Method for extracting instructions for monitoring and / or controlling a chemical plant from unstructured data

Abstract

The present invention belongs to the field of computer-implemented methods for monitoring or controlling chemical plants. The present invention relates to a computer-implemented method for monitoring and / or controlling a chemical plant, comprising: (a1) providing unstructured data containing instructions for monitoring and / or controlling the chemical plant, (a2) providing information about the chemical plant via an interface, the information comprising at least the geographical location of the plant or information about compounds processed in the plant, (b1) providing the unstructured data and the information about the chemical plant to a model suitable for extracting instructions from the unstructured data, (b2) obtaining the instructions from the model along with metadata, the metadata comprising the applicability of the instructions with respect to at least one of a time period, a geographical range, or a compound to be processed in the plant, and (c) outputting the instructions received from the model.

Description

Technical Field

[0001] The invention is in the field of computer-implemented methods for monitoring or controlling chemical plants. Background Art

[0002] Operating a chemical plant requires numerous actions to maintain a high level of safety and health for those working there, as well as environmental protection. For example, airborne emissions must be monitored, and different maintenance intervals for every part of the plant must be adhered to. These regulations are numerous and complex, as they include EU regulations, national laws, state laws, county rules, company rules, plant-specific regulations, and contracts. Navigating all of these regulations is time-consuming for plant operators, and some action items may be overlooked. Furthermore, if regulations change, it can be difficult to react promptly. Therefore, a system that can automate these tasks would be beneficial. However, this is a challenging task, as regulations are mostly unstructured data in the form of human-readable text.

[0003] WO 2017 / 129 636 A1 discloses a method for automatically determining patent infringement risks in chemical plants. However, this concept cannot be easily transferred to the above problem because the risk does not directly tell the plant operator what to do.

[0004] S. et al. published a system for converting complex rules in text into logic graphs in the Proceedings of the First Compliance Technology Workshop (http: / / ceur-ws.org / Vol-2049 / 08paper.pdf). However, creating relationships between words requires inputting predefined relationships, which requires considerable effort. Furthermore, the system cannot easily accommodate input in different languages.

[0005] E. Zamora et al., Journal of Chemical Information and Modeling, Vol. 24 (1984), pp. 176-188, disclose a method for extracting chemical reaction information from primary journal texts. However, this information is only stored in a database rather than being converted into instructions suitable for monitoring and / or controlling a chemical plant.

[0006] WO2019 / 023982 A1 discloses a database for storing industrial operation data obtained from various sources including unstructured data. However, no instructions for monitoring and / or controlling a chemical plant are generated from the database.

[0007] US 2008 / 0040298 A1 discloses a method for converting unstructured data related to a chemical reaction into structured data for storage in a structured database. However, no instructions suitable for monitoring and / or controlling a chemical plant are generated. Summary of the Invention

[0008] The object of the present invention is therefore to provide a method for monitoring or controlling a chemical plant that is fast, reliable, and easy to use in order to increase operational safety and minimize environmental impact. The method should be flexible so that it can be easily adapted to new requirements and quickly provide the necessary operations, reliably excluding anything that is not relevant in a particular case, thereby reducing the burden on the operator.

[0009] These objects are achieved by a computer-implemented method for monitoring and / or controlling a chemical plant, the method comprising:

[0010] (a1) providing unstructured data containing instructions for monitoring and / or controlling a chemical plant,

[0011] (a2) providing information about a chemical plant through an interface, the information including at least the geographical location of the plant or information about the compounds processed in the plant,

[0012] (b1) providing the unstructured data and information about the chemical plant to a model suitable for extracting instructions from the unstructured data,

[0013] (b2) obtaining the instruction from the model and metadata including applicability of the instruction relative to at least one of a time period, a geographic range, or a compound to be processed in the facility, and

[0014] (c) Output the instructions received from the model.

[0015] The invention further relates to a non-transitory computer-readable data medium storing a computer program comprising instructions for performing the steps of the method according to any of the preceding claims.

[0016] The invention further relates to the use of the instructions obtained in any preceding claim for monitoring and / or controlling a chemical plant.

