DE102022112188A1 - Arrangement and procedure for determining the status of a system - Google Patents

Abstract

An arrangement (1) for determining the state of a system (2) in which at least one physical quantity (p) changes over a period of time (T) is disclosed. The arrangement (1) comprises at least one sensor (8), a memory (10) and an evaluation unit (9). The at least one sensor (8) is designed to detect the at least one physical quantity/one of the at least one physical quantity (p) and to generate a plurality of measured values within the time period (T) based on a detected physical quantity (p). . Characteristics are stored in the memory (10), the characteristics being representative of at least one state of the system (2). The evaluation unit (9) is communicatively connected to the at least one sensor (8) and to the memory (10) and is designed to analyze at least part of the large number of measured values and/or their time course, using at least one of the characteristics To determine the deviation measure and to estimate a state of the system (2) based on the deviation measure. A corresponding method is also disclosed.

Description

The invention relates to an arrangement and a method for determining the state of a system in which at least one physical quantity that correlates with a state of the system changes over a period of time.

In practice, many systems are in use in which at least one physical quantity changes over a period of time. This change can be triggered from outside the system, for example by wind, accelerations, changes in air pressure, changes in temperature or the like. However, these changes can also be caused within the system, for example by a pump, a motor or other actuators. Other components of the system can also have an impact on at least one physical quantity, for example a source, a sink or pipes.

As an example, consider a pump system in which a fluid is moved from a source to a sink by means of a pump. The pump influences the physical quantity “pressure”. In the direction of movement of the fluid in front of the pump, the pump generates a negative pressure that sucks in fluid to be moved from the source. In the direction of movement after the pump, an excess pressure is created, which pushes the fluid towards the sink. Usually the pressure is not constant, but pulsates periodically, for example depending on the technology and operation of the pump. The pressure can, for example, follow a sine curve or a sine-like curve that pulsates around a mean value.

However, the negative or positive pressure generated does not only depend on the performance of the pump. What is also important is how the source can provide the fluid and how the sink can “take” the fluid. If the source does not provide sufficient fluid, the negative pressure can increase. For a sink that retrieves a lot of fluid, the overpressure will be lower than for a sink that retrieves little fluid. Temperature can also have an influence on pressure. The viscosity of the fluid can depend on the temperature and influence the pressure. If contaminants are present in the fluid, the pressure may also change depending on the contaminants.

The same applies to gases that are dissolved in the fluid, since gases, in contrast to the fluid, are compressible. Further influences on the physical size can be caused by disturbances within the pump system, such as a pump that is not working ideally, worn components of the pump, existing or developing restrictions in the pump's supply or discharge lines, leaks and much more. This short and non-exhaustive description shows that a physical quantity can depend on many factors. Some of these factors may be non-critical for the operation of the pump system, others may affect operational safety or at least indicate non-ideal operating conditions. In this respect, both with such pump systems and with other systems, it is desirable to know or be able to estimate the state of the system.

One possible approach to determining the status is to obtain as detailed a picture as possible of the system and its components. Given the many possible influencing factors, this also means that a large number of sensors would have to be used that record all conceivable influencing factors. However, the many sensors cause considerable additional costs, including through their procurement, installation and supply of energy. This also creates a complex arrangement that is difficult to maintain. In addition, such arrangements are usually relatively rigid and can only be adapted to changing installation situations with great effort.

The present disclosure is based on the object of providing an arrangement and a method with which it is possible to determine the status of a system with at least one physical quantity that changes over a period of time using simple means and at the lowest possible cost. It is advantageous if the arrangement or the method can be flexibly adapted to a wide variety of systems.

This task is solved by the combinations of features of the independent claims. Further refinements are disclosed in the respective subclaims.

It should be noted that the features listed individually in the claims can be combined with one another in any technically sensible manner (including across category boundaries, for example between method and device) and show further refinements of the invention. The description additionally characterizes and specifies the invention, particularly in connection with the figures.

It should also be noted that a conjunction “and/or” used here between two features and linking them together must always be interpreted in such a way that in a first embodiment of the subject matter according to the invention only the first feature can be present, in in a second embodiment only the second feature can be present and in a third embodiment both the first and the second feature can be present.

