken as the center, and the upper and lower floating range is set as twice the standard deviation. The resistance anomaly grade identification adopts a three-level classification method, and the mild anomaly corresponds to 10%-20% deviation from the normal interval, the moderate anomaly corresponds to 20%-40% deviation, and the severe anomaly corresponds to more than 40% deviation. The stirrer speed regulation parameter is determined according to the resistance anomaly degree, and the speed is recommended to be increased by 5%-10% to enhance the stirring intensity when the mild anomaly occurs, the speed is increased by 10%-20% when the moderate anomaly occurs, and the machine needs to be stopped for inspection when the severe anomaly occurs. The beneficial effect of this grading processing method is that corresponding adjustment measures can be taken according to the actual anomaly degree, which not only avoids the waste of energy caused by excessive adjustment, but also ensures timely and effective anomaly treatment.
[0025] In step S102, the change of pipeline pressure distribution when the stirring speed is abnormal is analyzed, and the change of oil fluid viscosity and the difference of oil circulating flow rate are monitored to obtain the shear stress distribution index in the pipeline.
[0026] The pressure value change of each measuring point in the pipeline under the abnormal state of stirring speed is obtained, the pressure gradient distribution in the axial and radial directions of the pipeline is measured by a pressure difference sensor, a pressure field distribution matrix under abnormal working conditions is established, and the pressure change rate between adjacent measuring points is calculated from the pressure field distribution matrix as the basic data for fluid dynamics state evaluation. According to the pressure change rate and the oil dynamic viscosity value measured by the online viscometer, the velocity gradient distribution of the fluid in the pipeline is calculated based on the linear relationship between the pressure gradient and the velocity gradient. If the oil dynamic viscosity value exceeds the preset viscosity range, a correction coefficient is used to adjust the velocity gradient calculation result to obtain the corrected fluid velocity field distribution data. The oil circulating flow rate measured values of different cross sections of the pipeline are obtained by a flowmeter, the fluid velocity field distribution data and the oil circulating flow rate measured values are calculated by difference, the spatial distribution deviation degree of the velocity field is determined, and the shear strain rate distribution of each position in the pipeline is derived from the spatial gradient of the velocity field distribution data. According to the shear strain rate distribution and the oil dynamic viscosity value, the shear stress distribution in the pipeline is calculated, the maximum position of the shear stress distribution is determined as the maximum shear stress value, the distribution unevenness coefficient is calculated by the ratio of the standard deviation to the mean value of the shear stress distribution, and the coefficient can be used as the shear stress distribution index in the pipeline to identify the spatial coordinates of the local stress concentration point.
[0027] For example, the pressure field distribution matrix establishment process requires high-precision differential pressure sensors to be arranged at key positions of the pipeline, which can monitor the axial and radial pressure changes in real time.
[0028] For example, in a 10-meter-long pipeline, a pressure measuring point is set every 1 meter along the axial direction, and measuring points are set at the center, middle position, and near the wall of the pipeline cross section in the radial direction. When the stirrer speed fluctuates abnormally, the pipeline inlet pressure may rise from the normal 2.5 MPa to 3.2 MPa, and the outlet pressure may correspondingly decrease from 1.8 MPa to 1.3 MPa. The sharp change in pressure gradient reflects the abnormal state of fluid resistance distribution. The calculation of the pressure change rate is obtained by dividing the pressure difference between adjacent measuring points by the distance, which directly reflects the energy loss distribution of the fluid in the pipeline.
[0029] In one possible implementation, there is a clear physical relationship between the pressure gradient and the velocity gradient. According to the basic principles of fluid mechanics, the pressure gradient drives the fluid motion and produces a corresponding velocity distribution. When the oil viscosity increases from 40 centipoise to 60 centipoise due to temperature changes, the velocity gradient under the same pressure gradient will decrease accordingly. At this time, a viscosity correction coefficient needs to be introduced to adjust the calculation results. The accurate calculation of the velocity gradient distribution is crucial for subsequent shear stress analysis, because the shear stress is directly dependent on the product of the velocity gradient and the fluid viscosity. The determination of the correction coefficient is based on the ratio of the actual viscosity value to the standard working condition viscosity value. This correction method can ensure the accuracy of the calculation results under different working conditions.
[0030] Specifically, the comparison and verification of the fluid velocity field distribution data and the measured flow rate values is an important link to ensure the accuracy of the calculation. The flow rate data measured by the flowmeter at different cross sections of the pipeline provides a verification benchmark for the velocity field calculation results. When the deviation between the theoretically calculated average flow rate and the measured flow rate exceeds 5%, the velocity field distribution needs to be recalibrated. The deviation degree of the spatial distribution of the velocity field is obtained by standard deviation calculation. The areas with large deviations usually correspond to unstable areas of fluid flow. In the derivation process of the shear strain rate distribution, numerical differentiation calculation is required for the gradient of the velocity field perpendicular to the flow direction. This process requires the velocity field data to have sufficient spatial resolution and accuracy.