[0017] The present invention further relates to a production monitoring and / or control system for monitoring and / or controlling a chemical plant, comprising:

[0018] (a) an input unit configured to receive unstructured data containing instructions for monitoring and / or controlling a chemical plant, and configured to receive information about the chemical plant, the information including a geographical location of the plant or information about compounds processed in the plant,

[0019] (b) a processing unit configured to provide the unstructured data and information about the chemical plant to a model adapted to extract instructions from the unstructured data, and metadata including applicability of the instructions with respect to at least one of a time period, a geographical range, or a chemical compound to be processed in the plant, and

[0020] (c) An output unit configured to output the instructions received from the model.

[0021] Preferred embodiments of the invention can be found in the description and claims. Combinations of different embodiments fall within the scope of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A possible implementation of the present invention is shown.

[0023] Figure 2 Another possible implementation of the present invention is shown. DETAILED DESCRIPTION

[0024] The present invention relates to a method for monitoring and / or controlling a chemical plant. Monitoring generally refers to observing and recording any operating state of a chemical plant. Operating states include internal parameters, i.e., those parameters relevant only within the plant, such as reactor temperature, pressure, power consumption, input or output material flows, agitator speed, valve status, vapor concentration in the plant air, and the number of people present. Operating states also include external parameters, i.e., parameters related to any interaction with the plant environment, such as the emission of chemical vapors, heat, sound, vibration, and light. Recording can involve storing raw data on a permanent data storage device or preparing files in a format required by the company or authorities.

[0025] Control generally means taking any action to change the operating state of a chemical plant. These actions can be direct, such as by changing the state of a valve, altering the temperature through additional heating or cooling, or by prompting an operator to take action, such as changing a filter or adjusting throughput.

[0026] A chemical plant is any facility that carries out chemical reactions to produce chemical compounds, to produce formulations by mixing chemical compounds, to increase the purity of chemical compounds, to make chemical compounds in different forms, or to package chemical compounds or formulations containing chemical compounds. In many cases, a chemical plant can house more than one of these activities. Examples of chemical plants include oil refineries; petrochemical plants such as steam crackers, ethylene oxide plants, carbon monoxide plants, and methanol plants; intermediate chemical plants such as those that produce acrylic acid, toluene diisocyanate, and tetrahydrofuran; plants that produce inorganic substances such as sulfuric acid, chlorine, or ferric chloride; plants that produce pharmaceuticals or agrochemicals; plants that produce food and feed such as aroma chemicals and nutritional formulations; plants that produce home and personal care chemicals and formulations; plants that produce polymers such as polyethylene, polystyrene, or polyethylene terephthalate; plants that produce dispersions; plants that produce pigments; plants that produce coatings and paints; and plants that increase the purity of chemical compounds, such as for analytical, pharmaceutical, nutritional, or microchip production.

[0027] The method according to the present invention includes (a1) providing unstructured data containing instructions for monitoring and / or controlling a chemical plant. Unstructured data is generally understood data that either does not have a predefined data model or is not organized in a predefined manner. Preferably, the unstructured data is in a format containing characters, such as ASCII code. Preferably, the unstructured data is human-readable. Examples of applicable formats are txt, pdf, html, xml, docx, rtf, odt, postscript, LaTeX, dvi, eml. If the unstructured data is only available in a format that does not contain characters, such as a paper scan bitmap, the data is preferably preprocessed to convert it into unstructured data containing characters, in particular into one of the preferred data formats. Various technologies are available, such as optical character recognition (OCR). If the unstructured data is available as a collection of various formats, it is preferred that they are converted to the same format.

[0028] Unstructured data can come from a variety of sources, including technical data sheets, manuals, plans, shipping orders, laws, directives, guidelines, scientific articles, reports. Preferably, the unstructured data originates from more than one source, such as at least two, at least three or at least four. Unstructured data often comes from multiple different parts of the same source, such as from multiple technical data sheets or multiple laws. For example, a technical data sheet may contain instructions to change a filter when emission values exceed a certain threshold or to renew lubrication when certain vibrations occur. A manual may contain instructions on what to do if a valve is blocked. Legal texts such as laws, directives and guidelines are often related to health and safety at work, environmental protection or resource management. They may contain obligations to record the concentration of certain compounds in the air or to reduce the discharge of warm water into a river during hot weather conditions. A scientific article or report may contain instructions to optimize operating parameters in order to increase product yield or reduce equipment wear.

