Industrial plant monitoring

Abstract

The present teachings relate to a method for integrating asset data from an industrial asset located within an industrial plant, the asset being communicatively coupled to a first processing layer, wherein the asset data is provided to a second processing layer via the first processing layer; the first processing layer is communicatively coupled to the second processing layer, and wherein the first processing layer and the second processing layer are configured in a secure network, the method comprising generating, at the second processing layer, technical context data related to the asset; providing, via the second processing layer, the technical context data to an interface to an external network; wherein the technical context data comprises one or more accessibility criteria for the asset data; the accessibility criteria comprising one or more rules and / or parameters to be adhered to by an external processing layer for receiving the asset data. The present teachings also relate to a method for preprocessing asset data, a computer software product, and an industrial plant system.

Description

TECHNICAL FIELD

[0001] The present teachings generally relate to computer-based monitoring and / or optimization of industrial plants. BACKGROUND

[0002] Industrial plants, such as process plants, include equipment that is operated to produce one or more industrial products. For example, the equipment can be machinery and / or heat exchangers that require monitoring and maintenance. Maintenance requirements can depend on several factors, including the operating time and / or load of the equipment, environmental conditions to which the equipment is exposed, etc. It is generally undesirable for the equipment to be shut down excessively or unscheduled, as it typically results in a stoppage of production, which can decrease the efficiency of the plant and can result in waste. Since the time period between two maintenances can vary, it can be difficult to schedule a shutdown of the equipment around the time when maintenance is actually required. Furthermore, safety is very important in industrial plants. To prevent unsafe situations from occurring, different equipment or portions of the plant can be monitored.

[0003] For example, an industrial plant, such as a chemical plant, can include equipment, such as reactors, storage tanks, heat exchangers, compressors, valves, etc., that are monitored using sensors. At least some of the equipment can be monitored and / or controlled to produce one or more industrial products. The monitoring and / or control can even be performed to optimize the production of the one or more products.

[0004] Industrial plants typically include a plurality of sensors that are distributed throughout the plant for monitoring and / or control purposes. Such sensors can generate a large amount of data. Thus, production, such as chemical production, can be a data-intensive environment. However, such data from multiple data sources has not been fully leveraged to date to increase the production efficiency of one or more plants.

[0005] Furthermore, data sources, such as sensors, can sometimes be distributed over or located remotely from multiple sites. In some cases, a site can be located in another city or even another country from one or more other sites. As can be appreciated, it can be challenging to utilize data from one or more data sources that are located remotely.

[0006] It is very attractive for industrial plants to apply new technologies in cloud computing and big data analytics. Unlike some other manufacturing industries, the process industry can follow higher safety standards. For this reason, the computing infrastructure is typically isolated, with highly restricted access to the monitoring and control systems. Due to this safety standard, latency and availability considerations can be at odds with a simple migration of embedded control systems to, for example, cloud computing systems. Thus, bridging the gap between highly proprietary industrial manufacturing systems and cloud technologies is one of the challenges in the process industry.

[0007] Thus, there is a need for a method for monitoring and / or analyzing data from an industrial plant, which can allow for a scalable and flexible integration of data from distributed data sources. SUMMARY

[0008] At least some of the problems inherent in the prior art will be solved by the subject matter of the independent claims appended hereto.

[0009] From a first point of view, a method for integrating asset data from an industrial asset located within an industrial plant can be provided, the asset being communicatively coupled to a first processing layer, wherein the asset data is provided to a second processing layer via the first processing layer; the first processing layer being communicatively coupled to the second processing layer, and wherein the first processing layer and the second processing layer are configured in a secure network, the method comprising:

[0010] generating, at the second processing layer, technical context data related to the asset;

[0011] providing, via the second processing layer, the technical context data to an interface to an external network;

[0012] wherein the technical context data comprises one or more accessibility criteria of the asset data; the accessibility criteria comprising one or more rules and / or parameters that can be adhered to by an external processing layer for receiving the asset data.

[0013] The method for integrating asset data as expressed above can also be understood as a data integration method of asset data, or even as a method for monitoring and / or analyzing at least a portion of asset data, wherein the asset data has been generated via an industrial asset within an industrial plant; the asset being communicatively coupled to a first processing layer, and wherein at least a portion of the asset data is provided to a second processing layer via the first processing layer; the first processing layer being communicatively coupled to the second processing layer, and wherein the first processing layer and the second processing layer are configured in a secure network.

[0014] The presented teachings can allow for a more efficient data processing via the second processing layer. Seamless data access can even be provided in different processing layers and cloud connections. The present teachings can bridge the gap between operational technology and information technology in highly advanced process industry environments. By aggregating contextual data in separate layers, the availability and performance of the first layer is not affected. Furthermore, data exchange with external processing layers outside of the secure network can be allowed while adhering to high safety standards of the chemical industry. By introducing different processing and storage system layers and communicatively coupling them, a large amount of data transmission and processing can be distributed over different layers, allowing for greater flexibility in the contextualization, storage and access of process applications. In some cases, even multiple plants can be accommodated via the second processing layer. Thus, a highly scalable, more reliable and more enhanced monitoring and / or control of industrial plants can be achieved. This way, even new technologies like serverless IaaS / PaaS / SaaS can be integrated into production environments, allowing for continuous application delivery and deployment. It should be understood that besides the first and second processing layer, even one or more further processing layers can exist. They can even be one or more additional processing layers between the first and second processing layer.

[0015] The present teachings can also allow for a more flexible handling of process applications. For example, even in multiple plants, the deployment of process applications that acquire plant or asset data can be streamlined for multiple assets. Furthermore, depending on the specific data required by a process application and the computational resources required to run such an application, the appropriate processing layer can be selected, adhering to high availability standards in certain industrial plants, such as chemical plants. For example, computationally intensive process applications that require the acquisition of plant specific data can be executed on the second processing layer, while process applications that require the acquisition of asset data and require low latency can be executed on the first processing layer.

[0016] In industrial plants, such as chemical production plants, there can be multiple distributed data sources. Some of these data sources or assets can also be distributed globally. Typically, the assets provide data that can be stored in a system or database, such as a plant information management system (“PIMS”). The plant typically also has a supervisory control and data acquisition (“SCADA”) system. The PIMS can be part of the SCADA, or they can be different systems. In this disclosure, when referring to the term PIMS, the PIMS and SCADA are alternatives of the same system are also assumed to be included within the scope of the term PIMS. Similarly, for the cases where these are two different systems, the data can be transmitted to one or both of them, so these alternatives should also be understood to be within the scope of the term. In some cases, a data source or asset can contain or produce more data than provided to the SCADA or PIMS. Thus, the internal data of the data source can be more extensive than the data sent to and / or stored in the PIMS. This data internal to the asset or asset data can be valuable for data analysis. For example, a change in internal data can be normal for the asset itself, but it can cause or indicate an important impact on another part of the plant that can not be apparent by observing the internal data and / or by observing the output data provided to the PIMS alone. Furthermore, if similar assets are operating in another plant, data from similar assets can also be used to optimize one or more such assets in the other plant. Optimization in this sense can mean any one or more of the following: training of a machine learning model, monitoring output signals, deriving one or more control or monitoring set points, and such functions. Thus, data from one asset can be leveraged to optimize other similar assets. Therefore, integrating asset data from different processing layers and / or assets can be valuable for process analysis.

[0017] To perform data analysis on data including asset data, a user operatively connected to the external processing layer can need to define the data analysis. The user can be a person responsible for performing the data analysis via the external processing layer, or it can be another computer processor that provides data from the external layer, or the user can be another computer processor itself for automatically running the data analysis using the data analysis. The user can even be the external processing layer itself with the capability to run the data analysis. Hereinafter, the term “user” will be used to refer to any of the above definitions or any combination thereof.