[0031] It should be noted that the calculation of the shear stress distribution combines two key parameters, shear strain rate and dynamic viscosity, and this calculation method is based on the constitutive relationship of Newtonian fluid. The maximum shear stress value usually occurs near the pipeline wall or in the flow separation area, and its numerical value is directly related to the degree of pipeline wall wear and the efficiency of fluid transportation. The distribution non-uniformity coefficient is obtained through statistical analysis of the shear stress distribution. This coefficient reflects the uniformity of the shear stress distribution in the pipeline, and a larger value indicates a more uneven distribution.
[0032] For example, the uneven distribution coefficient can be maintained at about 0.15 under normal conditions, and can rise to more than 0.35 under abnormal conditions. The identification of local stress concentration points is based on the detection of the mutation of the shear stress gradient, and these positions often correspond to the geometric mutation of the pipeline, elbows or valves and other parts. By accurately identifying these stress concentration points, important technical basis can be provided for the maintenance and optimization of the pipeline, effectively preventing the risk of pipeline fatigue damage and leakage caused by stress concentration.
[0033] In step S103, if the shear stress distribution index is abnormal, the suspension state and the deposition rate of the contaminated particles in the oil are detected. If the suspension state and the deposition rate exceed the corresponding preset threshold, the oil pollution diffusion degree is determined by the suspension state and the deposition rate.
[0034] According to the abnormal judgment result of the shear stress distribution index, the oil particle detection process is started. The size distribution and concentration distribution of the particles in the oil are scanned by a laser particle size analyzer, the particle suspension density data in different particle size ranges are obtained, and the quantitative evaluation parameters of the particle suspension state are established as the basic index for judging the pollution degree. The mass accumulation change of the contaminated particles at the bottom of the pipeline per unit time is measured by a sedimentation balance. The particle deposition rate value is calculated by extracting the particle settling characteristic parameters from the particle suspension density data. If the particle suspension density data exceeds the preset suspension threshold range, the suspension state is marked as abnormal, and it is judged whether the deposition rate value exceeds the preset deposition threshold. According to the threshold judgment result of the suspension state abnormality mark and the deposition rate value, the concentration gradient change of the contaminated particles in the oil is calculated based on the spatial distribution difference of the particle suspension density data. The particle diffusion coefficient is determined by the relationship between the concentration gradient and the time change rate, and the space-time distribution mode of the particle pollution in the pipeline is established. The propagation range and propagation speed of the pollution in the pipeline are calculated by the particle concentration gradient and the diffusion coefficient. According to the ratio of the propagation range to the total volume of the pipeline, the pollution diffusion degree is classified into different grades. If the propagation speed exceeds the preset control range, a high pollution diffusion grade mark is output, and the pollution source position and propagation path information are recorded.
[0035] For example, the working principle of the laser particle size analyzer is based on laser scattering technology. When a laser beam passes through an oil sample containing particles, particles of different sizes will produce scattered light at different angles.
[0036] Exemplarily, the scattering angle of 1-micron metal particles is significantly different from that of 10-micron contaminant particles, and the size distribution of the particles can be accurately determined by analyzing the scattering light intensity distribution. The acquisition of the suspension density data involves the statistics of the number of particles in each particle size range per unit volume of oil. When the concentration of particles in the range of 5-50 microns in the oil reaches more than 1000 per milliliter, it indicates that there is a significant pollution condition. This quantitative evaluation parameter establishes a direct correlation between the degree of pollution and the particle distribution characteristics, providing a reliable data basis for subsequent pollution diffusion analysis.
[0037] In one possible implementation, the sedimentation balance monitors the particle deposition process through the principle of precise weighing, and its working mechanism relies on the natural sedimentation behavior of particles in oil under the action of gravity. The particle sedimentation characteristic parameters include two key indicators, namely, the sedimentation speed and the sedimentation density, which are directly extracted from the suspension density data. When the suspension density data shows an increase in the proportion of large particles, the corresponding deposition rate will significantly increase.
[0038] For example, the deposition rate of heavy particles above 50 microns can reach 0.5 grams per square meter per hour, while the deposition rate of light particles below 20 microns is only 0.1 grams per square meter per hour. The threshold judgment process adopts a double standard, with a suspension threshold based on the particle concentration limit and a deposition threshold based on the mass accumulation rate limit. This double judgment mechanism can more comprehensively identify abnormal pollution states.
[0039] Specifically, the calculation of the particle concentration gradient is based on the difference analysis of the suspension density data at different spatial positions, and the change in particle concentration at different positions of the pipeline reflects the migration trend of the contaminant. The relationship between the concentration gradient and the time change rate follows the diffusion law, and the determination of the diffusion coefficient needs to consider the viscosity characteristics of the oil and the physical properties of the particles. When the particle concentration at the inlet position of the pipeline is 800 per milliliter, and the concentration at a distance of 5 meters from the inlet is reduced to 600 per milliliter, a significant concentration gradient is formed. The numerical value of the diffusion coefficient directly affects the speed and range of pollution propagation, and a larger diffusion coefficient means that the contaminant can be distributed more quickly throughout the pipeline. The establishment of the spatiotemporal distribution pattern combines the spatial position information and the time evolution characteristics of the particles, and this pattern can predict the distribution state of the contaminant in the future time.
