Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
  • G06Q50/04—Manufacturing
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B19/00—Program-control systems
  • G05B19/02—Program-control systems electric
  • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
  • G05B19/4185—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by the network communication
  • G05B19/41855—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by the network communication by local area network [LAN], network structure
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B19/00—Program-control systems
  • G05B19/02—Program-control systems electric
  • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
  • G05B19/41885—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by modeling, simulation of the manufacturing system
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B23/00—Testing or monitoring of control systems or parts thereof
  • G05B23/02—Electric testing or monitoring
  • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N20/00—Machine learning
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q10/00—Administration; Management
  • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q10/00—Administration; Management
  • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
  • G06Q50/10—Services

Definitions

• the present teachings relate generally to computer-based optimization of an industrial plant. Background art
• Industrial plants such as process plants comprise equipment that is operated to produce one or more industrial products.
• the equipment may, for example, be machinery and/or heat exchang ers that require monitoring and maintenance.
• a requirement for maintenance can depend on several factors that include operation time and/or load on the equipment, environmental condi tions that the equipment has been exposed to, and so forth.
• Undue or unplanned shutdown of the equipment is generally not desired as it often results in a stoppage of production, which can reduce the efficiency of the plant and can cause wastage. Since the time period between two maintenances can vary, it may be difficult to plan the shutdown of the equipment around the time when the maintenance is actually necessary. Additionally, safety is of high importance in industrial plants. In order to prevent unsafe conditions from occurring, different equipment or parts of the plant may be monitored.
• an industrial plant such as a chemical plant may comprise equipment such as reactors, storage tanks, heat exchangers, compressors, valves, etc., which are monitored using sensors.
• Plant equipment is often sourced from multiple external suppliers or manufacturers of the equipment. Some suppliers may even generate a computer model of their equipment.
• the computer model may be useful for estimating the health of the equipment and for forecasting its maintenance requirements.
• the model may even be capable of determining op timized operating parameters in response to the specific inputs. The latter use can also be of significant interest to the plant operator as the operator can thus use the model to optimize the operation of the equipment or plant.
• the model may even be helpful in determining safe condi tions for the equipment.
• the model may however comprise proprietary information or intellectual property of the suppli er, which may not be desirable for the supplier to be exposed to a third-party.
• the plant operator is usually interested only in the use of the model, i.e., model output in response to specific in puts.
• a deep knowledge behind the logic of the model might not be required by the plant opera tor for the purpose of benefiting from the model.
• Generating an accurate model may require significant effort and resources, so if the supplier is reluctant to exposing their know-how to third parties, they may avoid developing good models to mitigate the risk of leaking knowhow to third parties.
• a similar problem can also exist from the perspective of the plant operator.
• the industrial plant or even a specific plant equipment may generate a large amount of data when deployed in the plant.
• the data may have been collected and stored in one or more databases over a long peri- od of time.
• the data often comprises valuable information regarding response of the plant and/or equipment to different conditions.
• the data can be used, for example, to train machine learning (“ML”) models.
• ML models can be usable for predictions, safety enhancement, optimization, etc.
• the plant operator may not be willing to provide access to their plant- and/or equipment- data to external parties, including the supplier.
• the interface comprises a process data signal that is transmitted to the API of the model im age, and a model result data signal that is transmitted from the API of the model image; wherein the process data signal comprises data related to the usage conditions of the asset, and the model result signal is indicative of the response of the asset to at least some of the usage conditions, wherein the user container runtime environment is located within a user network environment which is isolated from the supplier network environment.
• At least one estimated value related to the behavior of the asset can be determined, whilst preserving data protection of the model as well as process data received as the process data signal. Moreover, deployment of the model on the user side can also be simplified.
• the asset is an equipment, for example, reaction chamber, heat ex changer, compressor, valve, sensor, etc.
• the asset can even be a group of equipment, for example, steam-cracker, distillation unit, boiler, cooling tower, etc.
• the asset is a product that is produced by a plant.
• the product can for example be a chemical product, a pharmaceutical product, or even any material that is used as a raw material by another plant in a value-chain.
• the product can be a polymer, an additive, an emulsion, a catalyst, etc.
• the usage conditions may be the operating conditions for the equipment, such as inputs that the equipment receives.
• usage conditions or operating conditions can be input conditions such as inlet water temperature and inlet water flow rate.
• the operation conditions may also include ambient conditions such as ambient temperature.
• the operating conditions may also include control conditions, such as controller set-points, for example, required steam outlet flow, outlet temper ature, etc.
• the usage conditions may be parameters under which the product was produced.
• tolerances exist for the end-product or the product produced. As a result, there can almost always be variation in the quality of the product.
• the conditions or parameters under which the product is produced can vary dependent upon time, for example, variations in temperature, pressure, amount of raw materials and so forth during production and/or storage can have a bearing on the specifications that the end-product.
• each of the parameters may be independent of at least some of the others, which result in the variation in the end-product specifications.
• the tolerances are usually tight to make sure that the user of the product can get a product with consistent specifications. Tight toleranc es often mean a more expensive production process and/or more wastage as the product batch lying outside the tolerance range may have to be discarded.
