DE202021001907U1 - Means for executing procedures for predictive maintenance and repair using sensors, computers and algorithms in process and factory automation - Google Patents

Abstract

Means for executing methods for predictive maintenance and repair using sensors, computers and algorithms in process and factory automation, the means including those from the means categories sensors, computers and software modules for algorithms, and thereby
- Sensors can be:
Temperature sensors; Acceleration sensors; Pressure sensors; Vibration sensors;
Shock sensors; photothermal sensors; Millimeter wave sensors; Infrared sensors;
Spectrometer; Uv sensors; Count rate sensors; Humidity sensors; magnetic inductive sensors; Roughness sensors; Ultrasonic sensors; Visible and infrared cameras;
Laser diodes; Pressure sensors; electrochemical sensors; Gas sensors; Distance sensors such as lasers or with ultrasound;
- Computers and components that work with them can be:
µ controller; Multicore processors; FPGA's (Filed Programmable Gate Arrays, ASIC's (application-specific integrated circuits), DSP (digital signal processors); circuits for current and voltage measurement IPC (Industrial PC); PC; hardware modules for algorithms; neural networks; fuzzy sets; expert systems; data acquisition systems with A / D conversion; interface modules; control hardware in the form of chips; COB (chip-on-board technology); interface modules for PLC and intelligent algorithms for signal evaluation and classifications / predictions; data loggers; hardware and software for radio processes such as WLAN / Wi-Fi , Bluetooth, GSM, UMTS, LTE,, LTEG4 +, 5G chips, Ethernet ASICs, WirlessHART chips, PROFIBUS ASICs, EtherCAT ASIC, HART ASICs, Ethernet ASICs
-Software modules for algorithms, in particular freely programmable and / or hardware modules, can be:
• algorithms
• Fundamental parameter analysis
• Feature extraction process
• Pattern recognition process
• AI expert systems
• Polynomial classifiers
• Artificial Neural Network
• Prediction algorithms such as linear regression
• Adaptive threshold detectors, especially for the acquisition of the relevant data
• Kalman filters and alpha-beta trackers
• Associative mapping procedure
• Retrieval method based on the minimum error square
• communication. EtherNet / IP ....... PROFIBUS, OPC UA and where the various special means specified for the aforementioned means categories are put together in different combinations of all special means adapted to the respective application.

Description

Predictive maintenance (predictive maintenance) learns from historical and, if necessary, maintenance-relevant data available in real time.

Through this and through the prognosis of future events, the question “What will happen when and where?” Can be answered.

Predictive maintenance techniques thus help to determine the condition of things that are in operation. They help to estimate when maintenance should be carried out. This approach promises cost savings compared to routine or time-based preventive maintenance strategies, as tasks are only carried out when they are justified and necessary. Therefore, the prediction is used to perform condition-based maintenance, which is made according to estimates of the deteriorated condition of an object.

Predictive algorithms as a branch of predictive maintenance have found their way into all technologies today. Prediction begins with the linear extrapolation of the linear regression, continues with the well-known alpha-beta tracker (GPS position determination in the car), the Kalman filter and on to artificial neural networks ( KNN ) with multiple hidden layers. Typical questions that one wants to predict are: Where does a vehicle have to steer? What will happen soon in the system or the machine? How long will the device continue to run flawlessly? Which spare parts do I have to keep available and when? What is the condition of the critical components and parts, such as an X-ray tube or a semiconductor detector?

Many of the predictions are based on events that take place in the past.

• • Die Datenerfassung
• • Das Reporting (was ist in der Anlage, was ist im Produkt geschehen, warum wird es passieren, was ist das Optimum)
• • Die Analyse der Daten (wann, wer, was, wie)
• • Das Monitoring (was passiert momentan in der Anlage, im Produkt)
• • Die eigentliche Prädiktion (Vorhersage), wann muss eine Aktion (Wartung) erfolgen
• • Übertragungssystem der Ergebnisse zur remote Auswertung mittels standardisierten Kommunikationsverfahren

The individual levels for the prediction are:

• • The data acquisition
• • Reporting (what is in the system, what has happened in the product, why will it happen, what is the optimum)
• • The analysis of the data (when, who, what, how)
• • Monitoring (what is currently happening in the system, in the product)
• • The actual prediction (prediction) of when an action (maintenance) must take place
• • Transmission system of the results for remote evaluation using standardized communication methods

From a mathematical point of view, these are, simply put, extrapolated regressions that place an optimal straight line after the mean minimum error square through the measurement points (or spectral components) recorded over time and can thus make the prediction with a high degree of probability.

