Method for monitoring the state of a machine, system, and computer program

Abstract

The invention relates to a method for monitoring the state of a machine, wherein the machine comprises a sensor system, which comprises a vibration and / or temperature sensor for measuring a vibration and / or a temperature of the machine as well as a trigger sensor, and the sensor system can be operated in a learning mode and in a state-monitoring mode.

Description

[0001] Method for monitoring the condition of a machine, system, computer program

[0002] The invention relates to a method for determining a machine condition, in particular for monitoring the condition of a machine. The invention further relates to a corresponding system and computer program.

[0003] Vibration-based or temperature-based condition monitoring typically uses sensors with cables, which generally results in quite expensive systems and methods. One of the main reasons for this is that the manual installation of such sensors, including the installation and protection of the cables, is time-consuming. More cost-effective condition monitoring can be achieved through the use of wireless vibration- or temperature-based condition monitoring systems. Such systems, especially the sensors, are usually battery-powered. The typical battery life of wireless battery-powered vibration sensors, for example, is approximately 5 years.

[0004] In such conventional systems, battery-powered wireless vibration sensors measure data on a schedule. For example, KPIs (Key Performance Indicators), such as ISO vibration severity (10-1000 Hz RMS), are measured every 4 hours, and raw data is measured every 24 hours. However, such measurements, performed on a fixed schedule, can result in inadequate data quality and detrimental condition monitoring for a wide variety of machines and applications, such as machines that do not operate continuously or machines with varying operating states, modes, or conditions, because the measurement data is often obtained when the machine is not running or is in an operating mode unsuitable for condition monitoring.

[0005] Against this background, DE 102021 129 363 B3 discloses a method that provides a condition monitoring mode in which a measurement is only performed if a machine signal obtained from a separate trigger sensor has previously exceeded a predefined trigger threshold. The trigger threshold is determined during a learning phase of the sensor system, in which learning data is measured and compared with a predefined standard trigger threshold. Depending on this predefined standard trigger threshold, an optimized trigger threshold for the sensor system is determined during the learning phase. However, it proved difficult and laborious to set the predefined standard trigger threshold to a value that ensures a properly functioning sensor system in condition monitoring mode.

[0006] US 2014 / 0 055 274 A1 describes a system for use in detecting a change in a mechanical system during its operation. The system includes a processor and a control system coupled to it.

[0007] Against this background, the object of the invention is to provide improved condition monitoring of a machine, in particular such that improved condition monitoring can be achieved for non-continuously operating machines or machines with different operating states or modes or conditions.

[0008] To achieve the object, the invention proposes a method for determining a machine condition, in particular for monitoring the condition of a machine, wherein the machine comprises a sensor system, wherein the sensor system comprises:

[0009] - a vibration and / or temperature sensor for measuring a machine vibration and / or a machine temperature, and

[0010] - a trigger sensor; wherein the sensor system is operable in a learning mode and in a condition monitoring mode, the method comprising the following steps:

[0011] — a) operating the sensor system in learning mode to determine a trigger threshold for the trigger sensor before operating the sensor system in condition monitoring mode, wherein the operation of the sensor system in learning mode comprises the following steps:

[0012] - in a learning measurement step, the trigger sensor is periodically activated according to a predefined time interval to measure a machine vibration and / or temperature sample of the machine each time and to store the machine vibration and / or temperature sample in a storage unit of the sensor system as learning data,

[0013] - in a trigger threshold calculation step, the learning data obtained in the learning mode are used to determine the trigger threshold to be used in the condition monitoring mode of the sensor system, in particular in step b);

[0014] — b) operating the sensor system in condition monitoring mode, wherein the operation of the sensor in condition monitoring mode comprises the following steps: -- in a trigger step, the trigger sensor measures a machine signal of the machine, wherein a comparison is carried out between the machine signal and the trigger threshold, wherein, depending on the comparison between the machine signal and the trigger threshold, the vibration and / or temperature sensor is transferred to a measuring state for carrying out a measuring step,

[0015] -- In the measuring step, the vibration and / or temperature sensor measures a machine vibration and / or temperature signal of the machine to determine the condition of the machine.

[0016] According to the invention, it is possible for the measuring step to be carried out in the condition monitoring mode of the sensor system as a function of the machine signal acquired by the trigger sensor and a previously determined trigger threshold. The trigger threshold is determined during a learning phase of the sensor system carried out in step a). In this case, it is possible for the trigger threshold to be adapted to the respective sensor system and the respective machine as well as to the environmental influences and noises to which the sensor system is exposed during its application. By using the trigger threshold, the data quality of the machine vibration and / or temperature signal for condition monitoring can be improved, since the amount of data measured at times when the machine is not operating or is not operating in a state useful for condition monitoring can be advantageously reduced.This makes it possible to collect data from valid machine states while minimizing the collection of data points outside of these states. At the same time, the overall energy consumption for condition monitoring can be reduced, thus extending the battery life of the sensors or ensuring that the energy harvesting power is sufficient to operate the sensors. In general, the present invention makes it possible to minimize the processing and transmission of unnecessary data, thereby reducing costs and saving energy.