[0018] A user can not be aware of the technical state of assets within an industrial plant. Industrial assets of a plant (or within a plant) or more generally referred to as assets, and the network infrastructure of a plant are typically secure environments that do not have or have limited access to data of the plant outside. A user with proper access rights can be provided with access to asset data generated by one or more assets in the plant, however the user can still lack information about other operational parameters within the plant network. Such operational parameters can include resource utilization, such as network load, central processing unit ("CPU") and / or controller load, memory utilization, and power usage within the infrastructure inside the plant, etc. In some cases, a user request can overload at least a portion of the plant infrastructure. For example, when a user request accesses or transfers asset data at the same time that a critical plant operation requires some of the same resources within the plant. In particular, if the request is resource intensive, the performance of the critical plant operation can be impacted due to the processing of the request. In a worse scenario, the plant infrastructure can crash due to the nature of the resource requirements of the user request. The safety and / or efficiency of the plant can thus be compromised. To prevent this, the present teachings propose to generate technical context data related to the asset at a second processing layer. The technical context data can then be provided to an external network via an interface, from which it can be provided to an external processing layer, for example, from which asset data can be requested. Mediation of the request via the second processing layer can prevent the first processing layer from becoming overloaded.

[0019] According to one aspect, the technical context data is generated at least partially using at least one of an asset network address, an asset CPU load, an asset memory such as random access memory ("RAM"), and at least a portion of a network path between the asset and the user. More generally, the a priori determined parameter can be any one or more characteristics of one or more computing resources and / or network resources between the asset and the user. Such parameters preferably indicate a physical limit of one or more computing resources and / or network resources between the asset and the user, such as available memory, network capacity, processing power, etc. The a priori parameters can be determined from current operating conditions of the one or more resources, or they can be determined from one or more past accesses and / or transfers of data from or around the asset.

[0020] According to one aspect, the a priori determined parameter is indicative of a temporal network load value, e.g. between assets and users, or between any parts of the network, which determines the network capacity for transferring and / or accessing asset data from the assets to the external processing layer or users. For example: in the morning of almost every workday, the network load can increase due to employees accessing the site or plant network substantially at the same time. During such high resource demand periods, the network can be heavily loaded, e.g. due to a large number of computers being booted up and starting applications requiring network access between each other within a short time period. Due to the excessive load, the network can fail. During such periods, the transfer and / or access of asset data can be affected by low throughput, or even can be intermittent or completely failed. Consequently, data analysis based on asset data can also be affected. Additional load due to the transfer and / or access of asset data can even cause or result in a collapse of the network and / or computing resources. Thus, the temporal network load value indicative of the estimated or calculated network capacity at a given time can be at least partially based on at least one of the a priori determined parameters of which the technical context data is generated. This can be used, for example, to appropriately prevent or prioritize the transfer and / or access of asset data from the assets to the users. The reliability of remote data analysis requiring asset data can thus be improved, and in some cases, even the efficiency of such analysis can be improved by minimizing the time of transferring asset data. With this aspect of the teachings and even further aspects, the analysis can thus be prioritized and optimized according to one or more resource capacities.

[0021] Additionally or alternatively, at least one iteration parameter can be used to at least partially generate the technical context data. The iteration parameter can be determined via the second processing layer by analyzing a response to the partial request, the response being indicative of an impact of the partial request on the plant performance. For example, the iteration parameter can be determined by analyzing or measuring the response. The response can be indicative of an impact of the partial request on one or more resources which are critical for the operation and / or efficiency and / or safety of the assets and / or any assets of the plant. For example, the impact can be estimated by analyzing the response in terms of a delay value in response to the partial request.

[0022] Alternatively, the response can even be calculated using changes in processing load and / or memory and / or network load. Processing load and / or memory can be measured at any one or more assets in the plant, or even one or more critical assets. A partial request can be generated by a second processing layer for accessing data from the asset. A partial request may include a request for a subset of asset data from the asset. The subset of asset data may be part of data requested by an external layer or user, or it may be test data used solely to assess resource capabilities. In any case, the partial request is a failsafe request generated by the second processing layer, making any impairment to plant performance unlikely to result in a real reduction in plant safety, and preferably also a reduction in efficiency. In other words, the partial request can only cause short-term changes in plant behavior, insufficient to result in a meaningful, practical, or significant reduction in plant safety. Therefore, the subset of asset data or test data is a small dataset that is essentially just sufficient or requires resources to be analyzed. Therefore, at least one iterative parameter can be used to generate technical context data. As previously mentioned, at least one iterative parameter can therefore be used to define one or more accessibility criteria. Thus, at least one iterative parameter indicates the resource allocation of the external layer when receiving asset data. It should be understood that resource allocation can be used for any one or more computing and / or network resources between assets and users.

[0023] Or more specifically, the method also includes:

[0024] - A partial request for accessing data from assets or asset data is generated via the second processing layer;

[0025] - Measure the response to a partial request; the response indicates the impact of the partial request on one or more computing and / or network resources;

[0026] -At least one iteration parameter is determined based on the response; where

[0027] Technical context data is generated at least partially using at least one iteration parameter.

[0028] According to another approach, the second processing unit can even generate a second partial request in response to a measurement of the response to the partial request. For one or more resources, the second partial request may have a higher resource requirement compared to the partial request. Therefore, iteration parameters can be determined via the second processing layer by measuring the second response to the second partial request. A second partial request, requiring more resources than the partial request, can be generated if the response to the partial request does not indicate a sufficiently large or substantially harmful effect of the partial request on one or more resources.

[0029] At least one iterative parameter can then be generated by measuring the response to the second part of the request. In some cases, any one or more iterative parameters determined from the partial requests and any one or more iterative parameters determined from the second part of the request can be included in at least one iterative parameter, such that the iterative parameters from each of the partial requests are used to define one or more accessibility criteria. By doing so, for example, if multiple asset transfers from a plant are required, the user or external processing layer can adjust the transfer and / or access of asset data. Thus, by managing requests for multiple assets, the transfer and / or access of asset data can be made more intelligent. The second processing layer can then obtain a more precise determination of one or more resource bottlenecks to provide the transfer and / or access of asset data to the external layer. For example, to satisfy asset data requests from multiple assets, the resource capacity of one or more resources required to satisfy the requests can be distributed in a way that can provide optimal transfer and / or access parameters for the overall data transfer.

[0030] Resource capacity distribution can be accomplished based on any one or more iterative parameters determined from the partial request and any one or more iterative parameters determined from the second partial request. The corresponding responses from the partial and second partial requests can provide a measure of the resource allocation required for each type of request. This can be used to allocate capacity from one request to another when data from multiple assets needs to be accessed simultaneously or approximately simultaneously. In any case, the process of determining at least one iterative parameter can be repeated by incrementally adjusting the partial requests so that the second processing layer can iteratively determine at least one iterative parameter. Therefore, there can be one or more further partial requests, i.e., more than two partial requests for determining resource capacity associated with different resource needs. This can not only be used to maximize resource allocation to accommodate requests from the external processing layer while ensuring sufficient resource capacity is reserved for plant operations, but also help the second processing layer and / or the external processing layer proactively adjust requests for asset data, particularly requests from multiple assets, in some of the situations described above.

[0031] Or more specifically, the method also includes:

[0032] - This depends on the response generating a second part request for accessing asset data via a second processing layer; wherein, the second part request requires more resources than the first part request;

[0033] - Measure the second response to the second part of the request; the second response indicates the impact of the second part of the request on one or more computing resources and / or network resources;

[0034] - At least one iterative parameter is determined based on the response and / or the second response; where

[0035] Technical context data is generated at least partially using at least one iteration parameter.