[0040] It should be noted that the calculation of the pollution spread range involves the analysis of the distribution of particle concentration in three-dimensional space, and the spread speed is determined by the rate of change of the concentration front position with time. When the pollution source is located in the middle of the pipeline, the pollution will diffuse to both upstream and downstream, forming a bidirectional propagation mode. The ratio of the spread range to the total volume of the pipeline provides a quantitative indicator of the degree of pollution diffusion, and when this ratio exceeds 30%, it is determined as moderate pollution diffusion, and when it exceeds 60%, it is determined as severe pollution diffusion. The control range of the spread speed is usually determined according to the design flow rate and cleaning capacity of the pipeline, and when the spread speed exceeds the design cleaning speed, the removal of the pollutants will become difficult. The record of the location of the pollution source is achieved by tracing the highest point of the concentration gradient, and the propagation path information is obtained by statistical analysis of the particle motion trajectory.
[0041] In step S104, the oil pollution diffusion path is analyzed according to the detected wear degree and oil temperature fluctuation, the pollution deposition area is determined based on the oil pollution diffusion path and diffusion degree, the concentration of abnormal particles in the oil in the area is detected and identified, the particle size distribution is obtained, and the pollution deposition feature vector is obtained.
[0042] The wear degree data detected by the pipeline inner wall wear sensor and the oil temperature fluctuation values monitored by the temperature sensor are obtained, the main path direction of the pollution flow is determined by the numerical correlation analysis of the wear degree and the temperature fluctuation, the temperature gradient change point is identified as the branch node position of the diffusion path, and the diffusion path length distribution from the pollution source to each branch node is calculated. According to the diffusion path length distribution and the branch node position, combined with the pollution spread range data, the concentration distribution mapping relationship of the pollutants in the pipeline is established, the boundary range of the main deposition area of high-concentration pollutants is determined, and the center coordinates and coverage area of each deposition area are obtained as the reference data for area positioning. The number of abnormal particles per unit volume in the oil in the deposition area is detected by a microscope particle counter, the size distribution range and the number proportion of each size segment of the particles are measured by particle size analysis screening, and if the number of abnormal particles per unit volume exceeds the preset abnormal threshold, the deposition area is marked as a key pollution area, and the concentration value and particle size distribution data of the abnormal particles are obtained. According to the concentration value and particle size distribution data of the abnormal particles, a multi-dimensional data matrix containing particle concentration, average particle size, particle size variance, and deposition density is constructed, the dimensional data is converted to a unified dimension value through data standardization processing, and the standardized particle concentration, average particle size, particle size variance, and deposition density are combined to form a pollution deposition feature vector as a numerical description of the pollution state.
[0043] For example, the wear degree data is obtained by arranging eddy current detection sensors on the inner wall of the pipeline, which can monitor the changes of the pipeline wall thickness and surface roughness in real time.
[0044] Exemplarily, when the wall thickness of a certain section of the pipeline is reduced from the original 8.5 mm to 7.8 mm, the wear degree data shows that there is significant material loss in this area. The monitoring of temperature fluctuation values is realized by distributed temperature sensors, and when the oil flows through the severely worn pipe section, due to the increase in fluid resistance and the intensification of turbulent flow, local temperature fluctuations will occur. There is a clear physical correlation between the degree of wear and the temperature fluctuations, and the irregular surface of the wear area will cause additional friction heat of the fluid, thereby causing temperature anomalies. This correlation analysis can identify the preferential deposition path of contaminants, as the concave-convex surface of the wear area is more likely to capture suspended particles.
[0045] In one possible implementation, the determination of the diffusion path is based on the principles of fluid dynamics and the migration rules of contaminants. Temperature gradient change points usually correspond to flow rate change areas, and these locations become key nodes for contaminant diffusion. When the main pipeline temperature is 45 degrees Celsius and the branch pipeline temperature drops to 40 degrees Celsius, the temperature difference will drive the migration of contaminants to the low-temperature area. The calculation of the diffusion path length involves the measurement of the distance from the pollution source point to each temperature gradient change point, which directly reflects the spatial range of contaminant propagation. The identification of branch nodes is crucial for understanding the diffusion pattern of contaminants, as the flow differentiation of contaminants occurs at these nodes, forming a multi-path diffusion pattern.
[0046] Specifically, the establishment of the concentration distribution mapping relationship needs to combine fluid mechanics and mass transfer theory. High-concentration contaminant accumulation areas are usually located at low-flow-rate pipeline bending parts or pipe diameter mutation positions, where the fluid residence time is longer, which is conducive to particle deposition. The determination of the deposition area boundary is based on the analysis of the contaminant concentration contour line, and when the contaminant concentration in a certain area exceeds 150% of the average concentration around it, the area is designated as the main deposition area. The reference data for area positioning include the geometric center coordinates and effective coverage area of the deposition area, which provide accurate spatial positioning information for subsequent particle detection. The size of the coverage area is directly related to the severity of the pollution, and the larger the area, the more extensive the pollution diffusion range.
