Intelligent process anomaly detection and trend prediction system

Abstract

The novel system includes an intelligent process anomaly detection and trend prediction system, which, according to some embodiments, is configured to train artificial intelligence and machine learning systems for anomaly prediction in industrial systems. In some embodiments, such an intelligent process anomaly detection and trend prediction system is configured to determine an estimated remaining usable lifetime of industrial assets. For example, in some embodiments, the system is configured to identify degraded portions and normal portions of a signal; separate the degraded portions from the normal portions; identify one or more patterns of the degraded and normal portions of the signal; and determine anomaly predictions based on the one or more patterns.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit and priority of U.S. Provisional Application No. 62 / 901,080, filed September 16, 2019, entitled “Intelligent Process Anomaly Detection and Trend Projection System,” the entire contents of which are incorporated herein by reference. Background Technology

[0003] Building digital infrastructure to capture the operation of a wide range of industrial assets opens up a vast space of possibilities. In many cases, the Internet of Things (IoT) or local sensor networks generate the data needed to initiate complex analyses of process development. Detecting anomalies in their very early stages is a crucial step toward optimal operation. Maintenance procedures should be initiated after early detection of degradation or anomalies. Typically, for various reasons (to prevent long downtime, maintenance teams busy with another operation, observed degradation being normal for a particular process and having some additional tolerance designed to be utilized, or even if degradation is anomalous, having some additional tolerance designed to be utilized), operators of equipment need to know how long they can keep it operational. Capturing patterns of degradation and estimating the future evolution of degradation is highly advantageous, and some embodiments of the present invention provide such capabilities. Summary of the Invention

[0004] Some embodiments of this disclosure provide computer-implemented platforms, systems, and methods with various exemplary technical improvements, including an improved intelligent process anomaly detection and trend prediction system. In some embodiments, the system is configured to receive a signal. In some embodiments, the system is configured to identify degraded portions and normal portions of the signal. In some embodiments, the system is configured to separate the degraded portions of the signal from the normal portions of the signal. In some embodiments, the system is configured to identify one or more patterns of degraded portions and normal portions of the signal. In some embodiments, the system is configured to determine anomaly predictions based on one or more patterns. In some embodiments, the system is configured to determine that the signal has no seasonal component.

[0005] In some embodiments, seasonality is a trend of change that occurs at specific, regular intervals within a process. In some embodiments, seasonality may cause quality variations as a piece of equipment operates throughout its "season" of life. An example of a change in seasonality in a trend could be a cutting blade becoming dull, which causes a steady increase in the drive motor amperage over time.

[0006] In some embodiments, the system is configured to determine an estimated remaining available lifetime based on one or more patterns. In some embodiments, the system is configured to perform a multi-resolution differential scheme. In some embodiments, the system is configured to generate one or more versions of a signal based on the multi-resolution differential scheme. In some embodiments, the system is configured to determine higher-order differences for each of the one or more versions of the signal. In some embodiments, the system is configured to identify one or more patterns based on the higher-order differences for each of the one or more versions of the signal.

[0007] In some embodiments, the system includes an intelligent process anomaly detection and trend prediction system configured to include a non-transitory computer-readable program memory storing instructions, a non-transitory computer-readable data memory, and a processor configured to execute the instructions. In some embodiments, the processor is configured to execute instructions to receive a signal. In some embodiments, the system is configured to identify degraded portions and normal portions of a signal. In some embodiments, the system is configured to separate degraded portions of a signal from normal portions of a signal. In some embodiments, the system is configured to identify one or more patterns of degraded portions and normal portions of a signal. In some embodiments, the system is configured to determine anomaly predictions based on one or more patterns. In some embodiments, the system is configured to determine that the signal has no seasonal component. In some embodiments, the system is configured to determine an estimated remaining usable lifetime based on one or more patterns. In some embodiments, the system is configured to perform a multi-resolution differential scheme. In some embodiments, the system is configured to generate one or more versions of a signal based on the multi-resolution differential scheme. In some embodiments, the system is configured to determine higher-order differences for each of the one or more versions of the signal. In some embodiments, the system is configured to identify one or more patterns based on the higher-order differences for each of the one or more versions of the signal.