[0036] Therefore, combining these two aspects, the method includes:

[0037] - A partial request for accessing data from assets or asset data is generated via the second processing layer;

[0038] - Measure the response to a partial request; the response indicates the impact of the partial request on one or more computing and / or network resources;

[0039] - Depending on the response, a second part request for accessing data from the asset or asset data is generated via a second processing layer; wherein the second part request requires more resources than the part request;

[0040] - Measure the second response to the second part of the request; the second response indicates the impact of the second part of the request on one or more computing resources and / or network resources;

[0041] - At least one iterative parameter is determined based on the response and / or the second response; where

[0042] Technical context data is generated at least partially using at least one iteration parameter.

[0043] As mentioned above, the data from the assets can be a subset of the asset data or it can be test data.

[0044] As discussed, technical context data includes one or more accessibility criteria for accessing asset data. Accessibility criteria may include one or more rules that must be followed by an external processing layer to be able to access or receive asset data. Alternatively or additionally, accessibility criteria may also include one or more parameters that the transmission and / or access of asset data should have or comply with.

[0045] It should be understood that the automatic specification or selection of one or more rules and / or parameters, including accessibility criteria, ensures that access to and / or transmission of asset data to an external transmission layer complies with these rules and / or parameters, and that such access and / or transmission does not affect any critical operations of the plant and / or any assets. In other words, by specifying one or more compliant rules and / or parameters, any significant plant or asset operations can be prevented from being adversely affected by access to and / or transmission of asset data to an external processing layer. Critical operations can be any operation or mode of operation of an asset, a group of assets, or the entire plant, which, when affected, could lead to a reduction in the security and / or efficiency and / or reliability of the plant and / or any assets associated with the plant. In essence, any critical operation of the plant will not be affected by access to and / or transmission of asset data performed using one or more rules and / or parameters.

[0046] Therefore, technical context data includes one or more accessibility criteria for asset data; accessibility criteria include one or more rules and / or parameters that can be followed by an external processing layer that receives the asset data so as not to affect any critical operations of the plant.

[0047] Furthermore, it allows users to access or transfer asset data without needing to know, at least within the plant, the operating or resource conditions. Therefore, the second processing layer can automatically adjust the accessibility conditions for asset data access and / or transfer based on plant conditions. This improves plant security. Additionally, it prevents the need for sensitive information about operating and / or resource conditions to leave the plant's secure network. For example, the outer layer doesn't need to know plant operating parameters. Instead, the second processing layer can specify one or more rules and / or parameters and provide them to the outer processing layer, ensuring that the outer processing layer only knows about one or more accessibility criteria for asset data based on its accessibility and / or transferability to the outer processing layer.

[0048] In some cases, data transmission and / or access may not be possible via a given network path using given characteristics (e.g., data rate). For example, a user might require a data rate higher than a given value to perform analysis, but infrastructure limitations might prevent this data rate from being achieved. This limitation could be latency, which can restrict total data throughput by affecting the bandwidth-latency product of the data link between the asset and the user. Therefore, in some cases, a user may request data transmission and / or access based on any one or more valid criteria provided by the second processing layer to the external processing layer using one or more accessibility criteria provided. In some cases, one or more accessibility criteria may be provided to the external processing layer after an access and / or transmission request is received at the second processing layer. This request may be initiated by the user. In some cases, the request may include one or more data transmission and / or access characteristics required by the user. Where the plant infrastructure and / or operating conditions can accommodate said characteristics, one or more accessibility criteria provided by the second processing layer to the external processing layer may include one or more data transmission and / or access characteristics required by the user. If one or more access characteristics better than one or more access characteristics requested by the user are possible, those better characteristics may be included in one or more access criteria provided to the external processing layer. The external processing layer can then decide whether to select better data transmission and / or access features from the accessibility criteria. In other cases, one or more accessibility criteria provided to the external processing layer by the second processing layer may only have one or more data transmission and / or access features requested by the user. This might be the case, for example, when the user request can be fulfilled without affecting critical plant operations or functions.

[0049] In some cases, technical context data may even include one or more performance parameters, such as latency values ​​or their estimates, between the asset and the user. Latency values ​​can be provided for the entire network path or a portion of the network path between the asset and the user. For example, in some cases, the total latency value may be dominated by a portion of the network or a bottleneck in the network path. In some cases, providing latency values ​​only for that portion of the network or bottleneck in the technical context data may be sufficient.

[0050] In some cases, the second layer can also apply additional contextualization to asset data. This additional contextualization can relate to the context available on the second processing layer. Contextualized contexts, such as plant identifiers, plant types, reliability indicators, or alarm constraints for industrial plants, can be added to the asset data. Alternatively, in some cases, even the technical asset structure of one or more plants, integrated sites, other asset management structures (e.g., asset networks), or application contexts (e.g., model identifiers, third-party exchanges) can be added to the asset data. This overall context can originate from functional locations or digital twins, such as digital piping and instrumentation diagrams of plant assets, 3D models, or scans using 3D coordinates. Alternatively, local scans from mobile devices linked to, for example, piping and instrumentation diagrams can be used for contextualization.

[0051] Contextualization refers to linking data points available in one or more storage units. These units can be persistent or non-volatile storage devices. Data points may be related to measurements or contextual information. At least one storage unit may be part of a first processing layer. Alternatively, at least one storage unit may be part of a second processing layer. Alternatively, at least one storage unit may be part of an external processing layer. Storage units can even be distributed across two or more processing layers. Links can be generated dynamically or statically. For example, predefined or dynamically generated scripts can generate dynamic or static links between information data points within a processing layer or across multiple processing layers. Links can be established by generating a new data object that includes the link data itself and storing this new data object in a new instance. If copies are stored elsewhere, any stored data points may be actively deleted. Therefore, any data points of a new data object copied from one storage unit to the same or another storage unit can be deleted to reduce storage space. Alternatively, links can be established by generating metadata objects with embedded links to address or access corresponding data points in distributed storage units. Therefore, any data point that can be addressed or accessed through a metadata object can remain in its original storage unit. Linking this information to form a new data object can still be performed, for example, on an external processing layer. For data retrieval, one can either directly access the data object or use the metadata object to address or access the data distributed across one or more storage units. Any operation on such data (such as an application) can either directly access the data or access a non-persistent image of the data, for example, from a cache or a persistent copy of the data.

[0052] According to one aspect, the first processing layer is associated with only a single plant (i.e., only an industrial plant). The first processing layer can therefore be a core processing system comprising one or more processing and storage devices. Such a layer can include one or more distributed processing and storage devices, forming a programmable logic controller (“PLC”) system and / or a distributed control system (“DCS”) with control loops distributed throughout the plant. Preferably, the first processing layer is configured to control and / or monitor chemical processes and assets at the asset level. Thus, the first processing layer is communicatively coupled to assets. The first processing layer can also monitor and / or control the lowest level of the chemical plant. Furthermore, the first processing layer can be configured to monitor and control critical assets. Alternatively or additionally, the first processing layer is configured to provide asset data to a second processing layer. The first processing layer can even be configured to provide processing data to the second processing layer. Asset data and / or process data can be provided directly or indirectly to the second processing layer.

[0053] The second processing layer may be associated with only one industrial plant, or it may be associated with more than one plant (e.g., a group of plants), or integrated with the latter. The second processing layer may include a process management system with one or more processing and storage devices. According to one aspect, the second processing layer is configured to manage data transfers to and / or from the first processing layer. The second processing layer may even host and / or coordinate processing applications. Such process applications can monitor and / or control one or more chemical plants or one or more assets. The process management system may be associated with one or more industrial plants.

