WO2025165342A1 - Predictive maintenance of equipment based on oil parameters - Google Patents

Abstract

System and methods are disclosed herein for predictive equipment maintenance. In some examples, oil features of oil (e.g., lubricating oil) being used by equipment can be received. Sensitive oil features from the oil features that are sensitive to oil degradation can be identified. A health indicator can be generated based on the sensitive oil features. The health indicator can be indicative of a health of the equipment. A failure prediction model can be generated based on the health indicator. In some examples, equipment maintenance can be implemented based on the failure prediction model.

Description

PREDICTIVE MAINTENANCE OF EQUIPMENT BASED ON OIL PARAMETERS FIELD OF THE DISCLOSURE _{[0001] This disclosure relates generally to equipment maintenance, and more specifically, to} predictive maintenance of equipment based on oil parameters. BACKGROUND OF THE DISCLOSURE _{[0002] Degradation or breakdown of oil (e.g., lubricating oil) used in equipment (e.g., machines)} can lead to equipment failure. This is because degraded oil impacts the performance and lifespan of the equipment. Lubricating oil is used in equipment, such as machinery, for several functions, for example, reducing friction, dissipating heat, sealing gaps, and/or carrying away contaminants. Lubricating oil is used in a gearbox of equipment for smooth and efficient operation. Gearboxes can include various gears, bearings, and other components that require use of the lubricating oil. Over time, as the gearbox operates, the lubricating oil within the gearbox undergoes degradation due to factors such as heat, pressure, and/or mechanical stress, as well as contamination. When the lubricating oil degrades or becomes contaminated, it can no longer effectively perform its intended function, which can result in equipment problems and/or failures. _{[0003] To reduce equipment problems and/or failures, predictive maintenance techniques can be} used to forecast equipment failure to predict when the equipment is likely to fail for proactive maintenance. Predictive maintenance starts with the equipment that needs to be maintained. This could be any type of industrial or mechanical equipment, ranging from manufacturing machines and pumps to turbines and engines. Oil condition monitoring (OCM) is a component of predictive maintenance for equipment that uses lubricating oil for operation. OCM refers to processes and/or a set of techniques that are used for assessing a condition of lubricating oil in the equipment. _{[0004] As part of OCM, oil analysis laboratories (OAL) are used for analyzing oil samples (of} the lubricating oil) to assess a condition of the lubricating oil in the equipment. OAL are used in order to address potential issues early and thus prevent equipment failures and downtime and extend the life of the machinery. OAL are used to check various factors that can affect oil’s performance and thus its condition. By examining these factors, a health of the equipment can be inferred, such that proper maintenance and corrective actions can be taken. Oil samples are typically sent to OALs on a regular schedule, as part of a planned maintenance program. A frequency of sampling and analysis depends on factors, such as a type of machinery, operating conditions, and/or manufacturer’s recommendations. There are drawbacks and/or challenges to using OAL for oil condition assessment, such as timing delay, cost, limited scope (e.g., only analyzing the oil condition of the lubricating oil and detecting wear particles, but failing to consider other critical factors, such as vibration analysis, temperature monitoring, electrical diagnostics, etc.), human interpretation, and a frequency and sampling of the lubricating oil (e.g., over-sampling or under-sampling can lead to inaccurate assessment of machinery condition, deciding on when and where to take oil samples requires accurate consideration, etc.). _{[0005] Sensors have been integrated into OCM and are being used for collecting real-time data} for identifying abnormal conditions and predicting equipment failure with greater accuracy than OAL, as well as being compatible with Industry 4.0, thereby solving the real-time predictive maintenance problem. SUMMARY OF THE DISCLOSURE _{[0006] Various details of the present disclosure are hereinafter summarized to provide a basic} understanding. This summary is not an extensive overview of the disclosure and is neither intended to identify certain elements of the disclosure nor to delineate the scope thereof. Rather, the primary purpose of this summary is to present some concepts of the disclosure in a simplified form prior to the more detailed description that is presented hereinafter. _{[0007] According to an embodiment, a computer-implemented method can include receiving oil} features of oil being used by equipment, identifying sensitive oil features from the oil features that are sensitive to oil degradation, and generating a health indicator based on the sensitive oil features. The health indicator can be indicative of a health of the equipment. The computer-implemented method can further include generating a failure prediction model based on the health indicator, and causing equipment maintenance to be implemented based on the failure prediction model. _{[0008] In another embodiment, a system can include one or more computing platforms} configured to identify sensitive oil features from oil features of lubricant oil being used in a gearbox of a transmission system that are sensitive to oil degradation and captured by one or more oil sensors, and generate a degradation trend based on the sensitive oil features. The degradation trend can be indicative of a degradation of the equipment. The one or more computing platforms can be further configured to generate a failure prediction model based on the degradation trend. The failure prediction model can be used to predict a degradation of the equipment. The one or more computing platforms can be further configured to determine whether gearbox maintenance is to be implemented based on the failure prediction model. _{[0009] In a further embodiment, a system can include an oil condition analyzer to receive oil} features of oil being used by equipment, identify sensitive oil features from the oil features that are sensitive to oil degradation, and generate a health indicator based on the sensitive oil features. The health indicator can be indicative of a health of the equipment. The system can further include a model generator to generate a failure prediction model based on the health indicator, and a maintenance alert component to output a maintenance alert for the equipment based on the failure prediction model. _{[0010] Any combinations of the various embodiments and implementations disclosed herein can} be used in a further embodiment, consistent with the disclosure. These and other aspects and features can be appreciated from the following description of certain embodiments presented herein in accordance with the disclosure and the accompanying drawings and claims. BRIEF DESCRIPTION OF THE DRAWINGS _{[0011] FIG. 1 is an example of a maintenance tool that can be used for predictive maintenance} of equipment based on a condition of oil. _{[0012] FIG. 2 is an example of a lubrication and gearbox system for which the maintenance tool} can be used for predictive maintenance of a gearbox. _{[0013] FIG. 3 is an example of a multi-feature graph for lubricant oil used in a lubrication and} gearbox system. _{[0014] FIG. 4 is an example of a table with definitions for each abbreviated nomenclature} property of a respective feature graph of the multi-feature graph. _{[0015] FIG. 5 is an example of a graph of monotonicity values for oil features. [0016] FIG. 6 is an example of a graph showing a contribution rate of each principal component} during a principal component analysis (PCA). _{[0017] FIG. 7 is an example of a graph showing a health of a component through a wear} evolution process with different stages: an initial “run-in” stage, a stable “steady” stage, and an “accelerated” stage. _{[0018] FIG. 8 is an example of a graph showing an accuracy of a failure prediction model. [0019] FIGS.9-10 are example graphs showing a comparison before and after prediction updates} of a failure prediction model. _{[0020] FIG. 11 is an example of a graph showing a prediction trend of oil performance before} actual gear failure happens. _{[0021] FIG. 12 is an example of a method for predicting maintenance of equipment. [0022] FIG. 13 is another example of a method for predicting maintenance of equipment. [0023] FIG. 14 depicts an example computing environment that can be used to perform methods} according to an aspect of the present disclosure. _{[0024] FIG. 15 depicts a cloud computing environment that can be used to perform one or more} actions according to an aspect of the present disclosure. _{[0025] FIG. 16 is an example of a graphical user interface (GUI) depicting a performance of} predictive maintenance using a failure prediction model and vibration data. DETAILED DESCRIPTION _{[0026] Embodiments of the present disclosure will now be described in detail with reference to} the accompanying Figures. Like elements in the various figures may be denoted by like reference numerals for consistency. Further, in the following detailed description of embodiments of the present disclosure, numerous specific details are set forth in order to provide a more thorough understanding of the claimed subject matter. However, it will be apparent to one of ordinary skill in the art that the embodiments disclosed herein may be practiced without these specific details. In other instances, well-known features have not been described in detail to avoid unnecessarily complicating the description. Additionally, it will be apparent to one of ordinary skill in the art that the scale of the elements presented in the accompanying Figures may vary without departing from the scope of the present disclosure. _{[0027] Embodiments of the present disclosure relate predictive maintenance of equipment. Gear} transmission systems are widely used in various industries and machinery including automobiles, industrial equipment, robotics, and wind turbines. Gear transmission systems include components and mechanisms that are involved in the transmitting of mechanical power and motion using gears. Gearboxes in gear transmission systems are components that are used to convert rotational energy (e.g., generated by blades of a wind turbine according to wind) into electricity. Any malfunction or failure in a gearbox can result in reduced energy production, or complete failure and/or shutdown of equipment. For example, wind turbines operate in demanding conditions, including exposure to wind, temperature variations, and heavy loads. These conditions impact the wear of gearbox components of the gearbox, for example, gears and bearings, which leads to deterioration of these components. The wear can result in stress concentration, and/or induce other failure modes, such as tooth crack and breaking. Stress concentration can weaken affected components and increase a risk of damage and/or failure. As wear progresses, tooth cracks and breaking can have cascading effects that can lead to equipment failure. In addition to economic consequences, there are safety concerns associated with equipment failure, for example, wind turbine failures. Catastrophic failures can lead to safety hazards for both maintenance personnel and a surrounding environment. Preparing and procuring replacement components for gearboxes (e.g., wind turbine gearboxes) can be time- consuming and costly as well. The lead time for replacements means that proactive maintenance and prediction are essential to avoid lengthy downtime. Thus, predictive maintenance techniques, including wear degradation monitoring and prediction, are essential for maintaining the reliability of equipment, such as wind turbines. These techniques involve continuous monitoring of gearbox conditions, analyzing wear trends, and forecasting when maintenance or component replacement is needed. _{[0028] Wear degradation monitoring techniques have been developed for assessing wear} degradation (e.g., of