[0047] It should be noted that the working principle of the microscope particle counter is based on optical imaging and image recognition technology, which can accurately identify and count abnormal particles of different sizes in the oil. The particle size analysis and screening method classifies particles through a series of standard aperture screens, thereby obtaining detailed particle size distribution information. When the number of particles with a particle size of 5-20 microns accounts for more than 60% of the total number of particles, it indicates the presence of typical wear particle pollution characteristics. The setting of the abnormal threshold is based on the statistical data of particle concentration under normal working conditions, and the normal concentration mean plus twice the standard deviation is usually used as the judgment standard. The marking of the key pollution area provides a clear target area for subsequent cleaning and maintenance, avoiding blind overall cleaning operations. The construction of the multi-dimensional data matrix integrates multiple characteristic parameters of particles into a unified data structure, and this method can comprehensively describe the complex characteristics of the pollution state. The data standardization processing adopts the Z-score method to convert parameters of different dimensions into dimensionless standardized values, so that each parameter is comparable.
[0048] For example, the unit of particle concentration is pieces per milliliter, and the unit of average particle size is microns. After standardization, both are converted into standardized scores. The formation of the pollution deposition feature vector combines the four key parameters into a four-dimensional vector. This vector representation method is convenient for subsequent pattern recognition and trend analysis, and provides a reliable mathematical basis for quantitative evaluation and prediction of the pollution state.
[0049] Step S105, statistical analysis of the pollution deposition feature vector obtains the pollution distribution pattern and deposition trend, and combined with the vibration transmission path of the equipment, the interference effect of the deposition feature vector on vibration transmission is analyzed to obtain the time sequence correlation of the abnormal vibration signal of the agitator to the pump valve caused by pollution deposition.
[0050] The statistical clustering method is used to group and classify the pollution deposition feature vectors, the distance distribution and the aggregation degree of each feature vector in the four-dimensional space are calculated, a spatial distribution pattern recognition matrix of pollution deposition is established, and the pollution concentration variation gradient and the particle size evolution law are extracted from the recognition matrix as quantitative indexes of the deposition trend. According to the quantitative indexes of the deposition trend, the vibration transmission path data detected by the acceleration sensor in the equipment vibration monitoring network is obtained, the vibration signal propagation path mapping from the agitator to the pump valve through the pipeline is established, the vibration amplitude and frequency characteristics of each propagation node are obtained, and the influence degree of the pollution deposition on the vibration signal propagation characteristics is determined. The numerical correlation between the pollution deposition feature vector and the vibration signal propagation characteristics is analyzed through the vibration signal propagation path mapping, if the pollution concentration exceeds the preset interference threshold, the attenuation ratio of the vibration amplitude and the frequency offset value are calculated, and the quantitative result of the interference effect of the pollution deposition on the vibration transmission between the devices is obtained. According to the interference effect quantitative result, the time sequence corresponding relationship between the agitator vibration signal and the pump valve vibration response is established, the delay time and the correlation strength of the vibration signals between the two devices are calculated through time domain correlation analysis, and the correlation coefficient and the transmission delay parameter of the vibration abnormal signal in time sequence caused by the pollution deposition are determined.
[0051] For example, the implementation of the statistical clustering method is based on the distribution characteristics of the pollution deposition feature vectors in the multidimensional space, each feature vector contains the numerical values of four dimensions of particle concentration, average particle size, particle size variance and deposition density.
[0052] For example, when the feature vector shows that the particle concentration in a certain area is high, the average particle size is medium, the particle size variance is low, and the deposition density is high, the vector will be classified as a typical heavy particle deposition pattern. The calculation of the distance distribution uses the Euclidean distance method, and the vectors with similar characteristics are closer in space, forming obvious aggregation clusters. The spatial distribution pattern recognition matrix converts these aggregation clusters into a two-dimensional mapping, which directly shows the spatial distribution law of pollution deposition. The pollution concentration variation gradient reflects the spatial variation trend of the pollution degree, and the particle size evolution law reveals the particle size variation characteristics of the particles in the deposition process.
[0053] In one possible implementation, the vibration monitoring network is composed of triaxial acceleration sensors arranged at the key equipment nodes, which can simultaneously detect vibration signals in x, y, z three directions. The mechanical vibration generated by the agitator is transmitted to the pump valve equipment through the connecting pipeline, which will be affected by the pollution deposition in the pipeline during transmission. The vibration signal propagation path mapping establishes the complete propagation link from the vibration source to the receiving point, including the direct propagation path and the indirect propagation path through the pipeline wall. The vibration amplitude monitoring of each propagation node shows that under normal working conditions, the vibration amplitude along the propagation path presents a regular attenuation, while the attenuation mode will change significantly when there is pollution deposition. The analysis of frequency characteristics shows that the pollution deposition will change the inherent frequency characteristics of the pipeline, leading to the amplification or suppression of certain frequency components.
