Method, device, system and readable storage medium for estimating prediction

Abstract

Methods, devices, systems, and readable storage media for estimating predictions are disclosed. The present application discloses methods, devices, and systems for obtaining an estimated prediction of at least one state, a method for creating a training model for an estimated prediction of at least one state, and a computer-readable medium. In order to also provide state information to one or more industrial devices for which no process data is available, a machine learning model is trained by using process data of a subset of the industrial devices, corresponding product data, and states obtained by performing remote monitoring analysis on the process data. When a user input including at least first information indicating at least one industrial device type is received, product data of one or more industrial automation devices having the same indicated industrial device type is retrieved and input into a training model that outputs one or more estimated states.

Description

Technical Field

[0001] The present invention relates to remote monitoring of industrial equipment, and in particular, to a method, device and system for obtaining an estimated predication of at least one state, a method for creating a training model for an estimated predication of at least one state, and a computer-readable medium. Background Art

[0002] The development of networking between computers and measuring devices (especially different sensors), which can communicate without user involvement, has increased the amount of data collected in equipment and processes. The Internet of Things (IoT) is the networking of physical devices (also called "connected devices" and "smart devices"), such as vehicles, household appliances, machines, computers, and other items embedded with electronic devices, software, sensors, actuators, and network connections that enable these objects to collect and exchange data. IoT allows objects to be remotely sensed and / or controlled on existing network infrastructure. The Industrial Internet of Things (IIoT) is the use of IoT technology in manufacturing. It combines machine learning and big data technologies, taking advantage of sensor data, machine-to-machine communication, and automation technologies that have existed in industrial environments for many years. IIoT also provides a mechanism for connecting industrial devices to cloud computing platforms and transmitting information to cloud platforms to perform various operations on industrial data (e.g., remote monitoring for example for predictive maintenance). However, not all industrial devices are configured to collect and transmit data. Therefore, for those industrial devices, no remote monitoring is available. Summary of the invention

[0003] The object of the invention is to facilitate remote monitoring by means of a method, a computer program product, a device and a system, which are characterized by what is defined in the independent claims. Further embodiments are disclosed in the dependent claims.

[0004] According to one aspect, with the help of the training model, it is also possible to obtain an estimated predication for an industrial device whose process data is unavailable, ie, an estimated remote monitoring result.

[0005] According to one aspect, a training model is trained using process data for a subset of industrial devices, corresponding product data, and predictions obtained by performing remote monitoring analysis on the process data.

[0006] According to an embodiment, a computer-implemented method for obtaining an estimated prediction is provided, the method comprising: receiving user input including industrial device type information, the industrial device type information at least indicating at least one industrial device type; using the industrial device type information to retrieve product data of one or more industrial devices from a product database including product data of multiple industrial devices, the one or more industrial devices having the same industrial device type as the type of at least one indicated industrial device; inputting the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected by a subset of the multiple industrial devices, a prediction of at least one state generated by a remote monitoring and analysis tool based on the process data, and product data of a subset of the multiple industrial devices; processing the input product data into one or more estimated predictions of at least one state through the training model; and outputting one or more estimated predictions of at least one state.

[0007] According to an embodiment, a computer-implemented method for creating a training model for an estimated prediction of at least one state is provided, the method comprising: retrieving process data collected on multiple industrial devices; retrieving product data of multiple industrial devices; performing remote monitoring analysis on the retrieved process data to obtain a prediction of at least one state; and using the product data, process data, and the prediction of at least one state as training data, training a machine learning model by: repeating the training until, by using the product data as input, the machine learning model outputs an estimated prediction of at least one state corresponding to the prediction of at least one state obtained from the remote monitoring analysis.

[0008] According to an embodiment, a non-transitory computer-readable storage medium is provided, which stores one or more instructions that, when executed by one or more processors, cause the one or more processors to perform at least the following operations: in response to user input including industrial device type information at least indicating at least one industrial device type, use the industrial device type information to retrieve product data of one or more industrial devices from a product database including multiple industrial devices, wherein the one or more industrial devices have the same industrial device type as at least one indicated industrial device type; input the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected by a subset of the multiple industrial devices, a prediction of at least one state generated by a remote monitoring and analysis tool based on the process data, and product data of a subset of the multiple industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of at least one state; and outputs one or more estimated predictions of at least one state.