components of the gearbox, or the gearbox itself). Vibration analysis techniques are being used for wear degradation monitoring of equipment (e.g., the gearbox). Vibration analysis, which can capture an impact caused by a local defect, is utilized for characterizing an equipment status (or health). However, it is difficult to sense a weak impulse feature at the early stage of equipment wear (e.g., gearbox wear) using vibration signals. In the context of vibration analysis, a weak impulse feature refers to a subtle or low-amplitude vibration event or signal that occurs in early stages of machinery (gearbox) wear or degradation. This weak impulse feature may be indicative of a developing problem or fault within the equipment or component, such as a gear, bearing, or other rotating part. Additional wear degradation monitoring techniques rely on acoustic emission (AE) to detect slight wear in gears, but such acoustic techniques are sensitive to noise and thus errors. Some wear degradation monitoring techniques use temperature information as an indicator to report a failure or fault of the equipment (e.g., the gearbox). However, a hysteresis of temperature changes becomes one of the most challenging problems in the application of thermal analysis and thus can lead to errors. _{[0029] Generally, rotating machinery, such as wind turbines, have a lubricant system (or oil} circulation system) for mitigating friction to minimize failure and/or downtime of wind turbines. As a working fluid in contact with oil-wetted components, lubricant oil contains health information (oil information) for the oil-wet components that can be used for predictive maintenance of equipment. Without transmission path attenuation and hysteresis effect, the oil information can be used to determine a wear status of the equipment (e.g., the gearbox) in real-time, which is favorable in wear degradation prognosis. Wear degradation prognosis refers to a process of predicting or forecasting a future progression of deterioration (e.g., wear of component or material) of equipment. Wear degradation prognosis is an aspect of predictive maintenance and reliability engineering and is used to estimate how a component or system will degrade over time and when maintenance or replacement will be necessary to prevent failures. _{[0030] Reliable (or accurate) wear degradation prognosis requires an accurate indicator for} characterizing a degradation process (e.g., of the lubricant oil), and a proper approach for predicting wear deterioration (e.g., wear trend prediction), among which wear degradation characterizing is a prerequisite of prognostics. Oil-related data has been used to track and understand how deterioration (e.g., wear) progresses in equipment over time. Information relating to a condition of the lubricating oil can be used to gain insight into a wear evolution process of the equipment (e.g., the gearbox). _{[0031] Some existing techniques for wear deterioration prediction have relied on detecting wear} states using debris morphology and size. Oil information from different dimensions allows for greater possibilities for characterizing wear degradation processes. However, there are technical issues and challenges that make accurate wear deterioration prediction difficult, such as information discrimination and evaluation. It is understood that a signal without deterioration information is not only difficult to characterize the wear evolution process, but also complicates wear degradation prognosis. _{[0032] In an Industry 4.0 environment, there is a push for the integration of advanced sensors} that can provide real-time data from multiple sources. This integration allows for a more comprehensive and multivariate approach to characterizing degradation (e.g., wear) of equipment. Industry 4.0, also known as a Fourth Industrial Revolution, is a transformative paradigm shift in manufacturing and industry that leverages advanced digital technologies to create “smart factories” and optimize various aspects of production and supply chains. However, it also presents technical challenges, such as dealing with a high level of redundancy among various types of oil data generated by these sensors, especially in the case of lubricant-related information. Managing and extracting meaningful insights from this redundant data can be a complex task in the context of wear analysis. _{[0033] Various techniques have been developed for predicting degradation of equipment, with} model-based strategies being a common approach. Model-based strategies rely on an understanding of specific wear mechanisms, but in practice, it is challenging, and sometimes impossible, to create highly accurate degradation models due to a complex and interconnected nature of factors influencing wear. As an alternative, data-driven strategies have been used as well. These approaches, which include artificial neural networks (ANN), support vector regression (SVR), recurrent neural networks (RNN), and variants, leverage historical degradation data without the need for complex physical models. Such models are useful when the relationship between wear and influencing factors is not well-defined. However, these data-driven strategies are limited and have technical drawbacks in their assumption that all data follows a similar (or same) statistical distribution. In reality, degradation patterns (e.g., wear patterns) can vary widely depending on operating conditions, usage, and other factors. This assumption can sometimes lead to technical challenges in accurately capturing degradation trends across different implementations and uses. Thus, there are significant challenges in degradation trend prediction. _{[0034] Equipment wear undergoes three distinct stages: an initial “run-in” stage, a stable} “steady” stage, and an “accelerated” stage where wear intensifies, and these stages collectively define or form the wear evolution process for the equipment. The “run-in” stage is the initial phase in a wear evolution process. During this stage, newly installed or freshly lubricated equipment components undergo a period of adjustment and initial wear as these components settle into respective operational states. This stage is characterized by gradual, relatively low levels of wear as the surfaces of moving parts adapt to each other. The “steady” stage represents a period of stability in wear evolution. In this stage, the rate of wear reaches a relatively constant level, and the equipment operates with consistent and predictable wear patterns. Data collected during the steady stage typically show smooth and consistent wear trends. The “accelerated” stage is a final phase in the wear evolution process. During this stage, wear and degradation of equipment components intensify significantly. Factors such as increased load, reduced lubrication effectiveness, or the accumulation of damage from previous stages can contribute to accelerated wear. Data collected in this stage often exhibit non-stationary behavior, indicating rapid and irregular changes in wear rates. Thus, data collected from these different stages exhibit diverse statistical distributions. For example, data from the steady wear stage tends to be relatively consistent, while data from the accelerated wear stage can exhibit significant non-stationary behavior. The problem arises because, during a model training process, there is often no prior information available about which stage of wear evolution the machinery is in. This diversity in data distribution among the different wear stages poses a challenge for the model's ability to generalize and accurately predict wear trends. _{[0035] Accordingly, existing prediction maintenance models used for degradation trend} prediction are unable to handle variations in data, such as wear data distribution across the different stages of the wear evolution process. Thus, existing models are not accurate enough for proactive maintenance. Failures in such models to provide adequate warning (in time) can lead to equipment failure, in some instances, catastrophic failure, such as loss of the equipment, or human life. In some examples, a maintenance tool is disclosed herein that can be used to predict when equipment is likely to fail (or require maintenance) by considering wear data distribution across different stages of a wear evolution process through use of health indicator information extracted from oil features of the oil under monitoring. The oil features can be captured or measured over the wear evolution process and used by the maintenance tool to provide a failure prediction model that can be used for proactive maintenance. The maintenance tool 100 can identify the most sensitive oil feature of the captured oil features and use these sensitive features to provide a health indicator that can be used to train a model (e.g., a time-series model) to provide a failure prediction model. The failure prediction model can be used for proactive equipment maintenance. Because the failure prediction model is generated (e.g., trained) on oil features that are most sensitive to oil degradation over the evolution process results in a forecasting model that is more accurate in predicting when equipment will fail and/or require maintenance. _{[0036] FIG. 1 is an example of a maintenance tool 100 that can be used for predictive} maintenance of equipment. The term “equipment” can refer to a physical object, device, system, asset, or machine designed for a specific purpose. Examples are presented herein in which the tool 100 is used for predictive maintenance of a gearbox of a transmission system (e.g., of a wind turbine) based on a condition of an oil (e.g., lubricating oil), but in other examples, the tool 100 can be used for predicting maintenance of other types of equipment. The tool 100 can be used to predict a wear of the equipment and thus provide trend degradation predictions. For example, the tool 100 can be used to monitor in real-time a degradation or wear of the gearbox (and thus the equipment in which it is being used), for example, one or more components of the gearbox, such as one or more gears, or other components, through analysis of oil (e.g., lubricant oil) being used by the gearbox. The tool can be used to evaluate a condition of the oil to determine a degradation trend for the equipment (e.g., determine how the equipment, such as the gearbox, is deteriorating over time). Using the degradation trend for the equipment, the tool 100 can construct a model to predict how the equipment will degrade and thus the model can be used for degradation trend prediction. The model can be used to alert or initiate maintenance of the equipment, according to one or more examples as disclosed herein. _{[0037] The tool 100 includes an oil condition analyzer 102 and model generator 104. The oil} condition analyzer 102 and the model generator 104 can be implemented using one or more modules, shown in block form in the drawings in the example of FIG.1. The one or more modules can be in software or hardware form, or a combination thereof. In some examples, the oil condition analyzer 102 and the model generator 104 can be implemented as machine-readable instructions for execution on a computing platform 106, as shown in FIG. 1. The computing platform 106 can include any computing device, for example, a desktop computer, a server, a controller, a blade, a mobile phone, a tablet, a laptop, a personal digital assistant (PDA), or other types of portable (or stationary) devices. The computing platform 106 can include a processor 108 and a memory 110. By way of example, the memory 110 can be implemented, for example, as a non-transitory computer storage medium, such as volatile memory (e.g., random access memory), non-volatile memory (e.g., a hard disk drive, a solid-state drive, a flash memory, or the like), or a combination thereof. The processor 108 can be implemented, for example, as one or more processor cores. The memory 110 can store machine- readable instructions (e.g., the oil condition analyzer 102 and the model generator 104) that can be retrieved and executed by the processor 108. Each of the processor 108 and the memory 110 can be implemented on a similar or a different computing platform. In some examples, the oil condition analyzer 102 can be implemented on a different