A computer implemented invention for processing logging data from one or more industrial packaging or labelling machine

Abstract

A computer-implemented method processes logging data from industrial packaging or labeling machines to identify anomalies. Each machine has multiple electronic components that generate logging data. A local processing unit (industrial personal computer, IPC) is positioned near the machines and includes an interface module, communication module, storage module, and processing module. The method involves the following steps: The interface module receives logging data from the machine components, which the storage module stores locally as raw data. The processing module then refines this raw data into a smaller, more manageable dataset by performing a data reduction process, filtering for relevant anomaly-related information. This process includes employing lossless data compression and determining correlations between different data elements to identify patterns and relationships. These relational datasets are stored alongside the refined dataset. Finally, the communication module transmits the refined dataset to a remote server for storage. This method significantly reduces the data volume while retaining crucial information for anomaly detection and analysis, optimizing data processing and storage efficiency.

Description

[0001] Title: A computer implemented invention for processing logging data from one or more industrial packaging or labelling machine

[0002] Description:

[0003] TECHNICAL FIELD

[0004] The present invention generally relates to the field of industrial automation and control systems, and more particularly to the field of data processing and anomaly detection in industrial packaging or labelling machines.

[0005] BACKGROUND

[0006] In the current state of the art, complex industrial packaging and labelling machines comprise a plurality of electronic components, such as PLCs, CPUs, microprocessors, smart sensors, encoders etc. which generate a vast amount of logging data. This data is crucial for monitoring, control, and maintenance purposes, containing detailed information about the performance and status of various machine components. Typically, the generated data is stored locally, creating a need for significant storage capacity due to the sheer volume of data. This approach necessitates heavy storage specifications, making it cumbersome and expensive to maintain. Furthermore, the process of debugging, maintenance, and inspection requires local presence, adding to operational inefficiencies.

[0007] T o address the storage issue, another approach has been to transfer the logging data to remote locations, such as cloud servers. However, this method demands substantial bandwidth to handle the large data transfers, which can be both cumbersome and costly. Additionally, the transfer process might introduce delays, impacting the timely availability of crucial data for remote diagnostics and maintenance. A third approach involves pre-selecting data that is likely to be relevant for remote storage and inspection. This method attempts to mitigate storage and bandwidth issues by reducing the data volume before transmission. However, it is challenging to determine in advance which data will be relevant, leading to potential loss of valuable information and missing correlations between different data elements. Each of the existing solutions has significant disadvantages. Local data storage requires extensive and expensive hardware, and it requires a lot of resources and time to process vast amounts or raw data. Remote data transfer demands high bandwidth and incurs ongoing costs. Pre-selecting data for transmission risks omitting critical information, compromising the effectiveness of predictive maintenance and error handling. These challenges highlight the limitations of current methodologies in managing and processing the vast amounts of data generated by industrial machines.

[0008] It is therefore a goal of the present invention to provide an improved system for processing logging data from industrial packaging machines such as shrink sleeve labelling machines, spouted pouches machines, cartooning machines, pressure-sensitive labelling machines, steam tunnels for shrink sleeve labelling machines and similar equipment. This system aims to efficiently reduce the data volume while preserving relevant information, thereby overcoming the above- mentioned disadvantages of the prior art at least in part.

[0009] SUMMARY

[0010] In a first aspect, there is provided, a computer implemented method for processing logging data from one or more industrial packaging or labelling machine for identifying anomalies, the machine comprising a plurality of electronic components for measuring and control of components within the machine, each component generating logging data, the method being performed by a processing unit, such as a local computer like an industrial personal computer, IPC, arranged to be located in the vicinity of the one or more industrial packaging or labelling machine, and comprising an interface module, a communication module, a storage module, and a processing module, the interface module being arranged to interface with the plurality of electronic components of the one or more industrial packaging or labelling machine, the communication module being arranged to communicate with a remote server, the storage module being arranged to store data, the processing module being arranged to store and execute computer program code to perform the method comprising the steps of: receiving, from the interface module, logging data from the plurality of electronic components of the one or more industrial packaging or labelling machine connected to the interface module; storing, by the storage module, the logging data received by the interface module as raw data locally in a local data storage module comprised in the storage module; processing, by the processing module, at least some of the raw data into a refined dataset, wherein the amount of refined dataset is at least one order of a magnitude smaller than the raw data; characterized by the processing module executing an algorithm comprising the steps of: performing a data reduction process by modelling the raw data to filter and select relevant logging data relating to anomalies, reducing the data volume into the refined dataset; employing lossless data compression on the refined dataset; determine correlations between different data elements comprised in the raw data, to identify patterns and relationships between the data elements, relating to anomalies, and storing these relational data sets alongside the refined dataset, and transmitting, through the communication module, the refined dataset to the remote server for remote storage in a remote data storage module comprised in the remote server.