[0029] Unstructured data contains instructions. These instructions are typically human-readable. Instructions can be direct or indirect. An example of a direct instruction is to measure and record the temperature of wastewater. An example of an indirect instruction is to take appropriate precautions when handling waste. Indirect instructions often must be combined with direct instructions from other data sources; in the previous example, these might be instructions for waste disposal. Instructions can be in one language or multiple languages, such as English, German, French, Spanish, Portuguese, Chinese, Japanese, Korean, Russian, or Arabic.

[0030] In many cases, unstructured data contains many instructions as well as further information that doesn't qualify as an instruction. This can even be the case if the unstructured data originates from a single source. A typical example is a factory manual containing hundreds of pages. Therefore, it's preferable to parse the unstructured data before feeding it to the model. This way, smaller sections are obtained that are most likely to contain no more than one instruction each. At the same time, data that clearly doesn't contain any instructions, such as formatting commands, can be removed. In simple examples, sentences or paragraphs might be examples of such smaller sections. However, more complex methods that identify logical units can be used. Various libraries are available to perform this type of parsing, such as PDFMiner.

[0031] The method according to the present invention comprises that (a2) provides information about chemical plant by interface, and this information at least comprises the geographical location of factory or the information of the compound processed in factory.Geographical location can comprise the country and / or state of GPS coordinates, factory location, to the distance or altitude of water area similar to river, lake or sea.The information about the compound processed in factory can comprise the chemical structure of compound, its each (such as monthly or annual) amount used, or the amount existing at factory place (such as in storage facility) at a specific time.Preferably, the information about compound comprises that this compound is used as reagent, intermediate or the information as product.

[0032] The method according to the present invention comprises (b1) providing unstructured data and information about the chemical plant to a model suitable for extracting instructions from unstructured data. The model is preferably a data-driven model. The data-driven model is a trained mathematical model that is parameterized according to training data to input unstructured data and output structured data, i.e., instructions for monitoring and / or controlling the chemical plant in this case. The data-driven model is preferably a data-driven machine learning model. The data-driven model can be a linear or polynomial regression, a decision tree, a random forest model, a Bayesian network, or a neural network, preferably a neural network. Even more preferably, the neural network is a recursive neural network, in particular a neural network comprising a long short-term memory (LSTM) or a gated recurrent unit (GRU).

[0033] Preferably, unstructured data is provided to the model in a vectorized form. A typical method for vectorizing unstructured data is term frequency-inverse document frequency (tf-idf). Using more advanced techniques for vectorizing unstructured data can achieve higher accuracy, particularly continuous bag of words (CBOW) or continuous skip grammars available in the Word2vec library.

[0034] The model may have been trained using historical data. Historical data in the context of the present invention refers to a dataset that includes instructions as unstructured data and associates them with instructions in a structured data format. Historical data can be generated by manually labeling data or storing user feedback. In the latter case, the pre-trained model extracts instructions from the unstructured data, provides them to the user, and then the user provides feedback on the results. This feedback can be in the form of a rating from poor to perfect, or it can be in the form of corrections to the results. Results with high ratings or corrections can be used as additional historical data to further train the model.

[0035] The more diverse the unstructured data is and the more detailed the instructions that need to be obtained from the model, the more historical data that needs to be obtained. If the task of the present invention has been performed manually for many plants before and the results are stored in a way that can be accessed in a systematic way, then a considerable amount of historical data may be available. However, this is often not the case. Even if a large amount of historical data is available, the historical data may be unbalanced or biased, for example, there may be only a few data sets for a specific parameter or a class related to metadata. Therefore, it is advantageous to artificially increase the amount of historical data by oversampling, for example by random oversampling, synthetic minority oversampling technique (SMOTE) or adaptive synthetic sampling (ADASYN).

[0036] The instructions obtained from the model can be in any machine-readable format, such as XML, JSON, or YAML. Instructions typically contain the snippets of information needed to monitor and / or control a chemical plant. Instructions typically include a subject, i.e., the content to be monitored or controlled, and the action to be taken on the subject. Instructions can also include timing information, such as the time period until the action must be taken or the frequency of such action. Instructions can also include information about the operator, i.e., the person who is required to execute the instruction, such as a safety officer or plant manager. As an example, for a filter exchange, instructions in XML format might look like the following.