• the usage conditions or production parameters of the product can be transmitted to a relevant user in the value chain.
• the user can thus adapt the process to be able to use the product more effectively. This can thus result in reduced cost and/or wastage by relaxing the tolerance range of the product.
• the usage conditions can thus be, concentration, density, weight, viscosity, strength, purity, or any such one or more parameters related to the product.
• the usage conditions may be conditions or parameters that influence the product.
• such con ditions can be one or more parameters such as: temperature, pressure, reaction time, or any other relevant parameter to which the product can be exposed to for use or further processing and/or for storage.
• the model result signal can thus be indicative of the response or behavior of the product when exposed to one or more usage conditions.
• the response can for example be indicative of one or more properties of the product, such as, composition, viscosity, density, structure, strength, etc.
• the result signal may be used to determine operating conditions for the asset, for example, by providing one or more control settings.
• the asset can thus be operated in an optimized manner via the determined operating conditions, for example achieved by the one or more control set tings.
• the method also comprises:
• This the containerized model may be used to either determine way in which the asset may be processed by performing the at least one control operation on the asset, or the asset may be used to perform the at least one control operation.
• the model can be used to leverage improvements at the user side without affecting data security of the model or that of the process data. It will be appreciated that the present teachings can thus provide a technical solution for en hancing data security not only for the supplier, but also for the user. Aspects of the teachings that work synergistically include the obfuscated containerized model, the container registry, and the API that is adapted to: receive the process data signal; and provide model result data signal. At least these features can prevent an unauthorized access to the model, and an unauthorized access to the process data.
• the term supplier is used to refer to the party that owns the model or the model logic.
• the term user is used to refer to the party that uses the model without requiring a detailed knowledge of the model logic.
• the supplier can hence be the manufacturer of the equipment, and the user can be the buyer of the equipment or the plant operator that uses the equipment.
• the user can even be a party that is considering sourcing the equipment from the supplier.
• the model can thus even be used by the user to evaluate the suit ability of the equipment on the user side prior to sourcing the equipment from the supplier. From the above it will be understood that when the asset is a product, the supplier may be the manu facturer of the product, and the user may be the buyer or prospective buyer of the product.
• the user container runtime environment is executed on a processing unit on the user side.
• the user network environment is isolated from the supplier network environment at least in terms of access, i.e., the user network does not have access rights to the supplier network, similarly the supplier network does not have access rights to the user network.
• the network environments can be physically isolated as different networks located at different locations.
• the networks may even be isolated using security features such as one or more firewalls and/or other security means. Each of the network environments are thus a secure network.
• a method for monitoring and/or controlling a plant equipment comprising:
• the interface comprises a process data signal that is transmitted to the API of the model im age, and a model result data signal that is transmitted from the API of the model image; wherein the process data signal comprises data related to the operating conditions of the equipment, and the model result signal is indicative of the response of the equipment to at least some of the operating conditions, wherein the user container runtime environment is located within a user network environment which is isolated from the supplier network environment.
• At least one estimated value related to the condition of the equipment can be determined.
• the container registry is preferably located within a shared environment.
• the shared environ ment is isolated from the user network environment and the supplier network environment at least in terms of access, i.e., the shared environment preferably has restricted or no access to the supplier network and/or the user network environments.
• the shared environment is however accessible by the from the user network environment and the supplier network environment.
• the user network environment may request access to the shared environment for re ceiving the image of the containerized model, the access is preferably secured via a multi-factor authentication, and/or any other secure access means to ensure that the container registry is not accessible by any unauthorized third party.
• the shared environment is located on the user side.
• the shared environment may be a virtualized isolated part of the user network environment.
• the shared environment may be part of a cloud service which is secured via a multi-factor authentication, and/or any other secure ac cess means to ensure that the container registry is not accessible by any unauthorized third party.
• the shared environment is located on the supplier side.
• the shared environment may be a virtualized isolated part of the supplier network environment.
• the applicant has realized the proposed teachings can provide a secure environment on the user side for the model that the supplier has developed for the asset.
• the risk of the logic of the model to be exposed on the user side can thus be reduced.
• Information or data security can thus be increased.
• any one or more of: the model execution logic, the process data signal and the model result signal can be secured on the user side without the information relat ed to these being transmitted to the supplier side.
• an always active connection between the supplier network environment and the user network connection can be avoided, which can further improve security and isolation between the two networks.
• the data related to the containerized model can either be an image of the containerized model itself which is transmitted to the container registry, or it may even be a flag signal indicating that a containerized model or a new version thereof is available.
• a signal may then be transmitted to the user container runtime environment, in response to which an image of the containerized model may be fetched either directly from model image already stored in the container registry, or the user environment may request that the model image be transmitted via the registry or the shared environment.
• the data connection to the shared environment may be disabled by either one or both by the user network environment and the supplier network envi ronment, and the flag signal may be cleared.
• the data connection may be reestablished in re sponse to a new flag signal indicating that a more recent version of the model is available.
