Lubricating oil pollution state analysis method based on measurement data probability distribution

Abstract

The invention belongs to the technical field of lubricating oil pollution monitoring and analysis, and particularly relates to a lubricating oil pollution state analysis method based on measurement data probability distribution. Comprising the steps of determining lubricating oil pollution factor indexes and a lubricating oil pollution grade set based on corresponding equipment lubricating oil pollution factors and pollution grades; establishing a membership degree matrix of the lubricating oil pollution elements to the lubricating oil pollution state; establishing a fuel oil pollution state change model under each pollution factor through single-factor lubricating oil pollution state monitoring data; a numerical calculation method is adopted to obtain time-varying data of pollution states caused by changes of different lubricating oil elements, and a weight vector is constructed; performing grouping training on an optimization model based on a machine learning method; and substituting a measured value into the model to determine a prediction result. According to the method, more complete data fitting analysis can be carried out based on less original data, and the effective lubricating oil fault analysis method on the basis of lower source data is improved.

Description

Technical Field

[0001] This invention belongs to the field of lubricating oil contamination monitoring and analysis technology, and in particular relates to a method for analyzing the state of lubricating oil contamination based on the probability distribution of measurement data. Background Technology

[0002] Lubrication systems are essential maintenance components for modern machinery, especially various types of power equipment and moving parts. The condition of the lubricating oil within the system directly affects the stability and safety of each mechanism. Good lubricating oil condition ensures efficient operation, effectively reduces structural wear, extends equipment life, and reduces failure rates. Currently, lubricating oil contamination detection mainly relies on a comprehensive assessment combining monitoring data with historical experience analysis. However, with the upgrading of lubrication systems and the increase in lubrication methods, lubricating oil composition, and contaminants, the efficiency of traditional analytical methods is limited. This is particularly true for lubricating oil contamination analysis in critical equipment systems, where high stability requirements and limited data on lubricating oil contamination status make it difficult to meet the needs of efficient analysis and processing. Summary of the Invention

[0003] The purpose of this invention is to provide a method for analyzing lubricating oil faults by combining measurement data with practical needs, enabling more complete data fitting analysis based on less raw data, and improving the effectiveness of lubricating oil fault analysis based on lower source data.

[0004] To achieve the above objectives, the present invention adopts the following technical solution.

[0005] A method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data, comprising:

[0006] S1. Determination of the lubricating oil contamination state range;

[0007] Lubricating oil contamination indexes are determined based on the corresponding equipment lubricating oil contamination elements and contamination levels. and lubricating oil contamination level set Determine the lubricating oil element indicators for different pollution levels based on prior data or measurement data under different pollution levels of equipment. Basic judgment range ;in It refers to the first Individual lubricating oil contamination factors; among which It refers to the first The state of lubricating oil contamination;

[0008] S2. Determine the pollution status membership matrix under different pollution elements.

[0009] For any pollution level, establish a membership matrix of lubricating oil pollution elements to lubricating oil pollution status. ;in , indicating the first The degree of membership of each pollution level corresponding to each lubricating oil pollution element. It refers to the first The first lubricating oil contamination element index is for the first The membership degree of each lubricating oil contamination state, and

[0010]

[0011] For the first Measurement values ​​of each lubricating oil contaminant element; It assesses the boundaries of the interval;

[0012] S3. Establish a fitting model of lubricating oil contamination status with various factors.

[0013] A database of lubricating oil contamination elements was established. Models of fuel contamination state changes under each contamination element were built using single-element lubricating oil contamination state monitoring data. A multi-order fitting model of lubricating oil contamination state with respect to the elements was then established. For any given element... The lubricating oil contamination state of each lubricating oil contamination element is represented by the element fitting model as follows:

[0014] ;in This refers to the monitoring time. It is the preset fitting order. It refers to the first The influence coefficient of the order, It refers to a random variable;

[0015] Different fitting orders were selected, and the monitoring data were used to perform fitting and solution to obtain the influence coefficients and random variable functions of each order. Error analysis was then performed to determine the best fitting parameters for each pollution element and obtain the corresponding fitting model.

[0016] S4. Establish a probability distribution model of lubricating oil contamination status.