[0037]

[0038] According to the present invention, in addition to instructions including at least one of a time period, a geographical range, or a compound to be processed in the plant, the model is able to extract further metadata from the unstructured data. Instructions may only apply during a specific time period, such as only during the winter or limited to the next few years. Instructions may only apply to plants within a specific geographical area (such as a country, state, town, or a location within a certain distance from a body of water or a settled area). Instructions may only apply to plants that process certain compounds (such as heavy metals, volatile organic compounds, explosives, or radioactive materials). Metadata can be used to select only those instructions that are relevant to a specific plant. To do this, corresponding information needs to be provided for each plant so that the metadata matches the information about the plant. If there is a mismatch, the instructions for that plant are deleted.

[0039] Thus, the model can extract instructions and tag them with information about which plant the instruction is relevant to. Preferably, the model can also tag instructions with information about the products the instruction involves. Preferably, the model can also tag instructions with information about which person in a specific plant (e.g., a safety consultant or maintenance team) the instruction is relevant to.

[0040] The model may output similar instructions from different parts of the unstructured data. The same instruction may be output twice or even more because it is contained in different data sources, such as a technical data sheet and a plant manual. In some cases, there are two instructions related to the same operating state of the plant, but requiring different actions. An example might be that national law requires the plant to limit its emissions of volatile organic compounds to the air to a certain value. At the same time, internal company documents require the plant to limit its emissions of volatile organic compounds to the air to a different value, which may be lower than the former. Therefore, instructions are preferably grouped, with each group containing all instructions related to the same operating state of the plant. Even more preferably, the groups are sorted, with the most relevant instructions placed first. Relevance is determined based on the most stringent action, such as the lowest emission limit or the shortest time period for a certain action. The rules used to determine relevance can be preset, or they can be entered by a user (e.g., a plant operator).

[0041] The method according to the present invention comprises (c) outputting the instructions received from the model. Outputting may refer to writing the instructions to a non-transient data storage medium, displaying them on a user interface, or transmitting them to a control unit that puts the instructions into physical action. Preferably, the instructions are output by displaying them on a user interface. The user interface is preferably adapted to receive a selection, modification, priority, or execution date for each or a group of instructions from a user (e.g., a plant operator). The instructions with associated user input may be stored on a permanent storage medium or transmitted to a control unit.

[0042] Preferably, the user interface has the functionality to list the instructions and sort them by certain criteria, such as the instruction's expiration date. Preferably, the user interface has the functionality to display only those instructions that are ranked highest within their group, which contains all instructions related to the same operational state of the plant. Preferably, the user interface has the functionality to display the instructions in a calendar, where each instruction is placed in the calendar according to its expiration date.

[0043] Preferably, the model is adapted to classify instructions related to associated actions. This classification can distinguish between actions for monitoring and actions for control. For actions related to monitoring, instructions are preferably transmitted to a control unit. The control unit can be connected to sensors that acquire information about the plant status. The control unit can be adapted to select the data required for the instructions from the corresponding sensors and store the results accordingly. The control unit can even be adapted to insert the data into a form template. Such a form template may be required for internal company documentation or may be submitted to officials such as regulatory authorities.

[0044] If instructions for monitoring or controlling a chemical plant are transmitted to a control unit, the instructions are typically converted into signals suitable for triggering the monitoring or control equipment. This conversion is typically performed in the control unit. However, other processing units can also be used for the conversion.

[0045] Preferably, the output of the instructions, along with metadata (if available), is stored in a database, preferably a graph database. The database associates the instructions and metadata with the factory and its information. Preferably, the database associates each instruction with its source. This allows the process according to the present invention to be performed on updated versions of unstructured data sources. After extracting new instructions, the old instructions can be replaced in the database. By associating the replaced instructions with the factory to which they relate, each factory can be informed of the update in a very short period of time. Therefore, preferably, only those instructions that are new or have changed relative to the last output for a particular factory are output. Alternatively, if certain changes occur at a factory, such as replacing a raw material with a different one, the necessary instructions can be extracted from the database. If certain changes occur at one or more factories, such as shifting production of a product from one factory in one region to a different factory in a different region, it is also possible to simulate the necessary actions. It is also conceivable to optimize the chain of production steps distributed across different factories by extracting instructions from the database for each scenario, thereby obtaining the optimal set of instructions, for example with respect to cost, environmental impact, or the time required to implement the instructions.