[0011] In a second aspect, there is provided, a system for processing logging data from one or more industrial packaging or labelling machines for identifying anomalies, the machine comprising a plurality of electronic components for measuring and control of components within the machine, each component generating logging data, the system comprising:

[0012] - a processing unit, such as a local industrial personal computer (IPC) arranged to be located in the vicinity of the one or more industrial packaging or labelling machines, the processing unit comprising:

[0013] - an interface module arranged to interface with the plurality of electronic components of the one or more industrial packaging or labelling machines,

[0014] - a communication module arranged to communicate with a remote server, - a storage module arranged to store data,

[0015] - a processing module arranged to store and execute computer program code to perform the method, wherein the processing module is configured to:

[0016] - receive logging data from the plurality of electronic components of the one or more industrial packaging or labelling machines via the interface module,

[0017] - store the received logging data as raw data locally in a local data storage module comprised in the storage module,

[0018] - process at least some of the raw data into a refined dataset, wherein the amount of refined dataset is at least one order of magnitude smaller than the raw data, by executing an algorithm that:

[0019] - performs a data reduction process by modeling the raw data to filter and select relevant logging data relating to anomalies, reducing the data volume into the refined dataset,

[0020] - employs lossless data compression on the refined dataset,

[0021] - determines correlations between different data elements comprised in the raw data, to identify patterns and relationships between the data elements relating to anomalies, and stores these relational data sets alongside the refined dataset,

[0022] - transmit the refined dataset to the remote server for remote storage in a remote data storage module comprised in the remote server via the communication module.

[0023] In a third aspect, there is provided a computer program product for processing logging data from one or more industrial packaging or labelling machines for identifying anomalies, the computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of, or an example of, the first aspect of the present disclosure.

[0024] One aspect of the present invention relates to a computer-implemented method for processing logging data from one or more industrial packaging or labelling machines for identifying anomalies. A computer-implemented method may be understood as a method performed by a computer system or network, involving software and hardware components to execute specific tasks. Processing logging data refers to the operations carried out to handle, analyze, and manipulate the data generated by the electronic components of the machines for various purposes, including anomaly detection. An industrial packaging or labelling machine is a type of machinery used in industrial processes to package or label products, often involving multiple electronic components for control and measurement.

[0025] The method comprises receiving logging data from the electronic components of the industrial packaging or labelling machine. In the context of the present disclosure, the term "packaging or labeling machine" refers to any equipment or machinery designed for the application, processing, or handling of labels, packaging materials, or containers in a production line. This includes, but is not limited to, machines for applying shrink sleeve labels, pressure-sensitive labels, spouted pouch systems, cartoning, sealing, and filling, as well as associated systems for product handling, configuration, and integration within packaging processes. The term is intended to encompass all types of machines that perform functions related to packaging or labeling, regardless of specific design, method of operation, or the type of packaging material used.

[0026] The method comprises receiving logging data from the electronic components of the industrial packaging or labelling machine through an interface module. An interface module is a component that enables communication between different systems or components within a system. The effect of this arrangement is that it allows seamless integration and data flow from the machine’s electronic components to the processing unit ensuring that all relevant data is captured for further processing.

[0027] In the context of the present disclosure, the term "processing unit" refers to any form of hardware, and optionally software, capable of processing data in various forms. This includes general-purpose computers, industrial (personal) computers (I PC), embedded systems, microcontrollers, or any combination of hardware and software components designed to execute, control, or facilitate data processing tasks. The processing unit may be implemented in different configurations and architectures, ranging from standalone devices to integrated systems, capable of performing the steps of the method of the present disclosure. It is not limited to specific hardware types, allowing flexibility to encompass industrial computers and other specialized devices that may be required for processing data in the described system and method. Through the present disclosure, processing unit may also be referred to as industrial personal computer, I PC, but the skilled person may appreciate that it may be interpreted as any type of processing unit. The method includes storing the received logging data as raw data locally in a local data storage module comprised in the storage module. A storage module is a hardware component or assembly that stores data, while raw data refers to unprocessed data in its original form. An effect of this arrangement is that it provides immediate availability of the raw data for local processing, reducing the need for immediate data transfer and thus conserving bandwidth.

[0028] The method involves processing at least some of the raw data into a refined dataset, where the refined dataset is at least one order of magnitude smaller than the raw data. A processing module is a hardware and software component that executes computer programs to process data and to this end comprises one or more hardware processors such as microcontrollers, PLCs, ASICs, CPUs, GPUs, etc. An effect of this arrangement is the significant reduction in data volume, which optimizes storage capacity and processing power by focusing only on the relevant information needed for anomaly detection.

[0029] The processing module executes an algorithm comprising performing a data reduction process by modeling the raw data to filter and select relevant logging data relating to anomalies, thereby reducing the data volume into the refined dataset. An algorithm is a set of rules or steps performed by a computer. An effect of this arrangement is that it ensures only pertinent data is retained, enhancing the efficiency and effectiveness of subsequent analysis and anomaly detection processes.

[0030] The method further employs lossless data compression on the refined dataset. Lossless data compression is a method of reducing data size without losing any information. An effect of this arrangement is that it further decreases the storage requirements and transmission bandwidth while ensuring that all original information can be perfectly reconstructed, maintaining data integrity.

[0031] The method determines correlations between different data elements comprised in the raw data to identify patterns and relationships between the data elements relating to anomalies and stores these relational data sets alongside the refined dataset. It is expressed that the proposed method not only provides for determining correlations between data elements in respect of anomalies, but may in stead or in addition also determine improvements of the system. For example, it may be determined that a certain data element is related to the efficiency of a the machine or a component thereof, which thus provides insight that certain modifications resulting in the change of that particular data element, e.g. a sensor output, will improve the performance of the machine, might reduce wear or provide other useful insights. Determining correlations may involve statistical analysis to find relationships between data points, but may alternatively or additionally also involve employing machine learning algorithms. An effect of this arrangement is that it enhances the understanding of underlying patterns and relationships, which is crucial for accurate anomaly detection and predictive maintenance.

