Method for monitoring the state of an asynchronous motor

Abstract

The invention relates to a method for monitoring the state of an asynchronous motor (AM) which has a slip (s) and to which a feed voltage (U) with a feed frequency (f) is applied, having the following steps: - defining two or more different operating ranges (B) of the asynchronous motor (AM) as a function of the slip (s) and as a function of the feed frequency (f); - assigning, to each of the defined operating ranges (B), at least one respective corresponding anomaly detector (Ai-1, Ai, Ai+1) which is adapted so as to detect a malfunction (E) of the asynchronous motor (AM) in the respective operating range (B); - ascertaining the current operating range (B) of the asynchronous motor (AM); - activating the at least one anomaly detector (A) which is assigned to the current operating range (B) and monitoring the state of the asynchronous motor (AM) using the active anomaly detector (A); and - if an anomaly is detected by the active anomaly detector (A), generating a corresponding signal.

Description

[0001]202218760 1 Description Method for condition monitoring of an asynchronous motor The invention relates to a method for condition monitoring of an asynchronous motor. Due to their cost-effective and robust design, asynchronous motors are widely used in industry: they are found in many applications, such as fans, saws, and pumps, from relatively small systems in the watt range to systems in the megawatt range. Asynchronous motors are therefore among the largest consumers of electrical energy generated worldwide. Due to the widespread use of asynchronous motors, there is enormous potential in terms of availability and reliability and the associated savings of time and money. In this context, the terms "condition monitoring" (CM) and "predictive maintenance" (PM) are used. ASM (= asynchronous motor(s)) operate according to the principle of electromagnetic induction, i.e.Interactions between currents or magnetic fields in the stator and rotor generate a torque that drives the rotor shaft. Motor faults such as eccentricity, particularly air gap eccentricity (when the rotation axes of the stator and rotor of the motor do not exactly coincide), broken rotor bars, defective bearings or misalignment change the amplitudes at frequencies characteristic of the faults at constant slip. The fault frequencies, in turn, are slip-dependent, i.e. a change in slip changes the frequency position of the fault frequencies. Motor faults can therefore be detected based on the change in the amplitudes of these characteristic frequencies in the frequency spectrum. The so-called motor current signature analysis (MCSA) can be used for this purpose, 202218760 2 see e.g. B. EP3961230A1 (Siemens AG; Zettner, Jürgen) 02.03.2022. For example, a Fast Fourier Transformation (FFT) is used.Fast Fourier Transformation (MCSA) is used to detect and quantify errors and operating states of the motor in the frequency domain. Classic MCSA is applied in the quasi-stationary case, i.e., at a nominally constant speed. An extension of MCSA is the analysis with short-term FFTs or wavelets to analyze processes in a time-resolved manner. There is also the approach of machine learning (or artificial intelligence, AI). State-of-the-art anomaly detection relies on learning phases at the ASM to be monitored, during which a so-called "good state" of the ASM exists, i.e., a state in which the ASM is "normal," i.e.,without significant errors (what is to be considered "normal" and what is to be considered a significant error must be defined in advance by an ASM operator); anomaly detection in the detection phase is then based on those characteristics that the anomaly detector was taught as "normal" during its learning phase. Typically, during the learning phase, also known as the training phase, the good state is learned under different operating conditions, in the hope that the mathematical model of the anomaly detector in the feature space covers all operating states of the good state. Frequency-controlled drives, as well as drive trains with grid machines, experience variations in rotor speed and torque due to the loads to be driven (= torque), temperature changes, aging, or open-loop control.As described in the theory of MCSA, the frequency spectrum of an asynchronous motor (position of the frequency lines) is highly dependent on the load condition, i.e. on the slip s. Furthermore, it is known from classical MCSA that frequency lines also vary with the frequency of the supply voltage applied to the ASM (“supply frequency f. supply The MCSA primarily distinguishes between slip-dependent motor-specific spectral lines, e.g., the Principal Slot Harmonics (PSHs), various slip-dependent error lines, and the motor-specific Winding Harmonics (WHs), which are multiples of the supply frequency. The following apply to the PSH: f PSH = [k•(R / p)•(1-s) ± ν] • f supply Equation (1) and for the typical faults: i) Broken rotor bars (Broken Bars BB): f BB1 = (1 ± 2k•s) • f supply Eq. (2) ii) Static / Dynamic rotor eccentricity (see ISO 20958:2013) f ecc1 = [1 ± (k / p)•(1-s)] • f supply Eq. (3) f ecc2= [(R ± n d )•(k / p)•(1-s) ± ν] • f supply Eq. (4) iii) Bearing damage to the rolling bearing