Method for generating estimation model for oil type, method for estimating oil type, and system for estimating oil type

Abstract

An object of the present invention is to provide a technique for identifying the type of oil.SOLUTION: According to an aspect of the present invention, there is provided a method for generating an estimation model for a type of oil, the method including generating an estimation model using an information processing system including an input unit, an output unit, a processing unit, and a storage unit, the information processing system including an estimation model generation unit configured to generate an estimation model, the estimation model generation unit being configured to take in absorption spectrum obtained by transmitting light having at least a part of wavelengths of 800nm from the wavelengths 2500nm through oil of a known type, and analyze the absorption spectrum using a dimension reduction algorithm to generate the estimation model for the type of oil.SELECTED DRAWING: Figure 9

Description

[Technical Field]

[0001] The present invention relates to an oil diagnostic technology, particularly to the maintenance of large machinery that uses industrial oils such as lubricating oil, insulating oil, and processing oil, and relates to a technology suitable for determining whether the correct oil is being used through measurements using sensors and for monitoring the machinery based on the results. [Background technology]

[0002] Diagnosing the properties of lubricants used in rotating parts such as bearings and gears is an important technology for maintaining large rotating machinery. Examples of large rotating machinery include wind turbine gearboxes, air compressors, ships, power generation turbines, construction machinery, agricultural machinery, cutting machines, pumps, and reduction gears for railway vehicles.

[0003] In addition to lubricating oil, insulating oil is used in transformers and other devices for electrical insulation, and diagnosing the properties of insulating oil is also important. Processing oils are also used during machining. There are processing oils for various purposes, such as cutting oil, press oil, heat treatment oil, rust prevention oil, and cleaning oil. In this specification, industrial oils such as lubricating oil, insulating oil, and processing oil are sometimes collectively referred to as oil.

[0004] Lubricating oils are classified into various types depending on their intended use, such as engine oil, turbine oil, hydraulic oil, bearing oil, sliding surface oil, gear oil, compressor oil, and cutting oil. Various additives are blended into the base oil (oil used as the base material) to ensure that each type of lubricating oil meets the required performance. Additives are also blended into other oils to achieve the properties required for each type.

[0005] Each oil product uses different base oils and additives in different concentrations. In many machines, it is necessary to select the appropriate oil to create the appropriate lubrication for the gears, bearings, and other parts.

[0006] In recent years, machine condition monitoring has often adopted a strategy to minimize the lifecycle costs of the machine. Large machines such as power generation turbines use large amounts of lubricating oil, and changing the lubricating oil requires stopping the machine, which has negative aspects such as power generation losses and production stoppages. In addition, it requires costs for purchasing and shipping new oil, oil change labor, and waste oil disposal, so it is desirable to use the lubricating oil for as long as possible. Refrigerant fluids in electric vehicles and data centers also undergo oil diagnosis, and replacement or hardware repairs are carried out. Similarly, the color and other aspects of insulating oil in transformers are monitored.

[0007] Recently, from the perspective of carbon neutrality, automobiles and other vehicles that use large amounts of petroleum-derived fuels are being electrified, and fuel demand will decrease in the future. However, there are often no alternatives for industrial oil, so there is a demand to minimize usage by extending oil change intervals, etc. This is because reducing oil consumption reduces carbon dioxide emissions. However, overlooking oil deterioration and contamination can lead to machine breakdowns.

[0008] Regarding the diagnosis of lubricating oil properties, we define and distinguish between "deterioration" and "contamination." Broadly speaking, it is necessary to diagnose two types of deterioration: (1) oxidative deterioration of lubricating oil over time, and (2) contamination by external contaminants such as water, dust, and wear particles.

[0009] (1) Oxidative degradation of lubricating oils includes degradation due to oxidation of the base oil and degradation due to the consumption of additives. Oxidative degradation of lubricating oils leads to a decrease in wear resistance, changes in viscosity and viscosity index, a decrease in rust prevention properties, and a decrease in corrosion prevention properties. As a result, it can accelerate wear and material fatigue in gearboxes. While it is desirable to use oil for as long as possible, any abnormal deterioration or contamination must be promptly replaced and the equipment inspected.

[0010] On the other hand, there is a background to the circulation of counterfeit lubricating oil. Furthermore, when changing lubricating oil, it is possible to unintentionally put in genuine oil or oil that is not the lubricating oil that should be used. When changing oil in a machine, new oil is generally poured in while a small amount of degraded oil remains in the machine. For this reason, if a different type of lubricating oil is poured in, additives may precipitate, which can cause machine malfunction. If lubricating oil with a different viscosity or type or concentration of additives is poured in, the thickness of the lubricating film during operation may change, and wear resistance and corrosion resistance may deteriorate, resulting in malfunction.

[0011] Counterfeit lubricants are generally manufactured at low cost, packaged in branded containers, and sold at a higher price to make a profit or at a lower price than genuine products. In this case, they may use inferior base oils, reduce the amount of additives, or use cheaper additives, resulting in inferior lubricant performance and machine failure.

[0012] To combat counterfeit lubricants, oil manufacturers have made it more difficult to copy lubricant containers. When the viscosity or color is clearly different from genuine oil, it can sometimes be detected using oil analysis and sensors that measure viscosity and color. When the container is indistinguishable and the viscosity and color are close to genuine oil, it is possible for an oil analysis company to perform composition analysis to identify the fake, but because it is time-consuming and expensive, composition analysis to determine whether the product is genuine is not commonly performed.

[0013] Service providers may monitor the condition of machines during maintenance services, but they may provide maintenance monitoring services on the assumption that genuine oil is used.

[0014] As one method of oil diagnosis, Patent Document 1 discloses the use of an optical sensor to obtain standardized standard values ​​for evaluating oil properties.

[0015] Patent Document 2 discloses that the concentration of an additive having a diphenylamine skeleton is quantitatively measured with high accuracy using a sensor that measures the optical properties of the lubricating oil. [Prior art documents] [Patent documents]

[0016] [Patent Document 1] Japanese Patent Application Publication No. 2024-072170 [Patent Document 2] Japanese Patent Publication No. 2022-118670 Summary of the Invention [Problem to be solved by the invention]

[0017] Using counterfeit lubricants or the wrong type of lubricant in machinery can cause machinery to malfunction.

[0018] An object of the present invention is to provide a technique for identifying the type of oil. [Means for solving the problem]

[0019] One aspect of the present invention is a method for generating an estimation model of an oil type, which uses an information processing system having an input device, an output device, a processing device, and a storage device, and the information processing system has an estimation model generation unit that generates an estimation model, and the estimation model generation unit takes in absorption spectrum data obtained by transmitting light of at least a portion of wavelengths between 800 nm and 2500 nm through oil of a known type, analyzes the absorption spectrum data using a dimensionality reduction algorithm, and generates an estimation model of the oil type.