[0032] The method includes transmitting the refined dataset to the remote server for remote storage in a remote data storage module comprised in the remote server through the communication module. A communication module is a component that enables data exchange between devices or systems over a network. An effect of this arrangement is that it allows for centralized data storage and analysis, facilitating remote monitoring and diagnostics while minimizing bandwidth usage due to the reduced data volume.

[0033] The inventors had foreseen that the traditional methods of handling large volumes of logging data from industrial packaging or labelling machines were not sufficient to address the challenges of storage capacity, bandwidth usage, and effective anomaly detection. By integrating a local industrial personal computer with specific modules, or referred to as algorithmic modules, which may be referred to as bricks, for interfacing, communication, storage, and processing, a method was developed that not only reduces the data volume significantly but also preserves the most relevant information. The surprising effect of this approach is that it enables efficient local data processing and storage, drastically minimizing the amount of data that needs to be transmitted to remote servers. This arrangement effectively balances the need for comprehensive data analysis and the practical constraints of storage and bandwidth, leading to a more robust and scalable solution for predictive maintenance and error handling in industrial environments.

[0034] In an example, a computer-implemented method for processing logging data from industrial packaging or labelling machines for identifying anomalies includes a processing module executing an algorithm that performs a data reduction process by modeling raw data to filter and select relevant logging data relating to anomalies, employs lossless data compression on the refined dataset, and determines correlations between different data elements comprised in the raw data, to identify patterns and relationships between the data elements relating to anomalies. It may be provided that this fixed and sequential application of data reduction followed by lossless data compression on the resulting refined dataset ensures the integrity and completeness of anomaly-detection data. An effect of this arrangement is that it preserves the full fidelity of pertinent data, which is beneficial for accurate anomaly detection and troubleshooting in industrial settings by ensuring no critical anomaly- related information is lost due to selective compression strategies. The data reduction step optimizes storage capacity and processing power by focusing only on relevant information, and subsequently, lossless compression further decreases storage requirements and transmission bandwidth while ensuring all original information can be perfectly reconstructed. The determination of correlations provides context and deeper insights into identified anomalies, supporting robust predictive maintenance.

[0035] In an example, the method further comprises a timing analyzer module that measures the time taken for a product to move from one location to another within the machine, based on sensor data. It may be provided that this feature allows for the precise monitoring of machine performance. An effect of this feature is the ability to track product movement and detect any deviations or delays, which is particularly relevant for packaging and labelling machines operating at high speeds. This enables timely interventions and adjustments to maintain optimal operation, thereby improving production efficiency and throughput by identifying bottlenecks, misalignments, or impending mechanical failures that affect production rates and product quality.

[0036] In an example, the system includes a feedback mechanism that allows users to provide input on the relevance of detected anomalies, which is then used to refine the processing algorithms. It may be provided that this feature allows for the continuous refinement of the anomaly detection processes, leveraging user insights to enhance system accuracy and effectiveness. An effect of this feature is that human operators, with their intimate knowledge of specific industrial processes, can guide the algorithms to distinguish between critical anomalies and benign variations. This leads to a more robust and contextually aware anomaly detection system, improving its ability to provide relevant insights by incorporating real-world operational context.

[0037] In an example, the method includes a correlation detection mechanism that identifies relationships between data elements such as motor information and sensor data, and stores these relational data sets alongside the refined dataset. It may be provided that this feature enhances the ability to detect and understand complex anomalies and interactions within the machinery. An effect of this feature is that it leads to more effective predictive maintenance and error resolution by providing deeper diagnostic insights into the interplay between motor performance and various sensor readings. This is particularly beneficial for diagnosing complex anomalies like misfeeds, jams, or wear in mechanical components that drive product movement in industrial packaging and labelling machines, thereby contributing to increased operational stability and reduced downtime.

[0038] In an example, a processing module comprises a plurality of algorithmic modules each defined by a configuration file such as a JSON file, enabling addition or modification of modules without reprogramming or manual intervention. It may be provided that configuration by JSON may be understood as the structured declaration of module parameters and interconnections in a human-readable and machineparsable format. An effect of this feature is that modules can be reconfigured remotely or automatically without recompilation of code, reducing downtime and enabling faster adaptation of algorithms to different machine types. The modular configuration also optimizes memory and processor allocation by loading only the required algorithms defined in the configuration file.

[0039] In an example, a communication architecture employs a domain-based routing mechanism that directs communication between algorithmic modules according to domain identifiers assigned to each module. It may be provided that a domain-based routing mechanism may be understood as a logical separation of communication paths between modules handling different classes of data, such as safety-critical control data and non-critical diagnostic data. An effect of this feature is that message traffic is efficiently organized and isolated, preventing cross-interference between unrelated processes and reducing latency by ensuring that only relevant data are exchanged within each domain.

[0040] In an example, the refined dataset transmitted to a remote server comprises metadata including a machine identifier, a station identifier, a sequence number and synchronized timestamps defining the chronological order of the processed data. It may be provided that such metadata may be understood as supplementary information describing the origin and temporal context of each portion of the refined dataset. An effect of this feature is the ability to reconstruct the complete operational history of a product or process instance across multiple machines, which improves traceability and supports correlation of anomalies occurring at different stages of the production line.