[0054] Specifically, there is a clear physical correlation mechanism between the pollution deposition feature vector and the vibration signal propagation characteristics. When a large number of pollution particles adhere to the inner wall of the pipeline, the effective inner diameter of the pipeline decreases, the fluid resistance increases, and the roughness of the pipeline wall changes, which jointly affect the propagation characteristics of the vibration signal. The establishment of the numerical correlation relationship is realized by the regression analysis method, taking the numerical values of each dimension of the feature vector as the independent variable and the vibration propagation parameters as the dependent variable. The setting of the interference threshold is based on the statistical data of the vibration propagation characteristics under normal working conditions. When the pollution concentration exceeds the threshold, the attenuation ratio of the vibration amplitude will increase significantly. The frequency offset value is calculated by comparing the frequency spectrum characteristics before and after the pollution, and the typical offset is the significant attenuation of high-frequency components and the relative enhancement of low-frequency components.
[0055] It should be noted that the establishment of the time sequence correspondence requires accurate time synchronization technology to ensure that the vibration signal collection of the agitator and the pump valve has a unified time reference. The time domain correlation analysis adopts the sliding window method to calculate the correlation coefficient of two vibration signals under different time delays. When the system is running normally, there is a stable delay time between the vibration signal of the agitator and the response signal of the pump valve, which is mainly determined by the signal propagation distance and the medium characteristics. The presence of pollution deposition will change the signal propagation speed, leading to a change in the delay time. The calculation result of the correlation intensity reflects the similarity between the two signals. Under normal working conditions, the correlation intensity is high, while the signal distortion caused by pollution will reduce the correlation intensity. The identification of vibration abnormal signal is based on the threshold judgment of correlation coefficient. When the correlation coefficient is lower than the preset threshold, it indicates that there is a vibration transmission anomaly caused by pollution deposition. The monitoring of the transmission delay parameter provides an important basis for the quantitative evaluation of the pollution degree. The increase of the delay time is positively correlated with the severity of the pollution deposition, which provides reliable technical support for equipment condition monitoring and preventive maintenance.
[0056] Step S106, identifying abnormal equipment according to abnormal signal time sequence correlation, collecting abnormal equipment operation parameters, and synchronously analyzing to obtain equipment operation state coupling, analyzing equipment fault chain reaction characteristics, and generating equipment chain abnormal distribution map according to the chain reaction characteristics.
[0057] According to the correlation coefficient and the transmission delay parameter of the vibration abnormal signal in the time sequence, the equipment node of the correlation abnormality is identified, the current, voltage, temperature, and pressure and other operation parameters of the equipment node of the correlation abnormality are obtained through the data collector, the real-time monitoring database of the abnormal equipment operation parameters is established, and the operation state parameter matrix of each abnormal equipment at the abnormal moment is obtained. The operation state parameter matrix is used for numerical calculation of the correlation degree of the operation parameters between the equipment, the coupling relationship strength of the operation state between the abnormal equipment is determined through the synchronism of the numerical change of the operation parameters, if the coupling relationship strength between the equipment exceeds the preset coupling threshold, it is judged that there is equipment operation state coupling, and the strength distribution and coupling direction information of the equipment operation state coupling are obtained. According to the strength distribution and coupling direction information of the equipment operation state coupling, the propagation path and propagation time sequence of the fault between the equipment are analyzed, the time delay length and propagation speed value of the fault from the source equipment to the target equipment are calculated, the position with the coupling strength exceeding the preset node threshold is identified as the key chain node, and the time and space propagation characteristic parameters of the fault chain reaction are obtained. The connection relationship map between the equipment is constructed through the time and space propagation characteristic parameters of the fault chain reaction, the time delay length, the propagation speed value and the chain node position are mapped into the physical distribution coordinates of the equipment, and the equipment chain abnormal distribution map containing the fault propagation direction, the propagation time sequence and the influence range is generated, which is used as a graphical reference for fault diagnosis and equipment maintenance.
[0058] For example, the abnormal equipment identification process is based on the deep analysis of the time sequence correlation of the vibration signal, when the correlation coefficient is lower than the normal threshold or the transmission delay parameter deviates significantly, it indicates that the corresponding equipment has operation abnormality.
[0059] For example, the normal correlation coefficient between the vibration signal of the agitator and the pump valve response signal is usually maintained at 0.85 or more, and when the pipeline is seriously contaminated and deposited, the coefficient may decrease to 0.6 or less, and the transmission delay time is prolonged from the normal 15 milliseconds to 25 milliseconds. The data collector connects the sensor interface of each equipment through industrial Ethernet, and real-time collects the key operation parameters such as current fluctuation, voltage change, temperature rise and fall, and pressure pulsation. The operation state parameter matrix organizes these multi-dimensional parameters in time sequence, forms a complete data structure reflecting the operation state of the equipment, and this matrix representation method provides a standardized data basis for the subsequent correlation analysis.
[0060] In one possible implementation, the correlation degree numerical calculation adopts the Pearson correlation coefficient method to quantify the degree of mutual influence between devices by comparing the change trends of different device operating parameters. The synchronization of the numerical changes in the operating parameters is a key technology for identifying the coupling relationship between devices. When the operating parameters of two devices exhibit a highly synchronized change pattern in time, it indicates that there is a strong coupling relationship between them.
[0061] For example, when the agitator current increases, causing the pipeline pressure to rise, the workload of the downstream pump device will increase accordingly, and its current consumption will also rise synchronously. This synchronization reflects the physical coupling relationship between the devices. The coupling threshold is set based on statistical analysis of historical operating data, and is usually determined by the average value of the correlation degree under normal operating conditions plus the standard deviation. The strength distribution describes the closeness of the coupling relationship between different device pairs, and the coupling direction information indicates the directionality of the influence transmission.