[0020] Another aspect of the present invention is a method for estimating oil type, in which the oil type estimation model represents the oil type using values ​​of at least two principal components, and a diagnostic system consisting of an information processing device implementing the estimation model is provided, wherein the diagnostic system inputs diagnostic object absorption spectrum data obtained by transmitting light of at least a portion of wavelengths between 800 nm and 2500 nm through the diagnostic object oil, values ​​based on the diagnostic object absorption spectrum data are input to the estimation model, and the type of the diagnostic object oil is estimated from the estimation model.

[0021] Another aspect of the present invention is an oil type estimation system including an information processing device that implements the estimation model. [Effects of the Invention]

[0022] According to the present invention, it is possible to provide a technology for identifying the type of oil. Problems, configurations, effects, etc. other than those described above will become clear from the following description of the embodiments. [Brief explanation of the drawings]

[0023] [Figure 1] Schematic diagram of an optical sensor that measures near-infrared absorption spectra. [Figure 2] FIG. 10 is a flowchart showing a process for generating a discrimination map used in a type estimation model for estimating the type of oil. [Figure 3] FIG. 2 is a graph showing the results of a principal component analysis of lubricating oils. [Figure 4] Schematic diagram of an optical sensor that measures near-infrared absorption spectra. [Figure 5] FIG. 4 is a flow chart showing a process for determining the type of lubricant and a process for monitoring and diagnosing the state of the lubricant; [Figure 6] FIG. 2 is a graph showing the results of a principal component analysis of lubricating oils. [Figure 7] FIG. 2 is a graph showing the results of a principal component analysis of lubricating oils. [Figure 8] FIG. 1 is a block diagram of an oil diagnosis system according to an embodiment. [Figure 9]3 is a flowchart of an oil diagnosis method according to an embodiment. [Figure 10] 3 is a flow chart of oil property estimation in an embodiment. [Figure 11] Graph showing the relationship between usage time and oil viscosity, total acid number, and antioxidant concentration. [Figure 12] FIG. 1 is a graph showing an example of a near-infrared spectrum. [Figure 13] 1 is a flowchart of an estimation model generation process according to an embodiment. [Figure 14] A graph showing an example of a near-infrared spectrum, differentiated twice. [Figure 15] 3 is a flowchart of a process for estimating oil properties in an embodiment. [Figure 16] FIG. 10 is a graph showing the consistency between predicted values ​​and actual measured values ​​using the estimation model of the embodiment. [Figure 17] 1 is a schematic diagram of a lubricant monitoring system for a wind turbine generator according to an embodiment; [Figure 18] Conceptual diagram of a rotating part equipped with a lubricant sensor. [Figure 19] FIG. 4 is a flow chart showing a lubricant diagnostic process according to an embodiment. [Figure 20] Graph showing an example of displaying results. DETAILED DESCRIPTION OF THE INVENTION

[0024] The embodiments will be described in detail with reference to the drawings. However, the present invention should not be interpreted as being limited to the description of the embodiments shown below. Those skilled in the art will easily understand that the specific configuration can be changed within the scope of the idea or purpose of the present invention.

[0025] In the configurations of the embodiments described below, the same parts or parts having similar functions are denoted by the same reference numerals in different drawings, and duplicated explanations may be omitted.

[0026] When there are multiple elements with the same or similar functions, they may be described using the same reference numeral with different subscripts. However, when there is no need to distinguish between multiple elements, the subscripts may be omitted.

[0027] The designations "first," "second," "third," etc. in this specification are used to identify components and do not necessarily limit the number, order, or content of the components. Furthermore, numbers used to identify components are used in different contexts, and a number used in one context does not necessarily indicate the same configuration in another context. Furthermore, this does not prevent a component identified by a certain number from also serving the function of a component identified by another number.

[0028] In order to facilitate understanding of the invention, the position, size, shape, range, etc. of each component shown in the drawings, etc. may not represent the actual position, size, shape, range, etc. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings, etc.

[0029] The publications, patents, and patent applications cited in this specification are incorporated herein by reference. In this specification, elements expressed in the singular include the plural unless otherwise clearly indicated in the context. In this specification, the wavelength range from 800 nm to 2500 nm is referred to as the near-infrared region.

[0030] A typical embodiment uses an information processing system including an input device, an output device, a processing device, and a storage device. The information processing system includes an estimation model generation unit that generates an estimation model, and the estimation model generation unit receives an absorption spectrum obtained by transmitting light with at least a portion of wavelengths between 800 nm and 2500 nm through an oil containing a base oil and additives, and a value reflecting the oil characteristics, and generates an estimation model using the value based on the absorption spectrum as an explanatory variable and a value reflecting the oil type as a response variable.

[0031] Another aspect of the present invention is a method for determining oil type using a diagnostic system consisting of an information processing device that implements the above-mentioned estimation model, in which an absorption spectrum obtained by transmitting light with at least a portion of wavelengths between 800 nm and 2500 nm through an oil to be diagnosed that contains base oil and additives is input to the diagnostic system, a value based on the oil type is input to the estimation model, and a value reflecting the characteristics of the oil to be diagnosed is obtained from the estimation model.

[0032] Another aspect of the present invention is an oil type diagnosis system including an information processing device that implements the above estimation model.

[0033] <Types of lubricants and additives> There are various types of lubricating oils, including engine oil, turbine oil, hydraulic oil, bearing oil, sliding surface oil, gear oil, compressor oil, and cutting oil.

[0034] Lubricating oils are composed of base oils and additives. Base oils, also called base oils, make up 80-90% of lubricating oils, and the performance of lubricating oils is determined by the base oil. Base oils are classified into mineral oils and synthetic oils. The American Petroleum Institute (API) classifies base oils into five groups. Mineral oils are oils distilled and refined from petroleum, and are inexpensive.

[0035] Mineral oils are further classified into paraffinic oil and naphthenic oil. Paraffinic base oil is often used in relatively low-cost lubricants, but it has inferior performance compared to synthetic oils and is not suitable for use at high temperatures or in harsh environments. Paraffinic oil is a base oil in which the paraffin carbon number of the contained components is 50% or more. Paraffinic base oils tend to crystallize under low temperature conditions, so pour point depressants are added. Naphthenic base oil is a base oil in which 30% or more naphthenic compounds are present. Although it has a low viscosity index, it has high solubility and excellent low-temperature fluidity.

[0036] Synthetic oils are base oils produced by chemical synthesis and are classified into synthetic hydrocarbon oils, ester oils, ether oils, silicone oils, and fluorine oils. Synthetic oils are used in automotive brake fluids, metalworking oils, and industrial and automotive lubricants, and have a wider usable temperature range than mineral oils. Therefore, they are used in conditions where lubricants using mineral base oils are prone to performance problems, such as low temperatures, high temperatures, high shear, resin resistance, and vacuum conditions. They are more expensive than mineral oils.

[0037] Synthetic hydrocarbon oils include polyalphaolefins (PAO), polybutene, alkylbenzenes, and cycloalkanes. Synthetic hydrocarbon oils generally have high viscosity indexes and thermal stability (especially at low temperatures). They also have low evaporation loss. They have excellent resistance to rubber and resins, making them suitable for a wide range of applications, but are unsuitable for use with natural rubber or ethylene propylene diene rubber (EPDM). Ester oils are compounds of fatty acids and alcohols, or fatty acids and glycerin. Ester oils include monoesters, diesters (DOS), polyol esters, phosphate esters, and silicate esters (silicates). Ester oils generally have excellent lubricating properties and thermal stability, and are characterized by a wide operating temperature range.