[0054] On the other hand, the second processing layer may include an intermediate processing layer or intermediate processing system, and optionally a process management system. The intermediate processing layer may be communicatively coupled to the first processing layer. In this case, the first processing layer and the process management system may be coupled via the intermediate processing system. The intermediate processing system may be configured to collect process or asset data provided by the first processing layer. The process management system may be configured to provide an interface to an external network for plant-specific data from one or more plants. The intermediate processing system may be associated with one or more industrial plants. By including the intermediate processing layer in the second processing layer, a further layer of security is added. It can decouple the sensitive first processing layer from any external network access. Furthermore, the intermediate layer can reduce the data transfer rate to the external processing layer through preprocessing and enhance data quality through contextualization, thereby allowing for more robust data processing. To provide access to and / or transmission of asset data, the load on the first processing layer due to the external processing layer can therefore be reduced. In some cases, the intermediate processing layer can even be used to perform data analysis specified by the external processing layer. Especially where data-intensive access and / or transmission to the external processing layer is difficult or impossible, asset data can still be utilized by performing analysis on the intermediate processing layer. Data analysis results from asset data can be sent to users via an interface. Options for performing data analysis locally on the middle tier can be specified in the technical context data. In some cases, the middle tier can be used to cache data from assets with at least high data throughput. Therefore, a middle tier separated from the first processing layer can better adapt to and satisfy resource-intensive requests from users without affecting the behavior of the first processing layer due to requests from external processing layers. This can further improve plant security and / or reliability while providing better access to asset data. Thus, asset data can be better utilized in a scalable and flexible manner, regardless of plant size or the distance between users and assets. The middle processing system and process management system may include one or more processing and storage devices.

[0055] Therefore, as explained in this disclosure, the second processing layer can be configured to contextualize process and / or asset-specific data. Contextualization via the second processing layer does not affect the performance of the first processing layer. This can also be advantageous for older plants, as they can be retrofitted with the second processing layer while keeping the first processing layer substantially unchanged. A typical core process system in older plants is built using older generation computer systems to implement the plant infrastructure. These older processing systems typically lack the computing power required to perform data-intensive tasks. Core processing systems in most industrial plants are rarely upgraded because they are highly integrated with several other components of the plant, and changes may require extensive testing to ensure the performance and safety of the new system. Therefore, many plants may continue to use systems that may be obsolete or legacy systems in relation to the latest technologies on the market. Contextualization can be achieved even for such plants by adding a separate second processing layer as a further system with higher performance. Therefore, this teaching provides a scalable and flexible way to implement contextualization, even in older plants. Furthermore, in some cases, the second processing layer, and in the case of implementing an intermediate processing system, can also enable data contextualization at the plant level rather than the asset level. Contextualizing data in the context of this disclosure involves adding contextual information to process or asset data or reducing data size by preprocessing process or asset-specific data. Adding context may include adding more information tags to the process or asset data. Preprocessing may include filtering, aggregating, normalizing, averaging, or inferring process or asset data.

[0056] On the other hand, one-way or two-way communication (e.g., data transmission or data access) can enable data flow between different processing layers. A data flow may include process or asset data from a first processing layer, which is passed to a second processing layer and contextualized thereafter, and communicated with an external processing layer. Contextualization can be performed on the second processing layer. In some cases, contextualization can even be performed on the external processing layer, or even on both the second and external processing layers. Furthermore, depending on the criticality of the process data, asset data, or plant-specific data, such data can be allocated for one-way or two-way communication. For example, one-way communication channels can be implemented to prohibit data communication from the second or external processing layer to critical assets. Such communication may only allow one-way communication from critical assets to the processing layer, but not necessarily the other way around. Therefore, access to at least some critical assets can be read-only, for example, allowing only the reading of asset data and not the sending of data to the asset.

[0057] The second processing layer disclosed herein can be implemented as a process management system. As discussed, the second processing layer can be communicatively coupled to an external processing layer via an external network. The second processing layer can even be configured to manage data transmissions to and / or from the external processing layer.

[0058] According to one aspect, the external processing layer can be at least partially implemented as a computing or cloud environment that provides virtualized computing resources (such as data storage and computing power). Additionally, in some cases where it is necessary to monitor and / or control multiple industrial plants operated by different parties, data or process applications affecting the industrial plants can be shared within such a cloud environment.

[0059] In another aspect, the second processing layer is configured to manage data transfers to and / or from external processing layers in real-time or on demand. It should be understood that data transfer is performed via any of one or more accessibility standards. Depending on the network and compute load to the external network or even down to the user's interface, real-time transfers may be buffered. On-demand transfers can be triggered in a predefined or dynamic manner, as per one or more accessibility standards.

[0060] On the other hand, a second processing layer and / or an external processing layer is configured to exchange data via a third-party management system. This can be achieved through secure connections such as VPNs or through integration with third-party or shared processing layers.

[0061] Assets can even be Internet of Things (“IoT”) devices or systems, or even systems comprising one or more IoT devices. More specifically, assets can be Industrial Internet of Things (“IIoT”) devices or systems, or even systems comprising one or more IIoT devices. For example, assets can be IoT sensors, or even cyber-physical systems (“CPS”). In this context, CPS encompasses any industrial system containing a network of interacting elements. Therefore, industrial systems that utilize intelligent mechanisms to establish a closer connection between the computational and physical elements of such systems (such as modern robotic systems and industrial control systems) fall within the scope of this term.

[0062] Therefore, this method can also have the following aspects.

[0063] Or more specifically, according to one aspect, the method also includes:

[0064] - Send technical context data to the external processing layer via the interface.

[0065] According to another aspect, the method also includes:

[0066] - Receive one or more selected accessibility criteria at the second processing layer.

[0067] One or more of the selected accessibility criteria are chosen from technical context data and are selected to be performed by an external processing layer.

[0068] According to another aspect, the method also includes:

[0069] - Receive asset data at the external processing layer

[0070] Asset data is sent via a second processing layer based on one or more selected accessibility criteria.

[0071] According to one aspect, technical context data is stored as historical technical context data in user-side memory or database, for example, via an external processing layer. The user or external processing layer can then use one or more past accesses and / or transfers of historical context data from asset data to access and / or transfer further asset data. Historical technical context data can even be stored in the external processing layer, and / or accessed via memory or database.

[0072] According to another aspect, the method also includes:

[0073] - Store at least some of the technical context data as historical context data via an external processing layer; and

[0074] - Receive one or more pre-selected accessibility criteria at the second processing layer.

[0075] One or more pre-selected accessibility criteria are chosen from historical technical context data, and the selection is performed by an external processing layer.

[0076] This can further accelerate the access and / or transmission of asset data. Historical context data can even include data related to the availability of assets and / or other resources in the plant (e.g., processing load, network bandwidth, latency) based on date and / or time. Therefore, the external processing layer can automatically adjust access to and / or transmission of asset data based on the historical availability of plant resources and / or assets. Thus, despite isolation from the plant, users can be able to adjust the utilization of data availability from assets.

[0077] Industrial assets can be any equipment associated with an industrial plant, or more simply, a plant. It should be understood that an asset can therefore be any equipment or any single piece of equipment capable of generating data that can be used to evaluate the performance of the asset and / or the plant. The data preferably includes measurement data indicating the values ​​of one or more process parameters of the asset and / or the plant, but it can even indicate one or more binary parameters, such as the state of a valve, or the “on” or “off” state of one or more pieces of equipment. The terms “industrial asset” and “asset” are used interchangeably in this disclosure to refer to any single piece or group of plant equipment. As previously stated, an asset can be any equipment capable of generating data (preferably in digital form) that can be used for plant monitoring and / or analysis. As mentioned, the term “asset” can even refer to a group of equipment, such as a robotic station comprising multiple motors, actuators, and sensors. Other non-limiting examples of assets are any one or more of the following, or combinations thereof: heat exchangers, reactors, pumps, piping, distillation columns, or absorption columns. Asset data can be data generated by the asset or data generated within the asset. For example, using the same example of a robotic station, asset data can be data from any one or more sensors of the robotic station. Asset data can even be a combination of data from multiple sensors and / or process parameters, or data from memory or even the controller of the robot station. Some non-limiting examples of process parameters are controller setpoints, output signals, settings, historical or recorded data, and any type of configuration data.