[0017] It is an advantage of the present invention that the improved condition monitoring that can be achieved by the learning mode according to the invention can be implemented in an automated manner that requires neither expert knowledge nor user input. This is achieved by the learning measurement step according to the invention, in which the vibration and / or temperature sensor is periodically activated after a predefined time interval to measure a machine vibration and / or temperature signal of the machine. These measured values ​​recorded at periodic times are stored in a memory unit of the sensor system as learning data and then used in the trigger threshold calculation step to derive a trigger threshold suitable for the sensor system. This is particularly advantageous for sensor systems that are produced in high volumes.Even when identical sensor systems are used for multiple identical machines, the individual operating conditions and environmental influences typically vary for each individual machine and sensor system. With the help of the present invention, an individually learned trigger threshold can be automatically determined for each individual sensor system and each individual machine, thereby improving the quality of condition monitoring and avoiding expert and user input.

[0018] According to the invention, the learning data periodically measured by the vibration and / or temperature sensor consists of a plurality of individual machine vibration and / or temperature samples, wherein the samples are taken at temporally separate points in time. By taking only individual samples at periodic points in time, the amount of acquired learning data can be reduced compared to the method known from DE 102021 129 363 B3. Consequently, it is possible to store the learning data locally in a memory of the sensor system, instead of transmitting the learning data to a system outside the sensor system, e.g., to a backend system, and storing it there.

[0019] According to a preferred embodiment of the present invention, the vibration and / or temperature sensor does not measure any machine vibration and / or temperature signal between the periodic activations during the learning measurement step. The vibration and / or temperature sensor can be deactivated between the periodic measurements, so that the power consumption of the vibration and / or temperature sensor can be reduced to a minimum.

[0020] According to a preferred embodiment of the present invention, the learning measurement step is carried out for a predetermined duration, wherein after the predetermined duration has elapsed, the learning data is analyzed, and depending on the analysis of the learning data, the learning measurement step is extended by a predetermined extension time. This approach is based on the realization that a certain number of samples must be measured in the learning measurement step in order to obtain usable learning data for determining the trigger threshold. By analyzing the learning data after the predetermined duration has elapsed, it can be checked whether the operating state of the machine is covered by the learning data. If the analysis shows that the learning data does not sufficiently cover the operating state of the machine, the learning measurement step can be extended by the predetermined extension time.The extension time allows more learning data to be collected and increases the probability that the learning data correctly covers the operating state of the machine.

[0021] According to a preferred embodiment of the present invention, the step of calculating the trigger threshold is performed by the sensor system, in particular by a computing device of the sensor system, preferably by a computing device of the vibration and / or temperature sensor. By performing the trigger threshold calculation step by a computing device of the sensor system, data transfer to external computing devices can be reduced. In particular, it is not necessary to first transmit the learning data to an external computing device, have the external computing device calculate the trigger threshold, and then transmit the calculated trigger threshold to the sensor system. Consequently, especially in installations with multiple sensor systems, network traffic can be reduced due to the calculation of the trigger threshold in learning mode.

[0022] According to the invention, in the step of calculating the trigger threshold, a metric is calculated that quantifies a variation in the learning data—wherein the metric particularly quantifies the variation in the learning data recorded while the machine was most likely in operating mode—and the trigger threshold is calculated as a function of the metric, preferably as a predetermined fraction of the metric. Calculating the metric that quantifies the variation in the learning data enables the derivation of the trigger threshold with little additional effort. The trigger threshold can be calculated as a function of the metric, e.g., as a predetermined fraction of the metric.

[0023] According to a preferred embodiment of the present invention, the learning data and / or the trigger threshold determined in the trigger threshold calculation step are sent to a backend system—in particular via a gateway unit. This optional embodiment has the advantage that the trigger threshold calculation performed locally by the computing means of the sensor system can be supplemented by computing means of the backend system. The backend system can be implemented, for example, using a cloud. Preferably, the determined trigger threshold is validated in the backend system. Validation in the backend system can be achieved by using the learning data obtained in the learning mode and by using:

[0024] - statistical methods, and / or

[0025] - machine learning, and / or

[0026] - artificial intelligence.

[0027] If the trigger threshold determined by the sensor system can be validated by the backend system, no adjustments are required, and the sensor system can operate in condition monitoring mode without further input from the backend system. On the other hand, if the trigger threshold calculated by the sensor system is not validated by the backend system, the backend system can calculate an appropriate trigger threshold based on the learning data provided to the backend system. The backend system may include more efficient computing means than the computing means of the sensor system, which can implement more precise algorithms and / or techniques for calculating the trigger threshold. Alternatively, if the trigger threshold calculated by the sensor system is not confirmed by the backend system, a user of the sensor system can be informed and prompted to manually set the trigger threshold for the sensor system.