[0038] Ether oils include polyalkylene glycols (PAGs) and phenyl ethers. PAGs, in particular, have excellent lubricity and a wide viscosity range. Silicone oils include polysiloxanes and silicate esters. Silicone oils have excellent heat, water, and cold resistance. They are resistant to sludge formation even when degraded, are hygroscopic, low-toxicity, colorless, and odorless, making them suitable for a wide range of applications. Fluorine oils are primarily composed of trifluoroethylene. They have excellent heat resistance and oxidation stability, but are expensive. The performance of lubricating oils, which are dependent on the base oil, includes lubricity (wear resistance), low torque, low-temperature resistance, heat resistance, and resin resistance.

[0039] As such, there are many types of base oil, and the composition and purity vary from product to product depending on the crude oil used as the raw material and the manufacturer.

[0040] Additives include antioxidants, rust inhibitors, antifoaming agents, viscosity index improvers, oiliness improvers, extreme pressure additives, detergent dispersants, pour point depressants, and emulsifiers. For many oils, the degree of additive consumption through use can be used to determine when to change the oil or as an indicator of mechanical abnormalities. The concentration of additives can be quantified by measuring changes in absorbance in the near-infrared range.

[0041] <Functions of additives> Lubricating oils are composed of base oils and additives. Base oils include mineral oils made from petroleum, high-performance synthetic oils, bio-oils made from plants, and biodegradable oils. Lubricating oil deterioration is an oxidation reaction involving oxygen. Antioxidants, a typical additive, are added to prevent base oil oxidation, and base oil oxidation usually begins when the antioxidants are depleted to a certain extent. When base oil oxidation begins, changes in viscosity occur, causing changes in the thickness of the lubricating film and other deterioration in lubrication properties, so an oil change is generally recommended. Therefore, by monitoring the level of antioxidant consumption, it is possible to estimate the remaining life of lubricating oil.

[0042] Among additives, extreme-pressure additives and anti-friction agents function to prevent wear on the sliding surfaces of parts. Because the consumption of these additives accelerates part wear, the remaining life of a lubricant is sometimes determined in terms of the concentration of the extreme-pressure additives and anti-friction agents. Therefore, by monitoring the degree of consumption of extreme-pressure additives and anti-friction agents, it is possible to estimate the remaining life of a lubricant. The lubricant's life is determined when either the amount or concentration of the antioxidant, extreme-pressure additive, or anti-friction agent reaches a threshold, and an oil change is generally recommended at this point.

[0043] Other additives such as rust inhibitors, antifoaming agents, viscosity index improvers, oiliness improvers, detergents and dispersants, pour point depressants, and emulsifiers are also known to be consumed as the lubricant is used, and thresholds are set based on the functionality of the lubricant and are sometimes used as indicators for determining when to change the oil. The type of oil is basically determined by the combination of base oil and additives.

[0044] <Identification of types and products using near-infrared absorption spectra> Through research by the inventors, it was discovered that it is possible to distinguish between lubricating oils with similar apparent colors and viscosities, which has previously been difficult to distinguish, and to determine which manufacturer's product a lubricating oil of unknown viscosity or viscosity is similar to, by measuring near-infrared light absorption spectra at wavelengths from 800 nm to 2500 nm using an optical sensor.

[0045] 1 is a diagram showing one form of optical sensor. Optical sensor 700 incorporates light source 101 that emits at least a portion of near-infrared light with wavelengths of, for example, 800 nm to 3 μm (3000 nm), and detector 103 that can detect the light emitted from light source 101. Light emitted from light source 101 (indicated by the arrow in the figure) passes through lubricating oil 102, which is the object to be measured, and the transmitted light is measured by detector 103. It is sufficient for optical sensor 700 to obtain an absorption spectrum of at least a portion of the wavelengths of near-infrared light with wavelengths of 800 nm to 3 μm (3000 nm).

[0046] Near-infrared spectroscopy measures the vibrational energy of atomic bonds in organic compounds. The near-infrared wavelength range only corresponds to overtones and combination tones of the fundamental tone, and since the fundamental tone is not detected, it is characterized by low absorption intensity. The reason for the low absorption is that overtones and combination tones are forbidden transitions with a low probability of occurring.

[0047] The advantage of near-infrared spectroscopy, which measures such low-probability forbidden transition absorption, is its excellent transparency and is less susceptible to the concentration saturation problem that occurs with ultraviolet-visible spectroscopy and mid-infrared spectroscopy. This means that there is no need to make the optical path length extremely short when measuring liquids such as lubricating oils, and an optical path length of approximately 2 mm to 20 mm can be selected.

[0048] Figure 2 is a flow diagram showing the process S1320 for generating a discrimination map used in a type estimation model for estimating the type of oil. First, a near-infrared absorption spectrum is obtained for the oil to be identified using the optical sensor shown in Figure 1. This near-infrared absorption spectrum is input (S1321). An example of a near-infrared absorption spectrum is shown in Figure 12.

[0049] Spectral analysis for quantifying additives in lubricating oil involves spectral preprocessing, as shown in the following representative example. In the first step, smoothing is performed to remove spectral noise (S1322). For example, a method is used in which the central moving average method is applied to a selected interval of approximately 3 to 20 adjacent points.

[0050] In the second step, a second-order derivative filter is applied to the spectrum, which allows for the extraction of wavelengths with large spectral changes, and allows for the extraction of peaks from spectra with complex overlaps and unclear peak positions, which is typical of near-infrared absorption spectra (S1323). Either of these two steps can be performed first. An example of a twice-differentiated near-infrared absorption spectrum is shown in Figure 14.

[0051] Next, multivariate analysis is performed on the preprocessed spectra (S1324). To identify the type and product of lubricant, principal component analysis (PCA), a type of unsupervised learning method using near-infrared absorption spectra, which contain a wealth of information about the composition of lubricants, is used. PCA is a method applied when quantitative data indicating the relationship between n samples and p variables is given. It is a method of extracting a small number of less correlated synthetic variables that best represent the overall variability from a large number of correlated p variables through dimensionality reduction. Using a dimensionality reduction algorithm such as PCA makes it easier to identify oils using near-infrared absorption spectra.

[0052] Principal component analysis can be performed using the following procedure. The near-infrared absorption spectra of genuine oil, lubricating oils with known product names, and lubricating oils to be identified are measured, and a two- or three-dimensional discrimination map is created by principal component analysis using the spectral data (S1325). This discrimination map is then implemented in the type estimation model. The near-infrared absorption spectrum of lubricating oils with unknown product names or whose authenticity is to be determined (the oil to be diagnosed) is then measured, and principal component analysis is performed using the spectral data. Based on the degree of match with lubricating oils with known product names, the product name of the lubricating oil to be diagnosed and its authenticity are identified. The processing of the oil to be diagnosed is basically the same as the processing in S1321 to S1324 in Figure 2.