f o = [(N b / 2)•(1 – (D b / D p )•cos(β)] • f rot Eq. (5) f i = [(N b / 2)•(1 + (D b / D p )•cos(β) / D p )] • f rot Eq. (6) where f BB1 Frequency at rotor bar breakage f supply Frequency of the supply voltage, e.g., 50 Hz f ecc1 Frequency of eccentricity (1st family) f ecc2 Frequency of eccentricity (2nd family) f PSH Frequency of the Principal Slot Harmonics (= PSH) f o Frequency of the bearing outer ring (o = outer) f i Frequency of the bearing inner ring (i = inner) f rot Rotation frequency of the rolling bearing k natural number p number of pole pairs = number of pole pairs R number of rotor bars = number of rotor bars s slip ν odd numbers 1, 3, 5, 7, ... = order of the harmonics of f supply n dIndex depending on the eccentricity type (static: n d = 0; dynamic: n d = 1, 2, 3, 4, ...) 202218760 4 N b Number of rolling elements D b / D p Characteristic diameters on the rolling bearing ß Contact angle Depending on the number of pole pairs P and the number of rotor bars, various modulation artifacts in the spectrum depend on the slip s of the ASM. Changes in the load and thus in the slip s lead to a superposition of different states and thus also a superposition of different frequency spectra. If an anomaly detector learns different operating states, caused either by a changing supply frequency f supplyand / or due to a changing torque (ASM with slip), in the learning phase as good states, a large number of motor-specific spectral lines and a large number of frequency bands change with the different operating states. In the event of a fault during a detection phase, there is therefore a relatively low probability of clearly finding a fault in the drive train. This is due to the fact that the amplitudes of motor-specific frequencies overlap with those of fault-specific frequencies in some operating states, see Fig. 6. Fig. 6 shows a horizontal time axis t, above which a motor speed v is plotted in the lower part of Fig. 6. motorof an electric motor. It can be seen that the motor in question runs through 12 different operating states 62, characterized by different motor speeds, twice, starting from a highest speed and then going to a lowest speed. The first run through the operating states takes place in the good state 63, the second run through the operating states takes place in a fault state 64. In the upper area of ​​Fig. 6, also plotted over the time axis t, a frequency spectrum of the electric motor resulting from an MCSA is plotted. It can be seen that the amplitudes of motor-specific frequencies 60 overlap with those of fault-specific frequencies 61 in some operating states. 202218760 5 This means: if all operating cases of the good state are taught into an anomaly detector, the anomaly detectors shown in Fig.6 and the presence of the associated amplitudes at the motor-specific frequencies are classified as "OK". If an error is then present with an indication at one of the error-specific frequencies, the separation from the good case is no longer necessarily given, since the amplitude of the error frequency was classified as an OK line, especially since the amplitudes at the error frequencies can be small compared to the amplitudes at the motor-specific amplitudes. There is therefore a need for a method with which a fault in the drive train can be clearly identified with a greater probability. One object of the present invention is therefore an improved MCSA. This object is achieved according to the invention by a method having the features specified in claim 1. The method is used for condition monitoring of an asynchronous motor.Due to its design, the asynchronous motor exhibits slip during operation: slip is defined as the speed difference between the stator's rotating field and the rotor's, usually specified as a percentage of the stator's rotating field. A supply voltage is applied to the asynchronous motor, which, as the operating voltage of the asynchronous motor, leads to a current flow in the motor, which generates the magnetic fields inside the motor that cause the rotor to rotate. The supply voltage, as an alternating voltage, has a frequency, the so-called supply frequency. According to the method, two or more different operating ranges of the asynchronous motor are defined as a function of slip and as a function of supply frequency: thus, two different operating ranges differ in terms of slip and supply frequency.Each of the defined operating ranges is assigned at least one associated anomaly detector, which is adapted to detect a fault condition of the asynchronous motor in the respective operating range. After it has been determined in which of the defined operating ranges the asynchronous motor is currently located ("current operating range"), the at least one anomaly detector assigned to the current operating range is activated. The state of the asynchronous motor is monitored by the active anomaly detector. If the active anomaly detector detects an anomaly in the asynchronous motor, a signal is generated indicating the detection of an anomaly. The invention proposes using an anomaly detector specifically trained for this range in different operating ranges of an asynchronous motor.The inventive approach has the following advantages: - The invention improves the