[0028] According to the invention, it is possible for the state of the machine to be determined by means of or based on the machine vibration signal and / or temperature signal of the machine measured by the vibration and / or temperature sensor, in particular in the condition monitoring mode of the sensor system in step b). Advantageously, the machine signal of the trigger sensor can be used to detect whether the machine is in operation, in particular whether the machine is in operation, so that a state of the machine can be determined. In particular, in step b), by comparing the machine signal (of the trigger sensor) with the trigger threshold, it is determined whether the machine is in operation, in particular whether the machine is in an operating state, so that the state of the machine can be determined.Preferably, only when it is determined that the machine is in such an operating state, the vibration and / or temperature sensor is triggered in such a way that the vibration and / or temperature sensor is activated and / or transferred to a measuring state for measuring the vibration and / or temperature signal.

[0029] It is conceivable that the condition of the machine relates, for example, to whether the machine has a defect or is showing signs of wear. Such effects can be detected in the machine's vibration and / or temperature signal or by means of the machine's vibration and / or temperature signal.

[0030] Preferably, it is conceivable for the vibration and / or temperature sensor to comprise at least one MEMS (microelectromechanical system) for measuring a machine vibration signal. Preferably, it is conceivable for the trigger sensor to comprise at least one MEMS (microelectromechanical system) for measuring a machine vibration signal.

[0031] In the context of the present invention, the trigger threshold may be a trigger threshold value.

[0032] According to one embodiment, it is preferred that in step b) (i.e. in the condition monitoring mode of the sensor system), depending on the comparison between the machine signal measured by the trigger sensor and the trigger threshold, the vibration and / or temperature sensor is transferred from an off state or idle state to the measuring state, so that a measuring step is carried out using the vibration and / or temperature sensor. Preferably, the vibration and / or temperature sensor is deactivated in the off state or idle state of the vibration and / or temperature sensor, so that the vibration and / or temperature sensor does not measure the machine vibration and / or temperature signal of the machine. It is particularly preferred that in the off state or idle state of the vibration and / or temperature sensor, the energy consumption of the vibration and / or temperature sensor is lower than in the measuring state.

[0033] According to the present invention, it is preferred that the machine is a non-continuously operating machine and / or a machine with different operating states or modes or conditions. For example, it is conceivable that the machine comprises or is a:

[0034] - a construction machine, in particular a crane,

[0035] - a mining machine,

[0036] - an agricultural machine,

[0037] - a processing machine,

[0038] -- Logistics machine, in particular storage and retrieval machine,

[0039] - Transport machine, especially subway, railway or train

[0040] -- Pump or conveyor in a pulp mill.

[0041] Alternatively, it may also be a continuously operating machine. According to one embodiment of the present invention, the sensor system may be part of the machine or attached to the machine or a component of the machine. The exact position of the sensor system may depend on the type of machine.

[0042] Preferably, the vibration and / or temperature sensor and the trigger sensor are wireless sensors that comprise energy storage and / or energy generation devices. For example, it is conceivable that the vibration and / or temperature sensor and / or the trigger sensor comprise one or more batteries. It is conceivable that the trigger sensor is a component or subcomponent of the vibration and / or temperature sensor. It is conceivable that the trigger sensor is used and / or operated at least partially independently of the remaining vibration and / or temperature sensor. However, it is also possible for the trigger sensor and the vibration and / or temperature sensor to be separate sensors according to the present invention.

[0043] According to a preferred embodiment of the present invention,

[0044] - the trigger sensor is designed to measure a machine vibration of the machine, wherein preferably the machine signal measured by the trigger sensor in the trigger step and / or in the learning trigger step is a trigger vibration signal of the machine; and / or

[0045] - the trigger sensor is designed to measure an electrical signal from the machine, in particular an electrical current and / or an electrical voltage, wherein the machine signal measured by the trigger sensor in the trigger step and / or in the learning trigger step is preferably an electrical signal from the machine. It is particularly preferred that the trigger sensor is a vibration sensor, in particular a MEMS vibration sensor, which is designed to measure a machine vibration of the machine. It is further preferred that the trigger sensor has a lower sampling rate than the vibration and / or temperature sensor. For example, it is possible for the sampling rate of the trigger sensor to be below 100 Hz, preferably at or below 50 Hz, particularly preferably at or below 10 Hz. This allows energy consumption to be kept low.It is particularly preferred that the trigger sensor consumes less energy during operation than the vibration and / or temperature sensor. According to an alternative embodiment, it is conceivable, for example, that the trigger sensor is an electrical sensor designed to measure an electrical quantity, e.g., a current and / or a voltage, of the machine.