[0041] In an example, the sequential algorithm executed by the processing module is defined as a deterministic pipeline comprising a data-reduction process that filters relevant data, a lossless compression of the resulting refined dataset, and a subsequent correlation analysis of data elements. It may be provided that a deterministic pipeline may be understood as an algorithmic structure in which each processing step is executed in a fixed, non-adaptive order and with reproducible results. An effect of this feature is the consistent generation of refined datasets with guaranteed information completeness, ensuring that the transmitted data can always be decompressed and correlated without uncertainty. This improves the reliability of remote analysis while maintaining efficient use of local storage and network bandwidth.

[0042] In an example, the local storage module is configured to retain both raw and refined datasets together with user input and timing data for a defined retention period, allowing retrospective analysis and retraining of models based on verified events. It may be provided that such retention may be understood as maintaining synchronized data layers of different abstraction levels for later reuse. An effect of this feature is that the system can regenerate improved models using the same consistent datasets, avoiding the need for repeated raw data collection and thereby reducing bandwidth and processing load.

[0043] In an example, the processing module employs a machine-learning algorithm that updates its parameters using the stored feedback data and timing correlations. It may be provided that such an algorithm may be understood as a continuously learning model executed locally that integrates new labelled data into its parameter set. An effect of this feature is that the detection of anomalies becomes progressively more accurate for the specific machine configuration, which improves the stability and efficiency of the data-processing system and reduces unnecessary data transmission. In an example, the method comprises a processing module that is arranged to comprise a plurality of algorithmic modules that are configurable, allowing the modification to one or more of the algorithmic module to change variables of the data reduction process, data compression mechanism, or correlation detection mechanism in place. It may be provided that this feature allows for configurability of the data processing, enabling the system to optimize the data handling processes according to specific requirements or changes in operational conditions. An effect of this feature is the enhanced efficiency in data processing, storage, and transmission, as the system can tailor its operations to the most relevant and current data needs.

[0044] In a further example, the modification of variables of one or more of the algorithmic modules can be changed dynamically, either as a self-learning mechanism or based on user input. . It may be provided that this feature allows the system to dynamically adapt and improve its performance through self-learning algorithms or user-directed adjustments. An effect of this feature is the ability to maintain optimal data processing and anomaly detection performance over time, even as the operating environment or data patterns evolve.

[0045] In an example, the method includes a data reduction process performed by the algorithmic modules based on regression analysis to filter and select relevant logging data relating to anomalies. It may be provided that regression analysis models are used to identify and prioritize the most significant data points. An effect of this feature is the reduction of data volume while preserving the integrity and usefulness of the information, which enhances the efficiency of data storage and subsequent analysis.

[0046] In an example, the method includes data compression performed by the algorithmic modules using a lossless compression technique to ensure no loss of relevant logging data. It may be provided that lossless compression techniques are applied to maintain the full fidelity of the original data while reducing storage space requirements. An effect of this feature is the efficient utilization of storage capacity without compromising data quality, which is particularly beneficial for detailed anomaly detection and troubleshooting. In an example, the method includes a correlation detection mechanism that identifies relationships between data elements such as motor information and sensor data, and stores these relational data sets alongside the refined dataset. It may be provided that statistical and analytical methods are used to uncover and record significant data relationships. An effect of this feature is the enhanced ability to detect and understand complex anomalies and interactions within the machinery, leading to more effective predictive maintenance and error resolution.

[0047] In an example, the method comprises a timing analyzer module that measures the time taken for a product to move from one location to another within the machine, based on sensor data. It may be provided that timing analysis is used to track product movement and detect any deviations or delays. An effect of this feature is the precise monitoring of machine performance, allowing for timely interventions and adjustments to maintain optimal operation.

[0048] In an example, the method comprises employing algorithmic modules that are configured using a JSON configuration file that allows easy addition and modification of algorithmic modules without manual intervention. It may be provided that JSON files are used to standardize and simplify the configuration process. An effect of this feature is the streamlined setup and maintenance of the system, reducing the need for specialized technical expertise and minimizing the risk of configuration errors.

[0049] In an example, the method includes the step of the refined dataset being transmitted to the remote server, which includes metadata such as the history and routing information of the processed messages. It may be provided that detailed metadata is appended to the refined dataset to facilitate tracking and analysis. An effect of this feature is the improved traceability and context for the transmitted data, enhancing the ability to perform remote diagnostics and historical analysis.

[0050] In an example, the method includes a local data storage module that retains raw data for a predefined period to allow for retrospective analysis in case of anomaly detection. It may be provided that raw data is temporarily stored to enable thorough investigation of detected issues. An effect of this feature is the provision of a robust fallback mechanism for detailed troubleshooting, ensuring that no critical data is overlooked during the anomaly resolution process. In an example, the method includes a communication module that employs secure communication protocols to ensure the integrity and confidentiality of the data transmitted to the remote server. It may be provided that encryption and secure transmission methods are used to protect data during transfer. An effect of this feature is the safeguarding of sensitive information against unauthorized access and data breaches, maintaining the confidentiality and reliability of the transmitted data.