[0008] In other embodiments, the intelligent process anomaly detection and trend prediction system includes a non-transitory computer-readable medium comprising one or more sequences of instructions that, when executed by one or more processors, cause one or more operations and / or configurations to be performed. In some embodiments, the intelligent process anomaly detection and trend prediction system is configured to receive a signal. In some embodiments, the system is configured to identify degraded portions and normal portions of a signal. In some embodiments, the system is configured to separate the degraded portions of the signal from the normal portions of the signal. In some embodiments, the system is configured to identify one or more patterns of the degraded portions and normal portions of the signal. In some embodiments, the system is configured to determine anomaly predictions based on one or more patterns. In some embodiments, the system is configured to determine that the signal has no seasonal component. In some embodiments, the system is configured to determine an estimated remaining usable lifetime based on one or more patterns. In some embodiments, the system is configured to perform a multi-resolution differencing scheme. In some embodiments, the system is configured to generate one or more versions of a signal based on the multi-resolution differencing scheme. In some embodiments, the system is configured to determine higher-order differences for each of the one or more versions of the signal. In some embodiments, the system is configured to identify one or more patterns based on the higher-order differences for each of the one or more versions of the signal. In some embodiments, the system is configured to use one or more patterns identified using training data in an AI model. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating an overview of some embodiments of the system.

[0010] Figure 2 This is a detailed description based on some embodiments. Figure 1 A flowchart detailing the steps for seasonal component testing.

[0011] Figure 3 This is a detailed description based on some embodiments. Figure 1 A flowchart detailing the steps for identifying normal and degraded signals.

[0012] Figure 4 This is a detailed description based on some embodiments. Figure 1 A flowchart detailing the steps for predicting future anomalies.

[0013] Figure 5 This is a detailed description based on some embodiments. Figure 1 A flowchart detailing the remaining usable lifetime prediction steps.

[0014] Figure 6 The illustration depicts a computer system that supports or operates intelligent process anomaly detection and trend prediction according to some embodiments. Detailed Implementation

[0015] Before explaining any embodiments of the invention in detail, it is to be understood that the invention, in its application, is not limited to the details of the construction and arrangement of the components set forth in the following description or illustrated in the following drawings. Some embodiments of the system are configured to be combined with some other embodiments, and all embodiments can be practiced or performed in various ways. Furthermore, it is to be understood that the wording and terminology used herein are for descriptive purposes and should not be considered limiting. The use of “comprising,” “including,” or “having,” and variations thereof herein is intended to cover the items listed thereafter and their equivalents, as well as additional items. Unless otherwise specified or limited, the terms “mounted,” “connected,” “supported,” and “coupled,” and variations thereof are used extensively and cover direct and indirect mounting, connection, support, and coupling. Additionally, “connected” and “coupled” are not limited to physical or mechanical connections or couplings.

[0016] The following discussion is presented to enable those skilled in the art to manufacture and use the system. Various modifications to the illustrated embodiments will readily be apparent to those skilled in the art, and the general principles detailed according to some of the illustrated embodiments are configured to be applied and / or combined with some other illustrated embodiments and applications without departing from the embodiments of the invention. Therefore, embodiments of the invention are not intended to be limited to the illustrated embodiments, but are to be given the widest scope consistent with the principles and features disclosed herein. The following detailed description is to be read with reference to the accompanying drawings, in which the same elements in different drawings have the same reference numerals. The drawings, not necessarily drawn to scale, depict selected embodiments and are not intended to limit the scope of embodiments of the invention. Those skilled in the art will recognize that the examples provided herein have many useful alternatives and fall within the scope of embodiments of the invention.