[0017] To achieve continuous monitoring and analysis, based on the lubricating oil contamination state variation model obtained in step S3, a numerical calculation method is used to obtain the data on the variation of contamination state over time caused by changes in different lubricating oil elements. A weight vector is then constructed based on the lubricating oil contamination element sensitivity coefficient of the corresponding equipment or based on the degree of influence of lubricating oil contamination elements on the lubricating oil contamination state. ,in It refers to the first The weights of each lubricating oil contamination element are then used to derive the probability distribution model of lubricating oil contamination status. ;in This refers to the probability distribution of lubricating oil in different contamination states under the input measurement values;

[0018] S5. Based on machine learning methods, the lubricating oil contamination state fitting model established in step S3 is used to improve the lubricating oil contamination state data. After being integrated with the actual measurement data, group training is carried out to optimize and adjust the fitting model parameters in step S3 and the weight vector in S4 until the accuracy of the lubricating oil contamination state probability distribution model reaches the best.

[0019] S6. Obtain the measured values, substitute them into the lubricating oil contamination state probability distribution model, determine the distribution probability of different contamination state levels, and take the one with the largest distribution probability as the actual contamination state prediction result.

[0020] In a further improvement or preferred implementation of the aforementioned lubricating oil contamination state analysis method based on the probability distribution of measurement data, step S3 further includes a step of temperature compensation for the liquid viscosity element in the lubricating oil contamination state monitoring data. Specifically, this refers to: compensating the liquid viscosity element in the monitoring data based on the temperature changes during the monitoring process. Influence on fluid viscosity Converted to a preset reference temperature Fluid viscosity at the following temperatures (such as the actual longest operating temperature of the lubricating oil or the standard operating temperature of the lubricating oil): Its conversion formula can be expressed as ;in The temperature influence coefficient was obtained through experiments.

[0021] Further improvements or preferred implementation steps of the aforementioned lubricating oil contamination state analysis method based on the probability distribution of measurement data, step S3 further includes a step of temperature compensation for the water content in the lubricating oil contamination state monitoring data. Specifically, this refers to using a polynomial fitting correction model based on experimental data to obtain the corrected expressions for humidity and temperature. The polynomial fitting correction model is expressed as follows: ;

[0022] ,

[0023] ;

[0024] ;

[0025] in, This refers to the corrected moisture content; This refers to the fitting parameter matrix of the fitted correction model. This refers to the fitting error in the j-th experiment. This refers to the fitting correction model parameter matrix corresponding to the j-th experiment. This refers to the water solubility coefficient of the lubricating oil in the j-th experiment. This refers to the moisture content adjustment coefficient. This refers to the lubricating oil temperature coefficient during the j-th experiment. This refers to the temperature regulation coefficient;

[0026] The experimental results were substituted into the fitting model in turn. The reasonableness of the parameter matrix of the correction model was optimized by the fitting error. After obtaining the fitting correction model that meets the error requirements, the measured moisture content index was substituted into the model for correction.

[0027] Further improvements or optimizations to the aforementioned method for analyzing the state of lubricating oil contamination based on the probability distribution of measurement data include the following steps: the pollutant element indicators need to be selected according to the specific circumstances, based on the quality and type of lubricating oil and the sensitivity of different lubricating oil equipment, and at least include: liquid viscosity element, water content element, and particulate contamination element.

[0028] Further improvements or preferred implementation steps to the aforementioned method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data, wherein the contamination status of the lubricating oil is determined according to different lubricating oils and different lubricating oil equipment specifications and grades, and includes at least: no contamination, light contamination, moderate contamination, and heavy contamination.

[0029] Further improvements or preferred implementation steps for the aforementioned lubricating oil contamination status analysis method based on the probability distribution of measurement data, wherein the error analysis in step S3 adopts one of the following analysis methods: mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R²) error. Attached Figure Description

[0030] Figure 1 This is a flowchart of a lubricating oil contamination status analysis method based on the probability distribution of measurement data;

[0031] Figure 2 This is a schematic diagram showing the relationship between common lubricating oil contamination elements and contamination status indicators. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0033] like Figure 1The diagram shows the basic flowchart of the lubricating oil contamination state analysis method based on the probability distribution of measurement data proposed in this application. This application mainly focuses on the lubricating oil contamination state analysis of various types of critical equipment with high stability. In order to ensure system stability, these critical equipment have high requirements for lubricating oil quality and often adopt redundant processing methods, such as regular replacement and advance treatment, to minimize the deterioration of lubricating oil contamination state. Therefore, the basic data on contamination state is relatively scarce. However, in equipment such as ships and ocean-going platforms, due to the uncertainty of voyage and resupply time, it is still necessary to effectively predict the lubricating oil contamination state in order to analyze and deal with possible lubricating oil contamination failures in advance.