[0009] According to an embodiment, a non-transitory computer-readable storage medium is provided, which stores one or more instructions that, when executed by one or more processors, cause the one or more processors to perform at least the following operations: retrieve process data collected on multiple industrial devices; retrieve product data of multiple industrial devices; perform remote monitoring analysis on the retrieved process data to obtain a prediction of at least one state; and use the product data, process data, and the prediction of at least one state as training data to train a machine learning model by: repeating the training until, by using the product data as input, the machine learning model outputs an estimated prediction of at least one state corresponding to the prediction of at least one state obtained from the remote monitoring analysis.

[0010] According to one embodiment, a device for obtaining estimated predictions is provided, the device comprising: a device for receiving user input including industrial device type information, wherein the industrial device type information at least indicates at least one industrial device type; a device for retrieving product data of one or more industrial devices from a product database including product data of multiple industrial devices using the industrial device type information, wherein the one or more industrial devices have the same industrial device type as at least one indicated industrial device type; a device for inputting the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected by a subset of the multiple industrial devices, a prediction of at least one state generated by a remote monitoring and analysis tool based on the process data, and product data of a subset of the multiple industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of at least one state; and a device for outputting one or more estimated predictions of at least one state.

[0011] According to one embodiment, a system for obtaining estimated predictions is provided, the system comprising at least: one or more industrial devices; a product database comprising product data on at least one or more industrial devices; and one or more devices, comprising: a device for receiving user input comprising industrial device type information, wherein the industrial device type information at least indicates at least one industrial device type; a device for retrieving product data of one or more industrial devices from a product database comprising product data of multiple industrial devices using the industrial device type information, wherein the one or more industrial devices have the same industrial device type as at least one indicated industrial device type; a device for inputting the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected by a subset of the multiple industrial devices, a prediction of at least one state generated by a remote monitoring analysis tool based on the process data, and product data of a subset of the multiple industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of at least one state; and a device for outputting one or more estimated predictions of at least one state. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Hereinafter, exemplary embodiments will be described in more detail with reference to the accompanying drawings, in which:

[0013] Figure 1 shows a simplified architecture of the system;

[0014] Figures 2 to 5 is a flowchart illustrating an example of functionality; and

[0015] Figure 6 is a schematic block diagram. DETAILED DESCRIPTION

[0016] The following embodiments are exemplary. Although the specification may refer to "one", "an" or "some" embodiments in multiple locations, this does not necessarily mean that each such reference is to the same (one or more) embodiments or that the feature only applies to a single embodiment. Individual features of different embodiments may also be combined to provide other embodiments.

[0017] The following uses a single unit, model, equipment and memory to describe different embodiments and examples, but the embodiments / examples are not limited to such solutions. Concepts known as cloud computing and / or virtualization can be used. Virtualization can allow a single physical computing device to host one or more instances of a virtual machine that appears and operates as an independent computing device, so that a single physical computing device can dynamically create, maintain, delete or otherwise manage the virtual machine. Device operations can also be distributed between multiple servers, nodes, devices or hosts. In a cloud computing network device, a computing device and / or a storage device provide shared resources. Some other technological advances, such as software defined networking (SDN), can migrate one or more functions described below to any corresponding abstract concepts or devices or devices. Therefore, all words and expressions should be interpreted broadly, and they are intended to illustrate embodiments rather than limit embodiments.

[0018] The general exemplary architecture of the system is as follows Figure 1 shown. Figure 1 This is a simplified system architecture, showing only some equipment (devices, devices, nodes) and functional entities, all of which are logical units, and their implementation methods and / or quantities may be different from Figure 1 Different than shown. Figure 1 The connections shown are logical connections; the actual physical connections may be different. It will be apparent to those skilled in the art that the system also includes other equipment, functional entities and structures, some of which are used in the system or in a part of the system for or in big data, data management and communications. These and the protocols used are well known to those skilled in the art and are not relevant to the actual invention. Therefore, there is no need to discuss them in more detail here.