[0017] Some embodiments of the present invention include various methods, means (including computer systems) for performing such methods, and computer-readable media containing instructions that, when executed by a computing system, cause the computing system to perform such methods. For example, some non-limiting embodiments include certain software instructions or program logic stored on one or more non-transitory computer-readable storage devices that tangibly store the program logic for execution by one or more processors of the system and / or one or more processors coupled to the system.

[0018] In the statistical literature, clear answers can be found regarding why using underrepresented categories of observation (e.g., anomalous events) presents insurmountable challenges when the objective is to estimate future behavior. Some embodiments of the present invention address these challenges.

[0019] Some embodiments of the system include one or more of the following stages:

[0020] I. According to some embodiments, the signal is examined to see if it contains seasonal components.

[0021] In some embodiments, the system is configured to use a Fourier transform to examine for seasonal components. In some embodiments, if the signal does not contain seasonal components, the system is configured to capture anomalous linear or exponential monotonic behavior. In some embodiments, if the signal does contain seasonal components, the system is configured to capture anomalous linear or exponential monotonic behavior and add a preprocessing step to account for the signal in terms of its Fourier components.

[0022] II. According to some embodiments, the degraded portion of the signal is identified and separated from normal behavior.

[0023] In some embodiments, the discovery of anomalous patterns is governed by statistical calculations of portions of the signal itself or portions of its transformed versions. In some embodiments, multi-resolution differencing schemes create multiple versions of the signal by downsampling the signal using integer factors. In some embodiments, the system is configured to schedule different differencing agents to operate on each signal representation. In some embodiments, the task of these agents is to find the first-order (or higher-order, depending on the pattern) difference of their corresponding signal representations. In some embodiments, different statistics of choice (e.g., conventional statistical methods) are then applied to the differencing time series to detect changes in the distribution. In some embodiments, the system is configured to divide the signal representation into samples of uniform size. In some embodiments, the chosen statistic is entropy. In some embodiments, the primary indicator of the pattern's identity is a sequence of decreasing differential entropy over the samples. In some embodiments, if such an identity or stability in the form of a trend pattern exists, then the differencing agent will expose it and it will be captured by the loss of entropy information. According to some embodiments, multiple large differences in the magnitude of these losses indicate the presence of multiple patterns within the degraded portions.

[0024] III. According to some embodiments, the strategies described in Phase II are used to identify patterns contained in normal behavior.

[0025] IV. In some embodiments, the above-described patterns are used as training data for models that predict the future development of anomalies (i.e., artificial intelligence, machine learning models).

[0026] In some embodiments, the system is configured to estimate future developments given a final deterioration pattern. In some embodiments, data from the final identity in Stage II becomes the training data for the model to be used. In some embodiments, several different modeling schemes are configured for extrapolation from that identity of device behavior. In some embodiments, if a feedforward ANN (Artificial Neural Network) is used, then to make it autoregressive (NARX), the system is configured to use the order of the differential agent as the amount of time delay. In some embodiments, a Long Short-Term Memory (LSTM) ANN is used to model the dynamic representation of the system. In some embodiments, the system is configured to use an LSTM as a stochastic machine, but uses a different method to infer the stochastic portion of the weights. In some embodiments, the system is configured to use variational inference to generate a distribution over the predicted sequence. In some embodiments, the system is configured to use a connection between the order of the differential agent that identifies the final trend pattern and the length of the input sequence of the LSTM.

[0027] In some embodiments, if more computational resources are available, various statistical techniques are used to compare sequences of patterns in degradation with those in normal operation and to determine which samples or variations of samples should be included in the training dataset.

[0028] V. According to some embodiments, the parameters of the overall design are inferred through mathematical optimization.

[0029] Generally, abstracting from the specificities of the embodiments, in some embodiments, the system is configured to involve design parameters (difference order, sample size of the sequences constituting the identity, number of LSTM units, and learning rate, etc.) where no prior knowledge exists. Furthermore, in some embodiments, different values ​​of these parameters significantly affect the results. Therefore, in some embodiments, estimators using these parameters (e.g., cross-validation) are employed. In some embodiments, the system is configured to use Bayesian optimization with an acquisition function (e.g., expected improvement).