It has been recognized that a large number of physical variables do not necessarily have to be recorded in order to determine the status of a system. Rather, it is sufficient to record at least one physical quantity with sufficient temporal resolution. In many application scenarios, recording one or two physical quantities is sufficient. The at least one physical quantity that changes over a period of time is recorded in such a way that a large number of measured values for the changing at least one physical quantity can be generated within this period of time. This large number of measured values is then compared with at least one characteristic that characterizes a state of the system. A measure of deviation is determined and a state of the system is estimated based on the measure of deviation.

To realize this basic idea, an arrangement is proposed which includes at least one sensor, a memory and an evaluation unit. The sensor is designed to detect the at least one physical quantity or one of the at least one physical quantity and to generate a large number of measured values depending on the detected physical quantity. Characteristics that are representative of at least one state of the system are stored in the memory. The evaluation unit is communicatively connected to the at least one sensor and to the memory. The evaluation unit obtains the large number of measured values from the sensor(s) and one or more characteristics from the memory. The evaluation unit is designed to analyze at least part of the large number of measured values and/or their time course and to determine a deviation measure using at least one characteristic. The deviation measure is then used to estimate the state of the system.

A “system”, the state of which can be determined by the arrangement of the present disclosure, can in principle be formed by a wide variety of systems. In this system, at least one physical quantity should change over a period of time. How exactly this change takes place is not of decisive importance. The change can occur periodically. But it is also conceivable that the physical quantity does not change with any defined pattern. In one embodiment, the system is a pump system in which a pump moves a fluid from a source to a sink. The fluid can also be moved in a circuit.

The “physical quantity” can in principle be formed by a wide variety of sizes. The physical quantity should be detectable by a sensor and have at least an indirect connection with a state of the system. In practice, however, these requirements can usually be met without any problems. Since many physical variables - as can be seen in the example mentioned in the introduction - can influence each other, it is also possible with the present disclosure to record several physical variables and to evaluate them together - for example in the sense of sensor data fusion. For this purpose, the assessment unit and the characteristics can be adjusted accordingly. It is also conceivable that, in addition to the at least one changing physical quantity, one or more stationary or quasi-stationary physical quantities are included in the status determination, for example a temperature that does not change or changes very slowly.

The “sensor” can also be designed in different ways, as long as at least one physical quantity or one of the at least one physical quantity can be detected and a large number of measured values can be generated within the period of time. Whether the sensor uses a capacitive, inductive, resistive, optical, thermoelectric or acoustic measuring principle, carries out differential or absolute measured value acquisition or is designed as a macro system, electronic system or MEMS (micro-electro-mechanical system), to name just a few conceivable examples name is not relevant to the present disclosure. In addition to the actual sensing element, the sensor can also have other components, for example amplifiers, filters (for example low pass, high pass, band pass), analog-digital converters, linearization elements, elements for galvanic isolation and/or the like. The sensor can have a (possibly replaceable) display and operating unit in which, in addition to output and input interfaces, other functions of the sensor and/or the arrangement can also be implemented.

The term “period of time” refers here to a period within which the at least one physical quantity changes. The change should occur at least to an extent that can be detected by the sensor. However, it is advisable if there is a sufficiently strong change in the physical quantity to be able to reliably carry out a comparison with the characteristic. In one embodiment, the change comprises at least 10% of the measuring range of the sensor, in another embodiment at least 25% of the measuring range, in a further embodiment at least 50% of the measuring range. It cannot be ruled out that the at least one physical quantity will continue to change outside of the time period; However, the at least one physical quantity should change at least within the period of time. Periods of time can also be “strung together,” meaning that after a period of time has elapsed, a new period of time would begin. The method according to the present disclosure would then be carried out repeatedly.

The period of time should not be too long. Periods of several minutes or even hours are likely to be unsuitable for many application scenarios in connection with the present disclosure. At the same time, the period of time should not be too short, as generating a large number of measured values could be problematic with very short periods of time. In the case of a periodically changing physical quantity, it may be advisable if the time span corresponds to the period length of the change or if the time span is chosen to be smaller than the period length. In one embodiment the time period is a maximum of 30 seconds, in another embodiment a maximum of 15 seconds, in a further embodiment a maximum of 5 seconds and in a still further embodiment a maximum of one second. In one embodiment, the time period is at least one millisecond long, in another embodiment at least 10 milliseconds, in a further embodiment at least 50 milliseconds and in yet another embodiment at least 100 milliseconds.