• the flag signal may even exist as two separate signals, one each at the user side and the supplier side, i.e., a supplier flag signal indicating to the container registry that a new model is available, and a user flag signal for indicating to the user side that a new model is available in the contain er registry.
• the security can be synergistically improved both for the model logic from being exposed on the user side, and the model execution logic, process data and model results from being exposed outside the user network environment.
• the image of an obfuscated containerized model may be removed or deleted from the container registry after the image has been received at the user container runtime environment.
• the user side or the user container runtime environment may provide a receipt signal or flag that the image has been received, in response of which the image may be removed from the container registry.
• the image in the registry may be automati cally deleted by the registry, or the user side may delete upon receiving it at the user container runtime environment, or the supplier side may delete the image, for example in response to the receipt signal or flag. This can further protect the model from an unauthorized access.
• the obfuscated containerized model is provided via the container regis try by transmitting data from the supplier network environment.
• the data related to the model is pushed to the container registry.
• a method for transmitting a model of an asset comprising:
• the model is an obfuscated containerized model comprising an API
• the supplier container runtime environment is located within a supplier network environment, transmitting, to a container registry, data related to the containerized model
• the container registry is accessible by a user container runtime environment, located within a user network environment, for receiving an image of the containerized model, the API of the containerized model image being interfacable with a model execution logic via an interface that comprises a process data signal that is transmitted to the API of the model im age, and a model result data signal that is transmitted from the API of the model image; the process data signal comprising data related to the usage conditions of the asset, and the model result signal being indicative of the response of the asset to at least some of the us age conditions, wherein the supplier network environment is isolated from the user network environment.
• the containerized model comprises obfuscated model logic (e.g. in C).
• the API Application Programming Interface
• the API can e.g., be a standardized API that forwards veri fied requests to the obfuscated model logic.
• the API could thus be REST according to OpenA- Pl, gRPC, some MQTT preferably with specified and well-documented message bodies de pending on model behavior.
• supplier container runtime environment is executed on a processing unit on the supplier side.
• the API could apply additional means to hide the contents of the con tainerized model.
• Such means could, for example, be authentication mechanisms or passing embedded secrets to the obfuscated model binary.
• the containerized model also comprises a Machine Learning (“ML”) module.
• the interface can also comprise a training data signal for training the containerized model image.
• the ML module if the containerized model image can be trained on the user side by providing user training data to the model image via the training data signal.
• the interface further comprises a training result signal for receiving training result data from the model image on the user side.
• the model may be, or it may comprise as the ML module, a prediction model that when trained using on the user side by providing the user training data to the model image via the training data signal may result in a user trained data driven model.
• Data driven model refers to a model that is at least partially derived from data, in this case from the user training data that may comprise historical data related to the asset, e.g., a product or equipment.
• a data driven model can allow describing relations that cannot be modelled by physio-chemical laws.
• the use of data driven models can allow to describe relations without solving equations from physio-chemical laws, e.g., related to the processes taking place within the respective production process. This can reduce computational power and/or improve speed. Additionally, the supplier may not need to know the specifics of the user to provide such a model usable at the user side.
• the data driven model may be a regression model.
• the data driven model may be a mathemat ical model.
• the mathematical model may describe the relation between provided performance properties and determined performance properties as a function.
• the containerized model may include a supplier trained data driven model, i.e., a model that has been trained using supplier data, such as supplier historical data, from the sup plier side.
• the supplier trained data driven model can provide a more holistic prediction at the user side without requiring to expose the supplier data to the user.
• the data-driven model preferably data-driven machine learning (“ML”) model or a merely data-driven model
• ML machine learning
• An untrained mathematical model refers to a model that does not reflect reaction kinetics or physio-chemical processes, e.g. the untrained mathematical model is not derived from physical law providing a scientific generalization based upon empirical observation. Hence, the kinetic or physio-chemical properties may not be inherent to the un trained mathematical model. The untrained model does not reflect such properties.
• Feature en gineering and training with the respective training data sets enable parametrization of the un trained mathematical model.
• the result of such training is a merely data-driven model, prefera bly data-driven ML model, which as a result of the training process, preferably solely as a result of the training process, reflects reaction kinetics or physio-chemical processes related to the respective production processes.
• the containerized model may even be a hybrid model.
• a hybrid model may refer to a model that comprises first-principles parts, so called white box, as well as data-driven parts as explained previously, so called black box.
• the containerized model may comprise a combination of a white-box-model and a black-box-model and/or a grey-box-model.
• the white-box-model may be based on physio-chemical laws, for example expressed as equations.
• the physio-chemical laws may be derived from first principles.
• the physio-chemical laws may comprise one or more of chemical kinetics, conservation laws of mass, momentum and energy, particle population in arbitrary dimension.
• the white-box-model may be selected according to the physio-chemical laws that govern the respective production process or parts thereof.
• the black-box-model may be based on historical data, such as the supplier historical data and/or the user historical data.
• the black-box-model may be built by using one or more of machine learning, deep learning, neural networks, or other form of artificial intelligence.
• the black-box-model may be any model that yields a good fit between the training data set and test data.
• the grey-box-model is a model that combines a partial theoretical structure with data to complete the model.
• the trained model may comprise a serial or parallel architecture.