[0030] VI. According to some embodiments, the development of anomalies is predicted.

[0031] In some embodiments, a number of regression techniques are used in the system. In some embodiments, the system is configured to use an estimate of the output sequence from an LSTM model and the prediction error. In some embodiments, from the resulting distribution, the system is configured to compute a path associated with the desired risk specification. In some embodiments, the system is then configured to derive a first hit time from both the intersection of this path and a threshold of the critical load of the asset of interest. Thus, in some embodiments, in addition to using this estimated path of the potential anomaly as an estimate of its future development, the system is configured to use it to obtain an estimate of the remaining available lifetime of the asset.

[0032] In some embodiments, the system is configured to use the same settings as above, but instead of using variational inference (which produces more accurate results), it is configured to infer the stochastic behavior of the LSTM parameters. In some embodiments, such a stochastic LSTM is trained to generate many, sometimes very different, probabilities in the future trajectory of the asset. In some embodiments, from this rich set of simulations, the system is configured to derive the most probable path, its standard error, and the path corresponding to the risk specified by the asset's operator.

[0033] Figure 1 This is a flowchart 100 illustrating an overview of some embodiments of the system. For example, according to some embodiments, Figure 1 The system is configured to provide an estimate of remaining available lifetime.

[0034] In some embodiments, Figure 1 In the system 110, the following are included: checking 102 whether the signal contains seasonal components; identifying 103 the degraded and normal portions of the signal; separating the degraded signal from the normal signal 104; identifying 105 the degraded signal pattern and the normal signal pattern; predicting 106 the future development of anomalies; and predicting 108 the remaining available lifetime estimate.

[0035] Figure 2 This is a detailed description based on some embodiments. Figure 1 A flowchart 102 detailing the steps for checking seasonal components. For example, according to some embodiments, Figure 2 The system is configured to determine whether the signal includes a seasonal component.

[0036] In some embodiments, Figure 2 In this system 201, the following are included: performing a Fourier transform on the signal 202; determining whether a seasonal component is detected in the signal; capturing anomalous linear or exponential monotonic behavior in response to the absence of a seasonal component in the signal 203; and considering the signal in terms of its Fourier components in response to the detection of a seasonal component in the signal 204.

[0037] Figure 3 This is a detailed description based on some embodiments. Figure 1 A flowchart 105 detailing the steps for identifying normal and degraded signals. For example, according to some embodiments, Figure 3 System 311 is configured to identify degraded signal patterns. For example, according to some embodiments, Figure 3 System 321 is configured to recognize normal signal mode.

[0038] In some embodiments, Figure 3In this system, systems 311-317 include: executing a multi-resolution differential scheme 312; creating multiple versions of a signal 313 by downsampling; finding the first (or higher) difference of each corresponding signal representation 314; performing statistical analysis on the differential data 315 to detect changes in the distribution; and identifying degraded signal patterns 316.

[0039] In some embodiments, Figure 3 In this system, systems 321-327 include: executing a multi-resolution differential scheme (322); creating multiple versions of a signal (323) through downsampling; finding the first (or higher) difference of each corresponding signal representation (324); performing statistical analysis on the differential data (325) to detect changes in the distribution; and identifying normal signal patterns (326).

[0040] Figure 4 This is a detailed description based on some embodiments. Figure 1 A flowchart 106 detailing the steps for predicting future anomalies. For example, according to some embodiments, Figure 4 Method 401 is configured to estimate future developments.

[0041] In some embodiments, Figure 4 In this system, 401-405 includes: inferring 402 unknown parameters through mathematical operations; using 403 signal patterns as training data for the model; and estimating 404 future development given the degraded final pattern.

[0042] Figure 5 This is a detailed description based on some embodiments. Figure 1 A flowchart 107 detailing the remaining usable lifetime prediction steps. For example, according to some embodiments, Figure 5 System 501 is configured to export the first hit time.