accuracy of anomaly detection for each speed / torque or operating point, as the variance of the features within an operating range is smaller. The smaller an "operating range" is, the smaller the proportion of real errors detected as "good" and the greater the accuracy: A smaller operating range can be evaluated better than a larger one. - The invention prevents an anomaly detector that is responsible for all operating ranges from recognizing all of them as operating ranges but being "blind" to errors that occur. This means that the probability of a type II error (false negative: "good message" even though an error is present) is minimized. - A minimum number of features must be calculated at any given operating point.The input values ​​of an anomaly detector, the so-called features, which are required for training an AI for anomaly detection, can be obtained in various ways. A first method for obtaining input values ​​is a method for diagnosing the condition of an electric motor to which an AC supply voltage with a supply frequency is applied, comprising the following steps: a) Acquiring a time series of current values ​​of a machine current flowing through the electric motor due to the supply voltage applied to the electric motor; b) Describing the time series with an autoregression model. ௧ ൌ ^^ ^ ^^ ௧ ^ ∑ ^ ^ ୀ^ ^^ ^ ^^ ௧ି^ of order p; c) forming a feature vector with p vector elements, where the p vector elements are defined by the p parameters a iof the autoregression model; and d) using the formed feature vector in an ML algorithm as input values, ie, as training data or as test data. Another method for obtaining input values ​​is a method for the condition diagnosis of an asynchronous motor to which a three-phase voltage with a supply frequency is applied, comprising the following steps: a) Acquiring a voltage time series U t the voltage and a current time series I t a three-phase motor current flowing through the asynchronous motor due to the voltage applied to the asynchronous motor; b) Calculating, using an FFT, a current spectrum I(f,t) and a voltage spectrum U(f,t) based on the recorded time series I t , U t ; c) Calculate one or more of the following time-dependent spectral quantities 1-6: 1) Current components I a (f,t), I b (f,t), where I a (f,t) and I b(f,t) by Park transformation from I R,S,T are formed; 2) total current amplitude A I (f,t) = √((I a (f,t)) 2 + (I b (f,t)) 2 ) 3) Phase angle φ ab (f,t) between current components: φ ab (f,t) = arctan(I a / I b ) 4) Phase angle φ UI (f,t) between current and voltage components: φ UI(f,t) = arctan(U(f,t) / I(f,t)) 5) instantaneous spectral power P(f,t) = U(f,t)*I(f,t) 6) admittance X(f,t) = I(f,t) / U(f,t) 202218760 8 d) forming a feature vector from one or more of the time-dependent spectral variables; and e) using the formed feature vector as input values ​​for an ML algorithm, i.e. as training data or as test data for an anomaly detector. In connection with a method for the condition diagnosis of an electric motor based on machine learning, in which the input values, the so-called features, which are required for training an AI for anomaly detection, are obtained from current or voltage samples, where e.g. For example, if one of the two aforementioned methods can be used, it was recognized that features should be selected in such a way that as much information as possible is contained in the features.Even an SOM (= self-organizing map), which uses either autoregression features or FFT frequency features, could train itself on noise and artifacts instead of error features, provided that an arbitrarily high number of features or many amplitudes in the noise equivalent are used for training. - The size of an individual anomaly model A(q,r), e.g. the map size of an SOM, is reduced by the invention, since less data is available for training per restricted operating range. Thus, training is simplified and prediction faster, and the storage of individual operating range models becomes more resource-efficient. Each individual model (anomaly detector) becomes more specific and smaller. The following describes the relationship between the slip s, the rotor speed n, the motor torque M, and the supply frequency f. S explained. The rotating field speed n Son the stator (unit revolutions per second) is a function of the supply frequency f S (Unit Hz) and the number of pole pairs p: n S = f S / p (I) 202218760 9 The supply frequency f S , which is also called stator frequency or mains frequency, can also be symbolized simply with the letter f, without a subscript S. The slip s is the difference between the rotating field speed n S on the stator (unit revolutions per second) and the rotor speed n (rotor), usually given as a percentage value related to the rotating field speed n S : s = (n S - n) / n S <=> n = n S (1 - s) (II) The greater the torque M required at the rotor, the more the rotor speed n decreases and the greater the slip s. The torque M of the motor is a function of the slip s. The relationship between the torque M of an asynchronous machine and the slip value s, the standstill slip s Kand the tipping moment M K gives approximately the Kloss equation: M = 2⋅M K / (s K / s + s / s K ) (III) In addition, the following