[0046] According to a preferred embodiment of the present invention, before the vibration and / or temperature sensor transitions to the measuring state and / or before the vibration and / or temperature sensor measures the machine vibration and / or temperature signal in the measuring step, a machine stabilization algorithm is carried out, in particular by means of computing means of the trigger sensor and / or computing means of a gateway unit and / or computing means of the machine, wherein the machine stabilization algorithm comprises:

[0047] - a waiting time is waited before the vibration and / or temperature sensor enters the measuring state and / or before the vibration and / or temperature sensor measures the machine vibration and / or temperature signal in the measuring step, and / or

[0048] - a preliminary analysis is carried out to verify or check that the machine is in operation and / or to verify or check that the machine is in an operating state, wherein preferably the vibration and / or temperature sensor only switches to the measuring state and / or begins measuring the machine vibration and / or temperature signal in the measuring step after the waiting time has elapsed and / or only if the preliminary analysis carried out shows that the machine is in operation and / or in an operating state. The machine stabilization algorithm is preferably carried out in step b) of a method according to an embodiment of the present invention, ie when the sensor system is operated in condition monitoring mode. For example, it is possible that the machine signal of the trigger sensor is analyzed as part of the preliminary analysis in order to check whether the machine is in operation.The measurement step is performed when the machine is in operation. For example, as part of the preliminary analysis, the machine's computing resources can be used to check whether the machine is in operation. When the machine is in operation, the measurement step is performed.

[0049] According to a preferred embodiment of the present invention, the learning data obtained in the learning mode are additionally used to determine one, some or all of the following parameters to be used in the condition monitoring mode, in particular in step b): - a duration for measuring the machine signal of the machine in the trigger step,

[0050] - a sampling rate for measuring the machine signal in the trigger step,

[0051] - a repetition rate or interval or time of day or time of week for performing the trigger step, in particular for measuring the machine signal of the machine in the trigger step,

[0052] - one or more parameters of the machine stabilization algorithm, in particular a waiting time. The parameter(s) are preferably determined using the learning data obtained in the learning mode in step a) and using: - statistical methods, and / or

[0053] - machine learning, and / or

[0054] - Artificial intelligence. Not only the trigger threshold, but also other parameters for the operation of the sensor system in condition monitoring mode can be determined or calculated based on the learning data obtained in learning mode.

[0055] According to one embodiment of the present invention, it is advantageously possible for the sensor system and / or the machine - preferably before and / or during operation of the sensor system in learning mode - to be provided with predefined values ​​for one, some or all of these parameters:

[0056] - a duration for measuring the machine signal of the machine in the learning trigger step,

[0057] - a sampling rate for measuring the machine signal of the machine in the learning trigger step,

[0058] - a repetition rate or interval or time of day or time of week for performing the learning trigger step, in particular for measuring the machine signal of the machine in the learning trigger step,

[0059] - one or more parameters of the machine stabilisation algorithm, in particular a duration of the waiting period.

[0060] The predefined values ​​of the mentioned parameters can be used in learning mode, especially in step a).

[0061] According to a preferred embodiment of the present invention, during and / or after the measurement of the machine vibration and / or temperature signal by means of the vibration and / or temperature sensor in the measuring step, an acceptance test is carried out by means of an acceptance criterion for the measured machine vibration and / or temperature signal, wherein the measured machine vibration and / or temperature signal is used to determine the machine condition only if the acceptance criterion is met.

[0062] According to a preferred embodiment of the present invention, it is conceivable that the acceptance test is carried out by computing means of the vibration and / or temperature sensor, wherein the measured machine vibration and / or temperature signal is only transmitted from the vibration and / or temperature sensor and / or the computing means of the vibration and / or temperature sensor to a gateway instance if the acceptance criterion is met. In particular, it is possible for the vibration and / or temperature sensor and / or the computing means of the vibration and / or temperature sensor not to transmit the measured machine vibration and / or temperature signal if the acceptance criterion is not met.This allows minimizing the energy consumption of the communication means used to transmit the measured machine vibration and / or temperature signal from the vibration and / or temperature sensor to the gateway instance, as it avoids transmitting unusable data or data of insufficient quality.