[0053] As the type estimation model, it is also possible to use an estimation model that uses the above-mentioned discrimination map and is machine-learned using spectral data as an explanatory variable and the discrimination map score as a target variable.

[0054] In addition to principal component analysis, methods that can be used to identify the type of lubricant and determine its authenticity include support vector machines, k-nearest neighbor methods, cluster analysis, quantum circuit learning, and discriminant analysis.

[0055] This type of lubricant type identification and authenticity determination can determine whether two or more types of oil samples are the same product or brand.

[0056] The procedure for identifying the type of lubricant oil will be explained using the example of principal component analysis mentioned above. Distance can be used to quantify the mutual similarity of the results of principal component analysis using near-infrared absorption spectrum data of multiple types of oil. Distances that can be used to explain similarity include Euclidean distance, Chebyshev distance, and Mahalanobis distance. These distances (d between the i-th sample and the j-th sample) i,j ) is defined as follows, where m is the dimension of the component.

[0057] The Euclidean distance is expressed by Equation 1.

[0058]

number

[0059] The Chebyshev distance is expressed by Equation 2.

[0060]

number

[0061] The Mahalanobis distance is expressed by Equation 3.

[0062]

number

[0063] This section explains how to identify the variety of an unknown oil. For example, if there are known types of oils A, B, C, D, E, F, and G, and an unknown type of oil Z, where oil Z is one of oils A to F, the near-infrared absorption spectra of these oils are acquired in the wavelength range of 1300 nm to 2500 nm, and principal component analysis is performed using the acquired spectral data to plot the scores of the first and second principal components. The oil with the score closest to oil Z is determined to be the oil of the same type as oil Z.

[0064] Figure 3 shows the results of principal component analysis of lubricating oils (the discrimination map created in Figure 2). The discrimination map, in which known types of oils A to G are plotted using two-dimensional principal components, is compared with the plot of unknown oil Z, which was subjected to a similar principal component analysis. According to the results of the principal component analysis in Figure 3, the plots of oil Z and oil B almost overlap, indicating that oil Z is the same type of oil as oil B.

[0065] Even for the same type of oil, characteristics may vary depending on usage conditions. In such cases, for example, for a known type of oil A, near-infrared absorption spectra are obtained for multiple samples, such as new oil, deteriorated oil, oil containing a small amount of water, and oil containing a small amount of foreign matter, and principal component analysis is performed to determine the distribution, or probability density, of oil A, and the Mahalanobis distance where oil A is distributed.

[0066] Additionally, for other oils of known type (B to F), the Mahalanobis distances that can be distributed are calculated using the same procedure as for oil A. For each oil score, a distribution range that includes a specified percentage of samples, such as 95%, is calculated. Then, for example, if the analysis result of an unknown type of oil falls within the 95% distribution range of any of A to H, it is determined to match that oil. Alternatively, if it falls within a specified distance from the outer edge of the 95% distribution range, it is determined to match that oil.

[0067] Alternatively, a method can be used in which the score of an oil of unknown type and the score of an oil of known type are calculated, and the oil with the smallest distance from the unknown oil is determined to be the same as the unknown oil.

[0068] Alternatively, a method can be used in which, if the distance between the score of an oil of unknown type and the score of an oil of known type is closer than a predetermined distance, the oil of unknown type is determined to be the same as the oil of known type.

[0069] <Measurement methods using optical sensors> The near-infrared absorption spectrum of a lubricating oil is measured by an optical sensor 700 having a light source that emits light with a wavelength in the near-infrared region and a detector that detects wavelengths in the near-infrared region. The lubricating oil to be measured is measured by a detector 703, which detects the light intensity of light emitted from a light source 701 after it has passed through a lubricating oil 702.

[0070] Figure 1 shown above shows one form of sensor, but it is also possible for the light emitted from the light source 701 to pass through the lubricating oil 702 and enter the detector 703 using means such as reflection by a mirror, focusing by a lens, or passing through a prism.

[0071] 4 is a diagram showing another form of the sensor. Optical sensor 700A has light source 101 and detector 103 arranged on the same plane, in contact with lubricating oil 102, with reflector 104 installed on the opposite side, and light emitted from light source 101 (indicated by the arrow in the figure) reflected by reflector 104 and received by detector 103. Even in this form, a lens, mirror, or prism may be installed along the optical path.

[0072] <Target lubricants> The subject of this study is lubricating oils containing base oils and additives. The base oils are mineral oils or synthetic oils, and the additives are one or more additives selected from antioxidants, rust inhibitors, antifoaming agents, viscosity index improvers, oiliness improvers, extreme pressure additives, detergents and dispersants, pour point depressants, emulsifiers, etc. Types of lubricating oils include engine oil, turbine oil, hydraulic oil, bearing oil, slideway oil, gear oil, compressor oil, cutting oil, insulating oil, cutting processing oil, press processing oil, heat treatment oil, rust preventative oil, and cleaning oil.

[0073] <Lubricant type determination and subsequent actions> Figure 5 shows an example flow of an oil maintenance use case that utilizes the oil type determination of the embodiment. When the lubricating oil in a generator or the like is changed (S401), optical sensor 700A is installed in the oil tank or oil piping as described below, and the near-infrared absorption spectrum of the lubricating oil is acquired immediately after the lubricating oil change to measure the oil to be identified (S402). As an alternative method, it is also possible to use optical sensor 700 or 700A to acquire the near-infrared absorption spectrum of a small amount of lubricating oil sampled from the machine.

[0074] Separately, using the method described in Figure 2, near-infrared absorption spectra are obtained for lubricating oils that should be used, genuine oils, and lubricating oils that are likely to be misused, and a discrimination map for various lubricating oils is created using principal component analysis (S403).

[0075] Using the near-infrared absorption spectrum data measured from the lubricant filled at the time of replacement and the near-infrared absorption spectrum data used to create the discrimination map for various lubricants, principal component analysis is performed in the same manner as when the discrimination map was created (S404). By comparing the results of the principal component analysis with the discrimination map, it is possible to confirm whether the lubricant filled at the time of replacement is the lubricant that should have been used or the genuine oil (S405). If it does not match the genuine oil, the machine owner, maintenance company, machine sales company, etc. are notified (S407) (authenticity determination process).

[0076] If it is confirmed as a result of the authenticity determination process that the lubricating oil filled at the time of replacement is the same as the lubricating oil that should have been used or the genuine oil, the machine is allowed to continue operating (S406).

[0077] If the authenticity determination process reveals that the lubricant filled during the replacement is not the same as the lubricant that should have been used or the genuine oil, the machine owner is notified to replace the lubricant with the lubricant that should have been used without operating the machine (S407).

[0078] Following the authentication process, when the machine is operated, it is effective to monitor the deterioration and contamination status of the lubricant. The near-infrared absorption spectrum data immediately after filling and the near-infrared absorption spectrum of the lubricant during use, for example, once per hour, are acquired and stored (S408). A type of machine learning technique called partial least squares regression (PLS) can be used to measure the consumption of additives in the lubricant (S409) and predict the next replacement time (S410). Evaluation of additive amounts using PLS will be explained later in Figure 13.