relationships apply: P el = η P mech (IV) P el = 3 UI cos φ (V) P mech = ω M = 2 π n M (VI) where P el active electrical power supplied to the motor η efficiency of the motor P mech Mechanical power delivered to the motor shaft U Supply voltage I Motor current φ Phase shift angle ω Angular frequency 202218760 10 From equation VI, the motor torque M is obtained by successively inserting equations II, I, IV and V as: The invention uses an anomaly detector for the voltage signals. This can be a SOM (self-organizing map) or an autoencoder, which detects deviations from undisturbed ("good") sinusoidal voltage waveforms. To measure the voltage / power quality, voltage feature data sets are created, e.g., voltage spectrum or autoregression features, and used to train a voltage anomaly detector. The idea of ​​the SOM is to generate a grid of X * Y neurons N. Each neuron N contains a weight vector of length L F, i.e., equal to the length of a feature vector. Determining the number of neurons (X * Y) is a compromise between the amount of data / computational effort on the one hand, and the desired reliability of an anomaly detector on the other. In tests already conducted, 20 x 20 = 400 neurons have led to very good results. Self-organizing maps, like most artificial neural networks, work in two modes: training and mapping. In training mode, an input dataset ("input space") is used to generate a map ("map space") as a lower-dimensional representation of the input data. In mapping mode, measurement data is classified using the generated map, e.g., as normal or anomalous. 202218760 11 Learning the SOM, i.e., training, is an iterative process. In the initial state, the weight vectors contain random values.The learning process now proceeds by assigning similar feature vectors from the training dataset to the same neuron. "Similar" is to be understood in the sense of a vectorial distance; depending on the application, the Euclidean distance, the cosine distance, or the Manhattan distance can be used. The goal of training is to represent a p-dimensional "input space," i.e., with p variables, as a 2-dimensional "map space." A "map space" consists of components called nodes or neurons, arranged on a hexagonal or rectangular grid with two dimensions. The number of nodes and their position on the grid are determined in advance based on the goal pursued by analyzing and investigating the "input space." Each node in the "map space" is assigned a weight vector that defines the node's position in the "input space."While the nodes remain anchored in their position in the map space, the weight vectors are approximated to the input data, i.e., the distance, e.g., the Euclidean distance, between the weight vector and the input data is minimized without destroying the topology generated by the map space. During the training phase, feature vectors that are "similar" to each other are grouped. This similarity is defined as the Euclidean distance, i.e., the smaller the distance between the feature vectors, the more similar they are. For each "group" of similar feature vectors, one or more representatives are selected, as is the maximum distance Dmax for each group. After the training phase is completed, the SOM generated in this way can be used to generate further "input space" vectors, i.e.,Observations in the "input space", here: time series of current values, are to be classified by identifying the node whose weight vector is closest to the "input space" vector, i.e., has the smallest distance metric, e.g., the Euclidean distance. A feature vector generated according to the invention with p vector elements, which were obtained using an AR method from a time series measured on an electric motor, is assigned to a neuron of the said SOM and a weight vector assigned to the neuron. In order to use the trained SOM as an anomaly detector, a function is required that assigns a degree to which the feature vector represents an anomaly to a feature vector Fν. For this purpose, the "quantization error" Eν is calculated for each feature vector Fν: Eν = | Fν – Gν| for all ν, i.e., for all feature vectors. Where |...| the previously defined vectorial distance and Gν the respective weight vector of the "optimal" neuron, i.e., the neuron whose weight vector G has the minimum distance to the feature vector. The quantization error E describes how well the respective feature vector can be mapped to the distribution of the data with which the SOM was trained. A high error indicates that the feature vector is very different from the training data and therefore describes anomalous motor behavior. In practice, an anomaly in the electric motor is detected if the quantization error Ex exceeds a fixed threshold value E. Threshold exceeds: Ex > E ThresholdAfter training, anomaly detection takes place as follows: A feature vector is generated from the input values, as described above. For this vector, the Euclidean distance to the learned "groups" is determined. The minimum value of all distances is determined. If this value is greater than Dmax, an anomaly is present. Advantageous embodiments and further developments of the invention are specified in the dependent claims. According to a preferred embodiment of the invention, a slip value range of