• serial architecture out put of the white-box-model may be used as input for the black-box-model or output of the black- box-model may be used as input for the white-box-model.
• parallel architecture a com bined output of the white-box-model and the black-box-model may be determined such as by superposition of the outputs.
• a first sub-model may predict at least one of the performance parameters and/or at least some of the control settings based on a hy brid model with an analytical white-box-model and a data-driven model that serves as a black box corrector trained on the respective historical data.
• This first sub-model may have a serial architecture, wherein the output of the white-box-model is input for the black-box-model, or the first sub-model may have parallel architecture. Predicted output of the white-box model may be compared with a test data set comprising a part of historical data. An error between the com puted white-box output and test data can be learned by the data-driven model and can then applied for arbitrary predictions.
• the second sub-model may have a parallel architecture. Other examples can be possible, too.
• machine learning may refer to a statistical method that ena bles machines to “learn” tasks from data without explicitly programming.
• Machine learning tech niques may comprise “traditional machine learning” — the workflow in which one manually se lects features and then trains the model. Examples of traditional machine learning techniques may include decision trees, support vector machines, and ensemble methods.
• the data driven model may comprises a data driven deep learning model. Deep learning is a subset of machine learning modeled loosely on the neural pathways of the human brain. Deep refers to the multiple layers between the input and output layers. In deep learning, the algorithm automatically learns what features are useful. Examples of deep learning techniques may in clude convolutional neural networks (“CNNs”), recurrent neural networks such as long short term memory (“LSTM”), and deep Q networks.
• CNNs convolutional neural networks
• LSTM long short term memory
• model execution logic shall be clear as such to those skilled in the art. The term re fers to any suitable hardcoded and/or software-based logic that can perform functions as de scribed herein. Dependent upon the model type and the use at the user side, a suitable model execution logic is interfaced with the API of the image. As it was discussed, the model execution logic may be used to provide the process data signal to the model image. Industrial plants can be data heavy environments, so the model execution logic can be used to provide relevant pro cess data to the model image. The model execution logic may also be used for training the con tainerized model. Furthermore, the model execution logic may also be used to receive the result data signal. Those skilled in the art will fully appreciate that the specifics of the model execution logic are not limiting to the generality of the present teachings.
• API or application programming interface is a computing interface which is used to define interactions between multiple software intermediaries.
• the API is used for interfacing the model image with the model execution logic.
• the API can further allow preventing exposing the model image on the user side by managing interactions between the model image and the model exe cution logic.
• the model execution logic receives process data from a process histori an (e.g. OSISoft PI) and transmits at least some of the process data via the process data signal to the model image API.
• the model execution logic then receives process results via the model result signal. At least some of the results can be transmitted back to the process historian, for example for further analysis and/or storage in a database.
• the user container runtime environment may be the same type as the supplier runtime envi ronment, or they may be different types yet compatible environments such that the model can be executed on both environments.
• the presence of a shared environment can further allow for more flexibility between the user runtime environment and the supplier runtime environment.
• runtime environment shall be clear as such to those skilled in the art. Flowever, as an example, the term may refer to a system or framework that provides an environment in which computer programs can run.
• network environment shall also be clear as such to those skilled in the art.
• the term may refer to a system, framework or infrastructure that comprises one or more data networks any suitable kind of data transmission medium, wired, wireless, or their combination.
• a specific kind of network is not limiting to the scope or generality of the present teachings.
• the network can hence refer to any suitable arbitrary interconnection between at least one communication endpoint to another communication endpoint.
• Network may comprise one or more distribution points, routers or other types of communication hardware.
• the inter connection of the network may be formed by means of physically hard wiring, optical and/or wireless radio-frequency methods.
• the network specifically may be or may comprise a physical network fully or partially made by hardwiring, such as a fiber-optical network or a network fully or partially made by electrically conductive cables or a combination thereof.
• a containerized model proposed herein may be realized by a virtualized machine environment such as those using Operating System (“OS”) level virtualization.
• the containerized model is realized using a container, for example a Docker container.
• a container is software package that often bundle its own software, libraries and configuration files.
• a con tainer is usually isolated from other containers, but the containers can communicate with each other through well-defined channels.
• the present teachings can be particularly suitable for ap plication in a value chain or even in serial production where an asset, or product, which is pro prised by a first plant is used by a second plant.
• the second plant can better adapt the production or process that utilizes the product according to the properties of the product that has been supplied to the second plant. This can result in one or more of: reduced costs, reduced wastage, and improved quality of the overall process involving the first and the second plant.
• the number of plants in the value chain can be more than two. Or more general ly, a user may be a supplier for another user downstream of the value chain, and so forth.
• Industrial plants or simply plants, comprise infrastructure that is used for an industrial purpose.
• the industrial purpose may be manufacturing of one or more products, i.e., process manufac turing done by a process plant.
• the product can, for example, be any product, such as a: chem ical, biological, pharmaceutical, food, beverage, textile, metal, plastic, semiconductor.
• the plant can be any or more of the: chemical plant, pharmaceutical plant, fossil fuel facili ty such as oil and/or natural gas well, refinery, petrochemical plant, cracking plant, fracking facil ity, and such.
• the product may even be an intermediate product which is used to produce an end-product, either in the same plant or by a user downstream of the value chain.
• the plant can even be any of the: distillery, incinerator, or power plant.