[0019] exist Figure 1 In the illustrated example, the system 100 includes a remote site 101 , a cloud platform 102 , one or more online industrial sites 103 , one or more offline industrial sites 104 , and one or more hybrid sites 105 .

[0020] Remote site 101 is described here as one or more sites that provide remote services (such as remote monitoring services). Remote site 101 can be a service center provided by a company that manufactures different industrial devices 131, 141 located in industrial sites 103, 104, but the solution is not limited to such sites.

[0021] In the example shown, the remote site 101 includes one or more devices 110 ( Figure 1Only one is shown in the figure), the device includes: one or more user interfaces 111 for user interaction; and a remote monitoring tool 112, the remote monitoring tool 112 includes an estimator unit 113.

[0022] The remote monitoring tool 112 is an analysis tool that may be based on data mining, self-learning artificial intelligence, mathematical models of the industrial processes used, and the like. The remote monitoring tool 112 may be a condition monitoring tool that is, for example, configured to identify changes in one or more condition parameters that indicate a developing fault. Another example of a remote monitoring tool 112 includes, for example, a remote performance optimization tool that identifies the source of a problem indicating increased energy consumption and / or increased emissions. The purpose and actual functioning of the remote monitoring tool and its internal structure are not relevant to the present invention, and therefore, the internal structure and actual functioning of the remote monitoring tool 112 will not be described in detail herein. It will be apparent to a person skilled in the art that any known or future remote monitoring tool may be used. However, in order to be able to perform virtual remote monitoring, Figure 1 The remote monitoring tool 112 shown in FIG. 1 includes an estimator unit 113. The estimator unit 113 is a trained machine learning model, the training and function of which will be described in more detail below. Figure 1 In the illustrated example of , the estimator unit 113 is included in the remote monitoring tool 114 , but it may also be a separate tool that operates independently of the remote monitoring tool.

[0023] exist Figure 1 In the example shown, the remote site 101 also includes a data storage device 114 and an apparatus 115, the apparatus 115 including at least a trainer unit 116 configured to train at least the estimator unit 113, as described below with reference to Figure 2 Will be described in more detail.

[0024] The data storage device 114 is a product data storage device, which contains at least basic information about the industrial devices Yn131 and Xn141. The industrial devices whose information is in the product data storage device may be industrial devices manufactured by a company that provides remote support via a remote site. The industrial devices Yn131 and Xn141 may be installed and / or waiting to be installed at the industrial sites 103, 104, 105. Examples of basic product information include product serial number, product type, product configuration, manufacturing date, components in the product with critical dates of the components, installation environment, expected load, industrial application (purpose of use), etc., which essentially depend on the industrial device.

[0025] One or more online industrial sites 103 include industrial devices Yn 131, which are configured to collect process data and transmit (send) it to a shared database 121 in a cloud platform. The process data includes measurement results on the industrial device and / or one or more operations of the industrial device and / or the environment of the industrial device, such as control signals and room temperature sent and / or received. One or more offline sites 104 include industrial devices Xn 141 that are not configured to transmit measurement results to a shared database, and one or more industrial devices Xn 141 in the offline site 104 may not even be configured to collect data. One or more hybrid sites 105 include two types of devices, namely, one or more devices configured to collect and transmit (send) process data, and one or more devices that are not configured to send process data and may not even be configured to collect any data. It should be understood that the industrial sites 103, 104, 105 here depict any industrial site including one or more industrial devices 131, 141, and there is no limitation on what constitutes an industrial device. A non-limiting list of examples of industrial sites includes power plants, manufacturing plants, chemical processing plants, power transmission systems, mining and mineral processing plants, upstream oil and gas systems, data centers, ships, transport fleet systems and utility plants. Industrial devices can be, for example, industrial automation devices. Industrial automation devices refer to electronic equipment for controlling one or more industrial processes. For example, industrial automation devices can control the position, speed, torque and / or direction of motors in conveyors, grinders, winding machines, pumps and / or fans. For example, industrial automation devices can control the speed and / or torque of motors by changing the frequency, current and / or voltage of the power supply to the motor. Naturally, the motor itself is an industrial device. A non-limiting list of examples of industrial automation devices includes drives, frequency converters, AC / DC converters, DC / AC converters, DC / DC converters, programmable logic controllers, switches, motion controllers or motion drivers, servo motors, soft starters, wind turbines and solar inverters.