Predictive maintenance also offers the advantage that, for example, not only a specific object is recognized in a signal or image (pattern recognition), but also, with a certain degree of trustworthiness, a whole series of successive signals and images can be used to statistically predict which signals or reactions will be will follow next.

That means, however, that the underlying systems must be continuously re-learned, with each learning cycle in turn increasing the stability of the prediction system.

The only figure in the drawing shows in an overview how predictive algorithms or classifiers are developed in industry.

A distinction is made between two phases: the offline teach-in process and online operation with taught-in algorithms.

In principle, what is to be predicted must be defined during the learning process.

The measurement points for the corresponding corner points of the prediction must be defined accordingly, as well as which measurement data are required and processed for this.

The data must be recorded (recording) and evaluated. The data recording must be carried out over longer periods of time with the appropriate sensors that are necessary for the prediction. Generating the first database often means weeks of measurements. As a rule, the measurements must be carried out up to the point of failure of the devices, components and machine parts in order to define the status of the actual prediction (setpoint). Features for characterizing the desired states must be found, which can be separated by means of a classifier and assigned to the predictive corner points. The feature extraction is available today because of the high Computing power is no longer absolutely necessary, but still recommended.

The classifier is taught in offline with the patterns or features and its functionality is checked. With neural networks ( KNN ) this can often take hours or days until the iterative solutions are satisfactory.

The classifier is then tested on untrained data that can be expected in the prediction. If the classifier does not meet the requirements, it is re-learned with additional templates or another decision-making function is selected. This process is the real "time waster" in development. All sensors, algorithms, software and hardware required for the prediction must be integrated into the device, machine, etc.

If it meets the requirements, the production algorithm is implemented in the hardware and software of the measuring system or the machine. The device is ready for inline or online production use.

In online operation (inline operation), the data and patterns are processed in real time and provide the result (prediction). The currently recorded data is saved for further optimization of the predictor (classifier). This data is then used from time to time to learn the algorithm in offline mode.

As the amount of data increases, the predictor becomes more and more robust and more “accurate” in its results. Newly determined parameters for the predictor are downloaded to the device in production and used in-line or online during further operation.

It should be expressly pointed out that, despite all the euphoria, the development of such systems only works as well as the underlying database is. Indeed, this means that you have to think carefully beforehand what the system should do.

In relative terms, the development requires a lot of time, because the development of such a prediction system does not work in passing.

As already mentioned above, predictive “cruise control” algorithms are increasingly used today, especially in automotive companies and aircraft.

For example, many sensors have been connected to the CAN bus in vehicles since the introduction of bus systems: ultrasonic sensors, temperature and infrared sensors, cameras and infrared cameras, acceleration sensors, radar sensors and other sensors.

The information from these sensors is transmitted to the central processing unit (CPU). There, the information is evaluated in real time and the corresponding control and monitoring functions are initiated: distance, speed control, compliance with lane markings, person recognition, anti-slip and much more. All of these methods use correlation methods and predictive algorithms based on KI methods (expert systems, artificial neural networks, etc.).

• Ob in der Fabrikautomation durch Druck-, Durchfluss-, Temperaturmessung, optische Sensoren zur Steuerung von Fertigungsprozessen oder in der Prozessautomation zur Regelung des Sauerstoffgehaltes im biologischen Klärbecken sowie zur Neutralisation in industriellen Wasseraufbereitungen, überall werden dieselben Verfahren und Algorithmen eingesetzt.

These algorithms are also used in modern asset management or process regulation:

• Whether in factory automation through pressure, flow or temperature measurement, optical sensors to control manufacturing processes or in process automation to regulate the oxygen content in biological clarifiers as well as for neutralization in industrial water treatment, the same processes and algorithms are used everywhere.

It is no coincidence that many companies have been dealing with these predictive control methods for more than 20 years and everyone equips their products and devices with the appropriate software, hardware and algorithms. This is also a fact that can be seen in an evolutionary manner and does not represent anything significantly new in terms of technology in Industry 4.0.