computer platform than the model generator 104. By way of further example, the oil condition analyzer 102 and the model generator 104 can be implemented in a cloud computing environment and thus allow for lubricant oil condition monitoring in an online setting (e.g., through the use of the cloud) _{[0038] The oil condition analyzer 102 can include a feature selector 112. The feature selector} 112 can implement feature selection based on oil features 114 for the oil. An oil feature refers to an oil parameter of the oil. An oil feature can include values (a data set of values) that are representative of a particular characteristic or property of the oil. Thus, the values or the data set of values can be referred to as oil parameter data. The oil parameter data can include values captured (or detected), in some instances, computed that can characterize a property or characteristic of the oil over time and thus can represent a time-varying signal. Thus, in some instances, the oil parameter data for the oil features 114 can be extracted based on the oil being used by the transmission system and provided as part of the oil features 114, as shown in FIG.1. _{[0039] A number of oil parameters can be monitored of the oil, including physical and chemical} properties of the oil. One or more sensors, devices and/or systems can be used to monitor the oil during use in the transmission system (e.g., the gearbox) to provide the oil parameter data and thus the oil features 114. By way of example, real-time parameters of the oil can be obtained by using _{one or more oil sensors, in some examples to provide one or more datasets ^ = [^^, ^^, … ,^^],} where N is a total number of sensor channels (e.g., oil sensors used for capturing a respective oil parameter of the oil) and ^_^stands for an i-th channel. Thus, the one or more oil sensors can be used for data acquisition. The one or more oil sensors can be mounted in an oil loop near an outlet of the transmission system. For example, real-time measurements of the oil parameters (or features) can include, but not limited to, a temperature, a moisture, , a kinematic viscosity, a viscosity index (VI), wherein VI is calculated from kinematic viscosity at about 40℃ according to ASTM D2270 and ISO 2909), a density, a permittivity, a count value of ferrous debris with different size, a count value of nonferrous debris with different size, and/or a particle count cleanness. In some instances, particle count may be sufficient as the oil parameter. _{[0040] Machine parts, including those in the transmission system, experience wear during} operation. This is a natural consequence of mechanical components rubbing against each other during normal use. Thus, a health condition of the transmission system (and consequently the equipment in which it is being used) is fluctuating or changing during operation as the equipment goes through a wear evolution process, as described herein. Thus, a performance of the transmission system and a condition of its components can vary due to factors like changing loads, temperatures, and operating conditions. Despite fluctuations in a system's health, there is an overall degradation trend. This means that, over an extended period, the system's condition (e.g., the gearbox) tends to deteriorate gradually. For example, this degradation can be due to cumulative wear and other factors that affect the performance and reliability of the transmission system. _{[0041] In some examples, the feature selector 112 can implement data smoothing to smooth the} oil features 114. In some examples, the oil condition analyzer 102 includes a data smoother for implementing data smoothing of the oil features. Data smoothing can be used by the oil condition analyzer 102 to remove noise from each channel (oil parameter data) using a moving averaging according to expression (1): ^ _{^ ^ ^^ ^^^ ^ ^(^) = ^} ^{( )} _^ ^{( )} _⋯^^ ⁽ _^^^^^ ⁾ ^ (1) wherein a feature of i-th ^ is an order of the moving averaging. _{[0042] Because not every feature (the oil} f_{or the oil provides information about} the degradation of the gearbox, the feature selector 112 can be used to select or identify features from the oil features 114 that are sensitive to the degradation of the oil to provide sensitive oil features 116. Thus, feature selection can be used to conduct a comprehensive evaluation about the information being provided by the oil features 114 relating to oil degradation. For example, the feature selector 112 can be used to ascertain a behavior of the oil features 114. The feature selector 112 can implement a feature selection method to evaluate a sensitivity to degradation for each channel (feature). _{[0043] For example, the feature selector 112 can determine a sensitivity of the oil features 114} to oil degradation. The feature selector 112 can determine for each oil feature of the oil features 114 a sensitivity (score) value that can be indicative of how sensitive that oil feature is to the oil degradation. The sensitivity value can quantify a degree to which a specific oil feature reacts or responds to changes in a condition or performance of a component (or equipment). Thus, the sensitivity value can serve as a measure of reactivity of the oil feature to the equipment (or component) degradation. A high sensitivity value can indicate that the oil feature is highly responsive to changes in the condition of the oil. In some examples, monotonicity can be used to describe or evaluate how a function changes values with respect to its input or how a sequence of values behaves. Thus, in some instances, the feature selector 112 can use monotonicity to evaluate the behavior of the oil features 114 for determining oil feature sensitivity (the sensitivity score value). For example, the feature selector 112 can use expression (2) to determine a monotonicity value as the sensitivity value indicative of how sensitive an oil feature is to oil degradation (e.g., gearbox degradation): ^_{^^^^^^^^^^^ =} ^ ^ ^_^^^ ^^^(^_^ ⁽^^^⁾^^_^ ⁽^⁾) ^_{^^ ^^^} ^ wherein, ^ is a total number of feature sequences and ^^^ ( ) function depicts a sign function, ^_^(^)denotes a k-th data point in a j-th feature sequence, and ^_^is a total length of the j-th feature. _{[0044] For example, an oil feature that is sensitive to oil degradation can have a monotonicity} value closer to or equal to “1.0,” whereas an oil feature that is less sensitive to oil degradation can have a monotonicity value that is closer to “0” or equal to “0.” For example, the feature selector 112 can apply a sensitivity threshold to identify a subset of features of the oil features 114 based on a corresponding computed sensitivity (e.g., monotonicity) value to provide the sensitive oil features 116. Thus, the sensitive oil features 116 can be features from the oil features 114 with a monotonicity value that can be equal to or greater than the sensitivity threshold. In a non-limiting example, the sensitivity threshold can be set to 0.3 to identify the sensitive oil features 116 having a sensitivity value greater than or equal to 0.3. _{[0045] In some examples, redundancy can exist after feature selection among the sensitive oil} features 116 (e.g., datasets of each of the sensitive oil features 116). Redundancy can refer to a presence of similar or correlated information among selected features, which can lead to data inefficiency and potentially obscure particular (e.g., key) information. To address this redundancy and extract most essential information from the sensitive oil features 116, a feature reducer 118 can be used of the oil condition analyzer 102. The oil condition analyzer 102 can include a health indicator (HI) engine 120 that includes the feature reducer 118. In other examples, the feature reducer 118 can be implemented separately from the HI engine 120. For example, the feature reducer 118 can implement Principal Component Analysis (PCA) for reducing a dimensionality of the sensitive oil features 116. PCA is a statistical technique used for both dimensionality reduction and information fusion. PCA works by transforming an original dataset (e.g., feature dataset) into a set of uncorrelated variables called principal components. These principal components can be linear combinations of the original dataset (the feature dataset) and can be used to capture most significant information in the dataset while reducing redundancy. Thus, PCA can be used to reduce data redundancy while information fusion by retaining key information from a dataset (the sensitive oil features 116). _{[0046] For example, the feature reducer 118 can normalize the sensitive oil features 116, wherein the normalized feature data can be defined as follows: ^ = [^^, ^^, … ,^^], wherein ^ is a normalized} feature of one of the sensitive oil features 116. It aims to ensures that all features are treated equally in the computations and prevents features with larger value ranges from dominating the results. The feature reducer 118 can calculate a covariance matrix using expression (3): ^ ₌ ^ ^{^} ^^^ ^ ^ (3) wherein ^ denotes the number of selected features (e.g., provided by the feature selector 112), and ^ is the normalized features. The covariance matrix describes the relationships and variances among different features. _{[0047] The feature reducer 118 can obtain eigenvalues [^^, ^^, … , ^^]with [^^ > ^^ > ⋯ ^^] and corresponding eigenvector [^^,^^, … ,^^] by eigenvalue decomposition. This process breaks down the covariance matrix into [^^, ^^, … , ^^] and their corresponding eigenvectors [^^,^^, … ,^^]. The eigenvectors the directions in the original data, while the} eigenvalues indicate the variances along directions.: ^_{^^ = ^^^^ (4). [0048] The feature reducer 118 can calculate a cumulative contribution rate (CCR) according to} expression (5): ∑ ^ ^_{^^ =} ^^^ _{^^ ∑} ^ (5), ^^^ _^^ wherein m is defined as a number of sensitive oil features 116) when the CCR _{exceeds 95%, as shown in FIG. 6. In top m eigenvectors [^^,^^, … ,^^]} are selected as the principal components. _{[0049] It should be noted that the way to determine the number of retained features using} expression (5) may not be absolute. Generally, m can also be empirically set to 1 to make the condition indicator more intuitive. _{[0050] Thus, the feature selector 118 can implement dimension feature reduction and feature} fusion to provide a reduced dataset for the sensitive oil features 116, which can be used by the HI engine 120. The reduced dataset can be referred to as reduced sensitive feature dataset 122, as shown in FIG. 1. For example, expressions (3)-(5) can be used to implement PCA. In this process, both information fusion and dimensionality reduction can be completed. For example, a 4-dimensional feature can be reduced to 1 dimension by performing PCA. In this process, redundancy between data can be reduced while key information of the data can be retained in a 1-dimensional matrix, which means that the 4-dimensional features (e.g., 4-dimensional matrix) can be merged into 1- dimensional features. Thus, the implementation of PCA can be summarized as follows (or in five (5) steps: data normalization, covariance matrix calculation, eigenvalue decomposition, selection of Principal components, and in some instances data projection, as disclosed herein. _{[0051] The HI engine 120 can provide an indication of health of the equipment that represents a} degradation of the equipment over time and thus over the wear evolution process. In some instances, the HI engine 120 can provide a degradation trend over time for the equipment indicating a health of the equipment as it degrades over time. As such, the health indicator engine 120 can provide HI data 124 for the equipment, which can be a health indicator for the equipment. The HI engine 120 can characterize a condition (e.g., health) of the equipment over time with respect to the wear evolution process, for example, across the different stages of the process, as disclosed herein. For example, the HI engine 120 can determine a condition indicator sequence ℎ to provide the HI data 124 which can be obtained according to expression (6): ℎ _{= [^^,^^, … ,^^]^ ∗ ^ (6), wherein [^^,^^, … ,^^] is the subset of … … ,^^] is calculated by expression} (4), ^ is determined as features, as disclosed herein. _{[0052] For