[0079] The amount of additives can also be evaluated using the technology disclosed in Patent Document 2 (JP 2022-118670 A). As described above, it is possible to determine the type of lubricating oil and perform condition monitoring and diagnosis processing. [Example]

[0080] An expected application example will be explained in association with the example flow in Figure 5. In this example, replaced oil is compared with one type of genuine oil. The engine oil is replaced in a 150 kVA diesel generator that is supposed to use genuine engine oil X (S401). Engine oil Y filled at the time of replacement and genuine engine oil X are sampled, and near-infrared absorption spectra in the wavelength range of 800 nm to 2500 nm are obtained using a small device equipped with optical sensor 700A shown in Figure 4. Measurements are taken 10 times for each oil. Principal component analysis is performed using the acquired spectral data (S404), and the results are plotted with the first principal component on the horizontal axis and the second principal component on the vertical axis.

[0081] Figure 6 shows an example of the results of principal component analysis of lubricating oil. As preprocessing for the principal component analysis, second-order differentiation of the spectral data and smoothing of 11 adjacent points were performed.

[0082] The results of the principal component analysis show that genuine engine oil X and engine oil Y filled at the time of change have overlapping plots, indicating that they are the same lubricating oil (S405). This result is notified to the machine owner, maintenance company, and machine sales company (S406), and the diesel generator is operated. Thereafter, the near-infrared absorption spectrum of the engine oil filled at the time of change is obtained once a day, and partial least squares regression analysis is performed using the near-infrared absorption spectrum data to quantify the antioxidant concentration in the engine oil (S408).

[0083] After 1000 hours, it was confirmed that the antioxidant concentration in the engine oil filled at the time of replacement had fallen to 30% of the initial value (S409), so the owner of the machine, the maintenance company, and the machine's sales company were notified to replace the engine oil filled at the time of replacement (S410), and the diesel generator's engine oil was replaced three days later (S401).

[0084] Immediately after the oil change (S401), a small amount of the newly filled engine oil is sampled, and a near-infrared absorption spectrum in the wavelength range of 800 nm to 2500 nm is obtained using a small device equipped with an optical sensor 700A (S402). Principal component analysis is performed using the obtained spectral data (S404), and it is confirmed that genuine engine oil X and the newly filled engine oil Y are the same lubricating oil (S405), and the diesel generator is then operated. [Example]

[0085] An expected application example will be explained in association with the example flow in Figure 5. In this example, replaced oil is compared with multiple types of oil. Seven types of gear oil are commercially available for use in wind turbine gearboxes. It is known that if a different gear oil product is filled when replacing these gear oils, additives may precipitate and the gearbox may malfunction.

[0086] For the seven types of gear oils A to G, near-infrared absorption spectra in the wavelength range of 1400 nm to 2000 nm are obtained using a small device equipped with optical sensor 700 shown in Fig. 1 (S402). Using the obtained near-infrared absorption spectrum data, smoothing of nine neighboring points is performed as preprocessing, and then principal component analysis is performed (S404), and the results are plotted with the first principal component on the horizontal axis and the second principal component on the vertical axis.

[0087] Figure 7 shows an example of the results of principal component analysis of lubricating oil. The gear oil in the gearbox of wind turbine H was replaced (S401), and the near-infrared absorption spectrum of the filled gear oil was obtained (S402). Gear oil A was to be used as the original oil in wind turbine H. When the near-infrared absorption spectrum data of the filled gear oil was subjected to principal component analysis (S404) together with the near-infrared absorption spectrum data of gear oils A to G (S403), it was determined that the filled gear oil was the same gear oil as gear oil A (S405).

[0088] The results, along with instructions to change the gear oil in the gearbox of wind turbine H, were notified to the owner of wind turbine H, its maintenance company, and the sales company of wind turbine H (S410), and one week later the maintenance company of wind turbine H carried out the oil change (S401). The gear oil that was changed again was confirmed to be the same as gear oil A by principal component analysis using near-infrared absorption spectrum data, and wind turbine H began operation the following day. [Example]

[0089] An example of an expected application will be explained below. The optical sensor 700 shown in Figure 1 was installed in the engine oil piping of a 300 kW diesel generator so that the near-infrared absorption spectrum of the oil could be collected. The near-infrared absorption spectrum is input into the following diagnostic system as needed.

[0090] <Diagnostic system configuration> FIG. 8 shows an example of a system used to diagnose engine oil in the embodiment. The diagnostic system 1200 in the embodiment is configured using a normal computer such as a server. The diagnostic system 1200 includes an input device 1210, an output device 1220, and a processing device 1230, and stores programs and data for processing in a storage device 1240. The storage device 1240 may be configured by combining known storage devices such as semiconductor memory and a magnetic disk device. The diagnostic system 1200 has the function of determining the type of oil and the function of monitoring the consumption of antioxidants in the oil.

[0091] The storage device 1240 includes an estimation model generation unit 1250 and an oil type estimation unit 1260. In this embodiment, the estimation model generation unit 1250 and the oil type estimation unit 1260 are included in the same device, but they can also be configured as separate devices. In this embodiment, a monitoring unit 1262 is optionally attached to the oil type estimation unit 1260, making it possible to monitor the properties of the oil during operation.

[0092] The estimation model generation unit 1250 includes a measurement database 1251 , a preprocessing unit 1252 A, a learning database 1253 , and a multivariate analysis unit 1254 .

[0093] The oil type estimation unit 1260 includes a preprocessing unit 1252 B, an estimation model 1261 , and a monitoring unit 1262 .

[0094] <Outline of processing by the diagnostic system> FIG. 9 shows an example of engine oil diagnosis using diagnostic system 1200. First, measurement data is acquired from actual engine oil, and measurement database 1251 is generated. The measurement database stores near-infrared absorption spectrum data measured with an optical sensor such as that shown in FIG. 1 or 4 for the oils to be compared (S1310). For example, near-infrared absorption spectrum data for genuine oil M, the engine oil to be used, gear oil N made by the same manufacturer as genuine oil M, and gear oil L made by a different manufacturer are stored in measurement database 1251.

[0095] As explained in Figure 2, the preprocessing unit 1252A of the estimation model generation unit 1250 is used to perform smoothing (S1322) and differentiation (S1323) of the measured data, to prepare preprocessed near-infrared absorption spectrum data, which is then stored in the training database 1253 as training data.

[0096] As explained in FIG. 2, the multivariate analysis unit 1254 uses principal component analysis (S1324) to create a discrimination map of oil types as shown in FIG. 3 from the training data (S1325), and generates an estimation model 1261 (S1320).

[0097] Using the obtained estimation model 1261, the oil type estimation unit 1260 estimates the type of oil using the method described above (S1330). That is, the diagnostic object absorption spectrum data is analyzed using a dimension reduction algorithm, and multivariate analysis using principal component analysis is performed to express the type of diagnostic object oil using values ​​of at least two principal components. This is then compared with an oil type estimation model that expresses the oil type using values ​​of at least two principal components. The results of oil type estimation S1330 are automatically notified to relevant parties by email.