The latest generations of devices, automats, machines and drives are equipped with additional sensors, which record a wide variety of parameters, such as temperature and pressure data, humidity, mechanical loads such as speed, lateral accelerations and much more, in order to ultimately provide a statement about the To meet the system and device status. In particular, an attempt is made to predict which device will require which service and, above all, when.

Another purpose of these measurements and predictions is now also used to monitor product liability: How often are statements made, the device no longer works, the quality is poor and the like.

In many cases it is often a question of the operation of the devices outside the permissible tolerance limits and specifications. And then we are back to the case of the car already mentioned above: Guarantee, warranty and service claims can be verified using the recorded control data. The exceedances of the Tolerances and their frequency lead to faster failure of the devices. The failure of the device and the components can be predicted because the tolerance is exceeded. However, here too, the measurement conditions must be defined in advance and the data recorded from which the prediction can be made.

An in-depth example of how predictive algorithms are developed can be explained in the case of devices for layer thickness measurement and material analysis.

• • Anwendung (z.B. Goldmessung oder Gold auf Kupfer)
• • Bestrahlungszeiten der Proben
• • Anzahl der Ein- und Ausschaltvorgänge der Röhre
• • verwendete Leistung (bis 4 kV, bis 50 KV)
• • Blindstrom (Maß für die Evakuierung der Röhre)
• • effektive Betriebsdauer
• • Betriebstemperatur
• • Außentemperatur
• • Umgebungsfeuchtigkeit
• • Temperaturschwankungen
• • Transport und Installation (Vibrationen)

Layer thickness measurements of coatings, alloys and material analyzes are carried out with, among other things, X-ray fluorescence devices. The most maintenance-intensive component in the coating thickness measuring device is the X-ray tube. The life of the X-ray tube depends on

• • Application (e.g. gold measurement or gold on copper)
• • Exposure times of the samples
• • Number of times the tube is switched on and off
• • Power used (up to 4 kV, up to 50 KV)
• • Reactive current (measure for evacuation of the tube)
• • effective operating time
• • Operating temperatur
• • Outside temperature
• • Ambient humidity
• • Temperature fluctuations
• • Transport and installation (vibrations)

Overseas transports are one of the most common breakdowns of these devices, e.g. due to the transport pallet being placed too hard or excessive vibrations during air and / or sea freight.

All of these causes have an impact on the service life of the X-ray tube and determine the time when the tube has to be replaced and the device has to be recalibrated. In principle, figuratively speaking, every exceeding of the tolerance of the data sheet shortens the life of the X-ray tube and at some point the tube has to be replaced.

As a rule, therefore, the X-ray fluorescence devices are now systematically serviced at least once a year according to a maintenance plan. It often happens that maintenance is either not necessary or that the worst case occurred before maintenance: the device fails unplanned, due to overuse or improper handling.

For this reason, predictive algorithms for maintenance prediction will be used in the new generations of X-ray fluorescence devices. To do this, however, it is necessary to integrate the appropriate sensors, hardware, algorithms and SW (HMI) into the X-ray fluorescence device and, in particular, to provide for remote maintenance in the usual communication structure of the automation pyramid (OSI model) with the corresponding bus systems and field buses.

Essential sensors for recording the state variables with regard to predictive maintenance for an X-ray fluorescence device are e.g. temperature sensors, pressure sensors, acceleration sensors, humidity sensors, hardware and software measures for counting rate determinations, predictive evaluation algorithms.

Once these sensors have been integrated, the definition of the database for the learning patterns and their corresponding data acquisition takes place (see the only figure in the drawing). It is important to record and catalog the characteristics that apply to the prediction. Defined target states with regard to the prediction to be decided are assigned to the patterns. Whereby the target states are usually to be determined by series of measurements; i.e. if you want to know the failure of the X-ray tube with regard to vibrations, corresponding measurements must be carried out under different stress conditions up to the failure (failure) of the tube and the individual states must be assigned to the temporal progressions in the predictor.

The patterns and target states are then taught to a classifier with the aim of predicting the next maintenance interval.

The learning itself is one of the most complex development steps. To stabilize the prediction, new measurement data have to be learned over and over again during operation. If you consider the points listed above that are necessary for the prediction and the global use of the devices in countries with the most varied of climatic conditions and locations, you can imagine the effort required to develop a predictive algorithm.

In the following, the means that can be used in procedures for predictive maintenance of devices for layer thickness measurement and material analysis will be discussed in more detail.

