DE102021129084A1 - Large-scale ecotoxicological early warning system with a freely movable bio-indicator and a compact electronic measuring and transmitting device - Google Patents

Abstract

This invention relates to an ecotoxicological early warning system for assessing water quality and rapid detection of pollutant inputs based on a wireless or wired compact sensor system and a bioindicator. The bio-indicator can be designed as a mussel or shrimp species or a combination thereof without any restrictions. Furthermore, this invention relates to a method for automatically classifying the behavior of the bioindicators based on machine learning approaches. This approach represents a technical advance over existing systems, which generate too many false alarms with poor detection of pollutant inputs due to overly simplified decision-making methods. This insecurity of existing systems does not create high added value for the end user. This invention is intended to remedy this and can be used for both large-area and continuous surveillance outdoors and in protected areas.

Description

technical field

This invention relates to an ecotoxicological early warning system for assessing water quality and rapid detection of pollutant inputs based on a wireless and compact sensor system and a bioindicator. Furthermore, this invention relates to a method for automatically classifying the behavior of the bioindicators based on machine learning approaches. This invention can be used for both wide area and continuous monitoring.

State of the art

A modern biological monitoring system must be able to measure continuously and in real time or with a sufficiently low latency. In addition, it should be able to be operated automatically and, above all, to be able to automatically send reliable alarm signals about the status of the water quality. Such a system not only brings tremendous benefits to the city's water supply systems, but also to commercial and industrial facilities, which contribute a great deal to environmental pollution. Continuous monitoring of water quality is based on the use of bioindicators such as mussels, crabs or fish, which react quickly to dangerous or significant changes in the environment. In the field of monitoring bio-indicators for monitoring water quality, there is already academic preparatory work and property rights applications, which are considered separately below.

State of the art in patents

RU2437093C1 presents a biomonitoring system characterized by Hall effect sensors, plethysmographs (heart rate monitors), stretch marks, optical sensors, as well as stationary video cameras for monitoring aquatic organisms. The utility model RU101838U1 by the partly same inventors describes a very similar biomonitoring system.

RU2452949C1 discloses a modification of the RU2437093C1 presented biomonitoring system with a focus on the measuring system. The measurement system consists of a tensoresistor mounted on a horizontal plate, an analog-to-digital converter and a microprocessor to process and record the electrical signal. The end of the tensoresistor is attached to the conch, allowing the movement of the conch to be sensed. According to the inventor, this system is said to have advantages over systems with a Hall effect sensor because the combination of Hall effect sensor and opposing magnets is said to impede the bioindicator, which is usually a clam, more. However, this hypothesis was not proven.

RU2595867C2 discloses a method for identifying and selecting biosensing organisms that can be used for rapid bioindication and biomonitoring of marine and fresh water, including drinking and wastewater.

State of the art in science

A system for monitoring heart contractions was developed by Depledge MH (Depledge MH, Andersen VV A computer-aided physiological monitoring system for continuous, long-term recording of cardiac activity in selected invertebrates .-- Biochem PhysioL, Vol.96 A, 1990, No 4, pp473-477) on biomonitoring of aquatic environments. This system contains a heartbeat sensor consisting essentially of an LED, a photoresistor and a device for amplifying and processing an electrical signal. The heartbeat sensor was attached to a crab shell.

• Kramer, Kees JM, and Edwin M. Foekema. „The „Musseimonitor^®“ as biological early warning system." Biomonitors and Biomarkers as Indicators of Environmental Change 2. Springer, Boston, MA, 2001. 59-87.
• Sluyts, Hilde, et al. „A dynamic new alarm system for use in biological early warning systems.“ Environmental Toxicology and Chemistry: An International Journal 15.8 (1996): 1317-1323.
• Polman, Harry JG, and H. A. Jenner. „Pulse-Chlorination^®, the best available technique in macrofouling mitigation using chlorine.“ PowerPlant Chemistry 4 (2002): 93-100.

A similar biological warning system was developed and produced in the Netherlands under the brand name Mosselmonitor, with some weaknesses that will be described in more detail below (De Zwart, D., KJM Kramer & HA Jenner (1995), Practical experiences with the early warning system 'Mosselmonitor ', Environ Toxicol, Water Qual. 10:237-247). This system has been used occasionally in academia, and also by some countries to control freshwater or near-shore seawater. In Budapest, Hungary, the Mosselmonitor system was used to monitor the quality of chlorinated drinking water. The Mossel Monitor is essentially a container that holds several shells (usually up to 8 shells). The shells are fixed in place and each shell has a Hall effect sensor on it and a magnet on the opposite side to measure the shell's degree of openness. If a certain threshold value is exceeded, an alarm signal is sent to the operator or user of the warning system, so that they can take further steps to control and improve the water quality if necessary. This limit value or this detection threshold is determined by an experienced employee. Parameters such as the degree of opening of the mussel and the duration of the closing and opening times play an important role here. There is a significant disadvantage of this system that the assessment of the threshold must be performed by a person with significant knowledge of aquatic biology and with the alert system. In addition, the threshold values can vary greatly depending on the location and time of year. See also the following relevant academic publications on the Mosselmonitor system:

• Kramer, Kees JM, and Edwin M. Foekema. "The "Musseimonitor ^® " as biological early warning system." Biomonitors and Biomarkers as Indicators of Environmental Change 2. Springer, Boston, MA, 2001. 59-87.
• Sluyts, Hilde, et al. "A dynamic new alarm system for use in biological early warning systems." Environmental Toxicology and Chemistry: An International Journal 15.8 (1996): 1317-1323.
• Polman, Harry JG, and HA Jenner. "Pulse- ^{Chlorination®} , the best available technique in macrofouling mitigation using chlorine." PowerPlant Chemistry 4 (2002): 93-100.

• - Llamas, R. A., James J. Niemeier, and Anton Kruger. „Curved spiral antennas for freshwater applications.“ 2015 IEEE Radio and Wireless Symposium (RWS). IEEE, 2015.
• - Llamas, Ruben A., et al. „Underwater Deployment and Performance of Curved Spiral Antennas in Mussel Backpacks.“ 2019 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting. IEEE, 2019.
• - Taylor, H. D., Anton Kruger, and J. J. Niemeier. „Embedded electronics for a mussel-based biological sensor.“ 2013 IEEE Sensors Applications Symposium Proceedings. IEEE, 2013.
• - Hodson, Hal. „Mussels with sensor backpacks taste the Mississippi for science.“ (2013): 21.