[0026] The product data storage device 114 describing one or more data storage devices and the process data storage device 121 describing one or more data storage devices can be any type of traditional or future data storage, including distributed and centralized data storage managed by any appropriate management system. Examples of distributed storage include cloud-based storage in a cloud environment (e.g., it can be a public cloud, a community cloud, a private cloud, or a hybrid cloud). The cloud storage service can be accessed through a co-located cloud computer service, a network service application programming interface (API), or through an application that utilizes an API, such as a cloud desktop storage, a cloud storage gateway, or a network-based content management system. In other words, the data storage device 114, 121 can be a computing device equipped with one or more memories, or a subsystem, or an online archiving system, the subsystem or system includes a computing device, which is configured to appear as a logical data storage device for storing data thereto and / or retrieving data therefrom (device, device). However, the implementation of the data storage device 114, 121, the way of data storage, retrieval and update, that is, the details of how the data storage device interfaces, and the location of the data storage are irrelevant to the present invention. For those skilled in the art, it is obvious that any known or future solution can be used.

[0027] Of course, even in Figure 1 Not shown, the system 100 includes one or more networks, through which one or more industrial devices 131 and / or one or more equipment 110, 115 at a remote site can be connected to the cloud platform 102 to access process data 121 and / or connect from the remote site 101 to the industrial sites 103, 105 for remote monitoring.

[0028] It should be understood that Figure 1 The systems shown in are examples only, and other ways of locating the different functions described may also be used. For example, the offline site 104 may include equipment configured to perform virtual remote monitoring by using the estimator unit as a separate unit (i.e., without any remote monitoring tools) and accessing product data on the device at the offline site. In another example, the product database 114 is also located at the cloud platform 102, and the equipment 115 including the trainer unit 116 is located at a site separate from the remote site 101.

[0029] Figure 2Example functionality of a device configured to create and / or (re)train an estimator unit is shown. More precisely, it describes an example of how a trainer unit operates. The example functionality describes the basic principles for creating, training and retraining an estimator unit when needed. In this example, it is assumed that there is enough process data available to be used as training data and validation data to create / update a training model. The training model shown utilizes machine learning, i.e., is based on machine learning. A variety of such algorithms can be used, and any of them or any corresponding future algorithm can be used as the basis for the training model. To give a few examples, machine learning can be classification using artificial neural networks or support vector machines or regression. Naturally, any other machine learning can be used, such as image learning, rule-based learning, and explanation-based learning.

[0030] Reference Figure 2 In step 201, process data is retrieved from a process database. Then, in step 202, industrial devices whose process data is to be retrieved are determined, and in step 203, product data of these devices are retrieved from a product database using the device information. Figure 1 In an example of, process data is retrieved, device Yn is determined, and product data of device Yn is retrieved. The product data may include all data of the device stored in the product database, or a predefined subset of the product data. A non-limiting list of predefined subsets includes country, customer, site, industrial application, installation date.

[0031] At the same time, in step 204, remote monitoring analysis is performed on the retrieved process data. The remote monitoring analysis is performed in a conventional manner using a remote monitoring tool, i.e., without any involvement of an estimator unit. Figure 1 As described above, any known or future remote monitoring tool may be used.

[0032] After receiving the results of the remote monitoring analysis, a training data set (training examples) is prepared. For example, in classification, product data is used as instances, process data is used as features, and predicted conditions (i.e., the results of the condition monitoring analysis) are used as classes. At step 205, the machine learning model is trained in an iterative manner using the training examples until the model meets the accuracy criteria in step 206. The accuracy criteria may be that the outputted estimated prediction of the condition should correspond to the prediction of the condition obtained from the remote monitoring analysis. The training may be supervised learning or semi-supervised learning. When the machine learning model is using an artificial neural network, the weights of the nodes in the artificial neural network-based model may be adjusted during the iteration.