[0043] In some embodiments, Figure 5 In this system, 501-505 include: obtaining 502 the output sequence from the Long Short-Term Memory (LSTM) model and an estimate of the prediction error; calculating 503 a path associated with the desired risk from the obtained distribution; and deriving 504 the first hit time from the intersection of this path and a threshold of the critical load of the asset of interest.

[0044] Figure 6 The illustration depicts a computer system supporting or operating an intelligent process anomaly detection and trend prediction system according to some embodiments. In some embodiments, the intelligent process anomaly detection and trend prediction system is configured to be operatively coupled to... Figure 6The computer system 610 shown, or the computer system 610 configured to include an intelligent process anomaly detection and trend prediction system, is described in some embodiments. In some embodiments, the computer system 610 is configured to include and / or operate and / or process computer-executable code of the aforementioned program logic, software modules, and / or one or more of the system. Additionally, in some embodiments, the computer system 610 is configured to operate and / or display information within one or more graphical user interfaces coupled to the intelligent process anomaly detection and trend prediction system. In some embodiments, the computer system 610 includes a cloud server and / or may be coupled to one or more cloud-based server systems.

[0045] In some embodiments, system 610 is configured to include at least one computer, which includes at least one processor 632. In some embodiments, the at least one processor 632 includes a processor residing in or coupled to one or more server platforms. In some embodiments, system 610 includes a network interface 635a and an application interface 635b coupled to at least one processor 632 capable of processing at least one operating system 634. Additionally, in some embodiments, interfaces 635a, 635b coupled to at least one processor 632 are configured to process one or more software modules 638 (e.g., enterprise applications). In some embodiments, software module 638 is configured to include server-based software and operate to host at least one user account and / or at least one client account, and to transfer data between one or more of these accounts using at least one processor 632.

[0046] Considering the above embodiments, it should be understood that the present invention can employ various computer-implemented operations involving data stored in a computer system. Furthermore, in some embodiments, the databases and models described herein can store analytical models and other data on and coupled to computer-readable storage media within system 610. Additionally, in some embodiments, the above-described applications of the system are configured to be stored on and / or coupled to computer-readable storage media within system 610. In some embodiments, these operations are operations requiring physical manipulation of physical quantities. Typically, although not mandatory, in some embodiments, these quantities take the form of electrical, electromagnetic, or magnetic signals, or optical or magneto-optical forms capable of being stored, transmitted, combined, compared, and otherwise manipulated. In some embodiments, system 610 includes at least one computer-readable medium 636 coupled to at least one data source 637a and / or at least one data storage device 637b and / or at least one input / output device 637c.

[0047] In some embodiments, the invention may be implemented as computer-readable code on a computer-readable medium 636. In some embodiments, the computer-readable medium 636 is any data storage device capable of storing data that can subsequently be read by a computer system (such as system 610). In some embodiments, the computer-readable medium 636 is any physical or material medium that can be used to tangibly store desired information or data or instructions and that can be accessed by a computer or processor 632.

[0048] In some embodiments, computer-readable medium 636 includes hard disk drives, network attached storage (NAS), read-only memory, random access memory, FLASH-based memory, CD-ROM, CD-R, CD-RW, DVD, magnetic tape, and other optical and non-optical data storage devices. In some embodiments, various other forms of computer-readable medium 636 transmit or carry instructions to computer 640 and / or at least user 631, including routers, private or public networks, or other transmission devices or channels, wired and wireless. In some embodiments, software module 638 is configured to send and receive data from a database (e.g., from computer-readable medium 636 including data source 637a and data storage 637b including the database), and data is received by software module 638 from at least one other source. In some embodiments, at least one of software modules 638 is configured within the system to output data to at least user 631 via at least one graphical user interface presented on at least one digital display.

[0049] In some embodiments, the computer-readable medium 636 may be distributed over a conventional computer network via a network interface 635a, wherein a system implemented by computer-readable code may be stored and executed in a distributed manner. For example, in some embodiments, one or more components of the system 610 are configured to send and / or receive data via a local area network (“LAN”) 639a and / or an internet-coupled network 639b (e.g., such as a wireless internet). In some other embodiments, networks 639a, 639b are configured to include a wide area network (“WAN”), a direct connection (e.g., via a universal serial bus port), and / or other forms of computer-readable medium 636, and / or any combination thereof.