[0078] In newer plant operations technology (“OT”) systems, it may be necessary to run one or more computer applications on a cloud computing platform. These applications may require data from one or more assets within the plant. Therefore, in some cases, it may be necessary to transfer asset data to the cloud platform to make the data accessible to the applications. It is understood that an external processing layer in this scenario could also be implemented within the cloud platform. The problem with older OT systems (particularly PIMS or SCADA) is their longer lifecycle, which often results in these plant systems employing older / legacy technologies. Typically, older OT systems are not designed to provide data in real-time or near real-time. Furthermore, such systems often lack streaming interfaces for streaming or retrieving data. Therefore, the only solution for users or external processing layers may be to periodically poll data requests to receive new asset data. The applicant has recognized that such polling interfaces can be highly inefficient.

[0079] The applicant has thus achieved a more efficient method for asset data transmission. In cases where not all asset data can be provided from the system in real-time or near real-time, a second processing layer can be configured to at least provide or transmit low-resolution asset data. It should be understood that this aspect is patentable in itself, at least due to the technical advantages outlined below. Combined with the remaining features of this teaching, this can provide further synergies, which may include at least the ability to pre-plan the transfer of asset data from a plant with computational bottlenecks by configuring and scheduling data transmission to an external processing layer or user using technical context data. This allows for better prioritization of data analysis tasks based on the availability of asset data.

[0080] The term "all asset data" can refer to asset data requested by an external processing layer, or it can refer to asset data required by an external processing layer. For example, a situation where all asset data cannot be provided in real-time or near real-time could be, for instance, when a second processing layer cannot retrieve all requested or required asset data from the assets in real-time or near real-time, respectively. Alternatively, a situation where all asset data cannot be provided in real-time or near real-time could even be a situation where all requested or required asset data cannot be transmitted in real-time or near real-time, respectively. Therefore, any computational and / or network bottlenecks can hinder real-time or near real-time data transmission.

[0081] Therefore, from another perspective, a method for preprocessing asset data from industrial assets located within an industrial plant, the assets being communicatively coupled to a first processing layer, wherein the asset data is provided via the first processing layer to a second processing layer; the first processing layer being communicatively coupled to the second processing layer, and wherein the first and second processing layers are configured in a secure network, the method comprising:

[0082] - Low-resolution asset data is provided via a second processing layer; where

[0083] Low-resolution asset data is a subset of asset data requested by an external processing layer, and the low-resolution data can be used by the external processing layer to initiate at least one or more data analyses.

[0084] Preprocessing methods can be implemented alone or in combination with other aspects. Therefore, industrial systems for preprocessing asset data, and software products for implementing preprocessing steps, can also be provided as standalone embodiments or in conjunction with other aspects of this teaching.

[0085] In some cases, low-resolution asset data may be the lowest-resolution portion of the asset data available to the external processing layer, while the remainder of one or more portions of the asset data can be provided at a later time. It should be understood that this can have the advantage that the external processing layer does not need to wait for the entire asset data to become available before data processing, such as data analysis, can begin. Therefore, the external processing layer can begin processing the lowest-resolution portion of the asset data, or coarse asset data, while the remaining asset data is received in the background or provided later. Similarly, low-resolution or coarse asset data can even be higher-resolution data compared to the lowest-resolution portion of the asset data available to the external processing layer. Therefore, if higher-resolution data than the minimum available resolution can be provided to the external processing layer, that data can be provided in the background or later when the remaining required asset data is provided. The terms used in the background may, for example, refer to caching or storing the remaining one or more portions of the asset data at a second processing layer, and / or caching or storing the remaining one or more portions of the asset data at the external processing layer. The remaining asset data may be provided all at once or in multiple transmission cycles. Preferably, the remaining asset data is prioritized for processing in one or more data blocks, each available to the external processing layer, without any pending asset data subsequently awaiting transmission to the external processing layer. In other words, the asset data is preferably subdivided into data blocks available to the external processing layer without any remaining asset data. It should be understood that by doing so, the data resolution of the asset data received at the external processing layer can be improved incrementally and seamlessly without waiting for the remaining data before continuing to process the received data. Therefore, although the data transmission rate is limited, data processing can thus be more efficient. The resolution is preferably the temporal resolution of the asset data.

[0086] Most signals from sensors are time-based, therefore at least the majority of asset data is time-series data. One or more techniques can be used to generate low-resolution asset data, for example, downsampling one or more signals included in the asset data. Specific techniques used to generate low-resolution asset data are not limited to the scope or generality of this teaching. Therefore, any technique that allows the generation of low-resolution asset data usable by an external processing layer can be used for this purpose.

[0087] According to one aspect, asset data transfer and / or access are initiated in response to variable-batch read history queries. This can be accomplished, for example, by varying the time frame of historical calls from the external processing layer to the network interface. By requesting asset data over a longer time period, the computational resources (query time / data points) required to transfer asset data from asset and / or storage units can be used more efficiently. While this may result in data transfers including older data points, this can reduce the computational load by making the overall transfer more directly provide better resolution for certain portions of the asset data, while still providing all other data points or eliminating the need for measurements at higher time frequencies or resolutions.

[0088] According to one aspect, the resolution of each portion of the asset data is adjusted based on the relevance values ​​calculated for the data analysis of the required asset data. According to another aspect, a machine learning (“ML”) model trained on asset data with a time span of approximately two years and hourly averages of data points is used to determine the data points of the asset data that need to be delivered at high resolution, while the remaining data points only require lower resolution. Similarly, depending on the asset, the machine learning (“ML”) model can be trained on asset data with a time span of approximately one year and hourly, half-hourly, or quarterly averages of data points. More generally, the machine learning (“ML”) model can be trained on asset data with a time span exceeding six months and the longest 1-day average of data points. The advantage of doing so is that, although portions of the asset data from the most recent hours are not available at the external processing layer, lower-resolution data from the long-term history of the asset data can be used to enhance analytical purposes.

[0089] According to another aspect, asset data usage is monitored to classify the relevance of specific data points based on specific use cases. Monitoring can be performed via a second or third processing layer. By doing so, the resolution of specific data portions can be determined to be relevant to a specific use case. Adjusting the resolution of data points in the asset data can make data transmission more efficient. For example, this can be achieved by monitoring the requested technical context as at least one boundary condition and calculating a cost function for how to transmit the requested asset data to, for example, an external processing layer. Furthermore, optimization algorithms can be applied to parameterize data polling at the second processing layer. According to one aspect, data usage is monitored substantially continuously or periodically, and the weights of the cost function are adjusted. Optimization can be rerun for different assets and / or different external processing layers.

[0090] In this context, the terminology, particularly the term "near real-time," can refer to a signal or data that includes a time delay of no more than 15 seconds, specifically no more than 10 seconds, and more specifically no more than 5 seconds between the generation of the signal / data and its transmission. Therefore, as an example, an asset dataset generated at a plant asset and then transmitted to an external processing layer and provided at a network interface within 15 seconds can be considered a near real-time transmission or a near real-time provision. Similarly, a transmission with a smaller time delay can be referred to as a real-time transmission.

[0091] Regarding technical context data, rules may be, for example, any time or period in which or near it is permitted to access and / or transmit asset data, via any one or a combination thereof, through any network path in which access and / or transmission of data is permitted.

[0092] The parameters can be any one or more of the following: data transfer rate, number of data packets, size of the dataset of asset data requested for access and / or transfer, size of one or more data packets combined into the asset data requested for access and / or transfer, and resolution of one or more data packets.