[0033] Satisfaction of the accuracy criterion in step 206 may be verified by using a test set separate from the training set to determine whether the output of the model corresponds to the predicted condition output by the remote monitoring tool.

[0034] When it is determined that the model is accurate, ie the model meets the accuracy criteria in step 206, the trained model is put into use in step 207. In other words, the estimator unit is ready to be used.

[0035] Depending on the implementation, once a training model is created, it can be continuously or periodically retrained / updated in response to explicit commands using the principles disclosed above.

[0036] Figures 3 to 5 Different functionalities of a device configured for virtual remote monitoring, ie a device comprising an estimator unit, are shown.

[0037] Reference Figure 3 , when user input including one or more device types as input information is received in step 301, in step 302, the device type is used as a search keyword and product data of the same type of device is retrieved from the product database and input to the estimator unit in step 303. The results from the estimator unit are then output in step 304. Before the results are output, they may undergo one or more statistical processes so as to output the results in a more user-friendly manner, similar to the way a remote monitoring tool would output its results. For example, when the remote monitoring tool is a condition monitoring tool, the output may be a generalized probability distribution of device conditions. Using drives as an example, the output may be "78% of installed drives are in good condition, 15% have a small risk of failure, and 7% are predicted to be in poor condition and require immediate maintenance". Another example includes a list of 105 drives that have a high risk of insulated gate bipolar transistor (IGBT) failure in the next two years, and this information may be used to replace the IGBT cells in the drives in the list, or at least perform condition checks at these sites.

[0038] Instead of the device type, the input information (industrial device type information) may be other information (industrial device type information) by which the device type can be determined, so that the device type can be used as a search keyword. For example, a device identifier may be received in step 301, and the corresponding device type may be determined from a product database. Another example includes receiving information identifying a site, and the device type in the site may be determined from a product database.

[0039] In another implementation, all input information is used as search keywords. If the input information includes site information in addition to the device type, then in step 302, product data of devices of the same type in the same site are retrieved.

[0040] Figure 4 An implementation is shown, wherein the estimator unit, ie the device configured for virtual remote monitoring, is further configured to offer the user the possibility to receive an estimated prediction of one or more states of one or more specific industrial devices.

[0041] Reference Figure 4 , when a user input including one or more device identifiers as input information is received in step 401, the device identifier is used as a search key and product data of the device is retrieved from a product database in step 402, and input to an estimator unit in step 403. The results from the estimator unit are then output in step 404. As described above, before the results are output, they may undergo one or more statistical processes.

[0042] In the above reference Figure 3 and 4 In the described examples, the device including the estimator unit is used for virtual remote monitoring, regardless of whether there is data available for traditional remote monitoring. In addition, in the examples, the device may not be configured to perform traditional remote monitoring, i.e., using a remote monitoring tool. The difference between the different remote monitoring is that virtual remote monitoring outputs an estimated prediction without knowing the actual process data, while traditional remote monitoring outputs a prediction based on the actual process data.

[0043] Figure 5 shows the functionality of a device configured for combined virtual and traditional remote monitoring, i.e., with a remote monitoring tool (called Figure 5 The predictor) and the estimator unit (called Figure 5 In addition, Figure 5 In the embodiment, the device type is used as an example of input information, but the example is not limited to such a solution. Any of the above disclosed examples and principles related to input information received from a user can use the above disclosed principles to Figure 5 process to achieve.

[0044] Reference Figure 5, when a user input including one or more device types as input information is received in step 501, in step 502, the device type is used as a search keyword, and product data of devices of the same type are retrieved from the product database. Then, in step 503, the device is acquired from the retrieved product data, and in step 504, it is determined whether there is available process data (or sufficient available process data) on the device. If there is no process data, or there is not enough process data, the product data of the device is input to the estimator in step 505. Then, the process proceeds to step 506 to check whether all devices whose product data have been retrieved in step 502 have been processed. If not, the process continues to step 503 to acquire another device.

[0045] If process data exists (step 504: yes), then the process data is retrieved and input to the predictor in step 507, and then the process continues to step 506 to check whether all devices have been processed.

[0046] Once all devices are processed, the results received from the predictor and / or estimator are output in step 508, possibly after one or more statistical processes. This time, the results may include predictions and estimated predictions. Depending on the implementation, the output results may indicate predictions and estimated predictions, for example by using markers or different colors.

