AI-supported detection system for the automatic identification of stored product and hygiene pests

Abstract

Main claim: AI-based detection system for the automatic identification of stored product and hygiene pests characterized in that a) the detection system consists of one or more sensors that communicate wirelessly or via a wired connection with a central evaluation unit. b) the sensors are mechanically composed of a sensor housing and one or more measuring chamber(s). c) all electronic components for acquiring measured values ​​from the sensors, for data analysis, for storing results and for communication are arranged in the sensor housing and are provided with a power supply. d) in a preferred embodiment, all functions for measurement acquisition from the sensors, data analysis, result storage and data communication are implemented via an application-specific programmed microcontroller with supplementary hardware components. e) each measuring chamber is in direct contact with the ambient air. f) in each measuring chamber one or more sensors for measuring temperature, humidity and VOC profile are arranged and each sensor is connected by wire to the measurement acquisition in the sensor housing. g) A measuring chamber design for detecting stored-product pests (for example, but not limited to: grain weevils, drugstore beetles, flat grain beetles, rice weevils, etc.) preferably, but not exclusively, consists of a pipe section for piercing a bulk material, which is connected to the gas phase in the bulk material via one or more openings. In a particular embodiment, this sensor can also be permanently installed in the storage container of the bulk material. h) a special embodiment consisting of several superimposed measuring chambers according to claim g) which are placed in the bulk material and thus enable the measurement of a VOC and moisture profile in the bulk material. i) any mechanically suitable material may be used as material for the construction of the measuring chamber housing, which does not cause relevant temperature measurement errors by introducing the surrounding outside temperature and which does not cause relevant humidity measurement errors due to material-related hygroscopy. k) the moisture content in the bulk material is used as an additional indicator of pest infestation (“metabolic water”). I) at selected measurement locations, stored product and hygiene pests are identified via the VOC profile of their species-specific, metabolically related body odor and / or the odor of their excretions. m) the presence of stored product and hygiene pests (for example, but not limited to: rats, mice, cockroaches, etc.) is identified from the measured VOC profile at the measuring point using algorithms, by means of a training phase in which the detection algorithms were determined in advance using an AI. n) When measuring the VOC profile of a pest species, the use of complex VOC sensor arrays is avoided by using a VOC sensor with a parameterizable heater. Here, the temperature in the sensor housing is increased in stages, and a VOC measurement is carried out after each temperature stage is reached. o) Higher-resolution VOC profiles can be achieved, for example, by ensuring that the maximum temperature in the sensor housing is only reached after several consecutive measurements. In this process, measurements are taken within a smaller temperature range during each individual measurement. VOC profiles generated in this way consist of as many individual values ​​as there are predefined temperature levels, multiplied by the number of measurement sequences until the maximum temperature is reached. p) for the simultaneous detection of various stored product and hygiene pest species, several differently trained and configured VOC sensors are installed in one sensor housing. q) the sensor electronics also handle the transmission of results to non-volatile storage media and / or via wired and / or wireless data communication to a central evaluation unit. r) the central evaluation unit communicates with the sensor electronics and performs, for example but not exclusively, the archiving of raw data, the evaluation of current measured values, the analysis of time series and trends, the detection of limit values ​​and other communication functions such as SMS and / or email.

Description

[0001] The invention relates to an AI-supported detection system for the automatic identification of stored product and hygiene pests. Fig. 1 shows schematically, by way of example and not exclusively, various versions of the detection system for the automatic identification of stored product and hygiene pests.

[0002] All currently available systems for the automatic early detection of stored product and hygiene pests are based on the detection of feeding traces and feces, as well as on trap systems (dead or live) with or without bait. These traps can be integrated into automated monitoring and reporting systems.

[0003] The detection system according to the invention utilizes the biogenic VOCs (bVOCs) produced by stored product and hygiene pests and / or their excrement, which are released into the environment by each active organism. It is used to support decision-making regarding the initiation of preventive or control measures within the framework of HACCP.

[0004] The measurement of the VOCs left behind during presence as well as on walking paths and locations is carried out temporarily or stationary via a parameterizable VOC sensor 3, a humidity sensor 4 and a temperature sensor 5 inside a measuring chamber 1 open to the outside 2

[0005] The possibility of measuring storage and hygiene pests in bulk materials is also possible with sensors in which a Fig. 2 or more measuring chambers axially one above the other Fig. 3 are arranged. The measuring chambers Fig. 3 are temporarily introduced into the bulk material 6 for measurement or are permanently installed there for stationary monitoring.

[0006] While the bVOC release of wood-destroying fungi and termites has already been the subject of scientific studies (Tiitta M. et al., 2009), almost no information is available on the VOC release of stored product and hygiene pests.

[0007] Current technology allows to avoid or subordinate the chemical identification of organism-specific VOCs via scientific preliminary investigations (Tiitta M. et al., 2009) by empirically determining the organism-specific bVOC profiles of the pests to be detected and / or their excrements under different practical conditions using one or more parameterizable VOC sensors.