[0098] If the result of the oil type estimation confirms that the oil is the same as the oil that should be used, the diagnostic system notifies permission to operate the diesel generator and then proceeds to oil property estimation. The monitoring unit 1262 collects actual measurement data every 12 hours and performs oil property estimation processing (S1340).

[0099] <Oil property estimation processing using diagnostic system> Figure 10 is a flow diagram of the oil property estimation process executed by the monitoring unit 1262. Oil measurement data is collected at any time using an optical sensor such as that shown in Figure 1 or 4 (S1341), and the oil property is estimated (S1342). The results of the oil property estimation process are automatically notified to relevant parties by email.

[0100] In this example, the viscosity of the new engine oil was 20 cP, and it is recommended to change the oil when the viscosity reaches 25 cP with use. The viscosity of this engine oil increases due to the consumption of the antioxidant. The engine oil contained a phenolic antioxidant, and its concentration in the new oil was 3.5 wt%.

[0101] The diesel generator was operated continuously. Every 12 hours, the near-infrared absorption spectrum of the engine oil was measured using the optical sensor shown in Figure 1. The wavelength resolution of the near-infrared absorption spectrum was 5 nm. For remote oil monitoring using an optical sensor, the technology described in Patent Document 2 and JP 2020-12690 A cited therein can be applied. Below, we will explain the technique using machine learning.

[0102] <Oil property estimation processing using machine learning> The process of estimating oil properties using machine learning is described below. The oil is used in an actual usage environment, and data on changes in properties over time is obtained. For example, the concentration of antioxidants in sampled engine oil is quantified using high-performance liquid chromatography (HPLC). The viscosity of the engine oil is also measured using any known method. The measured near-infrared absorption spectrum, antioxidant concentration, and viscosity are linked to time information and recorded in the actual measurement database 1251. Note that the time interval for obtaining near-infrared absorption spectrum data is an example, and any known analytical method for oil properties may also be selected.

[0103] As an example of data representation in the actual measurement database, a graph showing the relationship between usage time and oil viscosity, total acid number, and antioxidant concentration is shown in Figure 11. By preparing such relationships as reference data, viscosity can be calculated from antioxidant concentration.

[0104] Furthermore, the near-infrared absorption spectrum of the engine oil was measured using the optical sensor shown in Figure 1. These data were also linked to time and stored in the measurement database 1251.

[0105] Figure 12 shows an example of the near-infrared absorption spectrum of engine oil. In this example, the absorbance is shown for wavelengths from 1550 nm to 1950 nm. Multiple curves show the spectra after different periods of use.

[0106] Next, explanatory variables are generated using the measured near-infrared absorption spectrum, and an oil property estimation model is generated using the antioxidant concentration measured using HPLC as the response variable. In generating the oil property estimation model, model generation using known machine learning, in which explanatory variables are used as input and response variables are used as output, can be used. In this example, PLS regression analysis was performed.

[0107] 13 shows a flow diagram of the oil property estimation model generation process S3320. This process is performed by the estimation model generation unit 1250 using a general computer, with the processing device 1230 executing software.

[0108] First, the preprocessing unit 1252A reads out (S3321) near-infrared absorption spectrum data at a certain point in time from the actual measurement database 1251. The preprocessing of the near-infrared absorption spectrum in the preprocessing unit 1252A involves smoothing (S3322) by the Savitzky-Golay method (SG method) using data from nine adjacent points, and second-order differentiation of the spectrum (S3323).

[0109] Figure 14 shows a graph of the results of differentiating the spectrum twice. The curves show the derivatives of the spectrum at different times after use. Differentiation can emphasize the differences in the near-infrared absorption spectrum over time. Differentiation may not be necessary in some cases. Differentiation may also be performed only once. Alternatively, differentiation may be performed three or more times.

[0110] The twice-differentiated near-infrared absorption spectrum is stored in the learning database 1253 as an explanatory variable x, and the antioxidant concentration at the same time as the objective variable y.

[0111] The input of the near-infrared absorption spectrum (S3321), smoothing (S3322), and second-order differentiation of the spectrum (S3323) in Figure 12 are basically the same processes as the input of the near-infrared absorption spectrum (S1321), smoothing (S1322), and second-order differentiation of the spectrum (S1323) in Figure 2.

[0112] When estimating oil properties using machine learning, the multivariate analysis unit 1254 has an optional PLS regression analysis function in addition to multivariate analysis, and performs PLS regression analysis using the explanatory variable x and the response variable y (S3324) to generate an oil property estimation model 1261. In PLS regression analysis of spectral analysis, a large amount of data, namely light intensity at multiple wavelengths, is integrated into principal components, and a regression line is derived by two-dimensionally plotting the principal components and the response variable. After the PLS regression analysis, cross-validation is performed (S1325). The generated property estimation model is implemented in the monitoring unit 1262.

[0113] As mentioned above, when constructing an oil type estimation model using machine learning, similar machine learning can be performed using the twice-differentiated near-infrared absorption spectrum as the explanatory variable x and the score of the discriminant map dimensionally reduced by principal component analysis as the objective variable y.

[0114] 15 is a flow diagram of the oil property estimation process S1340 using the generated property estimation model. This process is performed by the monitoring unit 1262, which is a general computer, through software processing.

[0115] First, the near-infrared absorption spectrum of the engine oil whose properties are to be determined is acquired and input from the input device 1210 of the diagnostic system 1200 (S1341). The near-infrared absorption spectrum is acquired by the optical sensor shown in Fig. 1 or 3. One of the features of the embodiment is that the near-infrared absorption spectrum can be collected contactlessly and remotely.

[0116] The pre-processing unit 1252B performs pre-processing such as smoothing and second differentiation on the near-infrared absorption spectrum (S1342) in the same manner as when the property estimation model was generated in Fig. 13. The monitoring unit 1262 inputs the pre-processed near-infrared absorption spectrum into the oil property estimation model to estimate the antioxidant concentration (S1343).

[0117] The monitoring unit 1262 can have a characteristic conversion table (not necessarily in table form, but may be data showing the relationship between antioxidant concentration and viscosity as shown in Figure 11) that uses all or part of the actual measurement database 1251 to convert antioxidant concentration, viscosity, and total acid value into each other.

[0118] The viscosity of the engine oil is estimated based on the antioxidant concentration estimated by the oil property estimation model and the characteristic conversion table (S1344). Using the results of this analysis, it was found that the viscosity of the engine oil reaches 25 cP when the antioxidant concentration is reduced to 40% of the initial concentration.

[0119] Figure 16 is a graph showing the consistency between predicted and measured values ​​of antioxidant concentration using the property estimation model of the monitoring unit 1262. The reference value is the actual measured value of the sample, and the result of analysis with the PLS model is the PLS regression equation. The result of predicting from the actual measured value using this regression equation (estimation model) is the predicted value, and if the reference value and predicted value match well, a good PLS model has been constructed. Using Figure 15 as the calibration curve, the concentration of a sample with an unknown additive concentration can be quantified.