[0051] In an example, the method includes a processing module that uses machine learning algorithms to continuously improve the data reduction and anomaly detection processes based on historical data and user feedback. It may be provided that machine learning models are employed to refine and adapt the processing algorithms over time. An effect of this feature is the continuous enhancement of the system's accuracy and effectiveness, leveraging accumulated knowledge and user inputs to optimize performance.

[0052] In an example, the method comprises an interface module that supports multiple industrial communication protocols to interface with various types of electronic components within the packaging or labeling machine. It may be provided that the interface module is compatible with a wide range of industry-standard protocols. An effect of this feature is the increased versatility and interoperability of the system, enabling seamless integration with diverse machinery and electronic components.

[0053] In an example, the method comprises a user interface that allows users to configure and monitor the performance of the processing module and the algorithmic modules. It may be provided that a user-friendly interface is provided for system management. An effect of this feature is the enhanced accessibility and control for users, facilitating easier configuration, monitoring, and adjustment of the system's operations.

[0054] In an example, the method comprises the refined dataset with timestamps to allow for chronological analysis of the logging data. It may be provided that timestamps are added to the processed data to maintain temporal context. An effect of this feature is the ability to conduct time-based analysis and trend monitoring, providing valuable insights into the operational dynamics and historical performance of the machinery.

[0055] In an example, the method comprises a processing module that provides alerts to the remote server or local users when anomalies are detected based on the refined dataset. It may be provided that automated alert mechanisms are in place to notify relevant parties of detected issues. An effect of this feature is the prompt communication of critical information, enabling timely responses and mitigating the impact of potential problems.

[0056] In an example, the method comprises a domain-based routing mechanism to manage communication between algorithmic modules in different domains. It may be provided that domain-based routing is used to control and direct data flows between distinct system segments. An effect of this feature is the organized and efficient management of data traffic, ensuring appropriate and secure communication across different parts of the system.

[0057] In an example, the method comprises algorithmic modules that can process messages from multiple sources and aggregate the data before performing the data reduction process. It may be provided that multi-source data aggregation is implemented to consolidate input data. An effect of this feature is the comprehensive and unified processing of data from various origins, enhancing the robustness and reliability of the refined dataset.

[0058] In an example, the method comprises a feedback mechanism that allows users to provide input on the relevance of the detected anomalies, which is then used to refine the processing algorithms. It may be provided that user feedback is incorporated into the system to adjust and improve the algorithms. An effect of this feature is the continuous refinement of the anomaly detection processes, leveraging user insights to enhance system accuracy and effectiveness.

[0059] In an example of the computer-implemented method, the method may further comprise an operator interface module configured to allow operators to input information related to machine events, including maintenance activities, such as cleaning, incidents, or product breakage, wherein the inputted information is processed and stored alongside the refined dataset to enhance anomaly detection and system analysis.

[0060] It is to be understood that each feature described in the context of an example relating to one aspect of the invention, is similarly applicable to other aspects of the invention, including but not limited to a system, a computer-implemented method, a computer program product, and a data carrier signal. Each feature thus contributes to the overall functionality and technical advantages of the invention in its various embodiments.

[0061] Similarly, any advantages discussed in relation to specific features of an aspect or example of the present invention are equally pertinent to other aspects of the invention.

[0062] BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The present disclosure will be explained in more detail below by means of examples of a device according to the present disclosure shown in the drawings, in which:

[0064] Fig. 1 shows the steps of a method according to an aspect of the present disclosure;

[0065] Fig. 2 shows a system according to another aspect of the present disclosure.

[0066] DETAILED DESCRIPTION

[0067] Figure 1 illustrates a flowchart 100 that details the steps of the computer- implemented method for processing logging data from one or more industrial packaging or labeling machines. The method is designed for identifying anomalies and optimizing data processing. The flowchart is structured in a sequential manner, each step numbered, but it is expressed that the sequence may be changed and that some or all steps may be performed in a different order, or partly or fully in parallel.

[0068] The first step, labeled 101 , involves receiving logging data. This step signifies the action of collecting logging data from the various electronic components within the packaging or labeling machine. The interface module of the local processing unit, such as an industrial personal computer (IPC), facilitates this data reception,

[0069] In the context of the present disclosure, the term "packaging or labeling machine" refers to any equipment or machinery designed for the application, processing, or handling of labels, packaging materials, or containers in a production line. This includes, but is not limited to, machines for applying shrink sleeve labels, pressure-sensitive labels, spouted pouch systems, cartoning, sealing, and filling, as well as associated systems for product handling, configuration, and integration within packaging processes. The term is intended to encompass all types of machines that perform functions related to packaging or labeling, regardless of specific design, method of operation, or the type of packaging material used.

[0070] In the context of the present disclosure, the term "processing unit" refers to any form of hardware, and optionally software, capable of processing data in various forms. This includes general-purpose computers, industrial computers, embedded systems, microcontrollers, or any combination of hardware and software components designed to execute, control, or facilitate data processing tasks. The processing unit may be implemented in different configurations and architectures, ranging from standalone devices to integrated systems, capable of performing the steps of the method of the present disclosure. It is not limited to specific hardware types, allowing flexibility to encompass industrial computers and other specialized devices that may be required for processing data in the described system and method. Through the present disclosure, processing unit may also be referred to as industrial personal computer, I PC, but the skilled person may appreciate that it may be interpreted as any type of processing unit.