What a “multiplicity of measured values” means in concrete terms will depend on the physical quantity to be recorded and on the dynamics of a recorded change. The measured values should be so numerous that it is possible to determine the degree of deviation with sufficient reliability. For physical quantities that change very slowly or linearly, just a few measured values (e.g. 5 measured values) can be sufficient to meet this requirement. For quite dynamically changing physical quantities, hundreds or several hundred measurement values may be required. In one embodiment, the plurality of measured values comprises at least 10 measured values, in another embodiment at least 50 measured values, in a further embodiment at least 100 measured values and in a yet further embodiment at least 200 measured values. In principle, there is no upper limit to the number of measured values. In practical implementation, there are likely to be limit values. For most sensors, there is a minimum amount of time required to record a measured value. This recording period limits the number of measured values that can be recorded per period of time. Furthermore, in most systems there is likely to be a maximum number of measured values above which each additional measured value no longer provides any additional gain. Therefore, in one embodiment, the plurality of measured values comprises a maximum of 10,000 measured values, in another embodiment, a maximum of 5,000 measured values, in a further embodiment, a maximum of 1,000 measured values, and in a yet further embodiment, a maximum of 500 measured values.

A “state” of the system can refer to different operating states of the system. The state can describe normal operation in which all essential elements of the system are in a specified and/or usual operating range. This can also include, for example, that measured values are within a target range, that no components of the system are worn, that there are no defects and/or that there are no leaks or bottlenecks. The condition may also include one or more error operating conditions. In a fault operating condition, at least one element of the system is outside a target range. This can include, for example, that a bearing of a pump is worn, a supply line is leaking and/or clogged, a transported fluid is excessively contaminated with particles and/or dissolved gas, an electric motor has a large imbalance and/or a temperature outside a target range lies. It should be explicitly pointed out that the aforementioned examples only serve to explain possible states, but are not intended to limit the states to these specific examples.

In the arrangement according to the present disclosure, a state determination can be carried out repeatedly, preferably periodically. However, a status determination can also be triggered on an event-related basis, for example when the system starts, when a vibration limit value is exceeded or when there is an external trigger (for example from a higher-level control unit). Accordingly, in the present disclosure, a status check (with one or a few status determinations) and/or a status monitoring (with frequent status determinations) can take place.

A “characteristic” can in principle be formed by anything that is suitable for describing a state in the system. The specific design of a characteristic is likely to depend on the respective system, the changing at least one physical quantity and/or typical state images. A condition can be defined by one characteristic or a combination of several characteristics. In the aforementioned example of a pump system tems, the physical quantity “pressure” will change in a characteristic way over time. In this case, a characteristic could represent an approximation or some other description of the pressure curve over time. A bottleneck in a feed line or a drain line can result in an offset, so that a characteristic can be formed using an average value. Pressure surges within the system can be reflected in unusual peaks, so that a characteristic can describe the height and/or timing of a common peak. A defective pump bearing can manifest itself in increased noise, so one characteristic may include a level of a sound sensor. This short, non-exhaustive and not restrictive list shows how universal a characteristic can be and on what framework conditions the determination of a characteristic can depend.

The same applies to the “measure of deviation”, which describes how much the large number of measured values and/or their time course deviates from a characteristic. The lower the deviation measure, the closer a current state of the system is likely to be to a state that is associated with the respective characteristic. This in turn means that determining a measure of deviation will depend on a specific characteristic. For example, if a characteristic describes a specific target mean value of a physical quantity, determining a deviation measure should first include determining an average value and then comparing the determined mean value with the target mean value. If a characteristic includes a specific target spectral component, determining a measure of deviation should first involve a spectral analysis (for example using an FFT (Fast Fourier Transform) or using the Goertzel algorithm) and a subsequent comparison of the determined spectral component(s) with the Target spectral component includes. This short, non-exhaustive and non-limiting list shows what a deviation measure can be and how it can in principle be determined.

The arrangement can be implemented in various ways. In one embodiment, the arrangement is implemented entirely by hardware. In another embodiment, the arrangement is implemented by a combination of hardware and software. The hardware can include sensors, analog-digital converters, filters (for example high-pass, low-pass, band-pass filters), a processor (for example a microcontroller, digital signal processor or ASIC (Application Specific Integrated Circuit)) and/or a programmable logic circuit (for example an FPGA (Field Programmable Gate Array) or CPLD (Complex Programmable Logic Device)). Furthermore, one or more memories may be present, for example RAM (Random Access Memory), ROM (Read Only Memory) or flash memory, which other hardware components can access. Software can control the respective components of the arrangement, for example the processor or parameters of an analog-to-digital converter. This approach allows the arrangement to be flexibly adjusted even after production. Furthermore, individual elements can also be outsourced, for example to a cloud system.