[0047] As is evident from the above examples, devices for which the measurement data required for remote monitoring is not available or there is not enough available measurement data, will also receive estimates of the predicted state. Only a small portion of the industrial devices configured to collect and transmit process data is needed to provide virtual remote monitoring for all industrial devices for which product data is available. Even if the estimated predictions (remote monitoring results) are estimated rather than based on actual process data, the estimates contribute to predictive maintenance, using estimates based on similar devices, conditions, etc., it is easier to determine which devices may be experiencing problems and what those problems may be, so that maintenance can be focused on those devices. The same applies to remote performance optimization. For example, using estimated predictions of energy consumption, it is easier to save energy.

[0048] Furthermore, there is no need to increase the size of the product database to include fields for the process data.

[0049] The above Figures 2 to 5The steps and related functions described in the description are not in absolute chronological order, and some steps may be performed simultaneously or in an order different from the given order. Other functions may also be performed between steps or within steps. Some of the steps or a portion of the steps may also be omitted or replaced by a corresponding step or a portion of the steps. In addition, the described processes and steps within the processes may run in parallel, and features from one process (function) may be combined into another process (function).

[0050] The techniques and methods described herein may be implemented by various devices such that the devices are configured at least in part according to the above combination Figures 1 to 5 Any of the disclosed contents provides virtual remote monitoring, or creates / updates a training model to provide an estimated prediction - including implementing the above, for example, by means of Figures 1 to 5 The device / equipment / means of any one of the corresponding devices / units described in conjunction with the embodiments / examples includes not only the devices of the prior art, but also the devices for implementing, for example, Figures 1 to 5 Any one of the above-described estimator units and / or trainer units and / or devices may be implemented in hardware (one or more devices), firmware (one or more devices), software (one or more modules), or a combination thereof.

[0051] Figure 6 is a simplified block diagram showing some units of an apparatus 600, the apparatus 600 being configured to provide virtual remote monitoring or a corresponding computing device with an estimator unit or corresponding one or more training models, and / or comprising at least the above referenced Figures 1 to 5 The trainer unit described, or, if the functions are distributed in the future, the device 600 corresponds to the functions or some corresponding functions of the corresponding functions. In the example shown, the device 600 includes: one or more interfaces (IF) 601 for retrieving data from one or more different databases and for receiving information from other devices and / or sending information to other devices; one or more user interfaces (U-IF) 601' for user interaction including different human-machine interfaces; one or more processors 602, which are configured to use reference Figure 1 , Figures 3 to 5 The estimator unit described above implements the functions of a virtual remote monitoring device or a corresponding computing device, and / or implements at least one of the above combined Figure 1 and Figure 2The apparatus or corresponding computing device / equipment of the described trainer unit, or if a distributed scenario is implemented, the corresponding algorithm 603 is used to implement at least a part of the corresponding function as a sub-unit function; and one or more memories 604 that can be used to store the computer program code (i.e., the algorithm for implementing the function) required for the apparatus or corresponding computing device / equipment of the estimator unit and / or the trainer unit. The memory 604 can also be used to store the training model and other information.

[0052] In other words, a device (device, apparatus) configured to provide virtual remote monitoring or a corresponding computing device with an estimator unit or a device including at least one trainer unit or configured to provide the above reference Figures 1 to 5 The device / equipment of one or more corresponding functions described may be any device or device or equipment or node configured to perform the above one or more corresponding functions of the embodiment / example / implementation, and may be configured to perform functions from different embodiments / examples / implementations. The required units and subunits, such as the estimator unit, may be separate units, even located in another physical device, the distributed physical device forms a logical device that provides the function, or is integrated into another unit in the same device.