[0062] Specifically, the analysis of the fault propagation path is based on the topological structure of the coupling relationship between devices. The connection of devices with strong coupling relationship constitutes the preferred propagation path of the fault. The determination of the propagation timing is achieved by monitoring the time sequence of the occurrence of abnormal parameters of each device. The time delay length reflects the time interval required for the fault to affect the target device from the source device. The propagation speed value is calculated by the ratio of the fault influence distance to the propagation time, which directly reflects the urgency of the fault diffusion. The identification of the key interlocking nodes is based on the node degree and centrality analysis. Those devices nodes that have strong coupling relationships with multiple devices usually become the key hubs of fault propagation. When the coupling strength of a device exceeds the preset node threshold, the device is identified as a key interlocking node. The state changes of these nodes will have a significant impact on the entire device network.
[0063] It should be noted that the construction of the connection relationship map between devices adopts a mathematical representation method of directed graph,
[0064] The nodes represent the devices, and the edges represent the coupling relationship. The weight of the edge reflects the coupling strength. The mapping of the physical distribution coordinates combines the abstract relationship map with the actual device space layout, making the visualization analysis of fault propagation more intuitive. The fault propagation direction is displayed in the map by arrows, and the thickness of the arrows corresponds to the size of the propagation strength. The propagation timing information is expressed in the map by color coding or numerical annotation, allowing operators to quickly identify the development sequence of the fault. The representation of the influence range uses the radiation circle or shadow area method to intuitively display the range of devices that may be affected by the fault. This graphical reference tool provides an intuitive basis for fault diagnosis for maintenance personnel, enabling them to quickly locate the fault source, predict the fault development trend, and develop targeted emergency response measures, significantly improving the efficiency and accuracy of device maintenance, while providing scientific decision support for the development of preventive maintenance strategies.
[0065] In step S107, the node vibration frequency and abnormal signal time sequence correlation in the device chain abnormal distribution diagram are identified, the abnormal probability of each node is calculated, and if the abnormal probability of a certain node exceeds the preset threshold and is located at the starting position of the device vibration transmission path, the disturbance source is judged. The rotational speed of the stirrer production process and the pipeline flow distribution are adjusted in combination with the historical oil liquid circulation flow rate difference and the pollution particle deposition rate data.
[0066] The vibration frequency characteristics and abnormal signal time sequence correlation values of each node are extracted from the device chain abnormal distribution diagram, the amplitude distribution and main frequency component of the node vibration frequency are obtained through a spectrum analyzer, and the abnormal probability values of each node are obtained based on the weight calculation of the vibration frequency abnormality degree and the time sequence correlation deviation degree. According to the comparison between the abnormal probability values of each node and the preset abnormal threshold, if the abnormal probability of a certain node exceeds the preset abnormal threshold, the node is marked as a high-risk node, the position of the high-risk node in the vibration transmission sequence is determined through the vibration transmission path information in the device chain abnormal distribution diagram, and the high-risk node located at the starting position of the transmission sequence is judged as the vibration disturbance source. The historical records of the oil liquid circulation flow rate difference and the historical change data of the pollution particle deposition rate in the data storage are obtained, the corresponding historical flow rate difference values and deposition rate values are matched according to the device type and the abnormal probability value of the vibration disturbance source, the matching degree and the numerical deviation degree of the current abnormal condition and the historical record data are determined, the amplitude of the stirrer speed adjustment and the proportion of the pipeline flow redistribution are determined through the matching degree and the numerical deviation degree of the historical record data, and if the vibration disturbance source is a stirrer device, the stirrer speed is reduced and the flow distribution proportion of the branch pipeline is increased, and if the disturbance source is a pipeline node, the stirrer speed is increased and the flow distribution proportion between the main pipeline and the branch pipeline is adjusted.
[0067] For example, the vibration frequency characteristic extraction process needs to obtain the spectrum information of each node from the device chain abnormal distribution diagram, which reflects the vibration characteristic changes of the device in the abnormal state.
[0068] For example, the main frequency of the stirrer node under normal working conditions is usually concentrated in the range of 25-30 Hz, and when mechanical wear or abnormal load occurs, the spectrum will show an enhanced high-frequency component in the range of 50-60 Hz. The spectrum analyzer converts the time domain signal into a frequency domain representation by fast Fourier transform processing of the collected vibration signal, thereby identifying the abnormal frequency component. The amplitude distribution analysis shows that the vibration amplitude of the abnormal node usually increases significantly at a certain frequency band, and this increase mode is closely related to the fault type of the device. The weight calculation method considers the frequency abnormality degree and the time sequence correlation deviation degree, the frequency abnormality degree reflects the spectrum change of the vibration signal, and the time sequence correlation deviation degree reflects the time delay change in the signal transmission process.