slip values ​​that occur during operation of the asynchronous motor and a frequency value range of supply frequency values ​​that occur during operation of the asynchronous motor are determined as a prerequisite for the inventive operating point-specific design of the anomaly detector.The determined slip value range is divided into Q slip sub-ranges and the determined frequency value range is divided into R frequency sub-ranges (Q, R ^ ℕ^). Accordingly, Q x R operating ranges are defined, defined by one of the said slip sub-ranges and one of the said frequency sub-ranges. According to the invention, not one anomaly detector, but Q x R anomaly detectors A(q, r) are trained; an anomaly detector A(q, r) has a feature space with the dimensionality K, i.e., K specifies the number of features of the anomaly detector A(q, r). Each anomaly detector is valid for an assigned operating range. In a detection phase, a current operating range of the asynchronous motor is determined, then the anomaly detector assigned to the current operating range is activated, and the state of the asynchronous motor is determined by the active Anomaly detector monitored.If an anomaly is detected by the active anomaly detector, a signal indicating the anomaly is generated by the active anomaly detector. According to a preferred embodiment of the invention, in a detection phase, after determining the current operating range of the asynchronous motor, the anomaly detector assigned to the current operating range is activated, as well as one or more anomaly detectors assigned to operating ranges of the asynchronous motor adjacent to the current operating range. The state of the asynchronous motor is monitored by the active anomaly detectors. A resulting advantage is that each of the anomaly detectors knows the motor-specific frequencies learned for "its" operating point. In the good state, only these can occur, or the good state is optimally described in the K characteristics of the R frequencies according to the training.Amplitude changes at fault-specific frequencies in the event of a fault differ significantly from the amplitudes of the now frequency-fixed motor-specific frequencies for this operating point. The N anomaly detectors have the further advantage that majority decisions can also be made if, as is particularly the case in Fig. 6, the fault cannot be clearly detected at only a few overlap points between fault- and motor-specific frequencies, i.e., for certain operating points. However, the anomaly detectors at neighboring operating points, where this overlap does not exist to the same extent, can then reach "anomaly present" decisions with greater certainty. The presence of a fault can then be determined from the majority of decisions at the neighboring operating points, even if the actual exact operating point does not allow detection through a frequency overlap.There may also be errors that only have an impact in certain operating point ranges (cavitation would be an example: assume that cavitation only occurs when the following applies to the speed n: 3000 rpm > n > 2400 rpm), so that the majority decision of the adjacent anomaly detectors for this operating point range would then be decisive in order to detect with a high probability that an error must have occurred in the actual operating point (for example in the 2600-2650 rpm range, for which this anomaly detector has an overlap between the error and motor lines). 202218760 15 According to a preferred embodiment of the invention, several anomaly detectors are assigned to each of the defined operating ranges, wherein detection results from the assigned anomaly detectors are then available as a result vector. A majority decision is applied to the result vector to decide on an anomaly.According to a preferred embodiment of the invention, determining the current operating range of the asynchronous motor comprises the following steps: First, slip values ​​and supply frequency values ​​are determined during a time sequence. Based on the determined slip and supply frequency values, the current operating range of the asynchronous motor is determined. According to a preferred embodiment of the invention, the time sequence is divided into sub-units of shorter duration, with a slip measure being determined in each of the sub-units. An anomaly detector is only activated if the differences in the slip measures do not exceed a predetermined difference value; otherwise, an anomaly detector is not activated. Since determining the slip or the supply frequencies also requires a certain time period T, it may happen that the operating point changes within the detection time.By dividing a time sequence T into ν subunits of shorter duration Tν, a measure Ls,ν for the slip can be determined in each subsegment. The measure of the slip can be determined from the position of the principal slot harmonics (PSH) in the frequency domain or from the cos φ in the time domain. If the measures Ls,ν for the subsegments are not close enough to each other, the operating state will most likely change during T and no anomaly detector will be "armed". Alternatively, each anomaly detector A(n,m) is evaluated with the sub-operating points for Ls,ν, but an anomaly is only output if a sufficient