[0120] In the above example, the viscosity of the engine oil is finally determined and output from the output device 1220, but as will be described later, the total acid number, concentration, or usage time may also be output. Furthermore, the estimation model was generated using multivariate analysis using software that can execute PLS without using special hardware, but other well-known machine learning methods using a GPU (Graphical Processor Unit) or the like may also be employed.

[0121] In the above example, the near-infrared absorption spectrum was used as the explanatory variable x and the antioxidant concentration as the response variable y, but values ​​reflecting other oil characteristics (such as the concentration of other additives or the duration of use) can also be used as the response variable. Learning can also be performed using viscosity and total acid number, which indicate oil properties, directly as response variables. [Example]

[0122] This embodiment is an application of the configurations of embodiments 1 to 3 to a system and method for monitoring lubricating oil for a wind power generator. This embodiment is a monitoring system for lubricating oil supplied to a mechanical drive unit of a wind power generator. This system includes a diagnostic system 1200 shown in FIG. 8.

[0123] The memory device in the monitoring system stores near-infrared absorption spectrum data of various oils necessary for identifying and judging lubricating oils, as well as additive concentration data that stores the concentrations of additives in the lubricating oil in chronological order as references, and the diagnostic system 1200 estimates the time at which the additive concentration in the lubricating oil, determined from the near-infrared absorption spectrum of the lubricating oil, will reach a predetermined threshold value.

[0124] (1. Overall system configuration) Figure 17 shows a schematic diagram of a lubricant monitoring system for a wind power generator with a lubricant supply system. Inside the nacelle 3 of the wind power generator 1, there are a main shaft 31, a gearbox 33, a generator 34, and bearings such as yaw and pitch bearings (not shown), all of which are supplied with lubricant from an oil tank 37. The wind power generator also has the general components of a wind power generator, such as a hub 4, a nacelle bulkhead 30, a shrink disk 32, a main frame 35, a radiator 36, and a coupling 38.

[0125] 17, multiple wind turbines 1 are usually installed on the same site, and these are collectively referred to as a farm 200a, etc. Each wind turbine 1 is equipped with various sensors (not shown) in its lubricating oil supply system, and sensor signals reflecting the state of the lubricating oil are collected in a server 210 in the nacelle 3.

[0126] Furthermore, sensor signals obtained from the server 210 of each wind turbine generator 1 are sent to an aggregation server 220 arranged for each farm. Data from the aggregation server 220 is sent to a central server 240 via a network 230. Data from other farms 200b and 200c is also sent to the central server 240. Furthermore, the central server 240 can send instructions to each wind turbine generator 1 via the aggregation server 220 and server 210. The central server 240 basically has the functions of the diagnostic system 1200 shown in Fig. 8, and the system of the embodiment is capable of remotely monitoring oil.

[0127] (2. Sensor placement) 18 is a conceptual diagram of a rotating part equipped with a lubricant sensor. Lubricant is supplied to a rotating part 302 from a lubricant supply device 301 such as a pump. The lubricant supply device 301 is connected to an oil tank 37 and receives the supply of lubricant. The rotating part 302 is, for example, a gearbox 33 or any other general part where mechanical contact occurs, and is not particularly limited.

[0128] The optical sensor 304 is disposed in the lubricating oil flow path etc. in order to detect the state of the lubricating oil. A specific example of the optical sensor 304 is shown in FIG.

[0129] In this embodiment, a transparent measuring unit 303 is provided in a flow path (near the end of the lubricant path) branching off from the lubricant flow path connected to the lubricant drain port of the rotating part 302, and a portion of the lubricant is introduced into this measuring unit 303. An optical sensor 304 is then installed in the measuring unit 303. The measuring unit 303 is not provided in the main lubricant flow path in order to adjust the flow rate of the lubricant in the measuring unit 303 to a flow rate suitable for detecting the state of the lubricant. The lubricant discharged from the rotating part 302 returns to the oil tank 37 via a filter 305. Note that the filter 305 is not essential. The optical sensor 304 measures the near-infrared absorption spectrum of the oil. The state of the lubricant can be evaluated based on the temporal change in the near-infrared absorption spectrum of the lubricant.

[0130] In this embodiment, the optical sensor 304 includes an optical sensor equipped with a near-infrared light source and a light-receiving element. The optical sensor acquires the near-infrared absorption spectrum of the lubricating oil. The optical sensor 304 acquires the near-infrared absorption spectrum by transmitting near-infrared light from the near-infrared light source through the oil and detecting the near-infrared light that has transmitted through the oil with a light-receiving element that has sensitivity according to the wavelength. The acquired near-infrared absorption spectrum reflects the oil's absorptivity or transmittance for light of each wavelength. The acquired near-infrared absorption spectrum and estimation model are used to estimate the type of lubricating oil and determine the amount of additives remaining in the lubricating oil, and to perform a deterioration diagnosis and a remaining life diagnosis.

[0131] The quality of lubricating oil deteriorates with use and it no longer performs its original function. For this reason, maintenance such as replacement must be performed depending on the degree of quality deterioration. In order to know the timing of such maintenance, it is useful for efficient maintenance management to be able to monitor data collected by optical sensors 304 installed on-site from a remote location. Furthermore, filling with a lubricating oil different from the lubricating oil that should be used during an oil change can cause precipitation of the base oil and additives, so it is useful for efficient maintenance management to determine the type of oil when changing.

[0132] The data collected by the optical sensor 304 is collected, for example, in a server 210 in the nacelle 3, and then sent via an aggregation server 220 that aggregates data within the farm 200 to a central server 240 that aggregates data from multiple farms.

[0133] However, for analyses that require special equipment, such as LC (liquid chromatography), FT-IR (Fourier transform infrared spectroscopy), and NMR (nuclear magnetic resonance), samples of the lubricating oil must be collected as appropriate and analyzed using separately provided equipment. The results of these LC, FT-IR, and NMR measurements are also preferably stored as data in the central server 240, aggregated, and taken into consideration when determining the properties of the lubricating oil.

[0134] Furthermore, the data to be aggregated may include not only data related to lubricants but also data indicating the operating status of the wind turbine generator. For example, the wind turbine output value (the higher the value, the faster the rate of lubricant deterioration), the actual operating time (the longer the value, the faster the rate of lubricant deterioration), the machine temperature (the higher the value, the faster the rate of lubricant deterioration), the shaft rotation speed (the faster the value, the faster the rate of lubricant deterioration), etc. These can be collected from sensors of known configurations installed at various locations on the wind turbine generator, or from control signals from the device.

[0135] (3. Lubricant diagnosis flow) Figure 19 is a flow diagram showing the lubricant oil diagnosis process according to this embodiment. The process shown in Figure 19 is performed under the control of either the server 210, the aggregation server 220, or the central server 240 in Figure 17. In the following example, it is assumed that the process is performed by the central server 240. Functions such as calculation and control are realized by software stored in the server's storage device being executed by a processor, and the specified processes are realized in cooperation with other hardware. Note that functions equivalent to those configured by software can also be realized by hardware such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit).