A research group in Lowa, USA, has developed a mussel backpack that can record the behavioral patterns of freshwater mussels and transmit them wirelessly. However, the focus of this work was mainly on the electrotechnical side and less on the biological interpretation of the behavioral pattern. The electrotechnical aspects included the development and manufacture of an energy-saving measuring and transmission device, which was powered by a battery and attached directly to the bioindicator. In addition, the mentioned group investigated different antenna designs as well as the attenuation of the electromagnetic radio waves in the water in the low ultra-high frequency range. Some relevant publications of the mentioned group are listed below. In these, various antennas for underwater transmission were tested as well as the range of the specially manufactured shell PCBs (Printed Circuit Board):

• - Llamas, RA, James J. Niemeier, and Anton Kruger. "Curved spiral antennas for freshwater applications." 2015 IEEE Radio and Wireless Symposium (RWS). IEEE, 2015.
• - Llamas, Ruben A., et al. "Underwater Deployment and Performance of Curved Spiral Antennas in Mussel Backpacks." 2019 IEEE International Symposium on Antennas and Propagation and USNC-URSI Radio Science Meeting. IEEE, 2019.
• - Taylor, HD, Anton Kruger, and JJ Niemeier. "Embedded electronics for a mussel-based biological sensor." 2013 IEEE Sensors Applications Symposium Proceedings. IEEE, 2013.
• - Hodson, Hal. "Mussels with sensor backpacks taste the Mississippi for science." (2013): 21.

As explained in the state of the art, ecotoxicological early warning systems have so far consisted of permanently installed biosensors, in which predominantly mussels are used. The bioindicators are mostly glued to one side with a holder and wired in the laboratory. Due to the sensitive electronics, this configuration is mainly used indoors due to the weather. In addition, by gluing the bio-indicators to fixed brackets, their freedom of movement is significantly reduced. In the case of mussels, they can only filter, but they can no longer move freely with their mussel feet and position themselves in an optimal filtering angle or, if there is an increase in pollution, they cannot take a protective stance. Recording the movement of bioindicators is a relevant parameter for assessing their health and water quality and should be used in modern ecotoxicological early warning systems. The most relevant commercial early warning system to date based on freshwater mussels as bioindicators was developed in the Netherlands in the late 1990s under the brand name Mosselmonitor. With this system, a person still had to make the decision as to when there was a suspected case of alarming behavior by the mussels or an increased level of pollutants in a test liquid. Such an approach was contemporary and practical at the time, but it also contained an enormous source of error, high personnel costs and considerable differences from user to user in the evaluation of the data. Since the introduction of this device, the computing power of hardware components and the theoretical and practical knowledge of machine learning have increased enormously. In addition to this development, sensors such as accelerometers, Hall effect sensors, electrical sensors and cameras have also become more powerful, smaller, cheaper and more energy-efficient. The need for a reliable, fast and, if necessary, comprehensive ecotoxicological early warning system is high and is outlined below using a few examples. Such a system can be used as an early warning system for drinking water systems, which, in addition to known environmental pollution, can also be targets of terrorist attacks. In addition to drinking water monitoring, comprehensive monitoring of the lake and river quality is also indicated. For example, contaminated water from tunnels, shafts and pits from mines in the Ruhr area is chemically cleaned and fed into the Rhine. Previous systems can only measure the load selectively, while a continuous area-wide measurement would provide reliable values.

Disclosure of Invention

the solution of the problem

Thanks to the progress described in the previous section, it is possible to equip bioindicators with extremely compact sensors and compact electronics, hereinafter referred to as mussel backpacks. This sensor technology hardly disturbs the freedom of movement of the mussel and can be operated for months without battery charging.

• - Einem Bioindikator wie beispielsweise einer Muschel oder einer Krabbe.
• - Einem Muschelrucksack mit Recheneinheit, Batterie und Empfänger- und Sendeeinheit sowie Kommunikationsschnittstellen oder Analog-Digital-Wandler zum Auslesen von
• - Sensoren wie beispielsweise Akzelerometer Hall-Effekt-Sensoren, Dehnungssensoren, welche an der Muschel befestigt sind.
• - Einem Gateway zum drahtlosen Empfang der vom Muschelrucksack gesendeten Daten.
• - Ein Multiparametersonde zur Messung zusätzlicher relevanter Umgebungsparameter.
• - Ein Behältnis mit der Testsubstanz.
• - Ein Backend zur Datenspeicherung- und Abfrage.
• - Ein Machine Learning - Software zur Klassifizierung der Gesundheit der Bioindikatoren bzw. der Schadstoffbelastung der Testsubstanz.
• - Ein Frontend und eine Displayeinheit bzw. ein portables Gerät zur Visualisierung des Systemzustands und zur Befehlseingabe.
• - Ein automatische Alarmfunktion basierend auf einer Machine Learning gestützten Klassifikation zur Bewertung der Wasserqualität.

To remedy or to solve the problems mentioned, the invention of a large-scale ecotoxicological early warning system with a freely movable bioindicator and a compact electronic measuring and transmitting device is disclosed below. The system has the following features:

• - A biological indicator such as a clam or crab.
• - A shell backpack with a computing unit, battery and receiver and transmitter unit as well as communication interfaces or analog-to-digital converters for reading out
• - Sensors such as accelerometers, Hall effect sensors, strain sensors attached to the shell.
• - A gateway for wireless reception of the data sent by the shell backpack.
• - A multi-parameter probe for measuring additional relevant environmental parameters.
• - A container with the test substance.
• - A backend for data storage and query.
• - A machine learning software for classifying the health of the bioindicators or the pollutant load of the test substance.
• - A front end and a display unit or a portable device for visualizing the system status and for entering commands.
• - An automatic alarm function based on a machine learning-based classification for evaluating the water quality.

Ecotoxicology or environmental toxicology is an interdisciplinary science and a subfield of biology, toxicology, environmental chemistry and ecology. Ecotoxicology deals with the effects of substances or pollutants on the living environment. In this invention, an ecotoxicological early warning system is disclosed. This warning system sends an alarm as soon as the readings indicate that the test substance may be contaminated. Due to continuous monitoring, which takes place almost in real time due to low latency in the seconds or minute range, one can speak of an early warning system. In other methods for examining the pollution, samples are taken and examined at very long intervals of a few days to months. This significant latency can be detrimental to the health of living beings of all kinds. It is also possible that an increased pollution load was only present during the too long sampling intervals and therefore the cause of the consequential damage cannot even be clarified. The developed early warning system provides a remedy here. A machine learning model, which can reliably classify the states of the bioindicators and the pollutant load of the test substance, is being developed to aid in the decision-making process.