example, [^^,^} ^_{, … the eigenvector of covariance matrix, and[^^, ^^, … , ^^]} is the corresponding eigenvalue. Briefly, PCA evaluates the contribution rate of each lambda to _{determine how many} w_{ithin [^^,^^, … ,^^] will be retained, and ^ is the number of} retaining eigenvectors. The evaluation according to one or more examples as disclosed herein, or through an some instances, ^ is set as 1 to obtain a 1- dimensional condition indicator (or health indicator) for illustration visually. In some examples, ^ calculated by expression (5) can be 1 in some applications due to a strong relationship between various oil features. _{[0053] The HI engine 120 can form the condition indicator (CI) sequence, also named as health} indicator (HI) sequence (124). In other words, each value of the condition indicator sequence ℎ can be referred to as a HI value and thus a set of HI values can be provided by the HI engine 120 as the HI data 124, each value being associated with a point in time (e.g., minute). In some examples, the condition indicator sequence ℎ can be plotted over time. Thus, in some examples, the HI data 124 can be plotted to provide a time-series signal that is representative of a health of the equipment over the wear evolution process. The time-series signal can be referred to as a degradation signal (or trend), in some instances. In some examples, the health indicator engine 120 can cause the degradation trend to be rendered on an output device, for example, as disclosed herein. _{[0054] In some examples, the HI data 124 can be provided to the model generator 104 for} generating a failure prediction model 126 based on the HI data 124. The failure prediction model 126 can be used to predict when equipment (e.g., the gearbox) is likely to fail and use this information as predictive maintenance. In some examples, the HI data 124 can be referred to as a training dataset as it is used for training a model. The model generator 104 can be implemented as a model generation algorithm that can be configured to train the model to provide the failure prediction model 126. _{[0055] For example, the generator 104 can provide the failure prediction model 126 based on} modeling parameters 128. In the examples herein, the model generator 104 is used to train an Autoregressive Integrated Moving Average Model (ARIMA) to provide the failure prediction model 126. Thus, the modeling parameters 128 can include hyper-parameters for the ARIMA, for example, an auto-regressive (AR) parameter, an integration (I) parameter, and moving average (MA) parameter. The I parameter can be a difference of a time series, the AR parameter can be a weighted sum of lagged values of the series, and the MA parameter can be a weighted sum of lagged forecasted errors of the series. _{[0056] Using the modeling parameters 128, the model generator 104 can implement model} fitting or a parameter estimation process to find a best-fitting model parameter that minimizes a difference between model predictions and observed data. In response to model fitting, the model generator 104 can provide the failure prediction model 126. While examples are disclosed herein in which ARIMA is used a model type, in other examples, a different model type (e.g., another type of time-series data model) can be used as well. The model generator 104 can be used to construct (or build) a predictive maintenance model (e.g., the failure prediction model 126) for industrial transmission systems, for example. Thus, the model generator 104 can be used to provide a degradation trend prediction model as the failure prediction model 126 that can be used to provide alerts regarding equipment (or component) failure in an efficient and/or effective approach so that early warning can be provided to personnel or users. _{[0057] For example, the model generator 104 can implement model fitting to establish the} ARIMA according to expression (7): ^ ^ ^_{(^)^^ = ^ + ^^^^^^^ +} ^^_^ ^_^^^ wherein ^^{(^)}^_^denotes a indicator, e.g., integrated (I) part, and d is the differencing order, which denotes the number of differencing operations required to make the time series stationary. _{[0058] For example, differencing is performed to eliminate trends and seasonality in the time series. ∑} ^ ^_{^^ ^^^^^^} in expression (7) depicts a weighted sum of lagged values of the sequences (Auto- (AR) part), and p is the AR order, which indicates the relationship between the current observation and the past p observations. The AR order can determine the number and _{weights of autoregressive terms in the model. ∑} ^ ^_{^^ ^^} ^_^^^ in expression (7) can represent a weighted sum of lagged forecasted errors of the (Moving average (MA) part), and q is the MA order, which indicates the relationship between the current observation and the past q forecast errors (residuals). The MA order determines the number and weights of moving average terms in the model. ^ in expression (7) can represent a constant. The model parameters p, d, q can be set based on the modeling parameters 128. _{[0059] The model fitting implemented by the model generator 104 can include stationarity} checking by an augmented Dickey-Fuller test and differencing until data is stationary, calculating an autocorrelation function and a partial autocorrelation function, and then evaluating (or analyzing) lags, and determining the model parameters p, d, q, according to the results of autocorrelation function and partial autocorrelation function. _{[0060] The failure prediction model 126 can be used for predicting (forecasting) a degradation} trend of equipment, which can be used to determine when the equipment is likely to fail (and/or requires maintenance). For example, the oil condition analyzer 102 can receive new oil features for equipment under test (e.g., being tested or evaluated to determine whether it needs maintenance). The oil condition analyzer 102 can provide new HI data using the new oil features in a same or similar manner as disclosed herein, which can be received by a maintenance alert component 130, as shown in FIG.1. _{[0061] The maintenance alert component 130 can use the HI data and the failure prediction} model 126 to predict when the equipment is likely to fail (and/or require maintenance). While the example of FIG.1 illustrates the maintenance alert component 130 being implemented as part of the oil condition analyzer 102, in other examples, the maintenance alert component 130 can be implemented as a stand-alone module, or on a different system (or computing platform). _{[0062] For example, the ^ -steps trend prediction for condition indicators can be conducted by} failure prediction model 126. Here, ^-steps trend prediction refers to predicting a value of condition indicator within ^ time steps in the future. When the oil condition analyzer 102 receives new oil features, the model generator 104 can update the model parameters of the prediction model 126 based on the HI data through recursive method. The updated model parameters include the autoregressive order p and a moving average order q. The maintenance alert component 130 can determine whether these predicted condition indicators exceed the alert threshold 134 after the ^-step prediction, and if so, the maintenance alert component 130 can terminate and an alert will be sent to the user, or to initiate maintenance. If not, the above steps can be repeated. Alarm thresholds can be preset empirically or given based on statistical cases. For example, multiple failure cases of transmission systems of the same model can all occur when the status index is around 37. It can be considered that 37 is the failure threshold for this type of transmission system. _{[0063] In some examples, the maintenance alert component 130 can provide a maintenance alert} 136. The maintenance alert 136 can be provided to another system, device, apparatus, HMI, portable device (e.g., mobile phone) to alert a user (e.g., personnel) that maintenance may be required for the equipment. In some examples, the maintenance alert 136 can be provided to the equipment or system that controls the equipment to cause the equipment to seize operations (or enter a different operation state, for example, a state that causes less wear on the equipment). For example, the maintenance alert 136 can be provided to a control system 138 that controls the wind turbine to adjust a speed at which blades of the wind turbine rotate to reduce a rate of wear of components of the gearbox, and thus to prolong a life cycle of the gearbox. In some examples, maintenance alert 136 can be used by the control system 138 to cause the equipment (e.g., wind turbine) to enter a different operational state (e.g., stand-by state, off-state, a lower power generation state, a reduced operating state, etc.) During the operational state, maintenance of the equipment can be conducted. _{[0064] Accordingly, the maintenance tool 100 can be used to predict when equipment is likely} to fail (or require maintenance) based on wear data distribution across the different stages of the wear evolution process by the use of health indicator information extracted from oil features of the oil under monitoring. _{[0065] FIG. 2 is an example of a simplified lubrication and gearbox system 200 that can be used} to simulate gearbox wear of a wind turbine. For example, system 200 can be used for research, testing, or experimentation to assess how gearbox components, such as gears, bearings, and other components, wear and degrade over time under controlled conditions (e.g., replicating real-world conditions in which the gearbox will be used). The system 200 can communicate with the tool 100, as shown in FIG.1. Thus, reference can be made to one or more examples of FIG.1 in the example of FIG.2. In some examples, the system 200 includes the tool 100, in other examples, the tool 100 is implemented on a different system (e.g., in a cloud computing environment system, etc.). The system 200 includes an oil circulation (or lubrication) system 202 and a transmission system 204 (in some instances, referred to as drivetrain system). The oil circulation system 202 can manage a lubricant oil 206 that is being used to reduce friction, dissipate heat, and/or protect gears and/or bearings of a gearbox 208 of the transmission system 204. _{[0066] For example, the transmission system 204 can be used to control a rotational speed of a} wind turbine’s rotor to generate electricity efficiently. The transmission system 204 through the gearbox 208 can interact with an induction motor 210. The induction motor 210 can be used to provide torque to the gearbox 208, such as torque to an input shaft 212 of the gearbox 208. The induction motor 210 can drive the gearbox 208. In the example of FIG. 2, the induction motor 210 is used to supply mechanical power to the gearbox 208. The induction motor 210 acts as a driving force to rotate one or more components of the gearbox 208 to simulate a torque that would be provided by rotor blades of the wind turbine based on wind conditions. The induction motor 210 can operate at variable speeds and adjust its output to replicate a change in rotational speed of the blades of the wind turbine. For example, the induction motor 210 can be set or configured to have a given rotational speed (e.g., 10,000 rotations per minute (RPM)) to represent a targeted or nominal operating speed of the wind turbine). Wind turbines are designed to operate within a specific speed range to efficiently generate electricity. This speed corresponds to a rotation of the rotor blades, and is reflected by the driving force of the induction motor 210. _{[0067] The gearbox 208 can include a number of gears and shafts, such as an input shaft 212,} intermediate shafts 214-216, a test gear (or input gear) 218 and intermediate gears 220-224. The test gear 218 can be a specific type of gear that is used in machinery and mechanical systems, such as wind turbines and is being monitored according to one or more examples herein for failure. For example, the input gear 218 can be made from a type of steel alloy containing chromium (Cr) as one of its alloy elements. In some examples, the input gear 218 is a cylindrical gear, such as a cylindrical spur gear. If the input gear 218 has Cr and is a cylindrical spur gear, in some examples, the