[0050] In some embodiments, components of networks 639a, 639b include any number of user devices, such as personal computers, including, for example, desktop computers coupled via LAN 639a, and / or laptop computers, and / or any stationary, generally non-mobile internet-connected device. For example, some embodiments include a personal computer 640, a database 641, a server 642, or any other computing device coupled via LAN 639a, each of which can be configured for any type of user, including an administrator. Some embodiments include a personal computer coupled via network 639b. In some other embodiments, one or more components of system 610 are coupled to send or receive data via an internet network (e.g., network 639b).

[0051] For example, some embodiments include at least one user 631 who is wirelessly coupled and accesses one or more software modules of a system including at least one enterprise application 638 via an input and output (“I / O”) device 637c. In some other embodiments, system 610 may enable at least one user 631 to be coupled to access enterprise application 638 via I / O device 637c through LAN 639a. In some embodiments, user 631 may include user 631a coupled to system 610 using a desktop computer, laptop computer, and / or any fixed, generally non-mobile internet device coupled to the internet 639b. In some additional embodiments, user 631 includes mobile user 631b coupled to system 610. In some embodiments, user 631b may use any mobile computer 631c to wirelessly couple to system 610, including but not limited to personal digital assistants and / or cellular phones, mobile phones or smartphones, and / or pagers, and / or digital tablets, and / or fixed or mobile internet devices.

[0052] This document describes a subject matter addressing technological improvements in the field of artificial intelligence by providing improved methods for teaching ANNs and LSTMs (e.g., how to utilize multiple signals to identify anomalies in trends and create better predictive models based on these trends). This disclosure describes in detail how a machine comprising one or more computers (which include one or more processors and one or more non-transitory computers) can implement a system and its improvements relative to the prior art. Instructions executed by the machine cannot be executed in the human brain or derived by a human using pen and paper; instead, the machine needs to transform process input data into useful output data. Furthermore, the claims presented herein do not attempt to tie judicial exceptions to known conventional steps implemented by general-purpose computers; nor do they attempt to tie judicial exceptions by simply linking them to the technical field. In fact, the systems and methods described herein were unknown and / or did not exist in the public domain at the time of filing, and they offer advantages in technological improvements not known in the prior art. Moreover, the system includes non-conventional steps that limit the claims to useful applications.

[0053] It is understood that the system, in its application, is not limited to the details of the construction and arrangement of the components set forth in the foregoing description or illustrated in the accompanying drawings. The systems and methods disclosed herein fall within the scope of many embodiments. The foregoing discussion is presented to enable those skilled in the art to make and use embodiments of the system. Modifications to the embodiments and general principles shown herein can be applied to all embodiments and applications without departing from the embodiments of the system. Furthermore, it is understood that features from some embodiments presented herein can be combined with other features according to some embodiments. Therefore, some embodiments of the system are not intended to be limited to what is shown, but are to be endowed with the widest scope consistent with all the principles and features disclosed herein.

[0054] Some embodiments of the system present specific values ​​and / or setpoints. These values ​​and setpoints are not intended to be limiting, but are merely examples of higher and lower configurations, and are intended to assist those skilled in the art in creating and using the system.

[0055] In addition, as the applicant's own lexicographer, the applicant assigns additional meanings to the following terms:

[0056] "Substantially" and "approximately" when used in conjunction with values ​​cover a difference of 5% or less in the same unit and / or scale of the measured quantities. In some embodiments, "substantially" and "approximately" are defined as presented in some implementations, such as those described in the specification.

[0057] As used herein, “simultaneous” includes lag and / or latency associated with conventional and / or proprietary computers, such as the processors and / or networks described herein that attempt to process multiple types of data simultaneously. “Simultaneous” also includes the time it takes for a digital signal to travel from one physical location to another (via wireless and / or wired networks, and / or within processor circuitry).