In this invention, the assessment of the behavior of the bioindicators is not to be undertaken by people, but by a machine classification method. The learning of the classification algorithm takes place in a supervised manner (also compare supervised learning), i.e. the various classes that correspond to the behavioral patterns of the bioindicators are known. The accuracy of the classification can be increased through large amounts of usable data, a clever choice of parameters and sophisticated labeling of the data volumes. A detailed description of the machine learning methods used can be found below.

• 1: Schematische Darstellung des ökotoxikologischen Frühwarnsystems mit Frischwassermuscheln als Bioindikatoren, welche sich durch den kompakten und drahtlosen Sensorrucksack frei bewegen können.
• 2: Schematische Darstellung eines Bioindikators mit drahtlosem Sensorrucksack und Sensoren zur Messung des Verhaltens.
• 3: Schematische Darstellung der Platine inkl. Batterie im Sensorrucksack.
• 4: Detailliertere schematische Darstellung der Komponenten des Gateways (gestrichelte Linien) und seiner Interaktion mit den umliegenden Komponenten.
• 5: Schematische Darstellung der Verwendung der drahtlosen Muschelsensorik im Außenbereich zum Monitoring der Schadstoffbelastung des Wassers.
• 6: Schematische Darstellung einer weiteren Ausführung des Systems durch an der Wasseroberfläche schwimmende Sendeeinheiten.
• 7 Schematische Darstellung einer kabelgebundenen Ausführung des Frühwarnsystems
• 8 Schematische Darstellung des Funktionsprinzips des Muschelrucksacks mit zeitlicher Abfolge der einzelnen Aktionen.
• 9 Schematische Darstellung der bildbasierten Erkennung und Nachverfolgung der Bioindikatoren.
• 10 Schematische Darstellung des Funktionsprinzips des verwendeten Ansatzes zum Maschinellen Lernen.
• 11 Messergebnisse zur drahtlosen Übertragung des Muschelrucksacks unter Wasser in einem See.

The device presented here is explained in more detail below by way of example with reference to the accompanying drawings. Show it:

• 1 : Schematic representation of the ecotoxicological early warning system with freshwater mussels as bioindicators, which can move freely through the compact and wireless sensor backpack.
• 2 : Schematic representation of a bioindicator with wireless sensor backpack and sensors to measure behavior.
• 3 : Schematic representation of the circuit board including the battery in the sensor backpack.
• 4 : More detailed schematic representation of the components of the gateway (dashed lines) and its interaction with the surrounding components.
• 5 : Schematic representation of the use of the wireless shell sensor technology outdoors to monitor the pollution of the water.
• 6 : Schematic representation of another version of the system using transmitter units floating on the water surface.
• 7 Schematic representation of a wired version of the early warning system
• 8th Schematic representation of the functional principle of the shell backpack with the chronological sequence of the individual actions.
• 9 Schematic representation of the image-based detection and tracking of the bioindicators.
• 10 Schematic representation of the functional principle of the machine learning approach used.
• 11 Measurement results for the wireless transmission of the mussel backpack under water in a lake.

• - Einer Muschelsensorik (101), welche u.a. aus einem Mikroprozessor, Batterie, Sensoren, Analog-Digital-Wandlern, drahtgebundener und/oder drahtloser Ladeschaltung, einer Sende-Empfänger-Einheit, auch Transceiver genannt, sowie einer Antenne bestellt.
• - Einem Bioindikator (102) wie einer in der Skizze dargestellten Muschel, Krabben, Fischen oder Fischeiern.
• - Sensoren (103), welche nicht direkt im Muschelrucksack (101) untergebracht sind, sondern an einer geeigneteren Position zur Messung des Öffnungsgrads- bzw. Winkels der Muschelschalen oder der Bewegung. Besonders relevant sind Hall-Effekt-Sensoren in Kombination mit Magneten bzw. Sensoren zur Messung des elektrischen Feldes, um den Öffnungswinkel der Muschel zu ermitteln. Darüber hinaus kann ein Akzelerometer verwendet werden, um die Bewegung und Neigungswinkel des Bioindikators zu ermitteln. Es empfiehlt sich, nur jeweils einen Typ der genannten Sensoren zu verwenden, um den Bioindikator durch zusätzliche Gewichte und volumetrische Hindernisse nicht unnötig einzuschränken. Bei größeren Bioindikatoren können auch zusätzliche Sensoren vom gleichen Typ verwendet werden, um einem technischen Ausfall vorzubeugen bzw. statistisch belastbare Daten zu erhalten.
• - Einem Behältnis (114) für die Testsubstanz, sowie eine Ein- (115) und Ausleitung (116) und einem Boden aus Sand (104). Der Behälter kann im Wesentlichen als Aquarium bzw. Ausleitbecken ausgeführt werden, in einer weiteren günstigen Ausführung wird das System in Seen oder Flüssen verwendet. Für den Boden empfiehlt es sich, feinkörnigen Sand zu verwenden, sodass die Muscheln in der Lage sind, sich problemlos einzugraben, um eine ideale Filtrierposition einzunehmen. Im Schnitt ist die Verwendung von Sand mit einer durchschnittlichen Korngröße von bis zu wenigen mm zu empfehlen.
• - Gateway (107), welches bidirektional und drahtlos (106) mit dem Muschelrucksack kommunizieren kann. Das Gateway ist nicht beschränkt auf die Verwendung eines bestimmten Funkstandards. Mögliche Funktechnologien sind LoRaWAN, Sigfox, NB-loT, 2G, 3G, LTE/4G, 5G, Bluetooth, WiFi und andere. Bei der Wahl des Funkstandards gilt es zu beachten eine Technologie zu wählen, welche wenig Leistung verbraucht, jedoch über weite Strecken übertragen kann. LoRa stellt zum Zeitpunkt der Offenlegung dieser Beschreibung ein Vorreitertechnologie mit den genannten Eigenschaften dar und wird daher bevorzugt in dieser Erfindung verwendet. Das Gateway besteht aus einer Recheneinheit, einem Sendeempfänger zur bidirektionalen Kommunikation mit der Muschelsensorik sowie über einen Displayausgang zur Visualisierung des Systemstatus. Das Gateway wird in einem folgenden Abschnitt noch detaillierter beschrieben.
• - Machine Learning Software (109) als computergestützte Entscheidungshilfe, welche, ein Klassifizierungsverfahren zur Verhaltenserkennung der Muscheln durchführt. Bei dem hier vorgestellten maschinellem Lernansatz wird nur ein überwachtes Lernen bzw. supervised Learning durchgeführt. Die Klassen, welche jeweils verschiedenen Verhaltensmuster der Bioindikatoren entsprechen, z.B. offen, geschlossen, öffnet gerade, schließt gerade, verschließt sich stark wie etwa bei einer Muschel, sind bereits am Anfang bekannt. Bei den Algorithmen können verschiedene Ansätze verwendet werden wie „Convolutional Neural Network“, „Decision Tree“, „Segmentation“ etc.
• - Im Backend (108) werden die vom Gateway erhaltenen Messdaten gespeichert, ggf. prozessiert und als Schnittstelle auch API genannt dem Machine Learning Algorithmus bzw. dem Frontend zur Verfügung gestellt. Als Datenbanksysteme kommen sowohl SQL wie NoSQL (Not only SQL)-Systeme in Frage. Im Frontend (110) werden dem Endnutzer die Daten letztlich präsentiert. Dort kann auch programmiert werden, in welcher Form der automatische Alarm (112) erfolgt wie beispielsweise über SMS, Telefon, WhatsApp oder Email sowie aus deren Kombinationen. Mit einem portablen Gerät (111) wie z.B. einem Smartphone, Smartwatch, Tablet oder auch einem Desktop-PC wir das Frontend dem Endnutzer dargestellt. Über das portable Gerät ist auch eine Befehlseingabe möglich, z.B. das Konfigurieren der Alarmfunktionen.
• - Ein automatisches zuverlässiges und frühzeitiges Warnsignal (112) stellt einen enormen Mehrwert dieses Systems dar. Die Zuverlässigkeit leitet sich durch die Verwendung eines sophistizierten maschinellen Klassifizierungsverfahrens ab, dessen Genauigkeit über die Zeit durch Sammeln immer weiterer Daten noch erhöht werden kann.