[0136] When the central server 240 performs control, it has multiple wind power generators 1 under its control, so the following processing is performed for each wind power generator. This processing is basically a repetitive process, and the start timing is set by a timer or the like; for example, the processing starts at midnight every day (S601). The central server 240 can also perform the processing at any timing in response to instructions from an operator.

[0137] In step S602, the central server 240 checks when the lubricant needs to be replaced. The initial value of the replacement time can be calculated based on the physical properties, for example, by using the Arrhenius reaction rate, assuming that the lubricant is operating at its design temperature, and the remaining life can be set as an initial value. This replacement time can be updated later in step S610 based on actual measurement data.

[0138] If it is time to change the lubricant, the lubricant is changed in step S603. Since changing the lubricant is usually done by a worker, the central server 240 displays and notifies the worker when and what to change.

[0139] If it is not time to change the lubricant, in step S604, the central server 240 performs a diagnosis using sensor data. As sensor data, in addition to the near-infrared absorption spectrum of the lubricant obtained by the optical sensor, the oil temperature, oil pressure, and particle concentration contained in the lubricant, which can be measured using conventional technology, can be used. The data collected by the optical sensor 304 is sent to the central server 240, and the central server evaluates the characteristics of the lubricant, for example, by comparing the oil temperature, oil pressure, and particle concentration contained in the lubricant obtained from the sensor with predetermined thresholds.

[0140] If the diagnosis result in step S605 is abnormal, the lubricating oil is changed in step S603. If no abnormality is found, step S606 is performed. If the lubricating oil has been changed, the type of oil is subsequently determined using the method described above.

[0141] From the perspective of preventive and planned maintenance of wind turbines, it is desirable to detect when an incorrect lubricant is mistakenly filled during an oil change and replace it with the correct lubricant, and to perform predictive diagnosis of lubricant deterioration based on changes in the concentration of additives contained in the lubricant before it is determined that an abnormality exists.

[0142] In this embodiment, the time-series near-infrared absorption spectrum measured by the optical sensor is saved and the degree of deterioration of the lubricant is estimated based on the saved near-infrared absorption spectrum. The replacement time estimation result obtained by process S607 can be displayed as the lubricant diagnosis result (processes S608 and S609).

[0143] 20 shows an example of the display of the results of process S608. Since it is predicted that the viscosity will reach 200 in 50 months, the new replacement time can be set to a time before that (for example, half a month before). Process S610 completes one cycle of processing, and in process S602 of the next cycle, judgment processing is performed according to the new replacement time.

[0144] As described above, according to the technology described in the embodiments, an optical sensor having a light source and a detector that detects light emitted from the light source is used to obtain the absorption spectrum of oil in the wavelength range of 800 nm to 2500 nm, and data based on the absorption spectrum is input into an estimation model implemented in an information processing device to predict the type of oil, and based on the prediction result, the result is notified and a determination is made as to whether the machine using the oil can be operated.

[0145] According to this embodiment, the optical sensor accurately determines whether the correct lubricant has been filled when filling or changing the oil, and after confirming that the correct lubricant has been filled, the machine's operation and the condition of the lubricant can be monitored. This makes it possible to prevent machine breakdowns caused by using counterfeit lubricant or putting in the wrong type of lubricant.

[0146] Furthermore, according to the above embodiment, remote oil determination and diagnosis become possible, which reduces labor and costs. Furthermore, by applying dimension reduction using principal component analysis to the near-infrared absorption spectrum, it can be adapted to machine learning.

[0147] According to the above embodiment, efficient oil maintenance management can be realized, which reduces energy consumption, reduces carbon emissions, prevents global warming, and contributes to the realization of a sustainable society. [Explanation of symbols]

[0148] Diagnostic system 1200, estimation model generation unit 1250, oil type estimation unit 1260, actual measurement data collection process S1310, type estimation model generation process S1320, oil type estimation process S1330

Claims

1. An information processing system including an input device, an output device, a processing device, and a storage device is used, the information processing system includes an estimation model generation unit that generates an estimation model, The estimation model generation unit Acquire absorption spectrum data obtained by transmitting light of at least a portion of wavelengths between 800 nm and 2500 nm through a known type of oil; analyzing the absorption spectrum data using a dimensionality reduction algorithm to generate an estimated model of the oil type; How to generate an estimation model for oil type.

2. The dimensionality reduction algorithm performs multivariate analysis using principal component analysis. The method for generating an oil type estimation model according to claim 1 .

3. The dimension reduction algorithm analyzes the absorption spectrum data after one or more derivatives. The method for generating an oil type estimation model according to claim 1 .

4. the oil type estimation model expresses the oil type by values ​​of at least two principal components; The method for generating an oil type estimation model according to claim 1 .

5. The oil type estimation model is an estimation model generated by machine learning using the absorption spectrum data as an explanatory variable and the values ​​of the principal components as a response variable. The method for generating an oil type estimation model according to claim 4.

6. The estimation model generation unit acquires the absorption spectrum data for multiple samples of the same type of oil, the type of which is known. The method for generating an oil type estimation model according to claim 4.

7. A method for estimating an oil type using a diagnostic system comprising an information processing device that implements the estimation model according to claim 4, comprising: inputting, into the diagnostic system, diagnostic object absorption spectrum data obtained by transmitting light of at least a part of wavelengths between 800 nm and 2500 nm through the diagnostic object oil; inputting a value based on the diagnostic object absorption spectrum data into the estimation model; Estimating the type of the diagnostic target oil from the estimation model; How to estimate oil type.

8. Analyzing the diagnostic target absorption spectrum data using a dimension reduction algorithm, the dimension reduction algorithm performing multivariate analysis using principal component analysis, and expressing the type of the diagnostic target oil by values ​​of at least two principal components. The method for estimating the type of oil according to claim 7.

9. When a distance between the value of the principal component representing the oil to be diagnosed and the value of the principal component of an oil of known type represented by the oil type estimation model is within a predetermined range, the oil to be diagnosed is estimated to be the same type as the oil of known type. The method for estimating the type of oil according to claim 8.

10. The distance is at least one selected from Euclidean distance, Chebyshev distance, and Mahalanobis distance. The method for estimating the type of oil according to claim 9.

11. a value obtained by differentiating the diagnostic object absorption spectrum data one or more times is input into the estimation model; The method for estimating the type of oil according to claim 7.

12. If the estimated type of the oil to be diagnosed is not the same as a predetermined oil type, the system notifies the user of the estimated result of the type of the oil to be diagnosed and instructs the user to at least one of confirming the oil to be diagnosed and replacing the oil to be diagnosed. The method for estimating the type of oil according to claim 7.

13. If the estimated type of oil to be diagnosed is the same as a predetermined oil type, the system notifies and instructs the user to permit and start operation of the machine filled with the oil to be diagnosed. The method for estimating the type of oil according to claim 7.

14. After notifying and instructing the user to permit and start operation of the machine filled with the oil to be diagnosed, the oil properties of the oil to be diagnosed are monitored. The method for estimating oil type according to claim 13.

15. An oil type estimation system comprising an information processing device in which the estimation model according to claim 1 is implemented.

Images