[0071] In step 102, the received logging data is stored locally as raw data. This storage occurs within the local data storage module that is part of the storage module of the processing unit, alternatively, the actual storage itself may also be performed at a different, e.g. local or remote, location in a dedicated storage.

[0072] The next step, 103, involves processing the logging data. During this step, at least some of the raw data is processed into a refined dataset, significantly smaller in volume compared to the original raw data.

[0073] The reduction in data volume may be achieved through one or a series of data filtering and compression techniques designed to retain only the most relevant information needed for anomaly detection and other relevant information.

[0074] The refined dataset may be created by first modeling the raw data to filter out unnecessary or redundant information. For instance, logging data that consistently falls within normal operational ranges over time can be binned or aggregated into statistical summaries, such as mean, median, or standard deviation. This binning process compresses large quantities of routine data into smaller, representative values, thus reducing the overall dataset size. Only critical data points, such as deviations, sudden spikes, or patterns that indicate potential anomalies or other differences, may be preserved in their full detail.

[0075] Data reduction may also involve prioritizing the storage of key operational parameters that are indicative of machine health. For example, sensor data that directly correlates with known failure modes (e.g., motor temperature spikes, vibration frequencies, or pressure variations) may be kept in its original form or with higher precision. In contrast, less significant or auxiliary data (e.g., constant environmental conditions) might be abstracted or removed entirely.

[0076] After filtering, the processing module may employ lossless data compression techniques, described in more detail below, to further reduce the dataset size. This ensures that while the dataset volume is minimized, the integrity of the critical information is maintained. The refined dataset may also include calculated correlations between various data elements, providing insight into patterns and relationships that may relate to anomalies. These correlations may be stored alongside the refined dataset, enhancing the system's ability to perform targeted, predictive analysis and enabling the identification of complex fault scenarios. Thus, the refined dataset becomes a focused, highly informative subset of the raw data, optimized for remote transmission and analysis.

[0077] Step 104 is the data reduction process. During this step, the processing module may execute an algorithm to model the raw data, filtering and selecting relevant information associated with anomalies. This step may involve several layers of data analysis, where the algorithm distinguishes between routine operational data and potential indicators of anomalies. For instance, standard operational parameters, such as steady temperature readings or consistent motor speeds, may be aggregated into statistical summaries like mean values, ranges, or variance metrics. The data reduction process may also involve binning time-series data, where sensor readings are grouped into intervals that represent normal operational behavior. For example, if a sensor measuring motor vibration shows consistent values within a specific range, these readings could be summarized rather than stored individually, thereby reducing the dataset's volume.

[0078] The algorithm could also employ techniques such as regression analysis or machine learning models to identify patterns within the data that correlate with known fault conditions. In practical embodiments, data reduction may include filtering out noise or insignificant data points. For instance, a packaging machine might have sensors monitoring environmental conditions like room temperature or humidity, which might fluctuate naturally without affecting the machine's operation. Such fluctuations could be binned into broader categories (e.g., "normal," "slightly elevated," "high") or omitted if they do not impact performance. The refined dataset resulting from this process focuses on the most critical data, preserving only the information necessary for accurate anomaly detection while minimizing storage requirements.

[0079] Step 105 involves employing lossless data compression on the refined dataset to further minimize its size without losing any information. Several lossless compression techniques, such as the DEFLATE algorithm or Huffman coding, may be used here. These techniques work by identifying and encoding repetitive patterns or sequences within the dataset more efficiently. For example, if the machine produces periodic logging data with recurring values (e.g., "temperature stable at 70°C" every minute), the compression algorithm can replace these repetitions with a shorter representation. This not only conserves storage space but also streamlines the transmission of data. The choice of compression algorithm can vary depending on the nature of the data; in some cases, the system may use domain-specific compression methods optimized for industrial logging data formats. For example, run-length encoding may be suitable if the dataset contains long sequences of identical sensor readings. The goal of this step is to ensure that while the data volume is reduced, the information integrity is maintained, allowing the data to be perfectly reconstructed when decompressed for analysis.

[0080] In step 106, the processing module determines correlations between different data elements within the raw data to identify patterns and relationships related to anomalies. This step may involve statistical analysis, such as calculating correlation coefficients between sensor readings, or machine learning algorithms that can detect more complex interactions between multiple variables. For example, a rise in motor temperature might correlate with an increase in vibration frequency. By identifying and storing such relationships, the system may provide valuable insights into potential causes of machine malfunctions. These correlations are stored alongside the refined dataset to enhance future analyses. In practical embodiments, this could mean that when a certain pattern in sensor data, such as increased pressure in conjunction with an abnormal sound frequency, is detected, the system logs these findings for further evaluation. This relational data becomes a crucial part of predictive maintenance, allowing operators to identify recurring patterns that precede faults. The system might also adjust the correlation detection process dynamically, refining its analysis based on historical data and user feedback to improve its accuracy over time.

[0081] The final step, 107, involves transmitting the refined dataset to a remote server for storage through the communication module. To optimize the transmission, the communication module may use secure protocols like HTTPS or MQTT, ensuring data integrity and confidentiality during transfer. The reduced size of the refined dataset significantly lowers the required bandwidth, enabling quicker and more efficient data transmission, even in environments with limited network capabilities. For example, in a factory setting with multiple machines generating large volumes of data, the reduced dataset ensures that essential information is transmitted in near-real-time to a central server. Once received at the remote server, the dataset may be used for various purposes, such as long-term storage, further analysis, or integration into a broader monitoring system. In some implementations, the remote server may also provide feedback or alerts to local operators based on the transmitted data, facilitating timely interventions and maintenance actions.