A favorable embodiment of this invention is in 1 shown and characterized by the following features:

• - A shell sensor system (101), which includes a microprocessor, battery, sensors, analog-to-digital converters, wired and/or wireless charging circuit, a transceiver unit, also known as a transceiver, and an antenna.
• - A bio-indicator (102) such as a mussel, crab, fish or fish eggs shown in the sketch.
• - Sensors (103), which are not located directly in the mussel backpack (101), but in a more suitable position for measuring the degree of opening or angle of the mussel shells or the movement. Hall effect sensors in combination with magnets or sensors for measuring the electric field are particularly relevant in order to determine the opening angle of the mussel. In addition, an accelerometer can be used to determine the movement and tilt angles of the bioindicator. It is recommended to use only one type of each of the mentioned sensors in order not to unnecessarily restrict the bio-indicator with additional weights and volumetric obstacles. With larger biological indicators, additional sensors of the same type can also be used to prevent a technical failure or to obtain statistically reliable data.
• - A container (114) for the test substance, as well as an inlet (115) and outlet (116) and a floor made of sand (104). The container can essentially be designed as an aquarium or drainage basin. In another favorable design, the system is used in lakes or rivers. For the bottom, it is advisable to use fine-grained sand, so that the mussels will be able to burrow without any problems, in order to have an ideal filtration position. On average, the use of sand with an average grain size of up to a few mm is recommended.
• - Gateway (107), which can communicate bidirectionally and wirelessly (106) with the shell backpack. The gateway is not limited to the use of a specific radio standard. Possible radio technologies are LoRaWAN, Sigfox, NB-loT, 2G, 3G, LTE/4G, 5G, Bluetooth, WiFi and others. When choosing the radio standard, it is important to choose a technology that consumes little power but can transmit over long distances. At the time this description was published, LoRa represents a pioneering technology with the properties mentioned and is therefore preferably used in this invention. The gateway consists of a computing unit, a transceiver for bidirectional communication with the mussel sensors and a display output for visualizing the system status. The gateway is described in more detail in a following section.
• - Machine Learning Software (109) as a computer-aided decision-making tool, which carries out a classification procedure to recognize the behavior of the mussels. In the machine learning approach presented here, only supervised learning is carried out. The classes, which each correspond to different behavior patterns of the bioindicators, eg open, closed, just opening, closing just, closing strongly like a mussel, are already known at the beginning. Various approaches can be used for the algorithms, such as "Convolutional Neural Network", "Decision Tree", "Segmentation" etc.
• - In the backend (108), the measurement data received from the gateway are stored, processed if necessary and made available to the machine learning algorithm or the frontend as an interface, also known as API. Both SQL and NoSQL (Not only SQL) systems can be used as database systems. The data is ultimately presented to the end user in the front end (110). There you can also program the form in which the automatic alarm (112) takes place, for example via SMS, telephone, WhatsApp or email and combinations of these. We do this with a portable device (111) such as a smartphone, smartwatch, tablet or even a desktop PC Front end presented to the end user. Commands can also be entered via the portable device, eg configuring the alarm functions.
• - An automatic reliable and early warning signal (112) is an enormous added value of this system. The reliability derives from the use of a sophisticated machine classification method, the accuracy of which can be increased over time by collecting more and more data.