input gear 218 can be referred to as a 20Cr cylindrical spur gear. The connection of the input shaft 212 to the induction motor 210 can represent a connection of the rotor blades of the wind turbine to the gearbox 208 itself (e.g., in a real-world implementation). Thus, the input shaft 212 can be linked to a main rotor hub of the wind turbine, which supports and connects the rotor blades through the connection to the induction motor 210. The input shaft 212 can also be connected to the input gear 218. The input shaft 212 can transmit the mechanical power generated by the induction motor 210 to the gearbox 208, where it undergoes a series of gear reductions and transformations. For example, the mechanical energy can cause the input gear 218 to rotate, which causes the intermediate gears 220-224 to rotate as well. _{[0068] A brake 226 (e.g., a magnetic powder brake) can be connected to one of the intermediate} shafts, such as the intermediate shaft 216 and can be used to provide loading or resistance. For example, the brake 226 can provide to the gearbox 208 a given load (e.g., 10.4 Newton-meters (Nm)) that can be used as a stress test for the gearbox 208. Wind turbines have gearboxes that transmit a relatively slow rotational speed of a rotor to a higher rotational speed required by a generator. The loading provided by the brake 226 can represent a resistance that the gearbox 208 would encounter in a real-world scenario. The induction motor 210 and the brake 226 can be used in conjunction to simulate different loading conditions on the gearbox 208. By controlling an operation of the induction motor 210 and the brake 226, varying levels of mechanical load or resistance can be applied to the gearbox, mimicking real-world scenarios that the gearbox 208 may encounter, such as wind resistance on the wind turbine, or other external forces in different applications. _{[0069] In some examples, local meshing can be used, shown as reference numeral 244, between} the input (test) gear 218 and the intermediate gear 220 (an adjacent gear) to increase a contract stress to accelerate a wear evolution process of the input gear 218 and the intermediate gear 220, in some instances, the intermediate gears 220-224. For example, during operation, the gearbox 208 can be operated for a given period of time (e.g., 11,080 minutes) so that the input gear 218 can severely degrade (e.g., which can be confirmed through visual inspection, or sensor technology). The oil circulation system 202 pumps the lubricant oil 206 from an oil tank (or reservoir) 228 into the gearbox 208. For example, the oil circulation system 202 can be set with a given throughput rate (e.g., 2.5 liter per minute (L/min) to ensure proper operating conditions for the transmission system 204. The given throughput rate can represent a flow rate or a rate at which the lubricant oil 206 can be supplied to the gearbox 208. In some examples, the lubricant oil 206 is ExxonMobil Spartan EP 320, or some other type of wind turbine (WT) lubricant oil. In some examples, the lubricant oil 206 can be stored in oil tank 228 within a housing of the gearbox 208. A delivery pump 230 can be used to pump the lubricant oil 206 from the oil tank 228 into the gearbox 208. In some examples, a pressure regulator 232 can be used to monitor a pressure in a delivery pipeline 240 through which the lubricant oil 206 is being delivered. In additional or alternative examples, a delivery flow meter 234 can be used to monitor a flow rate of the lubricant oil 206 in the delivery pipeline 240. A return pump 236 can be used to pump the lubricating oil back into the oil tank 228 through a return pipeline 242 from the gearbox 208. In additional or alternative examples, a return flow meter 238 can be used to monitor a flow rate of the lubricant oil 206 in the return pipeline 242. _{[0070] A device 246 equipped with oil sensing capabilities (e.g., one or more oil sensors, for} example, in some examples, other sensors, for example, vibrational sensors) can be used to capture or acquire oil parameters of the lubricant oil 206. In some examples, the device 246 can be referred to as an oil sensing integrated terminal. The terminal 246 can be integrated into the oil circulation system 202 and can be used to monitor and gather data about a condition of the lubricant oil 206 within the system 200. The terminal can represent sensors, devices, and/or systems that can be used to monitor the condition of the lubricating oil being used at the wind turbine. In some examples, the terminal can include logic (e.g., machine-readable instructions) that can be used (e.g., executed by the terminal) to represent processing by the sensors, devices, and/or systems of captured oil test parameters of the lubricant oil at the wind turbine. In some examples, the terminal includes a data acquisition module and vibration signal acquisition module. The oil data sampling interval can be set to one (1) minute, while vibration data can be captured at one (1) time per minute, frequency of 5000 Hertz (Hz), acquisition period of ten (10) seconds. For example, from a brand new gear box to failure, the simulation took about 200 hours (hrs) in total with 11,180 data points being collected (or captured) to provide the oil and vibration data, respectively. _{[0071] In some examples, the oil parameters can include physical and chemical properties of the} lubricant oil 206 and thus the terminal can monitor a number of parameters of the lubricant oil 206 during operation. For example, twenty-three (23) different parameters of the lubricant oil 206 can be monitored. Each sensor that is being used provides respective oil parameter data, and in some instances, can be referred to as a channel. Thus, there can be a number of oil parameter datasets, each associated with a respective parameter of the lubricant oil 206. For example, if 23 channels are used, the 23 channels can include, for example, temperature, moisture, viscosity, DC, wear debris, cleanliness level etc. wear debris are detected within following range: 40-59 micrometers (^m), 60- 99 ^m, 100-299 ^m, 300-399 ^m, and >400 ^m, for ferrous channels and non-ferrous channels respectively. All data from the 23 channels can be collected at a given sampling interval (e.g., one (1) minute interval), in some examples, through RS485 for further analysis, such as by the tool 100, as disclosed herein. The oil parameter data for each oil parameter can include values captured (or detected), in some instances, computed that can characterize the oil parameter (or feature) of the lubricant oil 206 over time and thus can represent a time-varying signal. _{[0072] FIG. 3 is an example of a graph 300 with different types of oil features provided based} on oil parameter data that has been captured by sensors of the terminal 246, for example, during operation of the system 200. Thus, reference can be made to the example of FIGS.1-2 in the example of FIG. 3. In the example of FIG.3, 23 different properties (or features) of the lubricant oil 206 are shown (e.g., being provided by the device 246, for example). In some examples, the oil parameter data can be graphed over a period of time to provide a respective feature graph (identified with cardinal numerals, such as 1, 2, 3, and so on, and abbreviated nomenclature for the given property in the example of FIG.3). A table 400, as shown in the example of FIG.4 provides a definition for each abbreviated nomenclature property of each respective feature graph in the example of FIG. 3. Because the graph 300 illustrates multiple-features for the lubricant oil 206, the graph 300 can be referred to as a multi-feature graph. From the example of FIG. 3, it is apparent that one or more sensor signals (features) have little to no relative information on degradation of the input gear 218 (and thus the transmission system or the equipment (e.g., the wind turbine). _{[0073] According to the examples herein, the tool 100 can be used to determine a sensitivity of} each oil feature of the oil features 114, as shown in FIG.1. The tool 100 can be used to calculate or compute a sensitivity value for each oil feature to ascertain an importance of each oil feature of the oil features. FIG.5 is an example of a graph 500 of monotonicity values (sensitivity values) for the oil features (of FIG. 3) plotted in decreasing order (e.g., from highest to lowest). The calculated monotonicity of various oil parameters is listed in the example of FIG. 5. An x-axis of the graph 500 identifies each feature of the features (of FIG. 3) and a y-axis identifies a monotonicity value range, from 0.0 to 1.0. The tool 100 can be used to identify a subset of oil features 502 (referred to herein as sensitive oil features) of the oil features (of FIG. 3) using an associated or assigned sensitivity value according to the examples disclosed herein to provide sensitive oil features, such as the sensitive oil features 116, as shown in FIG.1. _{[0074] According to the result (e.g., FIG.5), the oil features which are sensitive to the component} degradation can be selected to characterize a health status according to the sensitivity threshold, as disclosed herein. In the example of FIG. 5, the FP1, FP2, FP3 and water content features can be selected and thus can be used to provide the sensitive oil features 502 based on the sensitivity threshold. FP1 can denote ferrous particles in a range of about 40-59 um, FP2 can denote ferrous particles in a range of about 60-99 um, FP3 can denote ferrous particles in a range of about 100-299 um. The sensitive oil features 502 are the most relative parameters to component condition. Consequently, the FP1, FP2, FP3 and water can be selected as condition monitoring key parameters (the sensitive oil features). _{[0075] As disclosed herein, the tool 100 can be used to reduce a redundancy (in some instances)} of the subset of oil features 502 while preserving key (or important) data to provide a reduced sensitive feature dataset, such as the reduced sensitive feature dataset 122, as shown in FIG.1. For example, the tool 100 can use PCA for data redundancy reduction. FIG.6 is an example of a graph 600 showing a contribution rate of each principal component during PCA. An x-axis identifies each principal component, and a y-axis identifies a contribution of a given principal component of principal components. As shown in the example of FIG.6, the first principal component (identified as “1’ in the example of FIG. 6) contributes a greatest amount during PCA. FIG. 6 illustrates a contribution of each component. It can be deduced from FIG. 6 that a first principal component dominates all contributions. Thus, a number of retaining eigenvectors m can be “1” since the first principal component could represent all components. Following PCA, as disclosed herein, in some instances, the tool 100 can determine a health of equipment. FIG. 7 is an example of a graph 700 illustrating a health of equipment determined by the tool 100 over a wear evolution process, as disclosed herein. An x-axis of the graph 700 represents a time (in minutes), and a y-axis of the graph represents a health indicator value (e.g., computed according to one or more examples, as disclosed herein). The graph 700 includes a degradation trend (signal) 702 representative of the health of the equipment over the wear evolution process: a run-in period 704, a steady operating period 706, and an accelerated period 708. _{[0076] FIG. 8 is an example of a graph 800 illustrating an accuracy of the failure prediction} model 126, as shown in FIG. 1. Thus, reference can be made to one or more examples of FIGS. 1- 7 in the example of FIG. 8. The failure prediction model 126 can be used to analyze a life cycle (e.g., a complete life cycle) of the lubricant oil 206, as shown in FIG. 2. For example, 11,080 data points in total (referred to as a dataset) can be captured or provided, and initial noise data can be excluded from the dataset. The dataset can be divided into a training dataset and a validation dataset. The training dataset can include about 2,000 data points to train and gain PCA parameters and provide a preliminary model. The training and PCA parameter gaining can be implemented by the model generator 104, as shown in FIG.1. The validation dataset can include about 9,080 data points for