[0063] According to a preferred embodiment of the present invention, the acceptance criterion comprises an acceptance threshold, wherein the acceptance test comprises monitoring whether the measured machine vibration and / or temperature signal, in particular during the measurement of the machine vibration and / or temperature signal, falls below the acceptance threshold for a predefined number of consecutive samples of the measured machine vibration and / or temperature signal, wherein preferably the acceptance criterion is not met if the measured machine vibration and / or temperature signal falls below the acceptance threshold for the predetermined number of consecutive samples, wherein preferably the acceptance criterion is met,if the measured machine vibration and / or temperature signal does not fall below the acceptance threshold during the duration of a measurement time interval for the specified number of consecutive samples. It is conceivable that the measurement time interval is a predefined measurement time interval. The measurement time interval can, for example, be chosen or selected by an operator and / or can depend on the machine and / or the operating conditions of the machine. For example, the measurement time interval can be in the order of seconds, e.g., 3 seconds or 5 seconds. Alternatively, it is conceivable that the duration of the measurement time interval is determined using the learning data obtained in the learning mode of the sensor system in step a). It is particularly preferredthat when the acceptance threshold of the measured machine vibration and / or temperature signal is undershot, the number of consecutive samples (e.g., the number of consecutive individual measured values) of the machine vibration and / or temperature signal is counted as long as the measured machine vibration and / or temperature signal remains below the acceptance threshold and / or does not rise above the acceptance threshold. It is conceivable that this count of consecutive samples of the measured machine vibration and / or temperature signal is canceled and / or reset when the measured machine vibration and / or temperature signal rises above the acceptance threshold. Preferably, when the measured machine vibration and / or temperature signal falls below the acceptance threshold again,The count of consecutive samples is restarted from zero. If the number of consecutive samples reaches the predefined number of consecutive samples, the acceptance criterion is preferably not met. Preferably, the acceptance criterion is met if the number of consecutive samples does not reach the predefined number of consecutive samples.

[0064] According to a preferred embodiment of the present invention, it is conceivable that the measurement of the machine vibration and / or temperature signal by means of the vibration and / or temperature sensor is terminated in the measuring step:

[0065] -- in response to the receipt of a further trigger signal by the vibration and / or temperature sensor, in particular from the further sensor and / or the computing device, wherein the further trigger signal indicates that the machine is not in the operating state and / or has transitioned from the operating state and / or that a stable operating state of the machine has ended; and / or

[0066] - in response to the vibration and / or temperature sensor receiving a further trigger signal from the trigger sensor, the further trigger signal indicating that the machine is not in the operating state and / or has transitioned from the operating state and / or that a stable operating state of the machine has ended; and / or

[0067] - when a measurement time interval has elapsed; and / or

[0068] - when a predefined total number of samples of the machine vibration and / or temperature signal have been obtained.

[0069] To achieve the above-mentioned object, the invention further proposes a system for determining a machine condition, in particular for monitoring the condition of a machine, wherein the system comprises a sensor system, wherein the system optionally comprises the machine, wherein the sensor system can be attached to the machine or is contained in the machine, wherein the sensor system comprises: - a vibration and / or temperature sensor for measuring a machine vibration and / or a temperature of the machine, and

[0070] - a trigger sensor; wherein the sensor system is operable in a learning mode and in a condition monitoring mode, wherein the system is configured to perform a method according to an embodiment of the present invention.

[0071] Preferably, the system according to the invention is a computer-implemented system comprising means for carrying out a method according to an embodiment of the present invention.

[0072] To solve the above-mentioned problem, the invention further proposes a computer program containing instructions which, when the computer program is executed by one or more computer means - in particular by a sensor system and / or by a machine and / or by a gateway unit and / or by a backend system - cause the one or more computer means to carry out a method according to an embodiment of the present invention.

[0073] For the system according to the invention and the computer program according to the invention, the same embodiments, advantages and technical effects can be achieved that were described in connection with the method according to the invention or in connection with an embodiment of the method according to the invention.

[0074] These and other features, characteristics, and advantages of the present invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings, which illustrate, by way of example, the principles of the invention. The description is provided for illustrative purposes only and is not intended to limit the scope of the invention. Reference numerals cited below refer to the accompanying drawings.

[0075] Fig. 1 shows an example of the operation of a non-continuous machine;

[0076] Fig. 2 shows a system according to an embodiment of the present invention;

[0077] Fig. 3 shows a schematic representation of the operation of a sensor system in a learning mode for determining a trigger threshold according to an embodiment of the present invention;

[0078] Fig. 4 shows a schematic representation of a method according to an embodiment of the present invention; Fig. 5 shows a vibration v over time t of a machine to illustrate an embodiment of a method according to the present invention;

[0079] Fig. 6A shows a method according to an embodiment of the present invention in which the measured data fails an acceptance test;

[0080] Fig. 6B shows a method according to an embodiment of the present invention in which the measured data passes an acceptance test.

[0081] Fig. 1 shows an example of the operation of a discontinuous machine to illustrate disadvantages of the prior art. In the time intervals 910, 911, 912, the machine operates in such a way that condition monitoring can be carried out. Outside of these time intervals, the machine is switched off or is in an operating state that is not suitable for condition monitoring. According to the prior art, measurements for condition monitoring would be carried out at preconfigured points 900, 901, 902, 903, 904, 905, e.g., at regular time intervals. However, the measurements at points 900, 901, 902, 903, 904 would be unsuitable for condition monitoring because they are carried out outside the time intervals 910, 911, 912, so that the overall result and quality of the condition monitoring would be poor. Such disadvantages can be overcome by the present invention.