2 outlines an advantageous embodiment of the invention with a freshwater mussel (201) as a bioindicator. Instead of the freshwater mussel, other types of mussels or crabs as well as fish can also be used. The mussel digs into the sandy subsoil (211) via its mussel feet (202) and mussel threads (203) in order to assume an ideal filtering position. The dimensions of the compact shell backpack (207) are approx. 30 mm (length) x 30 mm (width) x 15 mm (height). A negative (209) and a positive (210) charging cable lead out of the shell backpack, with which the battery can be recharged if necessary. The mussel backpack can be designed with various sensors to measure the environmental parameters for measuring electrical conductivity, pH value, dissolved oxygen content, turbidity or movement. In addition to the sensors that can be installed directly in the mussel backpack, sensors that are used expressly to measure mussel behavior are particularly important. These sensors include a Hall effect sensor (205), which is used by means of a magnet to determine the degree of opening of the mussel. Hall sensor and magnet are each attached to the opposite halves of the shell. A change in the degree of opening leads to a change in the magnetic field strength, which in turn induces a voltage signal in the Hall sensor that can be measured by an analog-to-digital converter (ADC). The ADC is located in the shell backpack. The Hall sensor is usually designed as an analogue sensor, which has to be supplied with current via two cables and returns an analogue signal (206) as a measured value, which correlates with the degree of opening of the mussel. In addition to the Hall sensor, an accelerometer can also be used, which can measure in up to 3 axes (x, y and z). Depending on the type, an accelerometer can not only determine changes in movement, but also an orientation that is static, such as the inclination of the bioindicator in the gravitational field. These accelerometers use surface-treated polysilicon, which in turn is suspended on polysilicon springs. The displacement of the structure in the gravitational field is measured using differential capacitors. The measured values are usually given in the units of acceleration, i.e. m/s^2 for all three directions in space and transmitted to the shell backpack. This transmission takes place via common communication standards, eg analogue or via I2C, SPI, UART, RS232, RS485 or other serial communication standards. The simple and widespread I2C protocol is currently used most frequently, followed by SPI. In addition to the sensors, the mussel should also have a marking (214) which can be used to detect the mussel by means of image acquisition. The marking can be made either in one color, multicolor, with a specific pattern or in the form of a shape. In addition to the accelerometer data, this approach offers an additional parameter for measuring the movement and the mussel behavior. At this point it should be pointed out that detecting the mussels via image recognition is only practicable if the degree of turbidity of the test substance is correspondingly low. A corresponding degree of turbidity can usually only be guaranteed indoors. In addition, real-time object detection and object tracking is relatively computationally intensive and requires the use of computers with minimal computing power comparable to single-board computers such as the RaspberryPi3 & 4 or the NVIDIA Jetson Nano. The marking and image recognition are therefore more suitable for using the system indoors and not for lakes or rivers. For example, the linear sensor SS39ET from Honeywell (Charlotte, North Carolina, USA) with a typical output voltage of 1.4mV/G can be used as a Hall effect sensor. The ADXL345 from AnalogDevices (Norwood, Massachusetts, USA) is a good accelerometer.

An advantageous design of the printed circuit board for the mussel sensors is in 3 sketched. The printed circuit board, also called PCB (=Printed Circuit Board), has a central processing unit, the microprocessor (302). In this version, this processor also has an integrated transceiver unit for wireless transmission and reception of data, as well as several analog-to-digital converters, with which, among other things, the charge status of the battery (304) or analog signals (310 & 313) from the external sensors can be detected. The external sensors can be powered via pins (308 & 309). The pins can either be connected directly to the supply line (designated Vdd or Vcc) or to digital outputs of the microprocessor. The latter version would have the advantage that the power supply to the external sensors can be switched off completely in standby mode. This enables particularly energy-saving operation. Sensors for measuring environmental parameters can be used on the printed circuit board such as a temperature sensor (305) or an accelerometer (311). The printed circuit board can also be equipped with other sensors, such as a sensor for measuring electrical conductivity, pH value, the proportion of dissolved oxygen or turbidity, as well as combinations of the measurement variables mentioned. The printed circuit board has a terminal (303) for installing the operating system and for programming the microprocessor. UART is preferably used as the communication standard here. The microprocessor with integrated transceiver is electrically connected to an antenna (312) to increase the range of the electromagnetic signal. Possible wireless transmission standards include, without limitation, Narrowband Internet of Things (NB-loT), LoRaWAN, Sigfox, Dash7 or Bluetooth low energy (BLE). Other radio standards can also be used, including Global System for Mobile Communication (GSM), Long Term Evolution (LTE), WiFi and others. However, the use of modern low-energy standards such as NB-loT or LoRaWAN, which were designed for loT applications with a long range, battery operation and small amounts of data, is particularly advantageous. The antenna can be implemented as a PCB antenna, as an SMT antenna (surface mount) or as an external antenna. Choosing a PCB antenna is not advantageous as the already very limited surface area of the printed circuit board is used for traces and integrated circuits. The most efficient way is to use a so-called "quarter-wavelength antenna", which is just a piece of wire, preferably with PVC insulation, with a length of exactly a quarter of the wavelength of the emitted electromagnetic radiation. With a particularly common frequency of 868MHz, this antenna length would correspond to approx. 8.6cm. In water, the electromagnetic wavelength is reduced as a function of electrical conductivity, and thus the antenna length as a function of electrical conductivity, which must be taken into account in the design (see VK5BR, Lloyd Butler. "Underwater radio communication." Originally published in Amateur Radio (1987).

A more detailed design of the gateway is in 4 shown. The gateway runs on a single board computer (401) that has sufficient processing power to perform real-time object detection and tracking using camera data. The cameras can be implemented either outside (412) or inside (413) of the container (408). The gateway is also equipped with a transceiver unit (402) for bidirectional communication with the shell sensor system. The electromagnetic signals are received or sent via antennas. These antennas can be placed either in the medium of air (407) or in water (409). The absorption of electromagnetic waves in the water is enormous, so that the range of the mussel sensors and the freedom of movement of the mussel can be extended enormously by skillfully placing the antennas. The use of an additional transceiver unit is only necessary if the single-board computer does not already have the wireless communication standards used. Currently, these standards are mostly limited to WiFi and Bluetooth, mostly in the 2.4 or 5 GHz range with high bandwidths. Both high bandwidths and high frequencies further reduce the already short range of electromagnetic signals in water compared to air. In an advantageous embodiment of this invention, LoRa is therefore used as the radio technology (see also the patent specifications US9525454B2 , EP3065302B1 and US9794095B2 about this technology). The gateway can either be supplied via a permanently installed power supply such as the 230V mains or independently of the mains. In the case of independent power supply, a 12V battery such as that used in cars or small solar systems can be used, for example. For energy self-sufficient operation, the gateway can also be equipped with a solar panel. The solar panel must be designed according to the average power consumption of the gateway and the average solar radiation at the installation site, taking into account a significantly reduced solar radiation in winter. In addition, a safety factor of 2 should be taken into account in order to continuously maintain the power supply. Tests have shown that 100W solar panels are sufficient for common gateways, but using 250W panels is safer. This estimate does not include the use of cameras. The end user or system operator is informed of the status of the system via a display unit (411), which can be designed as a smartphone, tablet or display or combinations thereof. The gateway must have Internet connectivity (404), which can be guaranteed via cellular, a local Ethernet socket (403) with an Internet connection or satellite Internet. Other wireless radio standards of the single-board computer (410) should enable the user to access the system via an app, for example. The gateway is able to communicate with a backend (409) via an internet connection. Data is stored on the backend and interfaces are provided for further processing by the frontend or the machine learning algorithm. Specifically, a Raspberry Pi 1, 2, 3 or 4 and probably all successors as well as the NVIDIA Jetson Nano or PCs with comparable power can be used for the single-board computer use. There are already numerous LoRa gateways for the RaspberryPi, which have an excellent range. These include the PG1301 LoRaWAN Concentrator from Dragino or the RAK7243 from Rakwireless.