• the plant can even be a combination of any of the above, for example, the plant may be a chemical plant that includes a cracking facility such as a steam cracker, and/or a power plant.
• the plant also comprises instrumentation that can include several different types of sensors for monitoring the plant parameters and equipment. At least part of the process data is generat ed via the instrumentation such as sensors.
• the method also comprises:
• the interface between the API of the model image and the API of the user model can thus be used for creating a model for the asset that is to be provided to an another user.
• the user model and the model image are contained in a user container composition.
• the user container composition is a well-defined set of containers that are speci fied to interact with each other in a certain way via the interface.
• the container composition can, for example, be a docker-corn pose file or a kubernetes deploy manifest.
• the interface comprises a process data signal that is transmitted to the API of the model im age, and a model result data signal that is transmitted from the API of the model image; wherein the process data signal comprises data related to the usage conditions of the asset, and the model result signal is indicative of the response of the asset to at least some of the usage conditions,
• model image is an obfuscated containerized model comprising an API, which model image has been received from a supplier container runtime environment via a container registry, and wherein the intermediate user container runtime environment is locat ed within an intermediate user network environment,
• the intermediate user container registry is accessible by a downstream user container runtime environment, located within a downstream user network environment, for receiving an image of the augmented model, the API of the augmented model image being interfacable with an augmented model execution logic via an interface that comprises a downstream user pro cess data signal that is transmitted to the API of the augmented model image, and a down stream user model result data signal that is transmitted from the API of the augmented model image; the downstream user process data signal comprising data related to the usage condi tions of the intermediate user asset, and the downstream user model result signal being indica tive of the response of the intermediate user asset to at least some of the usage conditions, wherein the supplier network environment, the intermediate user network environment, and the downstream user network environment are isolated from each another.
• the intermediate user container registry and the container registry can either be the same regis try which are located in a shared environment, or they may be different registries that are either located in the same shared environment, or in different environments shared between the sup plier and the intermediate user, and between the intermediate user and the downstream user respectively.
• a downstream user model execution logic is interfaced to the API of the augmented model image, either via the API of the intermediate user model, or via the API of the user model comprised in the augmented model image.
• the downstream user can then pass downstream user process data to the augmented model image similar to as was explained ear lier.
• the interface may also comprise a downstream model result signal.
• An advantage of the above teachings can be that the intermediate user can pass process data to the API of the of the containerized model image and use the model results to evaluate per formance at their end without needing to pass proprietary data to the downstream user.
• the downstream user can still benefit from the augmented model which comprises both the model of the asset as well as the contribution from the intermediate user.
• a computer program com prising instructions which, when the instructions are executed by a suitable computer processor, cause the processor to carry out the method steps herein disclosed.
• a non-transitory computer readable medium storing a program causing a suitable computer pro cessor to execute any method steps herein disclosed.
• a computer-readable data medium or carrier includes any suitable data storage device on which is stored one or more sets of instructions (e.g., software) embodying any one or more of the methodologies or functions described herein.
• the instructions may also reside, completely or at least partially, within the main memory and/or within the processor during execution thereof by the computer system, main memory, and processing device, which may constitute computer- readable storage media.
• the instructions may further be transmitted or received over a network via a network interface device.
• the computer program for implementing one or more of the embodiments described herein may be stored and/or distributed on a suitable medium, such as an optical storage medium or a solid state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the internet or other wired or wireless telecommunication systems.
• a suitable medium such as an optical storage medium or a solid state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the internet or other wired or wireless telecommunication systems.
• the computer program may also be presented over a network like the World Wide Web and can be downloaded into the working memory of a data processor from such a net work.
• a data carrier or a data storage medium for making a computer program element available for downloading can be also provided, which computer program element is arranged to perform a method according to one of the previously described embodiments.
• FIG. 1 shows a block-diagram of a network arrangement
• FIG. 2 shows a block-diagram of a network arrangement applied to a supply chain
• FIG. 3 shows a flow-chart depicting certain aspects of the present teachings
• FIG. 1 shows a network arrangement 100 pursuant to an aspect of the teachings.
• the receiving and transmitting parts are shown together. The method aspects will be explained below with reference to the network arrange ments.
• the arrangement 100 comprises a containerized model 101 which is provided within a supplier container runtime environment 105.
• the supplier container runtime environment 105 is located within a supplier network environment 109.
• the supplier network environment 109 in some cas es may be referred to as belonging to the supplier side.
• Supplier is used to refer to the party that has developed the model 101 and/or is interested in securing the logic of the model 101 from being exposed to third parties.
• the arrangement 100 also comprises a container registry 150 which is located in a shared envi ronment 159.
• the shared environment 159 is shown as a part of a cloud ser vice. However, the shared environment 159 in some cases may even be realized on the suppli er side.
• data related to the model 101 can be provided to the container registry 150.
• the data may either be an image or copy of the model 101, and/or it may be one or more flags indicating that a new version of the model 101 is available on the supplier side.
• a communication link may be a network connection, ei ther wired or wireless.
• the communications link may be at least partially be an in ternet connection, or in some cases it may even be an intranet connection.
• the supplier com munication link 121 may even be a mobile network supporting computer data transfer.