[0093] It should be understood that a secure network is a network or part thereof used for communication between at least some of the plant assets used in plant operations. Therefore, a secure network can be an intranet belonging to the plant. Typically, a secure network is located within the plant, but sometimes it can even extend beyond the plant's physical location. For example, if any or more of the plant-related databases, processing systems, or other computing services are implemented as one or more cloud-based services. The first and second processing layers are part of the secure network. A secure network can even be an isolated network, comprising more than two secure zones separated by firewalls. Such firewalls can be network-based or host-based virtual or physical firewalls. Firewalls can be hardware- or software-based to control incoming and outgoing network traffic. Here, predefined rules in the sense of whitelisting can define allowed traffic via access management or other configuration settings. Depending on the firewall configuration, secure zones may follow different security standards.

[0094] The external network can be at least partially a public network, such as the Internet. Alternatively or additionally, the external network can be at least partially another secure network isolated from the secure network within the provided second processing layer. It should be understood that in some cases, such as when two plants are interconnected via a dedicated private or non-public network, a plant's secure network can be isolated from the other plant's internal network. Therefore, users located in other plants may still face at least some of the same problems when accessing asset data from the plant, such as the user not knowing the plant's operating parameters. Therefore, this teaching can also be applied to solving similar problems in a group of plants interconnected via any type of public or private external network.

[0095] In another scenario, the first processing layer is configured in a first security zone via a first firewall, and the second processing layer is configured in a second security zone via a second firewall. To securely protect the first processing layer, the first security level can adhere to a higher security standard than the second security level. The security level can follow common industry standards such as those listed in Namur document IEC 62443. The second processing layer can provide further isolation via security zones. For example, the intermediate processing system can be configured in a third security zone via a third firewall, while the process management system can be configured in a second security zone via a second firewall. The third and second security zones can also be staggered in terms of security standards. For example, the third security zone can adhere to a higher security standard than the second security zone. This allows for higher security standards in the lower security zones of the first processing layer and lower security standards in the higher security zones of the second processing layer.

[0096] According to one aspect, the technical context data is generated using a machine learning (“ML”) model, such as a trainable neural network, which has been trained using historical access and / or transfer data associated with the asset. For example, the training data may include historical latency and performance data associated with the asset. The training data may include specific latency and performance data depending on multiple possible network paths used to transfer asset data from the asset to an external database or destination storage where the asset data is to be integrated. Alternatively or additionally, the training data may even include historical technical context data from one or more past accesses and / or transfers of the asset data. Alternatively or additionally, the training data may even include data from one or more historical partial requests used to determine at least one iteration parameter. The ML model may be executed at least partially on a second processing layer and / or an intermediate processing layer. The ML model may even be executed partially on an external processing layer. The ML model may even be used to identify at least one bottleneck and / or learn the characteristics of at least one bottleneck in at least one network path between the asset and an external processing layer or user. Therefore, the ML model may also be used to determine at least one characteristic of at least one bottleneck. The term "bottleneck" can refer to any type of limitation in computing resources required to provide users with access to and / or transmission of asset data. Therefore, a bottleneck can be a limitation in a network path, such as data bandwidth and / or latency. Alternatively, a bottleneck could even be a processing load limitation of any processing layer or processor through which asset data is processed. Further alternatively, a bottleneck could even refer to memory limitations or limited data storage capacity, such as limited random access memory ("RAM") or cache.

[0097] It should be understood that using the ML model trained as specified above can further enable the identification of bottlenecks in accessing and / or transferring asset data. Therefore, the ML model can be used as an intermediary to address data integration parameters between the user or external processing layer and the asset.

[0098] According to another approach, the training data is divided into: internal training data, which includes latency and performance data related to one or more possible internal network paths used to transfer asset data up to the interface with an external network; and external training data, which includes latency and performance data related to one or more possible external network paths used to transfer asset data from the interface to an external processing layer. Preferably, the external training data includes latency and performance data up to the external database or destination storage where the asset data is to be integrated. In some cases, the external training data may even be related to one or more partial paths between the interface and the external database or target storage. This may occur, for example, when the external training data is not fully populated or available. The advantage of splitting the training data into internal and external training data is that it increases the flexibility of training for changed paths. For example, if one or more external paths have changed, training can only be performed for the external paths. This can also save training time and allow for faster data integration. Similarly, machine learning (“ML”) models can even be split into internal ML models and external ML models. The internal ML model can be trained using internal training data, while the external ML model can be trained using external training data.

[0099] Another synergistic effect of dividing training data into internal and external training data is that the second processing layer can more flexibly determine the combination of internal and external paths, which provides the external processing layer with favorable accessibility criteria. Therefore, one or more such combinations can be provided to the external processing layer as part of the technical context data. Thus, the external processing layer can request access to the asset data using the most appropriate accessibility criteria.

[0100] The applicant further recognizes that this teaching may be particularly applicable to applications within a value chain (or even in continuous production where assets or products manufactured in the first factory are used by the second factory). Those skilled in the art will understand that the number of factories in a value chain can be more than two. Or more generally, a user can be a supplier to another user downstream in the value chain, and so on.

[0101] An industrial plant (or simply a plant) includes infrastructure for industrial purposes. Industrial purposes can be the manufacture of one or more products, i.e., process manufacturing completed by the processing plant. For example, the product can be any product, such as: chemical, biological, pharmaceutical, food, beverage, textile, metal, plastic, semiconductor. Therefore, the plant can be any one or more of the following: chemical plant, pharmaceutical plant, fossil fuel facilities such as oil and / or natural gas wells, refineries, petrochemical plants, cracking plants, fracturing facilities, etc. Those skilled in the art will understand that the plant also includes assets in the form of instruments, which can include several different types of sensors for monitoring plant parameters and equipment. Therefore, some of the asset data may be generated via instruments such as sensors.

[0102] From another perspective, a system for managing asset data of industrial assets may also be provided, the system comprising at least one processor, wherein any one of the at least one processor is configured to perform any method steps disclosed herein.

[0103] More specifically, an industrial plant system may be provided comprising a first processing layer and a second processing layer, the first processing layer being communicatively coupled to the second processing layer, and the first and second processing layers being configured in a secure network, wherein at least one industrial asset is configured to be communicatively coupled to the first processing layer, wherein the asset is configured to provide asset data to the second processing layer via the first processing layer, and the system further includes an interface to an external network, wherein...

[0104] The second processing layer is configured as follows:

[0105] - Generate technical context data related to the assets; and

[0106] - Provide technical context data to the interface;

[0107] The technical context data includes one or more accessibility criteria for asset data; the accessibility criteria include one or more rules and / or parameters that can be followed by an external processing layer that receives the asset data.

[0108] Therefore, this system is suitable for integrating asset data from industrial assets.

[0109] From another perspective, a computer program including instructions may also be provided, which, when executed by a suitable computer processor, cause the processor to perform the method steps disclosed herein. A non-transitory computer-readable medium storing a program that causes a suitable computer processor to perform any of the method steps disclosed herein may also be provided.

[0110] Computer-readable data media or carriers include any suitable data storage device on which one or more sets of instructions (e.g., software) embodying any one or more methods or functions described herein are stored. The instructions may also reside wholly or at least partially in main memory and / or in a processor during execution by a computer system, main memory, and processing device, which may constitute a computer-readable storage medium. The instructions may further be transmitted or received over a network via a network interface device.

[0111] The networks discussed herein can be any type of data transmission medium, wired, wireless, or a combination thereof. Specific types of networks are not limited to the scope or generality of this teaching.