validating the preliminary model and to continuously iterate parameters of the preliminary model, which can be implemented by the model generator 104, for providing the failure prediction model 126. During a training process, a PCA primary component can contribute about 95%, which shows a majority of data information is extracted by that component (e.g., see FIG.6). During a validation process, a prediction step can be set to about 600 points and updates to about 100 points by the model generator 104. The model generator 104 can automatically extract about 100 data points for iterating the preliminary model. _{[0077] FIGS. 9-10 are example graphs 900-1000 showing a comparison before and after} prediction updates. FIG. 11 is an example of a graph 1100 showing a prediction trend of oil performance before actual equipment (e.g., gearbox) failure occurs. When a degradation trend reaches about 10,600 minutes at 1102, the failure prediction model 126 can be used to prediction when equipment (or asset) failure can occur at about 11,200 minutes at 1104 based on an alarm threshold 1106. The alarm threshold 1106 in some instances can correspond to the alarm threshold 134, as shown in FIG. 1. After inspection, real failure happened at about 11,100 minutes at 1108. Because the failure prediction model 126 predicted that equipment failure would happen at about 1112 (at about 11,200 minutes) and actual equipment failure happened at about 1108 (11,100 minutes). The failure prediction model 126 can provide an alert 600 minutes (8.3 hrs) before the actual failure of the equipment occurs. As shown in FIG.11, the actual failure time and the predicted failure time are within a given percentage of error so that the predictions of equipment failure provided by the failure prediction model 126 is accurate so that the model can be used. _{[0078] In the example of FIG. 11, a degradation trend 1110 over a period of time (e.g., in the} example of FIG. 11 from about 0 minutes to about 10,600 minutes) can be generated based on measured oil parameters of the lubricant oil 206, as shown in FIG. 2. The degradation trend 1110 can be generated by the oil condition analyzer 102, as shown in FIG.1. The failure prediction model 126 can be used to forecast or predict a degradation of the lubricating oil based on the oil parameters for the lubricant oil 206 to provide a predicted degradation trend 1112 over a period of time, as shown in FIG. 11. The lubricant oil 206 can be monitored over a complete wear evolution process and the measured oil parameters can be used to provide an actual degradation 1114. As shown in FIG. 11, the actual failure time at 1108 and the predicted failure time at 1104 are within relative proximity of each other validating that the failure prediction model 126 is accurate to be used for equipment failure prediction. _{[0079] In some examples, when a time was at the time point corresponding to 1108, a machine} check can be performed to confirm that the machine failed. Therefore, it can be assumed that the point is a machine failure point, the alarm threshold. As can be seen from FIG. 11, when the time is at the time point corresponding to 1102, the forward forecast 600 points can reach the alarm threshold at the time corresponding to 1104. It should be noted that this may not mean that the device has failed at the 1102 corresponding point in time, but that the failure prediction model 126 performs the prediction action at the 1102 point in time and predicts that the equipment failure occurred at around 1104 (11,180 minutes), actual equipment failure occurred around 1108 (11,100 minutes). _{[0080] In view of the foregoing structural and functional features described above, example} methods will be better appreciated with reference to FIGS.12-13. While, for purposes of simplicity of explanation, the example methods of FIGS. 12-13 are shown and described as executing serially, it is to be understood and appreciated that the present examples are not limited by the illustrated order, as some actions could in other examples occur in different orders, multiple times and/or concurrently from that shown and described herein. Moreover, it is not necessary that all described actions be performed to implement the methods. _{[0081] FIG. 12 is an example of a method 1200 for generating a failure prediction model for use} in oil condition monitoring as part of predictive maintenance. The failure prediction model can correspond to the failure prediction model 126, as shown in FIG.1. Thus, reference can be made to one or more examples of FIGS.1-11 in the example of FIG.12. The method 1200 can begin at 1202 by receiving (e.g., at the feature selector 112, as shown in FIG. 1) oil features (or oil parameters) of oil (e.g., the lubricant oil 206, as shown in FIG.2). At 1204, sensitive oil features (e.g., the sensitive oil features 116, as shown in FIG. 1) can be identified from the oil features. At 1206, a health indicator (e.g., a degradation trend of the equipment) can be generated (e.g., by the HI engine 120, as shown in FIG.1) based on the sensitive oil features. At 1208, the failure prediction model can be generated based on the health indicator. In some examples, at 1210, the failure prediction model can be used to initiate or cause maintenance of the equipment. _{[0082] FIG. 13 is an example of another method 1300 for predicting maintenance of equipment} (e.g., the gearbox 208, as shown in FIG.2). Thus, reference can be made to one or more examples of FIGS. 1-11 in the example of FIG.13. The method 1300 can begin at 1302 through acquisition of lubricant oil data (e.g., the oil features 114, as shown in FIG.1) for oil (e.g., the lubricant oil 206, as shown in FIG. 2). At 1304, the lubricant oil data can be analyzed using feature selection based on monotonicity to identify sensitive oil data (e.g., the sensitive oil features 116, as shown in FIG. 1). At 1306, an asset (equipment) health indicator (e.g., the HI data 124, as shown in FIG.1) can be constructed (or generated) based on PCA (e.g., by the health indicator engine 120, as shown in FIG. 1). In the example of FIG. 13, steps 1304-1306 can refer to health condition monitoring 1308, in some instances. At 1310, degradation trend prediction can be implemented (e.g., in some instances by the alert component 130, as shown in FIG. 1) for the equipment (e.g., machinery) based on ARIMA, in some examples, corresponding to the failure prediction model 126, as shown in FIG.1. Step 1310 can refer to degradation trend prediction 132, in some instances. At 1314, predictive maintenance of the equipment can be initiated or caused based on the predicted degradation trend at step 1310 according to one or more examples disclosed herein. _{[0083] While the disclosure has described several exemplary embodiments, it will be understood} by those skilled in the art that various changes can be made, and equivalents can be substituted for elements thereof, without departing from the spirit and scope of the invention. In addition, many modifications will be appreciated by those skilled in the art to adapt a particular instrument, situation, or material to embodiments of the disclosure without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiments disclosed, or to the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the appended claims. Moreover, reference in the appended claims to an apparatus or system or a component of an apparatus or system being adapted to, arranged to, capable of, configured to, enabled to, operable to, or operative to perform a particular function encompasses that apparatus, system, or component, whether or not it or that particular function is activated, turned on, or unlocked, as long as that apparatus, system, or component is so adapted, arranged, capable, configured, enabled, operable, or operative. _{[0084] In view of the foregoing structural and functional description, those skilled in the art will} appreciate that portions of the embodiments may be embodied as a method, data processing system, or computer program product. Accordingly, these portions of the present embodiments may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware, such as shown and described with respect to the computer system of FIG. 14. Thus, reference can be made to one or more examples of FIGS. 1-13 in the example of FIG.14. _{[0085] In this regard, FIG. 14 illustrates one example of a computer system 1400 that can be} employed to execute one or more embodiments of the present disclosure. Computer system 1400 can be implemented on one or more general purpose networked computer systems, embedded computer systems, routers, switches, server devices, client devices, various intermediate devices/nodes or standalone computer systems. Additionally, computer system 1400 can be implemented on various mobile clients such as, for example, a personal digital assistant (PDA), laptop computer, pager, and the like, provided it includes sufficient processing capabilities. _{[0086] Computer system 1400 includes processing unit 1402, system memory 1404, and system} bus 1406 that couples various system components, including the system memory 1404, to processing unit 1402. Dual microprocessors and other multi-processor architectures also can be used as processing unit 1402. System bus 1406 may be any of several types of bus structure including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of bus architectures. System memory 1404 includes read only memory (ROM) 1410 and random access memory (RAM) 1412. A basic input/output system (BIOS) 1414 can reside in ROM 1412 containing the basic routines that help to transfer information among elements within computer system 1400. _{[0087] Computer system 1400 can include a hard disk drive 1416, magnetic disk drive 1418,} e.g., to read from or write to removable disk 1420, and an optical disk drive 1422, e.g., for reading CD-ROM disk 1424 or to read from or write to other optical media. Hard disk drive 1416, magnetic disk drive 1418, and optical disk drive 1422 are connected to system bus 1406 by a hard disk drive interface 1426, a magnetic disk drive interface 1428, and an optical drive interface 1430, respectively. The drives and associated computer-readable media provide nonvolatile storage of data, data structures, and computer-executable instructions for computer system 1400. Although the description of computer-readable media above refers to a hard disk, a removable magnetic disk and a CD, other types of media that are readable by a computer, such as magnetic cassettes, flash memory cards, digital video disks and the like, in a variety of forms, may also be used in the operating environment; further, any such media may contain computer-executable instructions for implementing one or more parts of embodiments shown and disclosed herein. A number of program modules may be stored in drives and RAM 1410, including operating system 1432, one or more application programs 1434, other program modules 1436, and program data 1438. In some examples, the application programs 1434 can include one or more modules (or block diagrams), or systems, as shown and disclosed herein. Thus, in some examples, the application programs 1434 can include the maintenance tool 100, as shown in FIG.1. _{[0088] A user may enter commands and information into computer system 1400 through one or} more input devices 1440, such as a pointing device (e.g., a mouse, touch screen), keyboard, microphone, joystick, game pad, scanner, and the like. These and other input devices are often connected to processing unit 1402 through a corresponding port interface 1442 that is coupled to the system bus, but may be connected by other interfaces, such as a parallel port, serial port, or universal serial bus (USB). One or more output devices 1444 (e.g., display, a monitor, printer, projector, or other type of displaying device) is also connected to system bus 1406 via interface 1446, such as a video adapter. _{[0089] Computer system 1400 may operate in a networked environment using logical} connections to one or more remote computers, such as remote computer 1448. Remote computer 1448 may be a workstation, computer system, router, peer device, or other common network node, and typically includes many or all the elements described relative to computer system 1400. The