[0034] Therefore, this application is analyzed based on the following main process:

[0035] Step S1: Determine the lubricating oil contamination state range;

[0036] As shown in Table 1, to facilitate the analysis of lubricating oil contamination status, the lubricating oil contamination element indicators must first be determined based on the corresponding equipment lubricating oil contamination elements and contamination levels. and lubricating oil contamination level set Based on prior data or measurement data under different contamination levels of equipment, the lubricating oil element indicators under different contamination levels are determined. Basic judgment range ;

[0037] in It refers to the first The above-mentioned pollutant element indicators are formulated based on common lubricant pollutant element indicators. In actual analysis, due to different lubricant qualities, types, and the sensitivity of different lubricant equipment, specific analysis is required according to the specific situation. Generally, at least the following should be included: liquid viscosity element, water content element, and particulate contamination element.

[0038] in It refers to the first The contamination status of lubricating oil varies. For different lubricating oils and different lubricating equipment, various industries and fields generally have corresponding standards for lubricating oil contamination status, which should generally be formulated according to the actual industry and equipment standards. These typically include: no contamination, light contamination, moderate contamination, and heavy contamination.

[0039] Table 1. Potential boundary relationships of lubricating oil contamination states corresponding to different lubricating oil contamination elements.

[0040]

[0041] S2. Determine the pollution status membership matrix under different pollution elements.

[0042] There is a correlation between a single lubricating oil contamination factor or a combination of multiple lubricating oil contamination factor indicators and the lubricating oil contamination state. This application expresses this correlation using a membership matrix. Since each contamination factor indicator involves a continuous range of measurement values, and given the limited historical monitoring data, it is often impossible to obtain a series of indicator parameter measurement results, this application simplifies the processing and avoids unnecessary calculations by incorporating specific simplifications. Specifically:

[0043] For any pollution level, establish a membership matrix of lubricating oil pollution elements to lubricating oil pollution status. ;in , indicating the first The degree of membership of each pollution level corresponding to each lubricating oil pollution element. It refers to the first The first lubricating oil contamination element index is for the first The membership degree of each lubricating oil contamination state, and

[0044]

[0045] For the first Measurement values ​​of each lubricating oil contaminant element; It assesses the boundaries of the interval;

[0046] During the specific analysis process, the interval boundaries are evaluated. The boundary interval values ​​are determined according to the pollution state level classification standard of lubricating oil pollution element indicators (as shown in Table 1). When further subdivision is required, the values ​​can be further divided according to the classification standard interval until the requirements are met.

[0047] S3. Establish a fitting model of lubricating oil contamination status with various factors.

[0048] Furthermore, to facilitate the generation of subsequent machine learning data, a database of lubricating oil contamination elements was established, and a model of fuel contamination state change under each contamination element was built using single-element lubricating oil contamination state monitoring data.

[0049] like Figure 2 As shown, considering the relationship between various common lubricating oil contamination elements and contamination state indicators, this application adopts a multi-order fitting formula to establish a lubricating oil contamination state fitting model with respect to the elements. For any... The lubricating oil contamination state of each lubricating oil contamination element is represented by the element fitting model as follows: ;

[0050] in This refers to the monitoring time. It is the preset fitting order. It refers to the first The influence coefficient of the order, It refers to a random variable;

[0051] Based on limited monitoring data, different fitting orders are selected, and the monitoring data is used to fit and solve the model to obtain the influence coefficients and random variable functions of each order. Error analysis is then performed to determine the best fitting parameters for each pollution element and obtain the corresponding fitting model. Error analysis can be performed using methods such as mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R²).