[0058] The use of "and / or," in relation to "A and / or B," means that one option can be "A and B," and the other option can be "A or B." This interpretation is consistent with the USPTO Patent Trial and Appeal Board's ruling in ex parte Gross (where the Board established that "and / or" means element A alone, element B alone, or elements A and B together).

[0059] As used herein, some embodiments detailed using the terms “capable” or “can” or their derivatives (e.g., the system display is capable of displaying X) are for descriptive purposes only and are to be understood as synonymous with “configured to” (e.g., the system display is configured to display X) used to define the boundaries of the system.

[0060] The foregoing detailed description should be read in conjunction with the accompanying drawings, in which the same elements in different drawings have the same reference numerals. The drawings, which are not necessarily drawn to scale, depict some embodiments and are not intended to limit the scope of embodiments of the system.

[0061] Any of the operations described herein that form part of this invention are useful machine operations. The invention also relates to devices or apparatuses for performing these operations. Apparatus, such as dedicated computers, can be specifically constructed for a desired purpose. When defined as a dedicated computer, the computer can also perform other processes, program executions, or routines that are not part of the dedicated purpose, while still being able to operate for the dedicated purpose. Alternatively, operations can be processed by a general-purpose computer that is selectively activated or configured by one or more computer programs stored in computer memory, cache, or obtained via a network. When data is obtained via a network, the data can be processed by other computers on the network (e.g., a cloud of computing resources).

[0062] Embodiments of the present invention can also be defined as machines that transform data from one state to another. The data can represent articles, can be represented as electronic signals, and can be manipulated electronically. The transformed data can, in some cases, be visually depicted on a display, thereby representing the physical object derived from the transformation of the data. The transformed data can be stored in a storage device in a general or specific format that enables the construction or depiction of physical and tangible objects. In some embodiments, manipulation can be performed by a processor. In such examples, the processor thus transforms data from one thing to another. Furthermore, some embodiments include methods that can be processed by one or more machines or processors that can be connected via a network. Each machine can transform data from one state or thing to another, and can also process the data, store the data in a storage device, transmit the data over a network, display the results, or pass the results to another machine. As used herein, computer-readable storage media refers to physical or tangible storage devices (as opposed to signals), and includes, but is not limited to, volatile and non-volatile, removable and non-removable storage media implemented in any method or technology for the tangible storage of information such as computer-readable instructions, data structures, program modules, or other data.

[0063] Although the method operations are presented in a specific order according to some embodiments, the execution of those steps does not necessarily occur in the listed order unless explicitly specified. Furthermore, other housekeeping operations may be performed between operations, operations may be adjusted so that they occur at slightly different times, and / or operations may be distributed across a system that allows processing operations to occur at various intervals associated with the processing, provided that the processing of the superimposed operations is performed in the desired manner and results in the desired system output.

[0064] Those skilled in the art will recognize that while the invention has been described above with reference to specific embodiments and examples, the invention is not necessarily limited thereto, and many other embodiments, examples, uses, alternative embodiments, examples, and modifications and deviations from them are intended to be covered by the appended claims. The entire disclosure of each patent and publication cited herein is incorporated by reference as if each such patent or publication were individually incorporated herein by reference. Various features and advantages of the invention are set forth in the following claims.

Claims

1. A system for providing anomaly detection and trend prediction of deterioration in industrial assets, comprising: One or more computers, the one or more computers including one or more processors and one or more non-transitory computer-readable media, the one or more non-transitory computer-readable media including instructions stored thereon, the instructions performing the following steps when executed by the one or more processors: Signals are received from sensors on assets coupled to an industrial process, the assets comprising multiple components; By analyzing the signal to obtain changes in trends occurring at regular intervals within the signal during the asset's lifespan, it is determined whether at least one of the plurality of components is a seasonal component that predictably deteriorates during the regular intervals. In response to determining that no seasonal component was detected in the plurality of components of the asset, the degraded portion of the signal and the normal portion of the signal are identified; Separate the degraded portion of the signal from the normal portion of the signal; Identify one or more patterns of the degraded portion of the signal and the normal portion of the signal; Based on one or more of the aforementioned patterns, predictions of future developments of anomalies are determined; as well as An estimate of the remaining usable life of the asset is determined based on predictions of future developments of the aforementioned anomalies and one or more of the aforementioned patterns.