[0046] Preferably, the computer-implemented method for monitoring and / or controlling a chemical plant comprises:

[0047] (a1) providing, via an interface, unstructured data containing instructions for monitoring and / or controlling a chemical plant,

[0048] (a2) providing information about a chemical plant through an interface, the information including at least the geographical location of the plant or information about the compounds processed in the plant,

[0049] (b1) providing unstructured data to a model suitable for extracting instructions from the unstructured data,

[0050] (b2) obtaining the instruction from the model and metadata including applicability of the instruction relative to at least one of a time period, a geographic range, or a compound to be processed in the facility,

[0051] (b3) grouping the instructions into groups, wherein each group contains all instructions related to the same operating state of the plant,

[0052] (c) outputting instructions received from the model to the user interface, and

[0053] (d) Receive user feedback on the instructions for further training of the model.

[0054] The present invention further relates to a non-transitory computer-readable data medium storing a computer program comprising instructions for executing the steps of the method according to the present invention. The computer-readable data medium comprises, for example, a hard drive on a server, a USB memory device, a CD, a DVD, or a Blu-ray disc. The computer program may contain all the functions and data required to execute the method according to the present invention, or it may provide an interface to allow parts of the method to be processed on a remote system (e.g., a cloud system).

[0055] The present invention further relates to a production monitoring and / or control system for monitoring and / or controlling material properties of a sample. Unless otherwise expressly described below, the descriptions relating to the method, including the preferred embodiments, also apply to the system. The system can be a computing device, such as a computer, tablet computer, or smartphone. Typically, the computing device has a network connection for communicating with other computing devices, such as a server or cloud network.

[0056] The production monitoring and / or control system according to the present invention comprises (a) an input unit configured to receive unstructured data containing instructions for monitoring and / or controlling a chemical plant. Preferably, the input unit comprises an interface, in particular a user interface that allows a user to select the unstructured data to be processed, for example, from a local or remote storage medium. According to the present invention, the input unit is configured to receive information about the chemical plant, the information including the geographical location of the plant or information about the compounds processed in the plant. The input unit can provide predefined options to select or enable free input. The input can have an interface to a database that contains data about the plant or preferably multiple plants, in particular all plants of a company or group of companies. The input unit can be implemented as a network service or a standalone software package. The input unit can form a presentation layer or an application layer.

[0057] The production monitoring and / or control system according to the present invention includes (b) a processing unit configured to provide unstructured data and information about the chemical plant to a model suitable for extracting instructions and metadata from the structured data, the metadata including the applicability of the instructions related to at least one of a time period, a geographical range, or a chemical compound to be processed in the plant. The processing unit can be a local processing unit, including a central processing unit (CPU) and / or a graphics processing unit (GPU) and / or an application-specific integrated circuit (ASIC) and / or a tensor processing unit (TPU) and / or a field programmable gate array (FPGA). The processing unit can also be an interface to a remote computer system (such as a cloud service).

[0058] The production monitoring and / or control system according to the present invention comprises (c) an output unit configured to output instructions received from the model. The output unit can be implemented as a network service or a stand-alone software package. The output unit can form a presentation layer or an application layer. Preferably, the output unit comprises an interface for outputting the instructions received from the model, in particular a user interface configured to display instructions for the factory. The user can then take necessary actions, such as adjusting production parameters or collecting sensor data. Preferably, the user interface is configured to receive feedback from the user about the instructions that can be used to further train the model. Alternatively, the output unit can include or have an interface with a device that automatically adjusts production parameters or collects sensor data. Preferably, the output unit has an interface to a database to store the instructions in a database, in particular a graph database. In another run of the production and / or control system, the database can be used to select those instructions that are new or have changed relative to the last run.

[0059] There are several ways to implement the present invention. One is Figure 1 . Unstructured data (10), which may or may not be filtered according to its relevance to a particular plant, is provided to a processing unit (11). The processing unit provides the unstructured data to a data-driven model trained on historical data. The processing unit obtains instructions from the model, which may or may not be grouped, wherein each group contains all instructions related to the same operating state of the plant. The instructions are provided to an output unit (12) which outputs the instructions, for example by displaying a list sorted by expiration date via a user interface (13). Each instruction may result in an action in the chemical plant (21) that is performed automatically by the control unit or manually (for example by a plant operator).