[0112] Computer programs for implementing one or more embodiments described herein may be stored and / or distributed on suitable media, such as optical storage media or solid-state media provided 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 telecommunications systems. However, computer programs may also be presented via networks like the World Wide Web and can be downloaded from such networks to the working memory of a data processor.

[0113] From another perspective, a data carrier or data storage medium may also be provided to make a computer program element available for download, the computer program element being arranged to perform a method according to one of the foregoing embodiments.

[0114] The word “comprising” does not exclude other elements or steps, and the indefinite articles “a” or “an” do not exclude a plural. A single processor, controller, or other unit can perform the functions of several items listed in the claims. The fact that certain measures are listed in mutually different dependent claims does not indicate that a combination of these measures cannot be used advantageously. Any reference signs in the claims should not be construed as limiting the scope. Attached Figure Description

[0115] Example embodiments are described below with reference to the accompanying drawings.

[0116] Figure 1 A block diagram of the system, including the processing layer, is shown.

[0117] Figure 2 A flowchart of one aspect is shown. Detailed Implementation

[0118] In industrial plants such as chemical plants, process industry production typically begins with an upstream product used to obtain further downstream products. In a typical plant or its layout, value chain production from one or more intermediate products to the final product is highly constrained and based on isolated infrastructure. This can hinder the adoption of new technologies such as IoT, cloud computing, and big data analytics.

[0119] Unlike other manufacturing industries, process industries may adhere to very high standards, particularly in terms of availability and security. For this reason, computing infrastructure is typically highly secure, e.g., unidirectional and isolated, with highly restricted access to the monitoring and control systems of such plants.

[0120] Typically, such industrial plants are embedded in the enterprise architecture in a siloed manner at different levels, thereby separating functions between operational technology and information technology solutions.

[0121] Level 0 relates to physical processes and defines the actual physical processes in a factory. Level 1 involves intelligent devices used to sense and manipulate physical processes, such as process sensors, analyzers, actuators, and related instruments. Level 2 relates to control systems—used to supervise, monitor, and control physical processes. Real-time control and software; DCS, Human-Machine Interface (“HMI”); Supervisory and Data Acquisition (“SCADA”) software are some typical components. Level 3 involves manufacturing operating systems used to manage production workflows to produce the required products. Batch management; Manufacturing Execution / Operations Management System (MES / MOMS); laboratory, maintenance, and factory performance management systems, data history records, and related middleware are typical components. Time frames for control and monitoring can be shifts, hours, minutes, or seconds. Level 4 involves business logistics systems used to manage business-related activities in manufacturing operations. Enterprise Resource Planning (ERP) is often the primary system and establishes basic factory production plans, material usage, transportation, and inventory levels. Time ranges may be months, weeks, days, or shifts.

[0122] Furthermore, this architecture may adhere to strict one-way communication protocols, disallowing any data flow into Level 2 or lower. There is no coverage of the company's or enterprise's external internet in this architecture. However, this model remains a fundamental concept in the field of cybersecurity. Within this context, the challenge lies in leveraging the benefits of cloud computing and big data while still preserving the advantages of the existing architecture: namely, high availability and reliability of the lower-level systems (Levels 1 and 2) of the control plant, and cybersecurity.

[0123] This teaching can enhance monitoring and / or control by systematically altering the framework, thereby introducing new capabilities consistent with the existing architecture. This disclosure can provide the process industry with a scalable, flexible, and usable computing infrastructure while adhering to high security standards. Furthermore, data and analytics can be leveraged across different plants while ensuring that plant performance is not unduly impacted by externally requested access and / or data transfers from assets located within the plant.

[0124] Figure 1 The diagram illustrates a system 100 or an arrangement including processing layers. A first industrial plant 101 is shown, which could be a chemical plant, for example. A chemical plant can be any manufacturing facility based on a chemical process (e.g., using a chemical process to convert raw materials into products). The system 100 shown includes two processing layers: a first processing layer in the form of a core process system 114 associated with plant 101, and a second processing layer 116, which could take the form of a process management system associated with plant 101, for example. The first processing layer 114, or the core processing system, is communicatively coupled to the second processing layer 116, allowing unidirectional or bidirectional data transfer. The core process system 114 includes a set of distributed processing units associated with the assets of chemical plant 101.

[0125] The first processing layer 114 and the second processing layer 116 are configured in a secure network, shown in this example schematically as two security zones separated by firewalls 118 and 120. The first security zone is located above the core process system 114 layer, where the first firewall 118 controls network traffic entering and leaving the core process system 114. The second security zone is located around the second processing layer 116, where the second firewall 120 controls network traffic entering and leaving the second processing layer 116. This isolated network architecture allows for the protection of vulnerable plant operations from unauthorized access or cyberattacks.

[0126] The first processing layer 114 provides asset data 122 of plant 101 to the second processing layer 116. The first processing layer 114 may also provide process- or asset-specific data of plant 101 to the second processing layer 116. Process- or asset-specific data may include value, quality, time, unit of measurement, and asset identifier. Further contextualization can be used to add information such as plant identifier, plant type, reliability indicators, or plant alarm limits. The second processing layer 116 is further configured to provide technical context data to an interface 126 of external network 124.

[0127] Technical context data includes one or more accessibility criteria for asset data. Accessibility criteria include one or more rules and / or parameters that external processing layer 150 must adhere to in order to receive asset data. External processing layer 150 may be located within second plant 102. Although the processing layer or security zone associated with second plant 102 is not shown in the figure, second plant 102 may have a similar layer setup to that associated with first plant 101. Alternatively, second plant 102 may have a different processing arrangement compared to first plant 101. Because users located in second plant 102 may not have a complete overview of the key operational parameters in first plant 101, data requests from assets (e.g., asset 12) may impact performance or security within first plant 101. In some cases, users may be applications running on external processing layer 150. External processing layer 150 may even be part of a cloud computing platform or service. Therefore, external processing layer 150 is not necessarily located in second plant 102. In some cases, external processing layer 150 may even be independent of any plant, unlike... Figure 1 The scenario shown illustrates an external processing layer 150 associated with the second plant 102. The external processing layer 150 could even be a separate remote computing service used to analyze asset data from one or more plants.

[0128] The second processing layer 116 is communicatively coupled to the external processing layer 150 via an interface 126 to an external network. In some cases, the external processing layer 150 may even be a computing or cloud environment that provides virtualized computing resources (such as data storage and computing power).

[0129] One or more rules and / or parameters included in the accessibility criteria may be automatically specified, or they may be selected such that access to and / or transmission of asset data to the external processing layer 150 performed in accordance with one or more of the aforementioned rules and / or parameters does not affect any critical operation of device 101 and / or any assets 10-12.

[0130] For example, to analyze the vibration of pump 11, a user may have requested measurement data from pump 11 via external processing layer 150. The measurement data may have a frequency of, for example, 10 kHz. Due to network latency, this data transmission may not be possible in real time. By applying technical context data, which may include a priori determined parameters rather than being immediately adapted to the user's request, and thus potentially affecting the performance of plant 101 and / or asset 11 and preventing the transmission of the required data, one or more feasible alternatives for accessing / transmitting asset 11 data can be provided to the user. In some cases, the technical context data can be generated using a machine learning (“ML”) model. The system can then, for example, via second processing layer 116, learn when such a transmission rate can be achieved. Alternatively or alternatively, alternative paths may be suggested, via which access and / or transmission with the requested characteristics may be possible. In some cases, the possibility of running the analysis or application locally (e.g., on second processing layer 116) may be offered to the user, thereby providing results to external processing layer 150. Accordingly, the second processing layer 116 and / or the external processing layer 150 may be configured to host and / or orchestrate process applications or analytics. In some cases, the second processing layer 116 may host and / or orchestrate process applications related to core plant operations, while the external processing layer 30 may be configured to host and / or orchestrate process applications related to non-core plant operations. Here, core plant operations may correspond to critical operations that allow plant 101 to operate in island mode without external network connectivity.