[0053] A device (device, equipment) configured to provide virtual remote monitoring, or a corresponding computing device with an estimator unit, and / or a device including at least a trainer unit, or a device configured to provide one or more corresponding functions may generally include one or more processors, controllers, control units, microcontrollers, etc. connected to one or more memories and various interfaces of the device / equipment. Typically, the processor is a central processing unit, but the processor may be another operating processor. Each or some or one of the units / subunits and / or algorithms described herein may be configured as a computer or processor or microprocessor, such as a single-chip computer element, or configured as a chipset, including at least a memory for providing a storage area for arithmetic operations and an operation processor for performing arithmetic operations. Each or some or one of the above units / subunits and / or algorithms may include one or more computer processors, application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), graphics processing units (GPUs), logic gates and / or other hardware components that have been programmed and / or will be programmed by downloading computer program code (one or more algorithms) to perform one or more functions of one or more embodiments / implementations / examples in such a way. The embodiments provide a computer program included on any client-readable distribution / data storage medium or memory unit or article of manufacture, which includes program instructions that can be executed by one or more processors / computers, which when loaded into the device / apparatus constitute an estimator unit and / or a trainer unit or any subunit. The program, also referred to as a program product including software routines, program fragments constituting a "program function library", applets and macros, may be stored in any medium and may be downloaded to a device. In other words, each or some or one of the above-mentioned units / sub-units and / or algorithms may be an element including one or more arithmetic logic units, a plurality of special registers and a control circuit.

[0054] Furthermore, a device configured to provide virtual remote monitoring, or a corresponding computing device having an estimator unit, and / or a device comprising at least a trainer unit, or a device configured to provide a combination of the above Figures 1 to 5The device (device, equipment) of one or more corresponding functional devices described may generally include volatile and / or non-volatile memory, such as EEPROM, ROM, PROM, RAM, DRAM, SRAM, dual floating gate field effect transistors, firmware, programmable logic, etc., and generally store content, data, etc. The memory or multiple storages may be of any type (different from each other), have any possible storage structure, and if necessary, be managed by any database management system. In other words, the memory or a portion thereof may be any computer-usable non-transient medium inside or outside the processor / device, in which case it may be communicatively coupled to the processor / device via various means known in the art. Examples of external memory include removable memory that is detachably connected to a device, a distributed database, and a cloud server. The memory may also store computer program code such as software applications (e.g., for one or more of the units / subunits / algorithms) or operating systems, information, data, content, etc., for the processor to perform steps associated with the operation of the device according to the example / implementation.

[0055] It will be obvious to a person skilled in the art that, as technology advances, the inventive concept can be implemented in various ways.The invention and its embodiments are not limited to the examples described above but may vary within the scope of the appended claims.

Claims

1. A computer-implemented method for obtaining a prediction or estimated prediction of at least one state, the method comprising: receiving a user input including industrial device type information, the industrial device type information at least indicating at least one industrial device type; retrieving product data of one or more industrial devices from a product database including product data of a plurality of industrial devices using the industrial device type information, the one or more industrial devices having the same industrial device type as the at least one industrial device type indicated, wherein the product data includes at least one or more of a product type, a product configuration, a manufacturing date, a component in the product having a critical date of the component, an installation environment, and an expected load; After retrieving the product data, for each industrial device for which product data was retrieved, determining whether process data is available for the industrial device; In response to process data being available, performing remote monitoring analysis on the process data to obtain and output one or more predictions of the at least one condition; In response to no process data being available, the following is performed: inputting the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected from a subset of the plurality of industrial devices, a prediction of at least one condition generated by a remote monitoring analysis tool based on the process data, and product data for the subset of the plurality of industrial devices; processing the input product data into one or more estimated predictions of the at least one state through the training model; and The one or more estimated predictions of the at least one state are output.

2. The computer-implemented method of claim 1 , wherein: In response to the industrial device type information including one or more industrial device identifiers, the retrieving includes retrieving product data of the one or more industrial devices for which the industrial device identifiers are received.

3. The computer-implemented method of claim 1 , wherein: The remote monitoring analysis tool is a condition monitoring analysis tool, and the at least one state is a condition.

4. The computer-implemented method of claim 1 , wherein: The remote monitoring and analysis tool is a remote performance optimization tool.