[0082] Fig. 2 shows a system according to an embodiment of the present invention. A vibration and / or temperature sensor 10 and a trigger sensor 20 are attached to or contained in a machine 1. Preferably, the vibration and / or temperature sensor 10 is at least one vibration sensor 10. In an alternative embodiment, however, the vibration and / or temperature sensor 10 is a temperature sensor. Preferably, the machine 1 is a non-continuously operating machine 1, e.g. a processing machine, an agricultural machine, a logistics machine, in particular a storage and retrieval machine, a transport machine, in particular a subway or a railway or a train or the like. The vibration sensor 10 and / or the trigger sensor 20 are preferably wireless devices that are used to monitor the condition of the machine 1.The vibration sensor 10 and the trigger sensor 20 comprise an energy storage and / or energy harvesting device. According to one embodiment of the present invention, the trigger sensor 20 can be part of the vibration sensor 10. Preferably, the trigger sensor 20 and the vibration sensor 10 comprise MEMS for measuring vibrations of the machine 1. A gateway unit 50 communicates with the trigger sensor 20 and / or the vibration sensor 10. It is conceivable that the gateway unit 50 is part of the computing device of the machine 1 or that the gateway unit is a gateway unit 50 separate from the machine 1. The sensors 10, 20 and / or the gateway entity 50 can form a mesh network or be part of a mesh network. It is conceivable that the gateway entity 50 is connected to a backend system and / or a network 60, in particular a telecommunications network. Furthermore, it is possible for the system to use a user device 70, e.g.a computer or a mobile device that displays data and / or the status of the machine. The respective devices of the system, in particular the vibration sensor 10, the trigger sensor 20 and / or the gateway unit 50, can have wireless communication means for communicating with each other and / or with the backend system and / or network 60. In particular, the sensor system comprising the vibration sensor 10 and the trigger sensor 20 can be operated in a learning mode for determining a trigger threshold for the trigger sensor 20 and in a condition monitoring mode in which the learned / determined trigger threshold is used. By means of the trigger sensor 20 and the vibration sensor 10, steps a) and b) of a method according to the present invention can be carried out. Embodiments thereof are described, for example, in connection with Figures 3, 4, 5, 6A and 6B.

[0083] Fig. 3 schematically illustrates an embodiment of the operation of the sensor system in learning mode for determining a trigger threshold for the trigger sensor 20 in step a) according to an embodiment of the present invention. First, a learning measurement step S21 is performed, in which the vibration and / or temperature sensor 10 is periodically activated after a predefined time interval to measure a machine vibration and / or temperature sample of the machine 1 and to store the machine vibration and / or temperature sample in a memory unit of the sensor system as learning data. In the learning measurement step S21, the vibration and / or temperature sensor 10 does not measure any machine vibration and / or temperature signal of the machine 1 between the periodic activations.The learning measurement step S21 is carried out for a predetermined duration, wherein after the expiration of the predetermined duration, the learning data is evaluated in an analysis step S22 and wherein, depending on the evaluation of the learning data, the learning measurement step is extended by a predetermined extension time. In particular, the learning measurement step is extended by the predetermined extension time if the analysis shows that the learning data do not sufficiently cover the operating state of the machine. After completion of the learning measurement step S21, either with or without the aforementioned extension time, and before operation of the sensor system in condition monitoring mode, the sensor system is transferred to a trigger threshold calculation step S23. In step S23 for calculating the trigger threshold, the learning data obtained in the learning mode is used to determine the trigger threshold to be used in the condition monitoring mode of the sensor system.The trigger threshold calculation step S23 is performed by the sensor system, in particular by a computing device of the sensor system, preferably by a computing device of the vibration and / or temperature sensor 10. In the trigger threshold calculation step S23, a metric is calculated that quantifies a variation of the learning data—wherein the metric particularly quantifies the variation of the learning data recorded while the machine was most likely in operating mode—and the trigger threshold is calculated as a function of the metric, preferably as a predetermined fraction of the metric.

[0084] Optionally, at the end of the trigger threshold calculation step S23, the learning data and / or the trigger threshold determined in the trigger threshold calculation step S23 are sent—in particular via a gateway unit 50—to the backend system 60. The determined trigger threshold can then be validated in the backend system 60.