The “memory” for the characteristics can also be implemented in a variety of ways. Volatile memory can be used in the same way as non-volatile memory. By way of example only, reference is made to RAM (Random Access Memory), MRAM (Magnetoresistive Random Access Memory), EEPROM (Electronically Erasable Read Only Memory) or flash memory. When using a volatile memory, it is advisable to protect its memory contents using a buffer element, for example a gold cap (or another capacitor) or an accumulator. The storage can also be connected via a remote access interface and, for example, outsourced to a cloud. The memory can be intended exclusively for storing characteristics. However, the memory can also be designed as a memory area of a larger memory in which additional content is stored.

In one embodiment, the arrangement additionally comprises a detection unit, wherein the detection unit is communicatively connected to the at least one sensor and to the memory and is designed to receive a plurality of measured values from the at least one sensor and to extract at least one characteristic from the plurality of measured values and store it in the memory. In this way, a system with a known state can be checked and the known state can be recorded and described with a characteristic / several characteristics. This procedure can be carried out, for example, when a system is commissioned for the first time. The known state can then be a “normal state”. However, it is also conceivable that error states are specifically brought about, for example by causing a leak, closing a valve or deliberately causing a shortage (for example of fluid sucked in by a pump). The characteristics extracted in this way offer the advantage that all relevant components of the system in which the arrangement used, the characteristics should be taken into account.

In one embodiment, the detection unit has a status input into which a known status of the system can be entered, an entered status being linked to an extracted characteristic and being able to be stored in the memory. In this way, the known status can be made known to the detection unit. An entered state can, for example, be a “normal state” or a (possibly deliberately induced) “error state”. An entered state can be represented in various ways, for example by a character string, a status code, a number or the like.

In one embodiment, the characteristics include a peak, a roundness of a peak, a slope, an average, a time period until a predefined value is reached, a time period for a sign change of a slope, a period length, a maximum, a minimum, an offset Spectrum and/or a harmonic. The height and/or temporal position of a peak can be easily determined, meaning that a peak can be used as a characteristic without much effort. The roundness of a peak can provide information about the behavior of the system, such as the nature of a peak trigger. For example, a pressure hammer in a pump system is likely to produce a sharp peak, while a leak will cause a peak to flatten. The roundness of the peak can be determined quite easily using a derivation. A slope can be easily determined using a derivative and can allow conclusions to be drawn about the behavior of the system, for example the behavior of an actuator or a sink. An average value, reaching a predefined value, a period length of a periodic change, a maximum, a minimum and/or an offset can also be easily determined and - depending on the design of the system - can provide important criteria for the state of the system. In the aforementioned example of a pump system, for example, a longer period length can indicate an unusually high load on the pump, for example due to too high a viscosity of the moving fluid. A change in sign of a slope can be determined via a second derivative and can provide a characteristic in a range of change in the physical quantity in which no clearly detectable peak or other clear distinguishing features can be extracted. A spectrum and/or a harmonic provides deep insights into the oscillation behavior of the system.

In one embodiment, one of the characteristics describes a target course of a physical quantity and/or measured values and the evaluation unit is designed to determine a deviation of the measured values and/or their time course from the target course as a measure of deviation. A target course can describe a time course of the physical quantity within the time period or a subsection of the time period that would be assumed, typical or favorable for a particular state. The target course can be represented in a variety of ways. The use of a table of values is just as conceivable as the functional description or an approximation. The deviation measure can describe a general difference between the target curve and measured values, for example “measured values deviate from the target curve by 20%” or “measured values correspond to the target curve by 93%”. However, the deviation measure can also address the differences between the target curve and the measured values more specifically, for example a time interval between peaks in the target curve or the measured values, a deviation in the roundness of a peak, a time interval between changes in the sign of the slope or a distance between mean values of the target curve and the measured values. In this way, detailed information about the status of the system can be obtained.

In one embodiment, the at least one sensor is designed to detect a pressure, a temperature, a vibration, a sound and/or other physical variables. In this way, the arrangement can be adapted to determine the status of a wide variety of systems.