[0008] Here, the bVOC emissions under different environmental conditions (temperature, humidity, body odor, walking route, hiding place, etc.) are used to train an AI that calculates an algorithm for the optimal identification of the pest from the totality of the measured individual VOC profiles.

[0009] To avoid the detrimental use of VOC sensor arrays (Nilsson K., 1996), the VOC composition is measured using a VOC sensor with a configurable heater. With such sensors, the air inside the sensor is heated gradually, and a VOC measurement is performed after each temperature step. This process creates a VOC temperature profile. The BME680 and BME688 VOC sensors from Bosch Sensortec, for example, offer 10 freely programmable temperature steps and can generate a 10-point VOC profile for each measurement. Higher-resolution VOC profiles are achieved, for example, by only reaching the maximum temperature in the measuring chamber after several consecutive measurements. In this case, a smaller temperature range is measured during each individual measurement.VOC profiles measured in this way on the BME680 / BME688 consist of 10 individual values ​​each multiplied by the number of measurement sequences until the maximum temperature is reached. By applying various measurement strategies, VOC profiles of almost any resolution can be provided for training an AI. Comparable sensors are also available and usable from other manufacturers.

[0010] The use of VOC sensors with configurable heaters offers many advantages over sensor arrays. The key advantage is the easy adaptation of the sensor electronics to different measurement tasks. This requires only changes to the measurement strategy, i.e., software modifications. In contrast, adaptations to VOC sensor arrays usually involve hardware changes, complex circuit board redesign, recalibration of the sensor group, and changes to sensor communication. Another advantage of VOC sensors with configurable heaters is their small footprint. The minimum space required for a VOC sensor array at / in a measuring point is approximately 300 mm. 2 VOC sensors with configurable heaters have a footprint of approximately 9 mm 2 .

[0011] The appropriateness of VOC measurements can also be determined via temperature sensor 5. Since VOC measurements are relatively energy-intensive, it should be ensured that the pests are actually active.

[0012] The measurement data acquisition and evaluation is preferably, but not exclusively, carried out by one or more microcontrollers. These programmable components handle the querying of the sensors, the calculation of the ambient humidity, and the identification of pests according to algorithms generated by an AI.

[0013] Each sensor Fig. 4 contains, in addition to the measuring chamber(s), at least one sensor electronics unit, via which the individual sensors in the measuring chamber(s) are read out via wires. Depending on the sensor design, additional functions (raw data calculation, execution of VOC detection algorithms and / or data buffering and / or communication and / or power supply) can be integrated into the sensor housing. Fig. 4.

[0014] The storage of measured values ​​from a Fig. 4a or more measuring chamber(s) Fig. 4b is carried out in the simplest case by sensor electronics 9 in a non-volatile storage medium (e.g. NV-RAM or SD card) Fig. 1a. This non-volatile memory can be replaced from time to time and read out for further analysis in a central evaluation unit with result output.

[0015] In a special embodiment, the sensor electronics 8 also performs evaluation and communication tasks, for example, by reporting critical findings and / or data sets to defined recipients via SMS or e-mail. Fig. 1c. This extended sensor electronics is usually, but not exclusively, integrated into the sensor housing together with a power supply 9.

[0016] When monitoring multiple, spatially separated sensors, the use of data concentrators can be useful. These data concentrators read the measured values ​​from the sensor electronics, either wired or wirelessly, and temporarily store them on a non-volatile storage medium. At specific intervals, the stored measured values ​​are read out and forwarded, either wired or wirelessly, to a central evaluation unit with results output. Fig. 1b and Fig. 1d.

[0017] Depending on the respective design of the sensor electronics, the function of the data concentrators can also be taken over directly by the central evaluation unit with result output via wired or wireless communication with the measuring sensors.

[0018] The central evaluation unit is preferably, but not exclusively, implemented as a specially programmed PC or microcontroller, supplemented by communication components. It handles, for example, but not exclusively, communication, archiving of raw data, evaluation of current measured values, analysis of time series and trends, and detection of threshold values.

[0019] In addition, the central evaluation unit reports critical findings, for example, via SMS or email to defined recipients. Using a suitable data format and a customized communication interface, the evaluation unit can be integrated as an extension into an existing monitoring system.

[0020] Areas of application for the detection system according to the invention for the automatic identification of stored product and hygiene pests include, for example, and not exclusively, hygiene monitoring in food processing plants (HACCP), in grain silos for the early detection of pest infestation, and in stable and storage hygiene in the agricultural sector. Literature: Tiitta M. et al. : Spectral and chemical analyzes of mold development on Scots pine heartwood. Eur J Wood Prod (2009) 67: 151-158 DE 102011108764 A1 : Device and method for determining the moisture content of wood (2011) Nilsson K. : ELECTRONIC NOSES FOR DETECTION OF ROT IN WOOD. The International Research Group on Wood Protection, Guadeloupe (1996) QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] DE 102011108764 A1