• the communication via the supplier communication link 121 may even be via a secure connection such as a Virtual Private Network (“VPN”) connection.
• VPN Virtual Private Network
• the containerized model 101 may be obfuscated for security, either within the supplier contain er runtime environment 105, or when transmitting data to or via the container registry 150.
• the containerized model 101 also comprises an API 102.
• the image 111 is an obfuscated containerized model image that has been transmitted via the container registry 150 through a user communication link 122.
• the user communication link 122 may even be either wired and/or wireless. Accordingly, the user communications link 122 may also be at least par tially be an internet connection, or in some cases it may even be an intranet connection.
• the user communication link 122 may even be a mobile network supporting computer data transfer.
• the communication via the user communication link 122 may even be via a secure connection such as a Virtual Private Network (“VPN”) connection.
• VPN Virtual Private Network
• the model image 111 also comprises a model image API 112 that is a copy of the model API 102.
• the model image API 112 is provided with an interface that comprises a process data sig nal 141 and a model result signal 142.
• the process data signal 141 is used to send data related to usage conditions of an asset.
• the asset is usually on the user side, so the user is often reluc tant to share process data generated by the asset with third parties.
• the process data often contains valuable information about the asset and the plant that comprises the asset.
• Such in formation can, for example, be used to optimize the performance of the asset and/or the plant, which can be used by the user for lowering costs and/or improving quality of products manufac tured by the asset or the plant.
• the model result signal 142 can provide such estimation of per formance by indicating the response of the asset to at least some of the usage conditions as specified by the process data signal 141.
• the user container runtime environment 115 is locat ed within a user network environment 119 which is isolated from the supplier network environ ment 109, i.e., the network environments 109 and 119 are isolated from each another. Being isolated means that resources on either network environments 109 and 119 do not have access rights to the other of the network environments.
• the proposed architecture and method of transmitting and/or receiving the model can thus allow improved security for the supplier and the user for their proprietary intellectual property.
• the supplier can provide a better model for the user’s purpose and the user can use the model with both parties enjoying better security.
• the process data signal 141 and the model result signal 142 are handled by a model execution logic 140, which the user can specify for feeding process data to the model image 111 and in terpreting model results.
• the interface may also comprise a training data signal 143 and a training result sig nal 144.
• the training data signal 143 may be used for training the model image 111 by providing training data on the user side.
• the training data can for example be historical data related to the asset, which is used to train a machine learning module in the model image 111.
• the training result signal 144 may be used for receiving training result data from the model image 111. The user can thus better adapt the model image 111 for modeling the asset on the user side.
• the training signals 143 and 144 can also be handled, as shown, via the model execution logic 140.
• FIG. 2 shows another network arrangement 200 pursuant to the present teachings.
• the receiving and transmitting parts are shown togeth er, which will also be used to explain the method aspects.
• the reference numbers from FIG. 1 are reused to refer to the components that can be the same as, or are similar to, those in the supplier-user example shown in FIG. 1.
• the containerized model 101 which is shown provided within the supplier container runtime environment 105.
• the sup plier container runtime environment 105 is located within the supplier network environment 109.
• the container registry 150 is shown as a common registry located in the shared envi ronment 159 in cloud. As it was discussed previously, the registry may even be split between sections of the supply chain, i.e., specific user-supplier groups.
• the shared environment 159 may either be a common environment as shown or different environments for different groups.
• the user network environment 119 in FIG. 2 can be considered as an intermediate user network environment.
• the image 111 of the containerized model 101 is shown provided in the user container runtime environment, which here can be referred to as an intermediate user container runtime environment 115.
• the intermediate user container runtime environment 115 is located within the intermediate user network environment 119.
• the intermediate user model execution logic 140 can be interfaced with the model image 111 for signals such as process- and/or training- data and results. The signals are not shown individually in FIG. 2, rather in a form of an intermediate logic bus 241.
• the intermediate user may be using the asset for producing an intermediate user asset, for ex ample by using the asset as a starting point for manufacturing the intermediate user asset.
• the intermediate user can cre ate an augmented model 222.
• the augmented model 222 comprises the model image 111 and an obfuscated containerized intermediate user model 261.
• the obfuscated containerized inter mediate user model has an intermediate API 262 that is used to interface the containerized in termediate user model 261 with the model image 111 , for example via an intermediate model data signal 225.
• the augmented model 222 may be comprised in a container composition.
• the intermediate user may transmit the augmented model 222 via an intermediate user con tainer registry, which in the example of FIG. 2 is shown as being the same as the container reg istry 150.
• a downstream user can which has access to the shared registry 150 can be provided an image of the augmented model 222 via a downstream communications link 223.
• the communication links were explained earlier with reference to FIG.1 .
• the downstream can optionally append the model further by creating a further augmented model 232. This may be the case when the downstream user wants to provide the further augmented model 232 to another user, for exam ple, another downstream user.
• the further augmented model 232 comprises an image of the augmented model 222, which is shown as model images 211 and 271 that correspond to con tainerized model image 111 and the containerized intermediate user model 261 respectively.
• the model images 211 and 271 are interfaced by signal 235 which corresponds to the interme diate model data signal 225.
• the downstream user creates a containerized downstream user model 281 which is interfaced via its API 282 to the augmented model image via a downstream model data signal 236.
• the further augmented model 232 is provided in a downstream contain er runtime environment 215 which is located within a downstream user network environment 219.
• a downstream logic bus 251 is shown, which may be similar to the intermediate logic bus 241 as discussed previously.