Usually at least 10 different situations for the failure of the tubes have to be recorded in order to learn a first classification / prediction. Both the individual parameters can be trimmed to prediction or Combinations of different measurements, e.g. only temperature and only humidity or a combination of temperature and / or humidity and / or pressure etc. Once the algorithm (prediction) has been determined, the classification / predictor method must be implemented in the X-ray fluorescence device in terms of hardware and software and the test phases begin. All devices sold should be equipped with the evaluated predictor. From the time of delivery, all measurement data are then recorded in the sense of prediction using a data recording medium (data logger) in correlation with the outputs of the predictor. Instead of a data logger integrated in the device, the measurement data can also be transmitted by radio (e.g. WLAN; GSM, UMTS; LTE; 5G) to a remote data center and the evaluations carried out there.

This also takes a corresponding amount of time. A prediction should be made on the basis of the individual states that have been taught-in or combinations thereof. The aim is to predict with a certain degree of certainty that a tube or detector will need to be replaced.

1. a) Feuchtesensor zur Erfassung von relativer und/oder absoluter Luftfeuchtigkeit und deren Schwankungen im Raum oder im Freien
2. b) Temperatursensor zum Erfassung der Umgebungstemperatur im Raum oder im Freien

The following procedure can be used for this and the following means, such as hardware, software and algorithms, can be used for this, whereby the sensors are decisive:

1. a) Humidity sensor for the detection of relative and / or absolute air humidity and its fluctuations in the room or outdoors
2. b) Temperature sensor for recording the ambient temperature in the room or outdoors

• c) Drucksensor zur Erfassung von z.B. barometrischem Druck, der in Verbindung mit Umgebungsfeuchte und Umgebungstemperatur den Zustand der Röntgenröhren vorhersagt.
• d) Messung der Spannungsverläufe und Erfassung der zeitlichen Dauer von unterschiedlichsten Spannungen akkumuliert über der Zeit mittels Timer-Bausteinen (Hardware oder softwaremäßige Realisierung). Clock Generator oder Oszillator der Hardware in Verbindung mit Zählsoftware. Spannungen über Einstellsoftware und Algorithmus oder hardwaremäßige Vergleichsschaltungen mit Gattern in Verbindung mit getriggerten Clock's.

The correlation between temperature and air humidity determines the service life of the X-ray tube and the X-ray detector. These states must be measured and correlated with the failure of the tube. This means that these measurements have to be carried out in different environments until the tube fails. The different temperature-humidity-time curves with the associated downtimes of the tubes are input values for the predictor.

• c) Pressure sensor for recording, for example, barometric pressure, which, in conjunction with ambient humidity and temperature, predicts the condition of the X-ray tubes.
• d) Measurement of the voltage curves and recording of the duration of different voltages accumulated over time by means of timer modules (hardware or software implementation). Hardware clock generator or oscillator in conjunction with counting software. Voltages via setting software and algorithm or hardware comparison circuits with gates in connection with triggered clocks.

• e) Beschleunigungssensor, Erschütterungssensor, Vibrationssensor dienen zur Erfassung von Schocks, azyklischen und zyklischen Erschütterungen sowie Vibrationen. Hier gibt es viele Hersteller.

This means that these measurements must be measured for all possible threshold voltages in correlation with time until the tube fails. The different accumulated voltage-time curves with the associated tube failure times are input values for the predictor.

• e) Accelerometer, shock sensor, vibration sensor are used to record shocks, acyclic and cyclic shocks and vibrations. There are many manufacturers here.

Usually a 3-axis accelerometer is used, which detects shocks and vibrations. The X, Y, Z accelerometer detects the direction of impact. Maximum limits of, for example, 0.01 to 10 G can be specified. The continuous course can also be measured and recorded.

• Aufprallmessung der G-Kräfte
• Aufprallrichtung
• Echtzeit-Warnmeldungen und Analyse historischer Daten
• Ereigniszeit und -ort

A shock sensor determines the following values:

• Impact measurement of the G-forces
• Direction of impact
• Real-time alerts and historical data analysis
• Event time and place

In this way, the causes of vibrations and improper handling of freight and equipment on the transport route by plane, ship or vehicle or combinations thereof can be determined so that an immediate response can be made and well-considered decisions can be made.