In a further advantageous embodiment of this invention in 5 outlined, the shell sensors and shells (501) are used area-wide along a river (507). As already outlined, the measurement data from the mussel backpacks can be received via gateways with or without an energy self-sufficient supply. In places with sufficiently good mobile phone coverage, you can do without the gateways and install the appropriate radio modules in the shell sensors. The advantage here is the use of NB-loT (Narrowband-loT) modules with low bandwidth and high range as well as good building penetration. Previous solutions have the sensitive disadvantage that they do not allow area-wide measurement, but usually only in a very small area. In the configuration proposed here, however, a river that stretches over hundreds of kilometers or a lake with an area of several thousand square kilometers can be tagged and controlled. A concrete design of this system is outlined using the installation of several mussel sensors in the river Rhine in Germany. Sewage treatment plants (502) and mines (505) discharge their waste water here. Certain substances such as medicines (birth control pills, blood fat reducers, painkillers or antibiotics) but also cleaning agents, wall paints and pollutants on objects such as clothing or pans can only be broken down with difficulty in the sewage treatment plant. Continuous monitoring by the sewage treatment plant is therefore indicated. In addition, however, it is also advisable to monitor the sewage treatment plant externally in order to further reduce the risk of excessive pollution. Mine operators such as RAG GmbH in North Rhine-Westphalia are obliged to take care of the contaminated sites caused by hard coal mining. Mine water, which collects as rainwater in tunnels, pits and shafts, makes up the largest proportion of the costs. The mine water must not rise so high that it mixes with the groundwater and pollutes it and makes it unusable. Therefore, the mine water has to be pumped out regularly and the pollutants cleaned, which usually takes place in a chemical process using hydrogen peroxide. However, both too high and too low a dosage of the clarifying agent causes polluted water. As in the case of the sewage treatment plant, additional monitoring by the industry to ensure high water quality makes sense here. Additional sources of pollution can also arise from precipitation (503) or people (504).

In another favorable version, which in 6 is shown schematically, the transmitter unit (603) is on the water surface. In this configuration, the distance in the water, i.e. between the shell and the water surface, is bridged. This increases the range from a few meters in water to a few kilometers in the air (without obstacles such as vegetation or buildings). In terms of its function and components, the transmission unit is essentially designed to be comparable to the mussel sensor system. The transmission unit contains a transceiver module (610) for bidirectional communication with a gateway (605), a microprocessor (609) for data processing and a battery (611) for power supply. The transmitter unit can also forward the collected measurement data directly via mobile radio (607) or satellite internet (606). The most energy efficient would be to use a LoRa gateway, closely followed by NB-loT in the cellular network. Using satellite internet is relatively power intensive for this application. The shell backpack can compared to 2 can be made significantly more compact because no battery or microprocessor is used at this point. On the bio-indicator, which is a shell in this version, there are only the components that absolutely have to be attached there. These include, for example, a heart rate sensor, a Hall sensor or an accelerometer and combinations thereof. The sensors on the mussel are supplied with current via a power and data cable (608) and the measurement data are read out.

7 Fig. 12 is a schematic representation of a further advantageous embodiment of the invention. Until now, the shell backpack (703) communicated wirelessly. However, cables are used in this configuration. The use of cables (704) has a significant disadvantage in that the shells (702) cannot move as freely in the container (701) as in the wireless configuration. This disadvantage can be eliminated by using thin, highly flexible cables. In addition to the reduced mobility of the shells, this system also offers some advantages. The sampling rate can be significantly increased down to a few seconds. The measured values can also be stored with higher accuracy. When using LoRaWAN, the payload (= transmitted data volume) must always be kept as lean as possible. The larger the payload, the longer a transmission takes. Well-known network servers such as TheThingsNetwork prescribe a maximum transmission time of approx. 30s per day. So if the amount of data is significantly reduced, are sent much more frequently every day. In addition, the use of batteries or rechargeable batteries is also eliminated. The shell sensors can be powered directly via the gateway (704). This weight then no longer hinders the mobility of the bioindicators. Code updates can also be loaded more quickly using such a system, since the microprocessors no longer have to be programmed individually, only the gateway. The gateway can be designed to be accessed via remote control software such as Teamviewer. Overall, such a configuration can significantly reduce development cycles and the system is also easier to maintain. Telephone cables and four-wire ribbon cables, which are often connected to RJ11 plugs/sockets and 6p4c plugs/sockets, are excellently suited as cables. Four cores are also exactly the number that are required, two for data transmission and two for power supply.

The functional diagram of the wireless shell backpacks is in 8th sketched. At the beginning, the measured values of the sensors are recorded (801). The measured values (802) include Hall voltage, acceleration, heart rate, temperature, but other measured values such as electrical conductivity, pH value, dissolved oxygen content and combinations of the measured values mentioned can also be used. To reduce the power consumption, it is then checked whether the measured values have changed by a certain value (803). To do this, the current measured values in the EEPROM (805) are read out (804). If the readings have not changed, then the microprocessor is put back into sleep mode (807). If the measured values have changed, they are transmitted wirelessly and stored in the EEPROM. After sending, the microprocessor is put into sleep mode. The microprocessor exits sleep mode after a defined time and is woken up (808). Some sensors also have a so-called interrupt function or you can simply expand the system with such an interrupt function by pulling the voltage signal to LOW or HIGH. LOW is usually the voltage 0V and High is a value usually between 2-5V.