[0131] Data or asset data from Pump 11 can even be provided in multiple data packets delivered at different times. These data packets can be cached at a second processing layer and / or an intermediate processing layer and / or an external processing layer. In some cases, asset data may initially be provided to the external processing layer in a low-resolution or coarse form, allowing data processing to begin without waiting for the complete asset data to become available at the external processing layer. The remaining asset data can be provided in one or more data packets, with each packet progressively increasing the resolution of the asset data at the external processing layer. In some cases, depending on the relevance of the asset data or pump data requiring analysis, the data packets may even have different resolutions than each other.

[0132] In some cases, one or more additional processing layers may also be present on the first factory 101 side, or on the second factory 102 side, or both. For example, an intermediate processing layer may be provided between the first processing layer 114 and the second processing layer 116. The intermediate processing layer may be communicatively coupled to the first processing layer 114 via a first firewall 118. Thus, the first processing layer 114 and the second processing layer 116 are communicatively coupled via the intermediate processing layer. An additional firewall may also be provided between the intermediate processing layer and the second processing layer 116. The intermediate processing layer may even reduce the data transfer rate to the external processing layer 150 by, for example, through preprocessing and contextualizing data quality enhancement to allow for more enhanced data processing.

[0133] In some cases, ML models can be executed at least partially on intermediate processing layers.

[0134] Figure 2 A flowchart 200 illustrates one aspect of this teaching. Technical context data associated with an asset (e.g., pump 11) is generated, for example, at a second processing layer 116 201. Technical context data is provided, for example, to an interface 126 via the second processing layer 116 202. Interface 126 is connected to an external network 124. Therefore, technical context data can be provided 203 to an external processing layer 150, which requires access to and / or transmission of data from pump 11 or asset data. The technical context data can be provided by transmission via network interface 126. The technical context data includes one or more accessibility criteria for asset data. Accessibility criteria include one or more rules and / or parameters that can be followed by the external processing layer 150 for accessing and / or transmitting pump data or asset data. Optionally, asset data can be received at the external processing layer 150 204 based on one or more selected accessibility criteria selected by the external processing layer 150 from the technical context data. Alternatively, in addition to steps 201-204 discussed herein, other aspects may be implemented, such as generating technical context data from at least one iterative parameter.

[0135] Various examples of methods for integrating asset data have been disclosed above, systems for managing asset data have been provided, and computer software products implementing any of the relevant method steps disclosed herein have been provided. However, those skilled in the art will understand that changes and modifications can be made to these examples without departing from the spirit and scope of the appended claims and their equivalents. It will further be understood that aspects from the method and product embodiments discussed herein can be freely combined. Certain exemplary embodiments of this teaching are summarized below.

Claims

1. A method for integrating asset data from industrial assets located within an industrial plant, said assets being communicatively coupled to a first processing layer, wherein, The asset data is provided to a second processing layer via a first processing layer; the first processing layer is communicatively coupled to the second processing layer, and wherein the first processing layer and the second processing layer are configured in a secure network defined by a firewall, comprising at least a first security zone and a second security zone, wherein the first security zone is located on the first processing layer and the second security zone is located around the second processing layer, and a first security level of the first security zone follows a higher security standard than a second security level of the second security zone, the method comprising: - Generate technical context data related to the asset at the second processing layer; - The technical context data is provided to the interface to the external network via the second processing layer; The technical context data includes at least one accessibility standard for the asset data; the accessibility standard includes at least one rule and / or parameter to be followed by an external processing layer for receiving the asset data.

2. The method according to claim 1, wherein, The technical context data is generated at least partially using at least one of the following: parameters that include the asset network address, asset CPU load, asset memory, and the network path between the industrial asset and the external processing layer.

3. The method according to claim 2, wherein, The asset storage includes random access memory.

4. The method according to any one of claims 1 to 3, wherein, The method further includes: - Generate at least a first part of a request for accessing the asset data via the second processing layer; - Measure at least a first response to the first part of the request; the first response indicates the impact of the first part of the request on at least one computing and / or network resource; - Determine at least a first iteration parameter based on the first response, wherein the first iteration parameter indicates the resource allocation of the external processing layer when receiving asset data; in The first iteration parameters are used to generate at least part of the technical context data.

5. The method according to claim 4, wherein, The method further includes: - Depending on the response, a second part of the request for accessing the asset data is generated via the second processing layer; wherein the second part of the request requires more resources than the first part of the request; - Measure a second response to the second part of the request; the second response indicates the impact of the second part of the request on the at least one computing and / or network resource; - Determine at least the first iteration parameter and / or the second iteration parameter, depending on the first response and / or the second response, wherein the second iteration parameter indicates the resource allocation of the external processing layer when receiving asset data.

6. The method according to any one of claims 1 to 3, wherein, The method further includes: - The technical context data is sent to the external processing layer via the interface.

7. The method according to any one of claims 1 to 3, wherein, The method further includes: - Receive at least one selected accessibility criterion at the second processing layer. The at least one selected accessibility criterion is selected from the technical context data and is selected to be executed by the external processing layer.

8. The method according to any one of claims 1 to 3, wherein, The method further includes: - Receive the asset data at the external processing layer. The asset data is transmitted via the second processing layer according to at least one selected accessibility criterion.

9. The method according to any one of claims 1 to 3, wherein, The method further includes: -Storing at least some of the technical context data from the technical context data as historical technical context data via the external processing layer; and - Receive at least one pre-selected accessibility criterion at the second processing layer. The at least one pre-selected accessibility criterion is chosen from the historical technology context data, and the selection is performed by the external processing layer.

10. The method according to any one of claims 1 to 3, wherein, The method further includes: - At the external processing layer, depending on the relevance of the asset data from at least one data analysis, low-resolution asset data or coarse-form asset data is received; wherein... The low-resolution asset data is a subset of the asset data requested by the external processing layer, and the low-resolution asset data can be used by the external processing layer to initiate at least one data analysis.

11. The method according to claim 10, wherein, The method further includes: - Receive second low-resolution asset data at the external processing layer; wherein The second low-resolution asset data is a subset of the asset data requested by the external processing layer, and wherein the second low-resolution asset data can be used by the external processing layer in conjunction with the low-resolution asset data to at least further process the at least one data analysis.

12. The method according to claim 11, wherein, The low-resolution asset data and the second low-resolution asset data have different resolutions from each other.

13. The method according to claim 5, wherein, The technical context data is generated using a machine learning ("ML") model that has been trained using historical access and / or transfer data associated with the asset and / or data from at least one historical partial request used to determine at least the first iteration parameter and / or the second iteration parameter.

14. The method according to claim 13, wherein, The machine learning model includes a trainable neural network.

15. A computer program product comprising instructions that, when executed by a suitable computer processor, cause the processor to perform the method steps of any one of claims 1 to 14.

16. An industrial plant system comprising a first processing layer and a second processing layer, the first processing layer being communicatively coupled to the second processing layer, and the first processing layer and the second processing layer being configured in a secure network defined by a firewall, comprising at least a first security zone and a second security zone, wherein, The first security zone is located on the first processing layer and the second security zone is located around the second processing layer, and the first security level of the first security zone follows a higher security standard than the second security level of the second security zone, wherein at least one industrial asset is configured to be communicatively coupled to the first processing layer, wherein the asset is configured to provide asset data to the second processing layer via the first processing layer, and the plant control system further includes an interface to an external network, wherein... The second processing layer is configured as follows: - Generate technical context data related to the asset; and - Provide the technical context data to the interface; The technical context data includes at least one accessibility standard for the asset data; the accessibility standard includes at least one rule and / or parameter to be followed by an external processing layer for receiving the asset data.

Images