• f) Zählerbaustein zur Erfassung der Ein- und Ausschaltvorgänge der Röhre. Dieser kann auch mit der Höhe der verwenden Betriebsspannung kombiniert werden (siehe c). Es muss bis zum Ausfall der Röhre gemessen werden, z.B. 15 x bei 4 keV für 540 Stunden und 20 keV x bei 40 keV für 400 Stunden. Verwendet werden kann der Hardwaretimer (Clock, Timer).
• g) Blindstromerfassung, Schaltung zur Messung des Stromes bei der Röntgen- Messung. Der Blindstrom ist die Energie, die im Verbraucher nicht umgewandelt wird und eine Blindleistung erzeugt, die eine zusätzliche Belastung für die Röntgenröhre darstellt. D.h. je höher der Blindstrom, desto schneller ist die Röhre zerstört. Zur Erfassung muss eine Schaltung im Röntgenschaltkreis integriert werden, der den Imaginärteil des Stromes misst, woraus die Blindleistung errechnet werden kann.
• h) Gassensoren zur Erfassung von Chlor und Chlordioxid und anderen korrosiven Gasen in der Umgebung. Der Gasgehalt wird kontinuierlich über der Zeit bis zum Ausfall des Röntgenfluoreszenz Gerätes gemessen. Die Gasgehalt-/Zeitverläufe sind die Eingabewerte für den Prädiktor. Derartige Gassensoren gibt es z.B. bei City Technology. Umgebungstemperatur und Umgebungsluftzusammensetzung haben entscheidenden Einfluss auf den Alterungsprozess der Röhre und des Röntgenfluoreszenzgerätes. Röntgengeräte in der Fabrik benötigen einen zusätzlichen Schutz gegen korrosive Gase (chlorhaltige Luft).

The timing of the vibrations, shocks and shocks must be recorded until the X-ray tubes fail. The different courses with the temporal failure points of the tube are the learning data for the classifier.

• f) Counter module for recording the switching on and off processes of the tube. This can also be combined with the level of the operating voltage used (see c). Measurements must be made until the tube fails, eg 15 x at 4 keV for 540 hours and 20 keV x at 40 keV for 400 hours. The hardware timer (clock, timer) can be used.
• g) Reactive current detection, circuit for measuring the current during the X-ray measurement. The reactive current is the energy that is in the consumer is not converted and generates reactive power, which represents an additional load for the X-ray tube. In other words, the higher the reactive current, the faster the tube is destroyed. For detection, a circuit must be integrated in the X-ray circuit, which measures the imaginary part of the current, from which the reactive power can be calculated.
• h) Gas sensors for the detection of chlorine and chlorine dioxide and other corrosive gases in the environment. The gas content is measured continuously over the time until the failure of the X-ray fluorescence device. The gas content / time curves are the input values for the predictor. Such gas sensors are available from City Technology, for example. Ambient temperature and ambient air composition have a decisive influence on the aging process of the tube and the X-ray fluorescence device. X-ray machines in the factory require additional protection against corrosive gases (air containing chlorine).

• i) Hardware und Software, die in der Lage sind, die Datenkomplexität zu erfassen, zu verarbeiten und zu klassifizieren. Dabei kann es sich um Kombination und/oder einzelne spezifische Hardware-Bausteine handeln, die oben angegeben wurden. Diese Hardware kann frei programmierbar sein (µ-Controller, DSP's (Signalprozessoren), PC) oder festgegossene Algorithmik in Form von ASIC's beinhalten. Beispielsweise sind DSP gut für Klassifikatoren oder Fouriertransformationen geeignet.
• j) Datenlogger oder remote System für die Aufzeichnung der Daten zur Beurteilung der Prädiktorleistung, sowie zum Gewinnen neuen Messreihen für das Einlernen des prädikitiven Algorithmus.

The reactive power over time in combination with the X-ray operating voltage must be combined with the failure of the tube and these are the input variables for the classifier.

• i) Hardware and software capable of capturing, processing and classifying the complexity of the data. This can be a combination and / or individual specific hardware modules that were specified above. This hardware can be freely programmable (µ-controller, DSPs (signal processors), PC) or contain fixed algorithms in the form of ASICs. For example, DSPs are well suited for classifiers or Fourier transforms.
• j) Data logger or remote system for recording the data to assess the predictor performance, as well as to obtain new series of measurements for teaching in the predictive algorithm.