[0052] Specifically, in the aforementioned single-factor lubricating oil contamination status monitoring data process, as operations proceed, the lubricating oil temperature will gradually rise or fall, directly altering the influence of lubricating oil-affecting factors on the lubricating oil contamination status. Therefore, ideally, data monitoring should be conducted at a constant temperature. However, this ideal situation is difficult to achieve and does not match the actual situation. Therefore, in the process of acquiring lubricating oil contamination status data, in addition to following single-factor variable monitoring, the temperature change of the lubricating oil over monitoring time should also be considered. In this application, temperature compensation is considered for single-factor monitoring data directly affected by lubricating oil temperature. Specifically:

[0053] For liquid viscosity factors, the temperature during the monitoring process should be considered. Influence on fluid viscosity Converted to a preset reference temperature Fluid viscosity at the following temperatures (such as the actual longest operating temperature of the lubricating oil or the standard operating temperature of the lubricating oil): Its conversion formula can be expressed as ;in The temperature influence coefficient was obtained experimentally.

[0054] Regarding the moisture content factor, since water molecules and lubricating oil are immiscible, they mainly exist in a mixed suspension state. Both are affected by temperature changes to varying degrees, and this effect also changes at different temperatures. Therefore, this application uses a polynomial fitting correction model based on experimental data to obtain the corrected expression for humidity and temperature. The polynomial fitting correction model is expressed as follows: ;

[0055] ,

[0056] ;

[0057] ;

[0058] in, This refers to the corrected moisture content; This refers to the fitting parameter matrix of the fitted correction model. This refers to the fitting error in the j-th experiment. This refers to the fitting correction model parameter matrix corresponding to the j-th experiment. This refers to the water solubility coefficient of the lubricating oil in the j-th experiment. This refers to the moisture content adjustment coefficient. This refers to the lubricating oil temperature coefficient during the j-th experiment. This refers to the temperature regulation coefficient;

[0059] The experimental results were substituted into the fitting model in turn. The reasonableness of the parameter matrix of the correction model was optimized by the fitting error. After obtaining the fitting correction model that meets the error requirements, the measured moisture content index was substituted into the model for correction.

[0060] S4. Establish a probability distribution model of lubricating oil contamination status.

[0061] To achieve continuous monitoring and analysis, based on the lubricating oil contamination state variation model obtained in step S3, numerical calculation methods are used to obtain the data on the variation of contamination state over time caused by changes in different lubricating oil elements. It is recommended that the lubricating oil contamination state variation model established in step S3 be fitted to obtain lubricating oil contamination state fitting data, which effectively avoids the problem of insufficient available contamination state data and establishes a more complete and comprehensive data chain of lubricating oil contamination state variation.

[0062] Furthermore, a weight vector is constructed based on the sensitivity coefficient of lubricating oil contamination factors for the corresponding equipment or based on the degree of influence of lubricating oil contamination factors on the lubricating oil contamination state. ,in It refers to the first The weights of each lubricating oil contamination element are then used to derive the probability distribution model of lubricating oil contamination status. ;in This refers to the probability distribution of lubricating oil in different contamination states under the input measurement values;

[0063] S5. Based on machine learning methods, the lubricating oil contamination state fitting model established in step S3 is used to improve the lubricating oil contamination state data. After being integrated with the actual measurement data, group training is carried out to optimize and adjust the fitting model parameters in step S3 and the weight vector in S4 until the accuracy of the lubricating oil contamination state probability distribution model reaches the best.