[0085] Fig. 4 schematically illustrates a method according to an embodiment of the present invention. In step S31, the sensor system is operated in learning mode to determine a trigger threshold for the trigger sensor 20. Step S31 may, for example, comprise steps S21, S22, and S23, as illustrated in Fig. 3. Once the trigger threshold for the trigger sensor 20 has been determined, the sensor system is transferred to condition monitoring mode, and step S32 is performed.Step S32 comprises the operation of the sensor system in the condition monitoring mode, wherein the operation of the sensor in the condition monitoring mode in step S32 comprises the following steps: In a trigger step, the trigger sensor 20 measures a machine signal of the machine 1, wherein a comparison is carried out between the machine signal and the trigger threshold, wherein, depending on the comparison between the machine signal and the trigger threshold, the vibration and / or temperature sensor 10 is transferred to a measuring state for carrying out a measuring step.

[0086] In the measuring step, the vibration and / or temperature sensor 10 measures a machine vibration and / or temperature signal of the machine 1 to determine the condition of the machine. In step S33, an acceptance test 303 is performed for the measured machine vibration and / or temperature signal using an acceptance criterion, wherein the measured machine vibration and / or temperature signal is only used to determine the condition of the machine 1 if the acceptance criterion is met. An embodiment of the acceptance test 303 is explained in more detail with reference to Figures 5, 6a, and 6b.

[0087] In Fig. 5, a vibration v of a machine 1 is shown over time t to illustrate an embodiment of a method according to the invention. In particular, an embodiment of step b) of a method according to the invention is shown in which the sensor system is operated in a condition monitoring mode. When the trigger sensor 20 is in a trigger state, the trigger sensor 20 measures a machine signal of the machine 1, in particular a machine vibration. In the trigger state, the trigger sensor 20 measures the machine vibration at a comparatively low sampling rate, for example in the order of magnitude of 10 Hz, in order to ensure low energy consumption. In the trigger step, a comparison is made between the measured machine signal and a trigger threshold so that an increase in the machine signal to and / or above the trigger threshold can be detected.The trigger threshold is a trigger threshold determined using the learning mode of the sensor system in step a). At a time 300, the trigger sensor 20 detects a vibration (i.e., a machine signal) that exceeds the trigger threshold. In response to this detection, the vibration sensor 10 is transitioned from a rest or off state to a measuring state, e.g., by a command from the trigger sensor 20 or the gateway unit 50. It is possible that—before the vibration sensor 10 begins measuring the machine vibration signal of the machine 1 to determine the machine state—a machine stabilization algorithm 302 is executed. The machine stabilization algorithm 302 may include waiting for a waiting time 301 before the vibration sensor 10 transitions to the measuring state and / or before the measurement of the machine vibration signal by means of the vibration sensor 10 begins.Alternatively or additionally, the machine stabilization algorithm 302 may include a pre-analysis to check or verify that the machine 1 is operating. For example, it is possible for the pre-analysis to include an analysis of the machine signal measured by the trigger sensor 20 to check or verify that the machine 1 is operating. If the machine stabilization algorithm 302 has been successfully run, the measurement of the machine vibration signal is performed using the vibration sensor 10 in the measuring step, in particular for a measuring time interval 200. In the measuring step, the vibration sensor 10 measures individual samples and / or measured values, which together form the machine vibration signal. During the measuring step and / or at the end of the measuring step, an acceptance test 303 may be performed for the measured machine vibration signal.Acceptance test 303 can ensure that the acquired data has sufficient or desired quality. Preferably, the measured data, i.e., the measured machine vibration signal, is only transmitted from vibration sensor 10 to gateway unit 50 or to backend system and / or network 60 if the acceptance test is successfully passed. This allows for a particularly energy-efficient method to be achieved. Preferably, the measured machine vibration signal is only used to determine the machine condition if acceptance test 303 is passed.

[0088] An embodiment of the acceptance test 303 is explained in more detail with reference to Figures 6A and 6B. The acceptance test 303 comprises an acceptance criterion, in particular an acceptance threshold 100. It is possible for the acceptance threshold 100 to be the same value as the trigger threshold or to correspond to the trigger threshold. During the acceptance test, it is monitored whether the measured machine vibration signal, in particular during the measurement of the machine vibration signal, falls below the decrease threshold 100 for a predetermined number of consecutive samples (in particular sample values) of the measured machine vibration signal. If the measured machine vibration signal falls below the acceptance threshold 100, the number of consecutive samples (e.g., the number of consecutive individual measured values) of the vibration signal is counted as long as the measured machine vibration signal remains below the acceptance threshold 100.It is conceivable that this count of the consecutive samples of the measured machine vibration signal is canceled and / or reset if the measured machine vibration signal again rises above the acceptance threshold 100. Preferably, when the measured machine vibration signal again falls below the acceptance threshold 100, the count of the consecutive samples is started again at zero.

[0089] If the number of consecutive samples reaches the predefined number of consecutive samples, the acceptance criterion is not met and the acceptance test is failed. This situation is shown in Fig. 6A. However, if the number of consecutive samples does not reach the predefined number of consecutive samples, the acceptance criterion is met and the

[0090] Acceptance test passed. This situation is shown in Fig. 6B. In this case, the measured vibration signal of the machine is further used to determine the condition of machine 1.