The protection claims 1 to 3 are explained by the entire content of the above description.

In connection with claim 3, it should also be noted: The different evaluation algorithms must be converted into a common fieldbus format in a standardized manner and transmitted to a PLC that works in the user's communication structure.

When converting the data, synchronization between the field device (XRAY device) and the PLC must be ensured. Software and modules of this kind are sold by the company Anybus in Sweden, for example.

List of reference symbols

5G

5th generation cellular network

ASIC

application-specific integrated circuit, custom chip

COB

Chip-on-board technology

DSP

Digital signal processors

FPGA

Field Programmable Gate Array

GPS

Global Positioning System

GSM

System for Mobile Communications

HARD

Highway Addressable Remote Transducer

IPC

Industrial personal computers

AI

Artificial intelligence

KNN

Artificial Neural Networks

LTE

Long Term Evolution: Name for the third generation cellular standard

LTE 4+

Extended LTE network

OPC UA

Open Platform Communications Unified Architecture

OSI

Open Systems Interconnection model

PM

Predictive maintenance

PROFIBUS

Process Field Bus

S7-1500 TM NPU

SIMATIC S7-1500 TM NPU Neural Processing Unit (NPU) opens up artificial intelligence for automation tasks

SoftPLC

Software-based programmable logic controller, e.g. as a PC plug-in card

UMTS

Universal Mobile Telecommunications System: Name for the 3rd generation mobile radio standard

Wi-fi

Wireless fidelity

WIRELESS INTERNET ACCESS

Wireless local area network based on the IEEE 802.11 standard.

Claims (3)

Means for executing methods for predictive maintenance and repair using sensors, computers and algorithms in process and factory automation, the means including those from the means categories sensors, computers and software modules for algorithms, and thereby - Sensors can be: Temperature sensors; Acceleration sensors; Pressure sensors; Vibration sensors; Shock sensors; photothermal sensors; Millimeter wave sensors; Infrared sensors; Spectrometer; Uv sensors; Count rate sensors; Humidity sensors; magnetic inductive sensors; Roughness sensors; Ultrasonic sensors; Visible and infrared cameras; Laser diodes; Pressure sensors; electrochemical sensors; Gas sensors; Distance sensors such as lasers or with ultrasound; - Computers and components that work with them can be: µ controller; Multicore processors; FPGA's (Filed Programmable Gate Arrays, ASIC's (application-specific integrated circuits), DSP (digital signal processors); circuits for current and voltage measurement IPC (Industrial PC); PC; hardware modules for algorithms; neural networks; fuzzy sets; expert systems; data acquisition systems with A / D conversion; interface modules; control hardware in the form of chips; COB (chip-on-board technology); interface modules for PLC and intelligent algorithms for signal evaluation and classifications / predictions; data loggers; hardware and software for radio processes such as WLAN / Wi-Fi , Bluetooth, GSM, UMTS, LTE,, LTEG4 +, 5G chips, Ethernet ASICs, WirlessHART chips, PROFIBUS ASICs, EtherCAT ASIC, HART ASICs, Ethernet ASICs -Software modules for algorithms, in particular freely programmable and / or hardware modules, can be: • algorithms • Fundamental parameter analysis • Feature extraction process • Pattern recognition process • AI expert systems • Polynomial classifiers • Artificial Neural Network • Prediction algorithms such as linear regression • Adaptive threshold detectors, especially for the acquisition of the relevant data • Kalman filters and alpha-beta trackers • Associative mapping procedure • Retrieval method based on the minimum error square • communication. EtherNet / IP ....... PROFIBUS, OPC UA and where the various special means specified for the aforementioned means categories are put together in different combinations of all special means adapted to the respective application. Means after Claim 1 , whereby additional modules for communication and remote maintenance, predictive remote maintenance include: • SoftPLC software that carries out a protocol conversion in a synchronized manner and can run on an industrial PC, • FieldbusBox as hardware and software for conversion to the corresponding fieldbus protocol, • Data logger, • Interface to the PLC IPC (industrial PC) for fieldbus connection Means after Claim 1 or 2 Fieldbuses are used as further means, which can be used individually or in combination: • PROFINET • PROFIBUS • HART • EtherCAT • EtherNet / IP • Fieldbus Foundation • OPC UA • CAN bus • Modbus • CC-Link • Wi-Fi • WLAN

Images