number of operating point-specific anomalies have accumulated, which is unlikely during a change. The invention is explained below with the aid of the accompanying drawing. It shows schematically and not to scale Fig.1 shows an embodiment of a device according to the invention for monitoring the condition of an asynchronous motor; Fig. 2 shows an embodiment of an artificial neural network; Fig. 3 shows a flow diagram of an embodiment of the method according to the invention; Fig. 4 shows the definition of the different operating ranges of the asynchronous motor; Fig. 5 shows the embodiment in which several anomaly detectors are responsible for one operating range; Fig. 1 shows an asynchronous motor AM (hereinafter also referred to simply as: electric motor or as: electric motor), which is electrically connected to a power grid N by means of a branch 12. The branch can, for example, be a three-phase electrical line that connects the three-phase power grid N to the connection terminals of the electric motor AM. The electric motor AM is designed as a drive machine of a drive train D, which also has a mechanical transmission G and a working machine W.A torque provided by the electric motor AM is transmitted via the transmission G to the working machine W. The working machine can be, for example, a conveyor belt, a roller, or a cylinder. In addition to the electric motor AM of the drive train D, further electrical loads 20 are connected to the power grid N. 202218760 17 A first sensor S1, which is arranged, for example, on the branch 12, detects voltage values ​​U. m an electrical supply voltage of the electric motor AM, e.g. a voltage between two phases of the branch 12 or between a phase of the branch 12 and a reference potential, and sends the detected voltage values ​​U m via a transmission medium 10, e.g., a data cable, to a device 15 for monitoring the condition of the asynchronous motor AM. A second sensor S2, which is arranged, for example, on the electric motor M, e.g., a shunt, a current transformer, or a Hall sensor, detects current values ​​I mof an electric current flowing through the electric motor AM and sends the current values ​​I mvia a transmission medium 11, e.g., a data cable, to the device 15 for monitoring the condition of the asynchronous motor AM. The current I flowing through the supply lines 11 is measured by a current sensor 14, e.g., at a measuring rate of 3200 measurements / sec. The supply voltage U applied to the asynchronous motor AM is tapped at the supply lines 11 by a voltage sensor 15. The current measured values ​​recorded by the current sensor 14 and the voltage measured values ​​recorded by the voltage sensor 15 are sent from the sensors 14, 15 to a computing unit 16. The computing unit calculates the slip and an FFT(I) from the recorded current and voltage measured values. The further method steps for determining the slip s are also carried out by the computing unit 16. For this purpose, the computing unit 16 has a processor 18 and a data memory 20.The data memory 20 stores software with an algorithm for determining the slip s from current measurements. This software is executed by the processor 18. In order for the method according to the invention to be feasible, current and voltage must be measured, and the slip, i.e., the actual speed, must be determined in order to be able to determine the operating point and the associated slip-dependent error frequencies. 202218760 18 The device 15 for monitoring the condition of the asynchronous motor AM has an anomaly detector 16, a user interface 17, and a data memory 19. The anomaly detector 16 has a trained AI function 18. The device 15 receives the detected voltage values ​​U. m and the recorded current values ​​I m and evaluates them using the trained AI function 18. The AI ​​function 18 analyzes the recorded measured values ​​U m , I mand ... defining two or more different operating ranges (B) of the asynchronous motor (AM) as a function of the slip (s) and as a function of the supply frequency (f); - Assigning to each of the defined operating ranges (B), at least one associated anomaly detector (A i-1 , A i , A i+1), which is adapted to detect a fault condition (E) of the asynchronous motor (AM) in the respective operating range (B); - determining a current operating range (B) of the asynchronous motor (AM); - activating at least one anomaly detector (A) assigned to the current operating range (B) and monitoring the condition of the asynchronous motor (AM) by the active anomaly detector (A); and - in the event of an anomaly being detected by the active anomaly detector (A), generating a corresponding signal. The device 15 reports a fault in the asynchronous motor AM if an anomaly has been detected. The reporting is carried out via the user interface 17, e.g., a display. Fig. 2 shows an embodiment of an artificial neural network 100, as used in the trained AI function 18.Alternative terms for "artificial neural network" are "neural network," "artificial neural network," or "neural network." 