logical connections, schematically indicated at 1450, can include a local area network (LAN) and a wide area network (WAN). When used in a LAN networking environment, computer system 1400 can be connected to the local network through a network interface or adapter 1452. When used in a WAN networking environment, computer system 1400 can include a modem, or can be connected to a communications server on the LAN. The modem, which may be internal or external, can be connected to system bus 1406 via an appropriate port interface. In a networked environment, application programs 1434 or program data 1438 depicted relative to computer system 1400, or portions thereof, may be stored in a remote memory storage device 1454. _{[0090] Although this disclosure includes a detailed description on a computing platform and/or} computer, implementation of the teachings recited herein are not limited to only such computing platforms. Rather, embodiments of the present disclosure are capable of being implemented in conjunction with any other type of computing environment now known or later developed. _{[0091] Cloud computing is a model of service delivery for enabling convenient, on-demand} network access to a shared pool of configurable computing resources (e.g., networks, network bandwidth, servers, processing, memory, storage, applications, virtual machines, and services) that can be rapidly provisioned and released with minimal management effort or interaction with a provider of the service. This cloud model may include at least five characteristics, at least three service models (e.g., software as a service (SaaS, platform as a service (PaaS), and/or infrastructure as a service (IaaS)) and at least four deployment models (e.g., private cloud, community cloud, public cloud, and/or hybrid cloud). A cloud computing environment can be service oriented with a focus on statelessness, low coupling, modularity, and semantic interoperability. _{[0092] FIG. 15 is an example of a cloud computing environment 1500 that can be used for} implementing one or more modules and/or systems in accordance with one or more examples, as disclosed herein. Thus, reference can be made to one or more examples of FIGS.1-14 in the example of FIG.15. As shown, cloud computing environment 1500 can include one or more cloud computing nodes 1502 with which local computing devices used by cloud consumers (or users), such as, for example, personal digital assistant (PDA), cellular, or portable device 1504, a desktop computer 1506, and/or a laptop computer 1508, may communicate. In some examples, the cloud computing environment 1500 can include a corporate network, as disclosed herein. Thus, the one or more nodes 1502 can be used to implement the maintenance tool 100, as shown in FIG. 1. In some examples, the portable device 1504, the desktop computer 1506 or the laptop computer 1508 is used to implement the maintenance tool 100. _{[0093] The computing nodes 1502 can communicate with one another. In some examples, the} computing nodes 1502 can be grouped (not shown) physically or virtually, in one or more networks, such as Private, Community, Public, or Hybrid clouds, or a combination thereof. This allows the cloud computing environment 1500 to offer infrastructure, platforms and/or software as services for which a cloud consumer does not need to maintain resources on a local computing device. The devices 1504-1508, as shown in FIG. 15, are intended to be illustrative and that computing nodes 1502 and cloud computing environment 1500 can communicate with any type of computerized device over any type of network and/or network addressable connection (e.g., using a web browser). In some examples, the one or more computing nodes 1502 are used for implementing one or more examples disclosed herein. _{[0094] In some examples, the cloud computing environment 1500 can provide one or more} functional abstraction layers. It is understood that the cloud computing environment 1500 need not provide all of the one or more functional abstraction layers (and corresponding functions and/or components), as disclosed herein. For example, the cloud computing environment 1500 can provide a hardware and software layer that can include hardware and software components. Examples of hardware components include: mainframes; RISC (Reduced Instruction Set Computer) architecture based servers; servers; blade servers; storage devices; and networks and networking components. In some embodiments, software components include network application server software and database software. _{[0095] In some examples, the cloud computing environment 1500 can provide a virtualization} layer that provides an abstraction layer from which the following examples of virtual entities may be provided: virtual servers; virtual storage; virtual networks, including virtual private networks; virtual applications and operating systems; and virtual clients. In some examples, the cloud computing environment 1500 can provide a management layer that can provide the functions described below. For example, the management layer can provide resource provisioning that can provide dynamic procurement of computing resources and other resources that are utilized to perform tasks within the cloud computing environment. The management layer can also provide metering and pricing to provide cost tracking as resources are utilized within the cloud computing environment 1500, and billing or invoicing for consumption of these resources. In one example, these resources may include application software licenses. Security provides identity verification for cloud consumers and tasks, as well as protection for data and other resources. The management layer can also provide a user portal that provides access to the cloud computing environment 1500 for consumers and system administrators. The management layer can also provide service level management, which can provide cloud computing resource allocation and management such that required service levels are met. Service Level Agreement (SLA) planning and fulfillment can also be provided to provide pre-arrangement for, and procurement of, cloud computing resources for which a future requirement is anticipated in accordance with an SLA. _{[0096] In some examples, the cloud computing environment 1500 can provide a workloads layer} that provides examples of functionality for which the cloud computing environment 1500 may be utilized. Examples of workloads and functions which may be provided from this layer include: mapping and navigation; software development and lifecycle management; virtual classroom education delivery; data analytics processing; and transaction processing. Various embodiments of the present disclosure can utilize the cloud computing environment 1500. _{[0097] FIG. 16 is an example of a graphical user interface (GUI) 1600 depicting a performance} of predictive maintenance using the failure prediction model 126, as shown in FIG. 1, relative to predictive maintenance that relies on vibration data for equipment (e.g., the gearbox 208, as shown in FIG. 2). Thus, reference can be made to one or more examples of FIGS. 1-15 in the example of FIG. 16. As shown in the example of FIG. 16, the vibration data shows an increase at 10,190 minutes, but the HI shows an increase in equipment wear at about 8,275 minutes compared with vibrational monitoring. In some examples, the maintenance alert component 1130 can provide the maintenance alert 136, as shown in FIG. 1 at about 1,915 minutes when the HI indicator value exceeds an alarm threshold (e.g., the alarm threshold 134, as shown in FIG.1). _{[0098] The present invention may be a system, a method, and/or a computer program product at} any possible technical detail level of integration. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention. The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read- only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire. _{[0099] Computer readable program instructions described herein can be downloaded to} respective computing/processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing/processing device. _{[0100] Computer readable program instructions for carrying out operations of the present} invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, configuration data for integrated circuitry, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++, or the like, and procedural programming languages, such as the “C” programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand- alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention. _{[0101] Aspects of the present invention are described herein with reference to flowchart} illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions. _{[0102] These computer readable program instructions may be provided to a processor of a} general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks. _{[0103] The computer readable program instructions may also be loaded onto a computer, other} programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks. _{[0104] The flowchart and block diagrams in the Figures illustrate the architecture, functionality,} and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions. _{[0105] The terminology used herein is for the purpose of describing particular embodiments only} and is not intended to be limiting of the invention. As used herein, for example, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “contains”, “containing”, “includes”, “including,” “comprises”, and/or “comprising,” and variations thereof, when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. In addition, the use of ordinal numbers (e.g., first, second, third, etc.) is for distinction and not counting. For example, the use of “third” does not imply there must be a corresponding “first” or “second.” Also, as used herein, the terms “coupled” or “coupled to” or “connected” or “connected to” or “attached” or “attached to” may indicate establishing either a direct or indirect connection, and is not limited to either unless expressly referenced as such. Furthermore, to the extent that the terms “includes,” “has,” “possesses,” and the like are used in the detailed description, claims, appendices and drawings such terms are intended to be inclusive in a manner similar to the term “comprising” as “comprising” is interpreted when employed as a transitional word in a claim. The term “based on” means “based at least in part on.” The terms “about” and “approximately” can be used to include any numerical value that can vary without changing the basic function of that value. When used with a range, “about” and “approximately” also disclose the range defined by the absolute values of the two endpoints, e.g. “about 2 to about 4” also discloses the range “from 2 to 4.” Generally, the terms “about” and “approximately” may refer to plus or minus 5-10% of the indicated number. _{[0106] What has been described above include mere examples of systems, computer program} products and computer-implemented methods. It is, of course, not possible to describe every conceivable combination of components, products and/or computer-implemented methods for purposes of describing this disclosure, but one of ordinary skill in the art can recognize that many further combinations and permutations of this disclosure are possible. The descriptions of the various embodiments have been presented for purposes of illustration, but are not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. _{[0107] Certain embodiments have also been described herein with reference to block illustrations} of methods, systems, and computer program products. It will be understood that blocks of the illustrations, and combinations of blocks in the illustrations, can be implemented by computer- executable instructions. These computer-executable instructions may be provided to one or more processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus (or a combination of devices and circuits) to produce a machine, such that the instructions, which execute via the processor, implement the functions specified in the block or blocks. _{[0108] These computer-executable instructions may also be stored in computer-readable} memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. The computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