[0082] Figure 2 illustrates a schematic representation of the system, designated as 200, for processing logging data.

[0083] At the core of the system is the local processing unit, 210, which may be an industrial personal computer (IPC), server, or the like, located near the packaging or labeling machines 220. This processing unit could contain multiple modules 211- 214, and at least an interface module 211 , communication module 212, storage module 213 and processing module 214, that work together to implement the data processing steps. There may be other modules present, which are not described in more detail and illustrated in Fig. 2 by outlining the blocks with a dashed line instead of a solid one. The IPC can for example include hardware and software components configured to interact with various machine elements, making it versatile for different operational environments.

[0084] The interface module, 211 , serves as the bridge between the IPC and external devices or systems, such as the electronic components 221-223, 225 of the packaging or labeling machine 220 or machines (in the figure there is shown only one such machine, but there may be more and each may have a dedicated separate or integrated local processing unit 210, but one local processing unit may also be used for two or more packaging or labeling machines.

[0085] The interface module is responsible for connecting with the external devices for collecting logging data from sensors, actuators, and other control elements within the machinery. It can interface with the components using various industrial communication protocols, such as Modbus, OPC UA, or Ethernet / IP. In certain implementations, the interface module may also support wireless communication to facilitate data collection in environments where cabling is impractical.

[0086] The communication module, 212, facilitates data transfer between the local processing unit and a remote server. It may employ a variety of communication methods, such as Ethernet, Wi-Fi, or cellular networks, to ensure reliable transmission of the refined dataset to remote storage or cloud-based analysis systems. In some embodiments, the communication module may incorporate secure data transmission protocols like HTTPS or MQTT, enhancing data integrity and confidentiality during transmission.

[0087] The storage module, 213, contains at least one, but may also contain two or more local data storage module(s), 213-1 , 213-2, where logging data received from the interface module may be stored. This storage module could use various forms of storage media, such as solid-state drives (SSD), hard disk drives (HDD), or even volatile memory, depending on the specific requirements of the application. The storage module allows for immediate access to raw logging data, which is beneficial for local processing and analysis.

[0088] The processing module, 214, is a dedicated or general processor(s) configured to execute computer program code that performs the data processing method. This module may include various functionalities, such as data reduction algorithms, lossless data compression techniques, and correlation determination processes. The processing module may utilize hardware accelerators, like GPUs or FPGAs, to enhance processing speeds, particularly when handling large volumes of data. Additionally, the processing module could support dynamic adjustments, where algorithmic modules can be modified or replaced based on evolving operational requirements or user input.

[0089] The remote server, 220, serves as the endpoint for the refined dataset transmitted by the communication module. It may include a remote data storage module, 221 , where the refined dataset is stored for further analysis, long-term archiving, or integration into larger data systems. In some practical embodiments, the remote server may also host machine learning models or analytical tools to interpret the refined dataset, providing feedback that could be used to optimize machine performance or predict maintenance needs.

[0090] Arrows depicted in Figure 2 illustrate the flow of data between the system's components, including the interface module, storage module, processing module, communication module, and the remote server. These arrows represent the pathways for data collection, processing, and transmission, showing the interaction between different parts of the system. The arrangement and interaction of these components may vary based on specific implementations, allowing for flexibility in the system design.

[0091] Based on this description, a skilled person may envision modifications and additions to the method and system disclosed, such as incorporating additional modules or alternative data transmission methods. These potential modifications fall within the scope of the appended claims, reflecting the system's adaptability to various industrial settings and data processing needs.

[0092] It will be clear that the intention of the above description is to shed light on the working of possible embodiments of the present invention, and not to limit the scope of protection of the invention. Starting from the description, a person skilled in the art is able to conceive of and use various embodiments that fall within the inventive concept and scope of protection of the present invention.

Claims

CLAIMS1. A computer implemented method for processing logging data from one or more industrial packaging or labelling machine for identifying anomalies, the machine comprising a plurality of electronic components for measuring and control of components within the machine, each component generating logging data, the method being performed by a local processing unit, arranged to be located in the vicinity of the one or more industrial packaging or labelling machine, and comprising an interface module, a communication module, a storage module, and a processing module, the interface module being arranged to interface with the plurality of electronic components of the one or more industrial packaging or labelling machine, the communication module being arranged to communicate with a remote server, the storage module being arranged to store data, the processing module being arranged to store and execute computer program code to perform the method comprising the steps of: receiving, from the interface module, logging data from the plurality of electronic components of the one or more industrial packaging or labelling machines connected to the interface module; storing, by the storage module, the logging data received by the interface module as raw data locally in a local data storage module comprised in the storage module; processing, by the processing module, at least some of the raw data into a refined dataset, wherein the amount of refined dataset is at least one order of a magnitude smaller than the raw data; characterized by the processing module executing an algorithm comprising the steps of: performing a data reduction process by modelling the raw data to filter and select relevant logging data relating to anomalies, reducing the data volume into the refined dataset; employing lossless data compression on the refined dataset; determine correlations between different data elements comprised in the raw data, to identify patterns and relationships between the data elements, relating to anomalies, and storing these relational data sets alongside the refined dataset, andtransmitting, through the communication module, the refined dataset to the remote server for remote storage in a remote data storage module comprised in the remote server.