[0091] Reference symbol

[0092] 1 machine

[0093] 10 Vibration and / or temperature sensor

[0094] 20 TriggerT rigger sensor

[0095] 50 Gateway Unit

[0096] 60 backend system and / or network

[0097] 70 user devices

[0098] 100 Acceptance threshold

[0099] 200 measurement time interval

[0100] 300 points

[0101] 301 Waiting time

[0102] 302 Machine stabilization algorithm

[0103] 303 Acceptance testAcceptance test

[0104] 900 - 905 points

[0105] 910 - 912 time intervals

[0106] S21 - S23 steps

[0107] S31 - S33 Steps t Time v Vibration x Measurements

Claims

Patent claims 1 . A method for monitoring the condition of a machine (1), wherein the machine (1) has a sensor system, wherein the sensor system comprises: - a vibration and / or temperature sensor (10) for measuring a machine vibration and / or a temperature of the machine (1), and - a trigger sensor (20); wherein the sensor system is operable in a learning mode and in a condition monitoring mode, the method comprising the following steps: a) operating the sensor system in the learning mode to determine a trigger threshold for the trigger sensor (20) before operating the sensor system in the condition monitoring mode, wherein operating the sensor system in the learning mode comprises the following steps: - in a learning measurement step, the trigger sensor (20) is activated periodically according to a predefined time interval to measure a machine vibration and / or temperature sample of the machine (1) each time and to store the machine vibration and / or temperature sample in a storage unit of the sensor system as learning data, - in a trigger threshold calculation step, namely before the sensor system is operated in condition monitoring mode, the learning data obtained in learning mode are used to determine the trigger threshold to be used in the condition monitoring mode of the sensor system, namely in step b), wherein a metric is calculated that quantifies a scatter of the learning data recorded while the machine was most likely in operation mode, and the trigger threshold is calculated as a function of the metric as a predetermined fraction of the metric;b) operating the sensor system in condition monitoring mode, wherein operating the sensor in condition monitoring mode comprises the following steps: in a trigger step, the trigger sensor (20) measures a machine signal of the machine (1), wherein a comparison is carried out between the machine signal and the trigger threshold, wherein depending on the comparison between the machine signal and the; Trigger threshold of the vibration and / or temperature sensor (10) is transferred into a measuring state for carrying out a measuring step, in the measuring step the vibration and / or temperature sensor (10) measures a machine vibration and / or temperature signal of the machine (1) for determining the state of the machine (1).

2. The method according to claim 1, wherein the vibration and / or temperature sensor (10) does not measure any machine vibration and / or temperature signal of the machine (1) in the learning measurement step between the periodic activations.

3. Method according to one of the preceding claims, wherein the learning measurement step is carried out for a predetermined duration, wherein after expiry of the predetermined duration the learning data are analyzed and wherein, depending on the analysis of the learning data, the learning measurement step is extended by a predetermined extension time.

4. Method according to one of the preceding claims, wherein the step of calculating the trigger threshold is carried out by the sensor system, in particular by a computing device of the sensor system, preferably by a computing device of the vibration and / or temperature sensor (10). Trigger thresholdTrigger threshold5. Method according to one of the preceding claims, wherein the learning data and / or the trigger threshold determined in the trigger threshold calculation step are sent to a backend system (60) - in particular via a gateway instance (50).

6. The method according to claim 5, wherein the determined trigger threshold is validated in the backend system (60) or wherein the determined trigger threshold is not validated by the backend system (60), wherein the backend system (60) calculates a suitable trigger threshold based on the learning data provided to the backend system (60).

7. Method according to one of the preceding claims, wherein the trigger sensor (20) is designed to measure a machine vibration of the machine (1), wherein preferably the machine signal measured by the trigger sensor (20) in the trigger step and / or in the learning trigger step is a Trigger vibration signal of the machine (1); and / or wherein the trigger sensor is designed to measure an electrical signal of the machine (1), in particular an electrical current and / or a voltage, wherein preferably the machine signal measured by the trigger sensor (20) in the trigger step and / or in the learning trigger step is an electrical signal of the machine (1).

8. A system for monitoring the condition of a machine (1), the system comprising a sensor system, the system optionally comprising the machine (1), the sensor system being attachable to the machine (1) or being contained in the machine (1), the sensor system comprising: - a vibration and / or temperature sensor (10) for measuring a machine vibration and / or a temperature of the machine (1), and - a trigger sensor (20); wherein the sensor system is operable in a learning mode and in a condition monitoring mode, the system being configured to carry out a method according to any one of the preceding claims.

9. A computer program comprising instructions which, when the computer program is executed by one or more computing means, namely by a sensor system and / or by a machine and / or by a gateway unit and / or by a backend system, cause the one or more computing means to execute a method according to one of claims 1 to 7.

Images