In one embodiment, the at least one sensor is designed to periodically carry out measurements and/or to carry out at least one measurement per second, preferably at least 10 measurements per second, particularly preferably at least 100 measurements per second, most preferably at least 200 measurements per second . With at least 10 measurements per second, the temporal behavior of many systems can be mapped in sufficient detail. With at least 100 measurements per second, the temporal behavior can be mapped in more detail. With at least 200 measurements per second, even more dynamic changes in a physical quantity can be reliably recorded.

In one embodiment, the evaluation unit includes a neural network, a least squares estimator, a correlator and/or one based on iterative and/or statistical models end estimator. The evaluation unit can generally be based on artificial intelligence (AI). Neural networks (also known as artificial neural networks) are a particularly powerful example. The neural network used here can have a wide variety of basic structures and can have been trained with measured values and classifications from a large number of similar systems. The respective state of the system can be entered into the neural network as an input variable. However, it is also conceivable that a separate neural network is provided and trained for each condition that can potentially be detected. Intermediate solutions are also conceivable, in which, for example, similar system states are jointly detected by a neural network. A least square estimator can be used, for example, when comparing the measured values with a target curve in order to compensate for measurement errors in the measured values and to approximate the measured values to a model function. A correlator can determine a quantitative agreement between a target curve and a measured value curve using a correlation. Model-based estimators can be used to estimate measured values and/or system-internal states that cannot be measured directly. For the iterative models, reference is made to the Kalman filter as an example. As an example of the stochastic models, reference is made to the Hidden Markov Model. Several approaches can also intertwine. For example, the measured values can first be approximated to a model function using a least-squares estimator, the resulting model function can be fed to a correlator and the result of the correlation can be fed to a model-based estimator. This allows the valuation unit to be flexibly adapted to a wide range of requirements.

In one embodiment, the evaluation unit is designed to carry out the status determination repeatedly in order to minimize the influence of measurement errors. It is conceivable that a decision is only made for a state when a certain proportion of successive state determinations deliver a specific state. In this way - particularly when determining an error condition - the probability of an incorrectly detected error can be reduced.

In one embodiment, the evaluation unit is designed to form a subset from the plurality of measured values and to use it to form the deviation measure, the subset preferably forming a temporal section, a frequency section and/or an amplitude section. In this way, determining the state can be simplified because less data has to be processed. The determination of a subset can depend on the respective characteristics. For example, if only the measurements around a peak are of interest, an appropriate subset can be formed in an area around the peak. For example, if only a certain harmonic is important for determining the status, a subset can be formed for the spectral component of interest. If, for example, only those measurement values that exceed or fall below a predefined limit value are important for determining the status, a subset can be formed for an amplitude section, namely for all measurement values above or below the limit value.

In one embodiment, the arrangement additionally comprises a reporting unit, wherein the reporting unit is designed to generate and send a message when a state requiring reporting is determined, wherein a state requiring reporting preferably includes an error state. In this way, a recognized system status can be reliably communicated. The arrangement can be used in an Industry 4.0 environment. This can enable and/or promote predictive maintenance of the tested/monitored system. What constitutes a “reportable condition” will depend on the particular system being tested/monitored. This can also include a normal state. However, since in most applications normal operation is/can be assumed, in one embodiment a “state requiring notification” is a state that deviates from normal operation. This may include an error condition, which may also include an emerging or impending error.

In one embodiment, the arrangement additionally has a communication interface that enables data exchange, the communication interface preferably being designed to send measured values, to send messages, to receive configurations, to receive characteristics and/or the like. Wireless interfaces as well as wired interfaces can be used for the communication interface. For wired interfaces, examples include Ethernet, Profibus, HART (IEC 61158) or Modbus. Examples of wireless interfaces include WLAN (IEEE 802.11), Bluetooth (IEE 802.15.1), LoRaWAN (LoRa-Alliance Industry consortium), NB-IoT/4G (3GPP release 13), 5G (3GPP Release 15) or WirelessHART ( IEC 62591). By sending measured values, a separation of measured value acquisition and measured value use can be achieved, ie the measured values can be recorded at one point and evaluated at another point. By sending messages, a recognized state can be communicated from the arrangement to the outside, for example to a higher-level control unit. By receiving configurations, the arrangement can be configured and/or controlled from outside, for example by a higher-level control unit. Receiving characteristics enables characteristics to be kept and/or updated at a central location, for example in a cloud system, and the characteristics of the arrangement to be made available, for example in an initialization step.