In an advantageous embodiment of this invention, real-time camera data is used to record the 'walking paths' or the movement behavior of the bio-indicators. For this image-based approach, a mark (214) is required, which is placed on the shell, as in 2 sketched. The tracking of the biological indicators is based on a simple flow chart, which is 9 is sketched. At the beginning, the video data of the camera or cameras (901) are converted into individual images, also known as frames. The images are then converted to an 8-bit gray scale (902). If necessary, a noise filter (903) can also be applied, or a soft character filter, which has proven to be advantageous for the subsequent edge detection (904). After edge detection, different shapes (905) can be recognized. Each shell must be marked with a unique shape such as a square, rectangle, circle, ellipse, star, or combinations thereof. After the respective shape has been identified, a region of interest is determined (906) and its focus or center point is stored either as a 2D coordinate or, if more than one camera is used, as a 3D coordinate. The region of interest (ROI) is tracked (907) throughout time and its center is reported as a reading (910). This configuration is particularly useful indoors where the biological indicators are in an aquarium or container with relatively optically clear water. The container selected should not be too large, otherwise the bioindicators and their markings appear blurred because the longer optical light path means that the light is more often deflected or absorbed by suspended matter in the container. Instead of shapes, colors can also be used for marking. An accelerometer can only be used to determine the inclination of the bioindicator and sudden movements. By detecting and tracking the path of the bioindicators using camera data, an extended movement pattern can be mapped, which can ultimately lead to increased accuracy in the classification of the behavioral pattern. The software for this method was written in Python in the OpenCV framework.

The classification of mussel behavior is to be made using methods of machine learning, also known as machine learning. With these methods, some approaches have emerged as particularly promising. These include the Support Vector Class, KNeighbors Classifier, Convolutional Neural Network and the Decision Tree Classifier. All four methods were tested during development, but the DecisionTreeClassifier provided the best results with over 96% accuracy and is therefore used in an advantageous embodiment of this invention. In the following it is disclosed how this classifier is used for the presented system. The decision tree method is a common way of regression or classification over a diverse data set. Decision trees are data mining methods or decision rules in the form of a tree. The result is a tree with a root and branches extending from it. The branches ver branch continuously at nodes. The branches finally end in leaves. These sheets then indicate the class membership or the decision.

• • Zuerst wird das Merkmal mit dem höchsten Informationsgehalt in Hinblick auf die Vorhersage des Labels (Zielvariable) selektiert.
• • Für jeden Wert, den das Attribut annehmen kann, wird anschließend ein Zweig des Baumes erstellt.
• • Für jeden neuen Knoten wird Schritt 1 und 2 wiederholt.

Decision trees are popular because they can represent rules in a simple and understandable way. The rules are processed hierarchically, ie one after the other in a fixed order and then end with a result. The algorithm is in 10 sketches and works on discrete variables as follows:

• • First, the characteristic with the highest information content with regard to the prediction of the label (target variable) is selected.
• • A branch of the tree is then created for each value that the attribute can assume.
• • Steps 1 and 2 are repeated for each new node.

The tree is complete when each node uniquely identifies a class. The last nodes then determine the class. These are also called leaves.

A popular machine learning framework is Tensorflow with Python support, which was also used in this invention. 10 outlines the flowchart of the Decision Tree method. Initially, the entire dataset is imported (1001) and then split (1002) into a training, validation, and test dataset. As the names suggest, the training data set is used to build the model and the validation data set is used to check that the model produced can make reliable predictions on unknown data. If this is the case, this model is said to have good generalization. Since the parameters are constantly adjusted by the developer in order to have good results with the training and validation data set, the model is also partially trained automatically with the validation data set, which limits the ability to generalize. Therefore, at the end of the day, you should always check the model against a completely unknown data set, the test data set. INPUT and OUTPUT (1003) are now extracted from the data record that has been read in. INPUT includes relevant environmental values such as Hall voltage, heart rate, temperature and the time during which OUTPUT describes the classification of the mussel behavior, i.e. the following classes in the example of the mussel as a bioindicator: 'Default' (1006), 'Opening' (1007), 'Closing ' (1008), 'Disturbed' (1009). It is also possible that other relevant classes may be included in the future or replace those mentioned. With the INPUT and OUTPUT data, the decision tree model (1004) can now be created, which can ultimately predict the classes (1005) mentioned for unknown INPUT data.

In a further advantageous embodiment, the wireless shell sensor system consists of the following components without limitation. An ATMEGA1284p from ATMEL (San Jose, California, USA) as the microprocessor, an RFM95Hope LoRa module from HopeRF (Shenzhen, China), a TMP116 temperature sensor from Texas Instruments, an ADXL345 accelerometer from Analog Devices (Norwood, Massachusetts, USA), and a Heart rate sensor based on a CNY70 IR reflectometer from Vishay (New York, USA). The shell sensor system can also be designed with comparable components with a comparable range of functions. The selection of the components mentioned is very large. However, it should be noted that the dimensions of the components should be as small as possible. The smallest design developed and manufactured in this invention measures approximately 28mm x 30mm x 3mm. The scientific publication by Burnett et al. (Burnett, Nicholas P., et al. "An improved noninvasive method for measuring heartbeat of intertidal animals." Limnology and Oceanography: Methods 11.2 (2013): 91-100.) as template (s. 2 . for a schematic sketch of the circuit and Table 1 for an order list of the components).

11 shows the measurement results of an experiment with the developed mussel sensor to verify wireless transmission under water. The LoRa module RFM95Hope from HopeRF was used as a transceiver unit and the LoRa gateway Dragino PG1301 (Dragino Technology Co., LTD. Shenzhen Dragino technology development co., LTD) with a Raspberry Pi4. The measuring point was the Echinger See (GPS coordinates approx.: 48.287, 11.631, Germany, federal state of Bavaria), the electrical conductivity was approx. 300 µS/cm. The Received Signal Strength Indicator (RSSI) was measured in dBm as a function of increasing depths of the mussel rucksack below the lake surface. As was to be expected from the physical point of view, RSSI decreases sharply with increasing immersion depth due to the attenuation of the electromagnetic waves in the water. However, it could be shown that the transmission worked without any problems up to an immersion depth of 70 cm.