2. The system as described in claim 1, One or more modes for identifying the degraded portion and the normal portion of the signal also include performing a multi-resolution differential scheme.

3. The system as described in claim 2, It also includes one or more versions of the signal generated based on the multi-resolution differential scheme.

4. The system as described in claim 3, It also includes determining one or more higher-order differences for each or more versions of the signal.

5. The system as described in claim 3, It also includes identifying one or more patterns based on the one or more higher-order differences of each or more versions of the signal.

6. A system for providing anomaly detection and trend prediction, comprising: One or more sensors, said one or more sensors being coupled to one or more physical assets, said one or more physical assets being configured to generate an output; One or more computers, the one or more computers including one or more processors and one or more non-transitory computer-readable media, the one or more non-transitory computer-readable media including instructions stored thereon, the instructions configuring the one or more computers to: One or more signals are received from the one or more sensors coupled to the asset, the asset comprising multiple components; By analyzing the signals to obtain changes in trends occurring at regular intervals within the signals during the asset's lifecycle, it can be determined whether at least one of the plurality of components is a seasonal component; In response to determining that no seasonal component was detected in the plurality of components of the asset, the degraded portion of the one or more signals and the normal portion of the one or more signals are identified; Separate the degraded portion of the one or more signals from the normal portion of the one or more signals; Identify one or more patterns of degraded portions and normal portions of the one or more signals; The one or more patterns are input into an artificial neural network to determine predictions of future developments of anomalies; as well as The estimated remaining usable lifespan is determined based on predictions of abnormal future developments.

7. The system as described in claim 6, in, One or more modes for identifying the degraded portion and the normal portion of the signal also include performing a multi-resolution differential scheme.

8. The system as described in claim 7, The instructions are also configured to generate one or more versions of the signal based on the multi-resolution differential scheme.

9. The system as described in claim 8, The instructions are also configured to determine the higher-order difference for each or more versions of the signal.

10. The system as described in claim 9, The instructions are also configured to identify one or more patterns based on the higher-order differences of each or more versions of the signal.

11. An anomaly detection and trend prediction system, comprising: One or more computers, the one or more computers including one or more processors and one or more non-transitory computer-readable media, the one or more non-transitory computer-readable media including instructions stored thereon, the instructions configuring the one or more computers to: Signals are received from sensors coupled to a physical asset, which is configured to generate mechanical operations within an industrial process, the asset comprising multiple components; By analyzing the signals to obtain changes in trends occurring at regular intervals within the signals during the asset's lifecycle, it can be determined whether at least one of the plurality of components is a seasonal component; In response to determining that no seasonal component was detected in the plurality of components of the asset, the degraded portion of the signal and the normal portion of the signal are identified; Separate the degraded portion of the signal from the normal portion of the signal; Identify one or more patterns of the degraded portion of the signal and the normal portion of the signal; The artificial neural network is trained using the one or more patterns, and the artificial neural network is configured to output anomaly predictions on a display based on the one or more patterns; as well as An estimate of the remaining usable life of the asset is determined based on the anomaly prediction.

12. The system as described in claim 11, in, The instructions for identifying one or more patterns of the degraded portion of the signal and the normal portion of the signal also include performing a multi-resolution differential scheme.

13. The system as described in claim 12, The instructions are also configured to generate one or more versions of the signal based on the multi-resolution differential scheme.

14. The system as described in claim 13, The instructions are also configured to determine the higher-order difference for each or more versions of the signal.

15. The system as described in claim 14, The artificial neural network is also configured to recognize the one or more patterns based on the higher-order differences of each or more versions of the signal.

Images