[0069] In one possible implementation, the calculation of the abnormal probability value employs a weighted average method, which synthesizes multiple abnormal indicators into a single probability value. The identification of high-risk nodes is based on a probability threshold comparison, and when the abnormal probability of a node exceeds a pre-set threshold, the node is marked as an object that requires special attention. The position ordering of the vibration transmission sequence reflects the propagation order of the vibration signal in the device network, and the node at the starting position is usually the original occurrence of the vibration disturbance. Accurate identification of the disturbance source is crucial for fault root cause analysis, because only by finding the true source of the fault can effective treatment measures be developed. The establishment of the transmission sequence is based on the physical connection relationship between devices and the time sequence of signal propagation, and this sequential representation method makes the complex device network relationship clear and visible.
[0070] Specifically, the matching process of historical data involves pattern recognition and similarity calculation techniques. The data storage stores a large number of historical operation cases, including flow rate variation patterns and particle deposition characteristics under different operating conditions. The matching of the current abnormal condition with the historical record is based on similarity analysis in multiple dimensions, including the device type where the abnormality occurs, the abnormality degree, the development trend, and other factors. The calculation of the matching degree employs mathematical methods such as Euclidean distance or cosine similarity to quantify the similarity between the current condition and the historical case. The numerical deviation degree reflects the difference between the current parameters and the historical normal values, and the larger the deviation, the more serious the abnormality. Effective use of historical record data can provide a mature solution for current abnormal treatment, avoiding the time cost of repeated experiments and adjustments.
[0071] It should be noted that the development of parameter adjustment strategies is based on the comprehensive analysis of historical experience and current conditions. The determination of the mixer speed adjustment range needs to consider the safe operating range of the device and the process requirements, and excessive speed adjustment may lead to new problems. The redistribution of pipeline flow involves the flow balance between the main pipeline and the branch pipeline, and the increase of flow in the branch pipeline can reduce the load pressure of the main pipeline, thereby alleviating the problem of increased resistance caused by pollution deposition. When the disturbance source is the mixer device, reducing the speed can reduce the generation of mechanical vibration, and at the same time, by increasing the flow of the branch pipeline, the overall flow demand can be maintained. When the disturbance source is the pipeline node, increasing the mixer speed can enhance the scouring effect of the fluid, which helps to remove the pollution deposits in the pipeline, and the adjustment of the flow distribution ensures the flow stability of the system. This differentiated adjustment strategy based on the type of disturbance source embodies the concept of precise control, which can take the most appropriate treatment measures for different abnormal root causes, significantly improving the effect and efficiency of abnormal treatment, while avoiding the negative effects of blind adjustment.
[0072] The above merely describes a specific implementation of the present specification, and those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described here. It should be understood that the protection scope of the present specification is not limited to this, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed in the present specification, and these modifications or replacements should be covered within the protection scope of the present specification.
Claims
1. A multi-source data fusion analysis method for lubricating oil production, characterized in that: include: Calculate the speed fluctuation variance based on the agitator speed signal, mark the state where the speed fluctuation variance exceeds the preset threshold as abnormal, and obtain the pipeline pressure distribution data corresponding to the abnormal state; analyze the pipeline pressure distribution data and the oil fluid viscosity data and the oil circulation flow rate data to calculate the shear stress distribution index in the pipeline; detect the oil particle suspension state and the pollution particle deposition rate based on the shear stress distribution index, and calculate the oil pollution diffusion degree based on the oil particle suspension state and the pollution particle deposition rate; determine the oil pollution diffusion path based on the oil temperature fluctuation data and the wear degree data, and determine the pollution deposition based on the oil pollution diffusion path and the oil pollution diffusion degree. area, detect the concentration and particle size distribution of abnormal oil particles in the pollution deposition area, and generate a pollution deposition feature vector; analyze the pollution deposition feature vector and the equipment vibration transmission path, and calculate the time series correlation of the vibration abnormality signal from the agitator to the pump valve; identify abnormal equipment according to the time series correlation of the vibration abnormality signal, collect the operating parameters of the abnormal equipment, analyze the equipment operating state coupling, and generate an equipment chain abnormality distribution map; extract the node vibration frequency and the abnormal signal time series correlation according to the equipment chain abnormality distribution map, calculate the node abnormality probability, determine the disturbance source based on the node abnormality probability and the equipment vibration transmission path, and adjust the agitator speed and pipeline flow distribution ratio.
2. The lubricating oil production multi-source data fusion analysis method according to claim 1, characterized in that: The calculating of the speed fluctuation variance according to the stirrer speed signal, marking a state where the speed fluctuation variance exceeds a preset threshold as abnormal, and obtaining pipeline pressure distribution data corresponding to the abnormal state, includes: The method includes collecting the voltage signal output by the agitator speed sensor and converting the voltage signal into digitized speed data; obtaining the pressure values of the multi-point pressure sensors in the pipeline and generating a time series correlation matrix between the digitized speed data and the pressure values; calculating the speed fluctuation variance within a continuous time window based on the time series correlation matrix; comparing the speed fluctuation variance with a preset threshold, and marking the state exceeding the preset threshold as abnormal; and extracting the pipeline pressure distribution data at the time corresponding to the abnormal state.