• the corresponding downstream process data signal and down stream model result signal in the downstream logic bus 251 may be handled by a downstream model execution logic 240.
• the downstream model execution logic 240 may be different from the model execution logic 140, or they may be similar.
• the downstream model image 211 also comprises a downstream model image API 212 that may be a copy of the model API 102.
• the proposed teachings can provide a secure way to provide models within a value chain.
• the further augmented model 232 can just be an image of the augmented model 222.
• model image and model execution logic on the respective sides are being executed on a pro cessing unit, which can either be the same processor or a distributed processing environment.
• a pro cessing unit which can either be the same processor or a distributed processing environment.
• the containerized model is generated using a processing unit on the supplier side.
• a specific architecture of the processing units is not essential to the scope or generality of the present teachings.
• FIG. 3 shows a flow chart 300 pursuant to the present teachings, e.g., illustrating a computer routine.
• it is received at the user container runtime environment 115 the image 111 of an obfuscated containerized model of an asset.
• the image 111 is provided via the container registry 150, which may be a part of the shared environment 159.
• the containerized model 101 comprises an API 102, so the image 111 also comprise an API 112.
• the containerized model 101 has been generated in the supplier runtime environment 105 located within the supplier network environment 109.
• it is interfaced with the model execution logic 140 the API 112 of the containerized model image 111.
• the interface comprises the process data signal
• the process data signal 141 comprises data related to the usage conditions of the asset, and the model result signal 142 is indicative of the response of the asset to at least some of the usage conditions.
• the user con tainer runtime environment 115 is located within the user network environment 119 which is isolated from the supplier network environment 109.
• the result signal 142 is used to determine operating conditions for the asset, which operating conditions may be used to determine one or more control settings for the asset.
• the control settings may be used to operate the asset for achieving optimized operating conditions, or the control settings may be related to processing the asset in an optimized manner, for example for achiev ing a desired performance of the asset.
• the asset may be a product such as a chemical product, or the asset may be an equipment.
• a similar flow-chart may be obtained for the supplier side when following the method aspects as discussed in the present disclosure.
• the model 101 in a first step it can be provided within the supplier container runtime environment 105, the model 101 of the asset.
• the model 101 is an obfuscated containerized model comprising the API 102.
• it is transmitted to the container registry 150 data related to the containerized model 101.
• the container registry 150 is accessible by a user container runtime environment 115 located within a user network environment 119.
• the container registry 150 is used for receiving the image 111 of the contain erized model.
• the API 112 of the containerized model image 111 is interfacable with the model execution logic 140 via the interface that comprises the process data signal 141 that is transmit ted to the API 112 of the model image 111 , and a model result data signal 142 that is transmit ted from the API 112 of the model image 111.
• the model result signal 142 is indicative of the response of the asset to at least some of the usage conditions.
• a method or computer-implemented method comprising: receiving at a user container runtime environment, via a container registry, an image of an obfuscated containerized model of an asset; wherein the containerized model comprises an API, and wherein the containerized model has been generated in a supplier runtime environ ment located within a supplier network environment, interfacing, with a model execution logic, the API of the containerized model image, wherein the interface comprises a process data signal that is transmitted to the API of the model image, and a model result data signal that is transmitted from the API of the model image; wherein the process data signal comprises data related to the usage conditions of the asset, and the model result signal is indicative of the response of the asset to at least some of the us age conditions, wherein the user container runtime environment is located within a user network environment which is isolated from the supplier network environment.
• a method or computer-implemented method for transmitting a model of an asset comprising:
• model is an obfuscated containerized model comprising an API
• supplier container runtime environment is located within a supplier network environment
• the container registry is accessible by a user container runtime environment, located within a user network environment, for receiving an image of the containerized model, the API of the containerized model image being interfacable with a model execution logic via an inter face that comprises a process data signal that is transmitted to the API of the model image, and a model result data signal that is transmitted from the API of the model image; the process data signal comprising data related to the usage conditions of the asset, and the model result signal being indicative of the response of the asset to at least some of the usage conditions, wherein the supplier network environment is isolated from the user network environment.
• a method or computer-implemented method for transmitting an augmented model comprising:
• model image is an obfuscated containerized model comprising an API, which model image has been received from a supplier container runtime environment via a container registry, and wherein the intermediate user container runtime environment is locat ed within an intermediate user network environment,
• the intermediate user container registry is accessible by a downstream user container runtime environment, located within a downstream user network environment, for receiving an image of the augmented model, the API of the augmented model image being interfacable with an augmented model execution logic via an interface that comprises a downstream user pro cess data signal that is transmitted to the API of the augmented model image, and a down stream user model result data signal that is transmitted from the API of the augmented model image; the downstream user process data signal comprising data related to the usage condi tions of the intermediate user asset, and the downstream user model result signal being indica tive of the response of the intermediate user asset to at least some of the usage conditions, wherein the supplier network environment, the intermediate user network environment, and the downstream user network environment are isolated from each another.
• a system comprising at least one computer processor which is configured to carry out the method steps of any of the clauses 1 - 13.
• Clause 15 A computer program product comprising instructions which, when executed by a suitable com puter processor, cause the processor to carry out the method steps of any of the clauses 1 - 13.