2. The computer-implemented method for processing logging data according to any of the previous claims, wherein the processing module comprises a plurality of algorithmic modules that are configurable, allowing the modification to one or more of the algorithmic module to change variables of the data reduction process, data compression mechanism, or correlation detection mechanism in place.

3. The computer-implemented method for processing logging data according to claim 3, wherein the modification of variables of one or more of the algorithmic modules can be changed dynamically, either as a self-learning mechanism or based on user input.

4. The computer-implemented method for processing logging data according to any of the previous claims, wherein the data reduction process performed by the algorithmic modules is based on regression analysis to filter and select relevant logging data relating to anomalies.

5. The computer-implemented method for processing logging data according to any of the previous claims, wherein the data compression performed by the algorithmic modules is a lossless compression technique (to ensure no loss of relevant logging data).

6. The computer-implemented method for processing logging data according to any of the previous claims, wherein the correlation detection mechanism identifies relationships between data elements such as motor information and sensor data, and stores these relational data sets alongside the refined dataset.

7. The computer-implemented method for processing logging data according to any of the previous claims, wherein the processing module furthercomprises a timing analyzer module that measures the time taken for a product to move from one location to another within the machine, based on sensor data.

8. The computer-implemented method for processing logging data according to any of the previous claims, wherein the algorithmic modules are configured according to a data exchange file format, and in particular one or more of a markup language or data serialization format, an more in particular, one or of JSON, XML, YAML, HTML and SGML,9. The computer-implemented method for processing logging data according to any of the previous claims, wherein the refined dataset transmitted to the remote server includes metadata such as the history and routing information of the processed messages.

10. The computer-implemented method for processing logging data according to any of the previous claims, wherein the local data storage module retains raw data for a predefined period (to allow for retrospective analysis in case of anomaly detection).

11. The computer-implemented method for processing logging data according to any of the previous claims, wherein the communication module employs secure communication protocols (to ensure the integrity and confidentiality of the data transmitted to the remote server).

12. The computer-implemented method for processing logging data according to any of the previous claims, wherein the processing module uses machine learning algorithms (to continuously improve the data reduction and anomaly detection processes based on historical data and user feedback).

13. The computer-implemented method for processing logging data according to any of the previous claims, wherein the interface module supports multiple industrial communication protocols to interface with various types of electronic components within the packaging or labeling machine.

14. The computer-implemented method for processing logging data according to any of the previous claims, wherein the system further includes a user interface (that allows users to configure and monitor the performance of the processing module and the algorithmic modules).

15. The computer-implemented method for processing logging data according to any of the previous claims, wherein the refined dataset includes timestamps (to allow for chronological analysis of the logging data).

16. The computer-implemented method for processing logging data according to any of the previous claims, wherein the processing module provides alerts to the remote server or local users when anomalies are detected based on the refined dataset.

17. The computer-implemented method for processing logging data according to any of the previous claims, wherein the system further includes a domainbased routing mechanism to manage communication between algorithmic modules in different domains.

18. The computer-implemented method for processing logging data according to any of the previous claims, wherein the algorithmic modules can process messages from multiple sources and aggregate the data before performing the data reduction process.

19. The computer-implemented method for processing logging data according to any of the previous claims, wherein the system includes a feedback mechanism that allows users to provide input on the relevance of the detected anomalies, which is then used to refine the processing algorithms.

20. A system (200) for processing logging data from one or more industrial packaging or labelling machines (220) for identifying anomalies, the machine comprising a plurality of electronic components (221 , 222, 223) for measuring andcontrol of components within the machine, each component generating logging data (231 , 232, 233), the system (200) comprising:- a local processing unit (210) arranged to be located in the vicinity of the one or more industrial packaging or labelling machines (220), the local processing unit (210) comprising:- an interface module (211) arranged to interface with the plurality of electronic components of the one or more industrial packaging or labelling machines,- a communication module (212) arranged to communicate with a remote server,- a storage module (213) arranged to store data,- a processing module (214) arranged to store and execute computer program code to perform the method, wherein the processing module is configured to:- receive logging data (231 , 232, 233) from the plurality of electronic components (221 , 222, 223) of the one or more industrial packaging or labelling machines (220) via the interface module (211),- store the received logging data (221 , 222, 223) as raw data locally in a local data storage module (213-1) comprised in the storage module (213),- process at least some of the raw data into a refined dataset, wherein the amount of refined dataset is at least one order of magnitude smaller than the raw data, by executing an algorithm that:- performs a data reduction process by modeling the raw data to filter and select relevant logging data relating to anomalies, reducing the data volume into the refined dataset,- employs lossless data compression on the refined dataset,- determining correlations between different data elements comprised in the raw data, to identify patterns and relationships between the data elements relating to anomalies, and storing these relational data sets alongside the refined dataset,- transmitting the refined dataset to the remote server (240) for remote storage in a remote data storage module (241) comprised in the remote server (240) via the communication module (212).

21. A computer program product for processing logging data from one or more industrial packaging or labelling machines for identifying anomalies, thecomputer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method of any of the previous claims 1-20.

Images