[0020] Cited non-patent literature

[0000] Tiitta M. et al., 2009 [0006, 0007] Nilsson K. : ELECTRONIC NOSES FOR DETECTION OF ROT IN WOOD. The International Research Group on Wood Protection, Guadeloupe (1996

[0020]

Claims

[1] Main claim: AI-based detection system for the automatic identification of stored product and hygiene pests characterized by , that a) the detection system consists of one or more sensors that communicate wirelessly or wired with a central evaluation unit. b) the measuring sensors are mechanically composed of a sensor housing and one or more measuring chamber(s). c) all electronic components for measuring values ​​from the sensors, for data analysis, for storing results and for communication are arranged in the sensor housing and are provided with a power supply. d) in a preferred embodiment, all functions for measuring values ​​from the sensors, for data analysis, for result storage and for data communication are implemented via an application-specifically programmed microcontroller with additional hardware components. e) each measuring chamber is directly connected to the ambient air. f) one or more sensors for measuring temperature, humidity and VOC profile are arranged in each measuring chamber and each sensor is connected by wire to the measured value acquisition in the sensor housing. g) a measuring chamber design for detecting stored product pests (for example, but not exclusively, grain beetles, bread beetles, flat-leaved grain beetles, rice beetles, etc.) preferably, but not exclusively, consists of a tube section for piercing a bulk material, which is connected to the gas phase in the bulk material via one or more openings. In a special embodiment, this sensor can also be permanently installed in the storage or storage container for the bulk material. h) a special embodiment consists of several measuring chambers arranged one above the other according to claim g), which are placed in the bulk material and thus enable the measurement of a VOC and moisture profile in the bulk material. i) any mechanically suitable material can be used for the construction of the measuring chamber housing, which does not cause any relevant temperature measurement errors due to the introduction of the ambient outside temperature and which does not cause any relevant humidity measurement errors due to material-related hygroscopy. k) the moisture content in the bulk material is used as an additional indicator of pest infestation (“metabolic water”). I) at selected measuring locations, the stored product and hygiene pests are identified using the VOC profile of their species-specific, metabolic-related body odour and / or the odour of their excretions. m) the presence of stored product and hygiene pests (for example and not exclusively: rats, mice, cockroaches, etc.) is identified from the measured VOC profile at the measuring point using algorithms, as the detection algorithms have been determined in advance in a training phase using an AI. n) When measuring the VOC profile of a pest species, the use of complex VOC sensor arrays is avoided by using a VOC sensor with a parameterizable heater. The temperature in the sensor housing is gradually increased, and a VOC measurement is performed after each temperature level is reached. o) Higher-resolution VOC profiles can be achieved, for example, by only reaching the maximum temperature in the sensor housing after several consecutive measurements. In this case, a smaller temperature range is measured during each individual measurement. VOC profiles generated in this way consist of as many individual values ​​as the specified temperature levels, multiplied by the number of measurement sequences until the maximum temperature is reached. p) for the simultaneous detection of different types of stored product and hygiene pests, several differently trained and configured VOC sensors are installed in one sensor housing. q) the sensor electronics also transfer the results to non-volatile storage media and / or via wired and / or wireless data communication to a central evaluation unit. r) the central evaluation unit communicates with the sensor electronics and, for example and not exclusively, carries out the archiving of raw data, the evaluation of current measured values, the analysis of time series and trends, the detection of limit values ​​and other communication functions such as SMS and / or e-mail. [2] Detection system for identifying stored product and hygiene pests according to claim 1.m) and claim 1.n) characterized by The BME680 or BME688, a multi-sensor IC from Bosch Sensortec, is used as a VOC sensor with configurable heating. The BME680 and BME688 each provide a temperature and humidity sensor, as well as a configurable VOC sensor with AI support, all in one housing. [3] Detection system for identifying stored product and hygiene pests according to claim 1.m) and claim 1.n) characterized bythat the algorithms for identifying the VOC profile of stored product and hygiene pests are determined via AI training with the software provided by Bosch Sensortec for the BME680, BME688 or another publicly available AI. [4] Detection system for identifying stored product and hygiene pests according to claim 1.m), claim 1.n), claim 2.) and claim 3.) characterized by that the sensor, evaluation and communication electronics consist of at least one application-specific programmed microcontroller in conjunction with additional hardware. [5] Detection system for identifying stored product and hygiene pests according to claim 4. characterized by that a power supply is located in the sensor housing. [6] Detection system for identifying stored product and hygiene pests according to claim 1 to claim 5 characterized bythat the wireless communication between the sensor(s) and the central evaluation unit is realized via a wireless "low-power" technology (LoRa, LoRaWAN, NB-IoT, LTE-M, Sigfox), if necessary in conjunction with an appropriate network infrastructure. [7] Recognition system according to claim 1.a) and claim 1.f) and / or claim 1.g) characterized by that a mobile (handheld) identification of the presence of stored product and hygiene pests is carried out by measurements at preferred locations and walking routes via their body odor and / or via species-specific odor markings and / or via the odor of their excretions.

Images