Landscapes

• Engineering & Computer Science (AREA)
• Business, Economics & Management (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Strategic Management (AREA)
• Economics (AREA)
• Human Resources & Organizations (AREA)
• Theoretical Computer Science (AREA)
• Manufacturing & Machinery (AREA)
• Quality & Reliability (AREA)
• Tourism & Hospitality (AREA)
• Marketing (AREA)
• General Business, Economics & Management (AREA)
• Entrepreneurship & Innovation (AREA)
• Automation & Control Theory (AREA)
• General Engineering & Computer Science (AREA)
• Development Economics (AREA)
• Operations Research (AREA)
• Game Theory and Decision Science (AREA)
• Health & Medical Sciences (AREA)
• General Health & Medical Sciences (AREA)
• Primary Health Care (AREA)
• Software Systems (AREA)
• Educational Administration (AREA)
• Evolutionary Computation (AREA)
• Data Mining & Analysis (AREA)
• Medical Informatics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Computing Systems (AREA)
• Mathematical Physics (AREA)
• Artificial Intelligence (AREA)
• Management, Administration, Business Operations System, And Electronic Commerce (AREA)
• Debugging And Monitoring (AREA)
• Stored Programmes (AREA)

Priority Applications (5)

Applications Claiming Priority (2)

Publications (1)

Family

ID=68917385

Family Applications (1)

Country Status (6)

Families Citing this family (5)

Citations (2)

Family Cites Families (7)

• 2019
  • 2019-12-13 EP EP19216211.3A patent/EP3836051A1/en not_active Withdrawn
• 2020
  • 2020-12-08 US US17/782,765 patent/US12346098B2/en active Active
  • 2020-12-08 WO PCT/EP2020/085133 patent/WO2021116126A1/en not_active Ceased
  • 2020-12-08 CN CN202080085407.2A patent/CN114830155A/zh active Pending
  • 2020-12-08 EP EP20819761.6A patent/EP4073722A1/en not_active Withdrawn
  • 2020-12-08 KR KR1020227019640A patent/KR20220114549A/ko active Pending
  • 2020-12-08 JP JP2022535731A patent/JP7755580B2/ja active Active

Patent Citations (2)

Also Published As

Similar Documents

Legal Events