[0064] S6. Obtain the measured values, substitute them into the lubricating oil contamination state probability distribution model, determine the distribution probability of different contamination state levels, and take the one with the largest distribution probability as the actual contamination state prediction result.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data, characterized in that, include: S1. Determination of the lubricating oil contamination state range; Lubricating oil contamination indexes are determined based on the corresponding equipment lubricating oil contamination elements and contamination levels. and lubricating oil contamination level set Determine the lubricating oil element indicators for different pollution levels based on prior data or measurement data under different pollution levels of equipment. Basic judgment range ;in It refers to the first Individual lubricating oil contamination factors; among which It refers to the first The state of lubricant contamination; S2. Determine the pollution status membership matrix under different pollution elements. For any pollution level, establish a membership matrix of lubricating oil pollution elements to lubricating oil pollution status. ;in , indicating the first The degree of membership of each pollution level corresponding to each lubricating oil pollution element. It refers to the first The first lubricating oil contamination element index is for the first The membership degree of each lubricating oil contamination state, and For the first Measurement values ​​of each lubricating oil contaminant element; It assesses the boundaries of the interval; S3. Establish a fitting model of lubricating oil contamination status with various factors. A database of lubricating oil contamination elements was established. Models of fuel contamination state changes under each contamination element were built using single-element lubricating oil contamination state monitoring data. A multi-order fitting model of lubricating oil contamination state with respect to the elements was then established. For any given element... The lubricating oil contamination state of each lubricating oil contamination element is represented by the element fitting model as follows: ;in This refers to the monitoring time. It is the preset fitting order. It refers to the first The influence coefficient of the order, It refers to a random variable; Different fitting orders were selected, and the monitoring data were used to perform fitting and solution to obtain the influence coefficients and random variable functions of each order. Error analysis was then performed to determine the best fitting parameters for each pollution element and obtain the corresponding fitting model. S4. Establish a probability distribution model of lubricating oil contamination status. To achieve continuous monitoring and analysis, based on the lubricating oil contamination state variation model obtained in step S3, a numerical calculation method is used to obtain the data on the variation of contamination state over time caused by changes in different lubricating oil elements. A weight vector is then constructed based on the lubricating oil contamination element sensitivity coefficient of the corresponding equipment or based on the degree of influence of lubricating oil contamination elements on the lubricating oil contamination state. ,in It refers to the first The weights of each lubricating oil contamination element are then used to derive the probability distribution model of lubricating oil contamination status. ;in This refers to the probability distribution of lubricating oil in different contamination states under the input measurement values; S5. Based on machine learning methods, the lubricating oil contamination state fitting model established in step S3 is used to improve the lubricating oil contamination state data. After being integrated with the actual measurement data, group training is carried out to optimize and adjust the fitting model parameters in step S3 and the weight vector in S4 until the accuracy of the lubricating oil contamination state probability distribution model reaches the best. S6. Obtain the measured values, substitute them into the lubricating oil contamination state probability distribution model, determine the distribution probability of different contamination state levels, and take the one with the largest distribution probability as the actual contamination state prediction result.

2. The method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data according to claim 1, characterized in that, Step S3 further includes a step of temperature compensation for the liquid viscosity element in the lubricating oil contamination state monitoring data, specifically referring to: compensating for the temperature variation during the monitoring process. Influence on fluid viscosity Converted to a preset reference temperature Fluid viscosity at the following temperatures (such as the actual longest operating temperature of the lubricating oil or the standard operating temperature of the lubricating oil): Its conversion formula can be expressed as ;in The temperature influence coefficient was obtained through experiments.

3. The method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data according to claim 1, characterized in that, Step S3 further includes a step of temperature compensation for the water content in the lubricating oil contamination status monitoring data. Specifically, this involves using a polynomial fitting correction model based on experimental data to obtain the corrected expressions for humidity and temperature. The polynomial fitting correction model is expressed as follows: ; ， ； ； in, This refers to the corrected moisture content; This refers to the fitting parameter matrix of the fitted correction model. This refers to the fitting error in the j-th experiment. This refers to the fitting correction model parameter matrix corresponding to the j-th experiment. This refers to the water solubility coefficient of the lubricating oil in the j-th experiment. This refers to the moisture content adjustment coefficient. This refers to the lubricating oil temperature coefficient during the j-th experiment. This refers to the temperature regulation coefficient; The experimental results were substituted into the fitting model in turn. The reasonableness of the parameter matrix of the correction model was optimized by the fitting error. After obtaining the fitting correction model that meets the error requirements, the measured moisture content index was substituted into the model for correction.

4. The method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data according to claim 1, characterized in that, The pollutant element indicators need to be selected according to the specific circumstances, based on the quality and type of lubricating oil and the sensitivity of different lubricating oil equipment. At least the following should be included: liquid viscosity element, water content element, and particulate contamination element.

5. The method for analyzing lubricating oil contamination status based on the probability distribution of measurement data according to claim 1, characterized in that, The lubricating oil contamination status is determined according to different lubricating oils and different lubricating oil equipment specifications and grades, and includes at least: no contamination, light contamination, moderate contamination, and heavy contamination.

6. The method for analyzing the contamination status of lubricating oil based on the probability distribution of measurement data according to claim 1, characterized in that, In step S3, the error analysis employs one of the following methods: mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), or coefficient of determination (R²) error.

Images