3. The lubricating oil production multi-source data fusion analysis method according to claim 1, characterized in that: The method of determining an oil contamination diffusion path based on the oil temperature fluctuation data and the wear degree data, determining a contamination deposition area based on the oil contamination diffusion path and the oil contamination diffusion degree, detecting the concentration and particle size distribution of abnormal oil particles in the contamination deposition area, and generating a contamination deposition feature vector includes: Collect wear degree data output by the pipeline inner wall wear sensor and oil temperature fluctuation data output by the temperature sensor; analyze the correlation between the wear degree data and the oil temperature fluctuation data to determine the flow direction of pollutants; identify the gradient change points in the oil temperature fluctuation data to generate the branch node positions of the diffusion path; calculate the path length distribution from the pollution source to the branch node position; combine the path length distribution with the oil pollution diffusion degree to generate a pollutant concentration distribution map; determine the high concentration area in the pollutant concentration distribution map, and generate the center coordinates and coverage area of the pollution deposition area; detect the concentration and particle size distribution of abnormal particles in the pollution deposition area, and generate a multidimensional data matrix containing concentration, average particle size, and particle size variance; standardize the multidimensional data matrix to generate a pollution deposition feature vector.
4. The lubricating oil production multi-source data fusion analysis method according to claim 1, characterized in that: The analyzing the pipeline pressure distribution data, the oil fluid viscosity data, and the oil circulation velocity data to calculate the shear stress distribution index in the pipeline includes: Collect the pressure value change in the pipeline pressure distribution data; generate a pressure field distribution matrix; calculate the pressure change rate between adjacent measuring points in the pressure field distribution matrix; combine the oil fluid viscosity data and the pressure change rate to generate a fluid velocity gradient distribution; combine the oil circulation flow rate data to calculate the spatial deviation of the fluid velocity gradient distribution and generate a shear strain rate distribution; generate a shear stress distribution index based on the shear strain rate distribution and the oil fluid viscosity data.
5. The lubricating oil production multi-source data fusion analysis method according to claim 4, characterized in that: The detecting the oil particle suspension state and the pollution particle deposition rate according to the shear stress distribution index, and calculating the oil pollution diffusion degree based on the oil particle suspension state and the pollution particle deposition rate, includes: Based on the shear stress distribution index, the oil particle detection process is started; the size distribution and concentration distribution of particles in the oil are scanned to generate particle suspension density data; the sedimentation characteristic parameters in the particle suspension density data are calculated to generate the pollution particle deposition rate; the particle suspension density data is compared with the pollution particle deposition rate to generate the pollution particle concentration gradient; based on the pollution particle concentration gradient, a particle diffusion coefficient is generated; and the oil pollution diffusion degree level is generated by combining the particle diffusion coefficient and the pollution particle concentration gradient.
6. The lubricating oil production multi-source data fusion analysis method according to claim 1, characterized in that: The analyzing the pollution deposition characteristic vector and the equipment vibration transmission path, and calculating the time series correlation of the vibration abnormality signal from the agitator to the pump valve, includes: Process the pollution deposition feature vector to generate a pollution deposition spatial distribution matrix; extract the concentration change gradient in the pollution deposition spatial distribution matrix; collect the vibration amplitude and frequency characteristics in the equipment vibration transmission path; analyze the relationship between the pollution deposition spatial distribution matrix and the vibration amplitude to generate vibration signal interference effect data; and generate a time series correlation coefficient of the agitator and pump valve vibration signals based on the vibration signal interference effect data.
7. The lubricating oil production multi-source data fusion analysis method according to claim 1, characterized in that: The method of identifying abnormal equipment according to the time series correlation of the abnormal vibration signal, collecting operating parameters of the abnormal equipment, analyzing the coupling of equipment operating states, and generating an equipment chain abnormality distribution map includes: Process the timing correlation of the abnormal vibration signal to identify abnormal equipment nodes; collect the operating parameters of the abnormal equipment nodes to generate an operating status parameter matrix; analyze the operating status parameter matrix to generate the coupling relationship strength between devices; generate the fault propagation path and timing based on the coupling relationship strength; map the fault propagation path and timing to generate an equipment chain abnormality distribution map.
8. The lubricating oil production multi-source data fusion analysis method according to claim 7, characterized in that: The method of extracting the correlation between the node vibration frequency and the abnormal signal time series according to the equipment chain abnormality distribution diagram, calculating the node abnormality probability, determining the disturbance source based on the node abnormality probability and the equipment vibration transmission path, and adjusting the agitator speed and pipeline flow distribution ratio includes: Extract the correlation between the node vibration frequency and the abnormal signal time series in the equipment chain abnormality distribution diagram; calculate the amplitude distribution of the node vibration frequency to generate the node abnormality probability; compare the node abnormality probability with the preset threshold and mark the high-risk node; analyze the position of the high-risk node in the equipment vibration transmission path to determine the source of the disturbance; match the disturbance source with the historical oil circulation flow rate data to generate the agitator speed adjustment amplitude and pipeline flow distribution ratio.
Citation Information
Cited By
Inspection and detection experiment device and system for well entering materials for oilfield construction site
CN121723202A
Deposition early warning method and system for sonic nozzle gas flow standard device
CN121740199A