In one embodiment, at least one of the characteristics characterizes normal operation and/or at least one of the characteristics characterizes error operation. In this way, normal operation and error operation can be recognized and differentiated.

• 1 eine Darstellung eines Ausführungsbeispiels einer Anordnung gemäß der vorliegenden Offenbarung, das bei einem Pumpensystem eingesetzt ist,
• 2 ein Diagramm mit einem beispielhaften Verlauf eines Drucks und
• 3 ein Ablaufdiagramm mit Schritten eines Ausführungsbeispiels eines Verfahrens gemäß der vorliegenden Offenbarung.

Further features and advantages of the invention result from the following description of exemplary embodiments which are not to be understood as restrictive, which are explained in more detail below with reference to the drawing. This drawing shows schematically:

• 1 a representation of an embodiment of an arrangement according to the present disclosure used in a pump system,
• 2 a diagram with an exemplary pressure curve and
• 3 a flowchart with steps of an exemplary embodiment of a method according to the present disclosure.

1 shows a schematic representation of an exemplary embodiment of an arrangement according to the present disclosure. The arrangement 1 is used to determine the status of a system 2, the system 2 being formed by a pump system. The pump system comprises a pump 3, which moves a fluid from a source 4 via a supply line 5 and a discharge line 6 to a sink 7. The pump 3 generates a negative pressure in the supply line 5, which sucks in the fluid from the source 4, and an excess pressure in the discharge line 6, which pushes the fluid to the sink 7.

The arrangement 1 comprises a plurality of sensors 8, namely a first pressure sensor 8', a second pressure sensor 8'' and a vibration sensor 8'''. The first pressure sensor 8′ detects a pressure in the discharge line 6 and the second pressure sensor 8″ detects a pressure in the supply line 5, the second pressure sensor 8″ being optional. The vibration sensor 8''' is designed as a structure-borne sound sensor and detects a vibration of the pump 3. The sensors 8 each detect a physical quantity, here pressure and vibration, and generate a large number of measured values from the detected physical quantity. The sensors 8 are communicatively connected to an evaluation unit 9 and transmit the large number of measured values to the evaluation unit 9 via this connection. The evaluation unit 9 is additionally connected to a memory 10 in which characteristics are stored. The characteristics are representative of at least one state of the system 2. A reporting unit 11, which is also connected to the evaluation unit 9, can communicate a status that requires reporting from the arrangement. For this purpose, the reporting unit 11 can have a communication interface (not shown).

In 2 an exemplary course of a pressure p is shown, as detected, for example, by the pressure sensor 8 '. The pressure runs sinusoidally and periodically oscillates around an average value p _M with a period length T. In the present exemplary embodiment, this period length T can be equated with the time period within which the physical quantity “pressure” changes. It can be seen that several time periods are strung together, ie after one time period has elapsed, a new time period begins. Within this period of time, the pressure sensor 8′ records a large number of measured values for the physical quantity “pressure”. An example should make this clear: The period length T is 1 second. The sensor needs 5 milliseconds to detect the pressure. This means that the sensor can record a total of 200 measured values within the period.

When operating according to the arrangement 1 A method can be used as described in the exemplary embodiment 3 is shown. In step S1, the physical variables “pressure” and “vibration” are first detected by the sensors 8, namely by the first pressure sensor 8', the second pressure sensor 8'' and the vibration sensor 8'''. A large number of measured values are generated based on the recorded physical variables, with each of the sensors 8 generating a large number of measured values. In step S2, at least one characteristic is loaded from the memory 10 by the evaluation unit 9. In step S3, the measured values (or at least some of them) and their time course are analyzed. Only the measured values of one sensor 8 can be used, while the measured values of the other sensors are used for a plausibility check. In step S4, the measured values (or part of them) and their time course are compared with the loaded characteristic and a deviation measure is generated that shows a difference between the measured values and the characteristic describes. In step S5, a state of the system is determined using the deviation measure. Steps S3 to S5 could be carried out by an AI of the evaluation unit 9 without the individual steps being explicitly distinguishable from one another. After step S5, the process returns to step S1 and the process is restarted with the acquisition of further measured values.

With regard to further advantageous embodiments, to avoid repetition, reference is made to the general part of the description and to the attached claims.