Claims

CLAIMS The invention claimed is: 1. A computer-implemented method comprising receiving oil features of oil being used by equipment; identifying sensitive oil features from the oil features that are sensitive to oil degradation; generating a health indicator based on the sensitive oil features, the health indicator being indicative of a health of the equipment; generating a failure prediction model based on the health indicator; and causing equipment maintenance to be implemented based on the failure prediction model.

2. The computer-implemented method of claim 1, wherein the identifying comprises: determining a sensitivity score for each oil feature that indicates how sensitive that oil feature is to oil degradation; and identifying the sensitive oil features based on the sensitivity score for each oil feature.

3. The computer-implemented method of claim 2, wherein the determining comprises a monotonicity value as the sensitivity score for each oil feature, and the sensitive oil features being identified based on the monotonicity value.

4. The computer-implemented method of claim 2, wherein the identifying comprises evaluating the sensitivity scores relative to a sensitivity threshold to identify the sensitive oil features that have the sensitivity score that is greater than or equal to the sensitivity threshold.

5. The computer-implemented method of claim 1, wherein the generating the health indicator comprises applying a principal component analysis (PCA) to the sensitive oil features to reduce a dimensionality of the sensitive oil features to provide a reduced sensitive feature dataset, the health indicator being generated based on the reduced sensitive feature dataset.

6. The computer-implemented method of claim 5, the generating the health indicator further comprises determining a condition indicator sequence, wherein each value of the condition indicator sequence is indicative of the health of the equipment at time instance over a period of time, the health indicator being provided based on the condition indicator sequence.

7. The computer-implemented method of claim 6, wherein the generating the failure prediction model comprises using the condition indicator sequence and modeling parameters to train a model over a number of iterations to provide the failure prediction model.

8. The computer-implemented method of claim 7, wherein the failure prediction model is a Autoregressive Integrated Moving Average Model (ARIMA).

9. The computer-implemented method of claim 7, wherein the modeling parameters comprises an auto-regressive (AR) parameter, an integration (I) parameter, and moving average (MA) parameter for the ARIMA.

10. The computer-implemented method of any preceding claim, wherein the causing comprises generating a maintenance alert using the failure prediction model.

11. The computer-implemented method of claim 10, wherein the equipment is first equipment and the causing comprises causing the maintenance alert to a control system to adjust an operational state of second equipment being monitored from a given operational state to a different operational state.

12. A system comprising: one or more computing platforms configured to: identify sensitive oil features from oil features of lubricant oil being used in a gearbox of a transmission system that are sensitive to oil degradation, the oil features being captured by one or more oil sensors; generate a degradation trend based on the sensitive oil features, the degradation trend being indicative of a degradation of the equipment; generate a failure prediction model based on the degradation trend, the failure prediction model predicting a degradation of the equipment; and determine whether gearbox maintenance is to be implemented based on the failure prediction model.

13. The system of claim 12, wherein one or more computing platforms are configured to cause maintenance of the gearbox in response to determining that maintenance of the gearbox is needed.

14. The system of claim 12 or 13, wherein the failure prediction model is a Autoregressive Integrated Moving Average Model (ARIMA).

15. The system of claims 12-14, wherein the one or more computing platforms are further configured to: determine a sensitivity score for each oil feature that indicates how sensitive that oil feature is to oil degradation; and identify the sensitive oil features based on the sensitivity score for each oil feature.