5. A non-transitory computer-readable storage medium storing one or more instructions that, when executed by one or more processors, cause the one or more processors to at least perform the following operations: In response to user input including industrial device type information indicating at least one industrial device type, retrieving product data of one or more industrial devices from a product database including product data of a plurality of industrial devices using the industrial device type information, the one or more industrial devices having the same industrial device type as the indicated at least one industrial device type, wherein The product data includes at least one or more of a product type, a product configuration, a manufacturing date, a component in the product having a critical date for the component, an installation environment, and an expected load; After retrieving the product data, for each industrial device for which product data was retrieved, determining whether process data is available for the industrial device; In response to process data being available, performing remote monitoring analysis on the process data to obtain and output one or more predictions of at least one state; In response to no process data being available, the following is performed: inputting the retrieved product data into a training model, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected by a subset of the plurality of industrial devices, predictions of at least one condition generated by a remote monitoring analysis tool based on the process data, and product data for the subset of the plurality of industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of the at least one condition; and The one or more estimated predictions of the at least one state are output.

6. An apparatus for obtaining a prediction or estimated prediction of at least one state, the apparatus comprising: means for receiving user input including industrial device type information, the industrial device type information indicating at least one industrial device type; means for retrieving product data of one or more industrial devices from a product database including product data of a plurality of industrial devices using the industrial device type information, wherein the one or more industrial devices have the same industrial device type as the at least one industrial device type indicated, wherein the product data includes at least one or more of a product type, a product configuration, a manufacturing date, a component in the product having a critical date of the component, an installation environment, and an expected load; means for determining, after retrieving the product data, for each industrial device for which product data was retrieved, whether process data is available for the industrial device; means for performing remote monitoring analysis on the process data in response to the process data to obtain and output one or more predictions of the at least one condition; means for inputting the retrieved product data into a training model in response to no process data being available for the industrial device, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected from a subset of the plurality of industrial devices, predictions of at least one condition generated by a remote monitoring analysis tool based on the process data, and product data for the subset of the plurality of industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of the at least one condition; and Means for outputting the one or more predictions or the one or more estimated predictions of the at least one state.

7. A system for obtaining a prediction or estimated prediction of at least one state, the system comprising at least: one or more industrial installations; a product database comprising product data at least regarding the one or more industrial devices; as well as One or more devices, including: means for receiving user input including industrial device type information, the industrial device type information indicating at least one industrial device type; means for retrieving product data of one or more industrial devices from a product database including product data of a plurality of industrial devices using the industrial device type information, the one or more industrial devices having the same industrial device type as the at least one industrial device type indicated, wherein the product data includes at least one or more of a product type, a product configuration, a manufacturing date, a component in the product having a critical date of the component, an installation environment, and an expected load; means for determining, after retrieving the product data, for each industrial device for which product data was retrieved, whether process data is available for the industrial device; means for performing remote monitoring analysis on the process data in response to the process data to obtain and output one or more predictions of the at least one condition; means for inputting the retrieved product data into a training model in response to no process data being available for the industrial device, wherein the training model utilizes machine learning, and wherein the training model has been trained using process data collected from a subset of the plurality of industrial devices, predictions of at least one condition generated by a remote monitoring analysis tool based on the process data, and product data for the subset of the plurality of industrial devices, and wherein the training model processes the input product data into one or more estimated predictions of the at least one condition; and Means for outputting the one or more predictions or the one or more estimated predictions of the at least one state.

8. The system according to claim 7, wherein: The system further comprises: a process database, wherein at least one of the one or more industrial devices is configured to collect process data and transmit the process data to the process database; and One or more devices, including: means for retrieving process data collected on a plurality of industrial devices from the process database; means for retrieving product data for the plurality of industrial devices; means for performing remote monitoring analysis on the retrieved process data to obtain a prediction of at least one condition; and A device for training a machine learning model into the training model using the product data, process data and the prediction of the at least one state as training data by repeating the training until the machine learning model outputs an estimated prediction of the at least one state corresponding to the prediction of the at least one state obtained from the remote monitoring analysis by using the product data as input.

9. The system according to claim 7, wherein: The one or more industrial devices include one or more of a drive, a frequency converter, an AC / DC converter, a DC / AC converter, a DC / DC converter, a programmable logic controller, a switch, a motion controller, a motion drive, a servo motor, a soft starter, a wind turbine, and / or a solar inverter.

Images