Finally, it should be expressly pointed out that the exemplary embodiments described above only serve to discuss the claimed teaching, but do not limit it to the exemplary embodiments.

Reference symbol list

1

arrangement

2

system

3

pump

4

source

5

supply line

6

Derivation

7

sink

8th

sensor

9

Valuation unit

10

Memory (for characteristics)

11

Reporting unit P025n

Claims (14)

Arrangement (1) for determining the state of a system (2) in which at least one physical quantity (p) changes over a period of time (T), comprising: at least one sensor (8) which is designed to detect the at least one physical quantity/one of the at least one physical quantity (p) and to record a plurality of measured values based on a detected physical quantity (p) within the time period (T). generate, a memory (10) in which characteristics are stored, the characteristics being representative of at least one state of the system (2), and an evaluation unit (9), which is communicatively connected to the at least one sensor (8) and to the memory (10) and which is designed to analyze at least part of the large number of measured values and/or their time course, using at least one of the characteristics to determine a deviation measure and to estimate a state of the system (2) based on the deviation measure. Arrangement according to Claim 1 , which additionally comprises a detection unit, wherein the detection unit is communicatively connected to the at least one sensor (8) and to the memory (10) and is designed to receive a plurality of measured values from the at least one sensor (8) and from the plurality of To extract at least one characteristic from measured values and store it in the memory (10). Arrangement according to Claim 2 , characterized in that the detection unit has a state input into which a known state of the system can be entered, an entered state linked to an extracted characteristic being able to be stored in the memory. Arrangement according to one of the Claims 1 until 3 , characterized in that the characteristics are a peak, a roundness of a peak, a slope, an average, a time period until a predefined value is reached, a time period for a change in sign of a slope, a period length, a maximum, a minimum, an offset, include a spectrum and/or a harmonic. Arrangement according to one of the Claims 1 until 4 , characterized in that one of the characteristics describes a target course of a physical quantity and/or measured values and the evaluation unit (9) is designed to determine a deviation of the measured values and/or their time course from the target course as a measure of deviation. Arrangement according to one of the Claims 1 until 5 , characterized in that the at least one sensor (8) is designed to detect a pressure, a temperature, a vibration and / or a sound. Arrangement according to one of the Claims 1 until 6 , characterized in that the at least one sensor (8) for periodically carrying out measurements and / or for carrying out at least one measurement per second, preferably at least 10 measurements per second, particularly preferably at least 100 measurements per second, most preferably at least 200 Measurements per second is formed. Arrangement according to one of the Claims 1 until 7 , characterized in that the evaluation unit (9) comprises a neural network, a least squares estimator, a correlator and/or an estimator based on iterative and/or statistical models. Arrangement according to one of the Claims 1 until 8th , characterized in that the evaluation unit (9) is designed to carry out the status determination repeatedly in order to minimize the influence of measurement errors. Arrangement according to one of the Claims 1 until 9 , characterized in that the evaluation unit (9) is designed to form a subset from the plurality of measured values and to use it to form the deviation measure, the subset preferably forming a temporal section, a frequency section and/or an amplitude section. Arrangement according to one of the Claims 1 until 10 , which additionally comprises a reporting unit (11), wherein the reporting unit (11) is designed to generate and send a message when a state requiring reporting is determined, wherein a state requiring reporting preferably includes an error state. Arrangement according to one of the Claims 1 until 11 , which additionally has a communication interface that enables data exchange, the communication interface preferably being designed to send measured values, to send messages, to receive configurations and / or to receive characteristics. Arrangement according to one of the Claims 1 until 12 , characterized in that at least one of the characteristics characterizes normal operation and / or at least one of the characteristics characterizes error operation. Method for determining the state of a system in which at least one physical quantity changes over a period of time (T), preferably for operating an arrangement (1) according to one of Claims 1 until 13 , comprising: detecting (S1) the at least one physical quantity/one of the at least one physical quantity (p) and generating a plurality of measured values within the time period (T) by means of at least one sensor (8), loading (S2) of at least one state of the System (1) characterizing characteristic from a memory (10), analyzing (S3) at least a part of the multitude of measured values and / or their time course, generating (S4) a deviation measure based on at least a part of the multitude of measured values and / or their time course and the at least one characteristic and determining (S5) a state of the system (2) based on the deviation measure.

Images