Reference List

101

Sensor backpack

102

bioindicator

103

sensor

104

Floor

105

display unit

106

wireless communication

107

Gateway

108

backend

109

machine learning

110

front end

111

input

112

alarm

113

Multi-parameter measuring probe

114

measuring container

115

test water discharge

116

test water drainage

117

camera

201

shell

202

clam foot

203

shell threads

204

magnet

205

Hall effect sensor

206

sensor cable

207

Sensor backpack

208

wireless communication

209

Charging cable negative

210

Charging cable positive

211

Floor

212

accelerometer

213

Electrode for wireless charging

214

mark

301

circuit board

302

microprocessor

303

programming pins

304

battery

305

temperature sensor

306

charging circuit

307

heart rate circuit

308

Power cable 1 external sensors

309

Power cable 2 external sensors

310

Signal 1 external sensors

311

accelerometer

312

antenna

313

Signal 2 external sensors

314

charging cable 1

315

charging cable 2

301

circuit board

302

microprocessor

303

programming pins

304

battery

305

temperature sensor

306

charging circuit

307

heart rate circuit

308

Power cable 1 external sensors

309

Power cable 2 external sensors

310

Signal 1 external sensors

311

accelerometer

312

antenna

313

Signal 2 external sensors

314

charging cable 1

315

charging cable 2

401

single board computer

402

receiver unit

403

Ethernet port

404

internet connectivity

405

power connection

406

electric generator

407

antenna in air

408

Aquarium

409

antennae in the water

410

wireless connectivity

411

display unit

412

Webcam in air

413

Webcam underwater

501

Shell with sensors

502

sewage treatment plant

503

precipitation

504

persons

505

Industry

506

gateways

601

water tank

602

bioindicator

603

transmitter unit

604

electronics

605

Gateway

606

satellite internet

607

cellular

608

power and data cables

609

microprocessor

610

transceivers

611

battery pack

701

water tank

702

bioindicator

703

shell backpack

704

Cable

705

Gateway

801

Measurement of the sensor values

802

sensors

803

Check if sensor value changes > X%

804

Reading the EEPROM

805

EEPROM

806

Wireless sending

807

sleep mode

808

Awaken

809

Writing to EEPROM

901

Conversion video stream to images

902

Conversion to grayscale

903

noise filter

904

edge detection

905

forms detection

906

Determination of the region of interest (ROI)

907

Tracking ROI

908

Output of the ROI as a measured value

100

Database is imported

1002

Database is scaled and divided

1003

X and Y are read from the database and transferred to 1004

1004

Tree decision model is created

1005

classes

1006

class Default

1007

Class opening

1008

class closing

1009

Class Disturbed

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• RU 101838 U1 [0003]
• RU 2452949 C1 [0004]
• RU 2437093 C1 [0004]
• RU 2595867 C2 [0005]
• US 9525454 B2 [0018]
• EP 3065302 B1 [0018]
• US 9794095 B2 [0018]

Claims (15)

Large-scale ecotoxicological early warning system with a freely movable bio-indicator and a compact electronic measuring and transmitting device, also abbreviated as early warning system, which includes the following features: - At least one biological indicator - A mussel backpack attached to the bioindicator. - External sensors not attached to the shell backpack to capture environmental parameters - Gateway for wireless communication with the shell backpacks. - An edge device running machine learning software to automatically classify mussel behavior. - Automatic alarm function An early warning system according to claim 1 characterized in that at least one biological indicator is used which belongs to a mussel species, crab species, fish species, fish eggs and combinations thereof. An early warning system according to claim 1 characterized in that the mussel backpack attached to the bio-indicator is either wireless or wired and a combination of wireless and/or wired mussel backpacks in the same system is possible. An early warning system according to the previous claim , characterized in that both the wireless and the wired mussel backpack are equipped with the following sensors or combinations thereof: temperature, pressure, electrical conductivity, turbidity, dissolved oxygen content, pH value, accelerometer, gyrometer or magnetometer for measuring the Movement of the shell, Hall sensor with magnet for measuring the opening angle of the shell halves. An early warning system according to the two previous claims , characterized in that external sensors, i.e. sensors not housed in the shell backpack, are attached to the shell to record its behavior, consisting of the following sensors or combinations thereof: Hall sensor with magnet, accelerometer, gyrometer and magnetometer. An early warning system according to the previous three claims , characterized in that the shell backpack is equipped with at least one or a combination of the following antennas for wireless communication in the ultra-high frequency range of 300-3000MHz: PCB (Printed Circuit Board) antenna located directly in the conductive layer of the printed circuit board , surface mount (SMT) antenna or a monopole antenna. An early warning system according to the previous four claims , characterized in that the shell backpack is equipped with a rechargeable battery, which can be charged either wired or wirelessly via a coil and a charging circuit mounted on the printed circuit board. An early warning system according to claim 3 characterized in that it is equipped with at least one gateway for wireless and wired communication with at least one shell backpack. An early warning system according to the previous claim , characterized in that the gateway essentially consists of an edge device and a transceiver module, which have the following features or combinations thereof: Wi-Fi (from IEEE 802.11 Wi-Fi 4 upwards), Bluetooth (Bluetooth Classic & Bluetooth Low Energy), Ethernet connection, High Definition Multimedia Interface (HDMI) output, Ethernet connection, CPU and a GPU for processing image data. An early warning system according to the previous two claims , characterized in that the gateway can be powered by solar power with common voltages of 5, 12, 24 and 48V and a buffer battery in the outdoor area and with a multi-band mobile radio module with support for 2G, 3G, LTE, 5G and NB-loT or combinations thereof or satellite internet. An early warning system according to claim 1 characterized in that this is used both in an artificial container such as an aquarium as well as outdoors over large areas and distances in rivers, lakes, ponds or seas and combinations thereof. An early warning system according to claim 1 characterized in that a program based on machine learning approaches for classifying the behavior of the individual bioindicators as well as their collective behavior is executed on the edge device. An early warning system according to the previous claim , characterized in that an alarm is sent to the user by telephone, SMS, e-mail, WhatsApp or push notification, and combinations thereof, as soon as the classification method determines that the bioindicators follow a behavior pattern which indicates an increased pollution of the surrounding test substance. An early warning system according to claim 1 characterized in that, in addition to the mussel backpacks, this can be equipped with at least one multi-parameter probe for measuring one of the relevant environmental parameters temperature, pH value, electrical conductivity, turbidity, dissolved oxygen content and combinations thereof to improve the accuracy of the classification process. An early warning system according to claim 1 characterized in that it is equipped with a display unit consisting of a monitor, a smartphone, a tablet, a smartwatch or combinations thereof for visualizing the system status and for maintenance directly at the measuring point.

Images