[0060] Figure 2 An alternative implementation is depicted in . Unstructured data (10) is provided to a processing unit (11). The processing unit provides the unstructured data to a data-driven model trained on historical data. The processing unit (11) obtains instructions from the model along with metadata, the metadata including at least one of a time period, a geographical range, or a compound to be processed in the plant. The processing unit (11) provides these data to an output unit (12). The output unit selects instructions relevant to each of the plants (21, 22, 23) by comparing the metadata with information about each of the plants (21, 22, 23) obtained from a database (31). The database (31) may also contain information about which instructions have been given to the plants (21, 22, 23), so that the output unit can select only those instructions that have not been given to any of the plants (21, 22, 23) or are updated versions. A plant manager of each of the plants (21, 22, 23) can take the required actions to monitor and / or control the plant based on the instructions received from the output unit (12).

Claims

1. A computer-implemented method for monitoring and / or controlling a chemical plant, comprising: (a1) providing unstructured data containing instructions for monitoring and / or controlling the operating status of a chemical plant, (a2) providing information about the chemical plant through an interface, the information including at least the geographical location of the chemical plant or information about compounds processed in the chemical plant, (b1) providing the unstructured data and information about the chemical plant to a model adapted to extract instructions and metadata from the unstructured data, wherein the model is trained using historical data including the unstructured data that has been labeled and / or user feedback, (b2) obtaining, from the model, the instructions extracted from the unstructured data and the metadata, the metadata including applicability of the instructions with respect to at least one of a time period, a geographical range, or the chemical compound to be processed in the chemical plant, and tagging the instructions with at least one of information about the chemical plant, information about the product, and information about a person associated with the instructions; and (c) outputting the instruction received from the model, wherein information marking the instruction matches the metadata.

2. The method according to claim 1, wherein The model is a neural network.

3. The method according to claim 2, wherein: The neural network includes long short-term memory.

4. The method according to any one of claims 1 to 3, wherein The instruction is output on the user interface.

5. The method according to claim 4, wherein The user interface is adapted to receive user feedback on the instructions for further training of the model.

6. The method according to any one of claims 1 to 3, wherein Information about the chemical plant is provided, the information including at least the geographical location of the chemical plant or information about the compounds processed in the chemical plant.

7. The method according to any one of claims 1 to 3, wherein Metadata is obtained from the model, the metadata comprising the applicability of the instruction relative to at least one of a time period, a geographic range, or the chemical compound processed in the chemical plant.

8. The method according to any one of claims 1 to 3, wherein The instructions obtained from the model are grouped, wherein each group contains all instructions related to the same operating state of the chemical plant.

9. The method according to any one of claims 1 to 3, wherein Only those instructions are output that are new or have changed with respect to the last output to the particular chemical plant.

10. A non-transitory computer-readable data medium storing a computer program comprising instructions for executing the steps of the method according to any one of claims 1 to 9.

11. Use of the instructions obtained according to any one of claims 1 to 10 for monitoring and / or controlling a chemical plant.

12. A production monitoring and / or control system for monitoring and / or controlling a chemical plant, comprising: (a) an input unit configured to receive unstructured data containing instructions for monitoring and / or controlling an operating state of a chemical plant, and configured to receive information about the chemical plant, the information including a geographical location of the chemical plant or information about compounds processed in the chemical plant, (b) a processing unit configured to provide the unstructured data and the information about the chemical plant to a model adapted to extract the instructions and metadata from the structured data, and to tag the instructions with at least one of information about the chemical plant, information about the product, and information about the person associated with the instructions, wherein the model is trained using historical data including the unstructured data that has been tagged and / or user feedback, and the metadata includes applicability of the instructions with respect to at least one of a time period, a geographical range, or the chemical compound to be processed in the chemical plant; and (c) an output unit configured to output the instruction received from the model, wherein information marking the instruction matches the metadata.

13. The production monitoring and / or control system according to claim 12, wherein: The input unit includes an interface for receiving unstructured data to be processed, and the output unit includes an interface for outputting the instructions.

14. The production monitoring and / or control system according to claim 12, wherein: The output unit includes a user interface, and the output unit includes a user interface.

15. The production monitoring and / or control system according to any one of claims 12 to 13, wherein: The output unit has an interface to a database to store the instructions in the database.

16. The production monitoring and / or control system according to any one of claims 12 to 13, wherein: The output unit has a user interface configured to receive feedback from the user regarding the instructions that can be used to further train the model.

Images