202218760 19 The artificial neural network 100 has nodes 120, ..., 132 and edges 140, 141, 142, where each edge 140, 141, 142 is a directed connection of a first node 120, ..., 132 to a second node 120, ..., 132. While in general the first node 120, ..., 132 and the second node 120, ..., 132 are different nodes 120, ..., 132, it is also possible that the first node 120, ..., 132 and the second node 120, ..., 132 are identical. For example, in Fig. 6, edge 140 is a directed connection from node 120 to node 123, and edge 142 is a directed connection from node 130 to node 132. An edge 140, 141, 142 from a first node 120, ..., 132 to a second node 120, ..., 132 is also called an "incoming edge".: ingoing edge; incoming edge) with respect to the second node 120, ..., 132 and as an "outgoing edge" with respect to the first node 120, ..., 132: "incoming edges" of a node are directed edges whose destination point is the node, "outgoing edges" of a node are directed edges whose origin is the node. In this embodiment, the nodes 120, ..., 132 of the artificial neural network 100 are arranged in layers 110, ..., 113, wherein the layers may have an intrinsic order introduced by the edges 140, 141, 142 between the nodes 120, ..., 132. In particular, the edges 140, 141, 142 may only exist between adjacent layers of nodes.In the illustrated embodiment, there is an input layer 110 that has only nodes 120, 121, 122 without an incoming edge, an output layer 113 that has only nodes 131, 132 without an outgoing edge, and hidden layers 111, 112 between the input layer 110 and the output layer 113. In general, the number of hidden layers 111, 112 can be chosen arbitrarily. The number of nodes 120, 121, 122 in the input layer 110 is usually related to the number of input values ​​of the neural network, and the number of nodes 131, 132 in the output layer 113 is usually related to the number of output values ​​of the neural network. In particular, each node 120, ..., 132 of the neural network 100 can be assigned a (real) number as a value. Here, x means (n)i denotes the value of the i-th node 120, ..., 132 of the n-th layer 110, ..., 113. The values ​​of nodes 120, 121, 122 of the input layer 110 are equivalent to the input values ​​of the neural network 100, and the values ​​of nodes 131 and 132 of the output layer 113 are equivalent to the output values ​​of the neural network 100. In addition, each edge 140, 141, 142 can have a real number as a weight, in particular a real number in the interval [-1, 1] or the interval [0, 1]. Here, w denotes (m,n) i ,j the weight of the edge between the i-th node 120, ..., 132 of the m-th layer 110, ..., 113 and the j-th node 120, ..., 132 of the n-th layer 110, ..., 113. In addition, the abbreviation w (n) i ,j defined as w (n,n+1) i ,j. In particular, to calculate the output values ​​of the neural network, the input values ​​are propagated through the neural network. Specifically, the values ​​of nodes 120, ..., 132 of the (n+1)-th layer 110, ..., 113 can be calculated based on the values ​​of nodes 120, ..., 132 of the n-th layer 110, ..., 113 as follows: x (n+1) j = f( Σ i x (n) i ∙ w (n) i ,j). Here, the function f is a transfer function (also referred to as an “activation function”). Well-known transfer functions are step functions, sigmoid functions (e.g., the logistic function, the generalized logistic function, the hyperbolic tangent, the arctangent function, the error function, the smoothstep function), or rectifier functions. The transfer function is mainly used for normalization purposes. 202218760 21 In particular, the values ​​are propagated layer by layer through the neural network, where values ​​of the input layer 110 are given by the input to the neural network 100, values ​​of the first hidden layer 111 can be calculated based on the values ​​of the input layer 110 of the neural network 100, values ​​of the second hidden layer 112 can be calculated based on the values ​​of the first hidden layer 111, etc. To determine the weights w (m,n) i ,jfor the edges, the neural network 100 must be trained using training data. In particular, training data includes training input data and training output data (denoted by t i ). For a training step, the neural network 100 is applied to the training input data to generate calculated output data. In particular, the training data and the calculated output data have a number of values ​​equal to the number of nodes of the output layer. In particular, a comparison between the calculated output data and the training data is used to recursively adjust the weights in the neural network 100 (backpropagation algorithm). In particular, the weights are changed according to where γ is the learning rate, and the numbers δ (n) j can be calculated recursively as follows: δ (n) j = (Σ k δ (n+1) k ∙ w (n+1) j ,k ) ∙ f'(Σi x (n) i ∙ w (n) i ,j ) based on δ (n+1) j, if the (n+1)-th layer is not the output layer, and δ (n) j = (x (n+1) k - t (n+1) j) ∙ f'(Σ i x (n) i ∙ w (n) i ,j ) 202218760 22 if the (n+1)-th layer is the output layer 113, where f' is the first derivative of the activation function and t (n+1)j is the comparison training value for the jth node of the output layer 113. Fig. 3 shows a flow diagram of an embodiment of the method according to the invention. In a first step 31, two or more different operating ranges of the asynchronous motor are defined as a function of the slip and as a function of the supply frequency. In a second step 32 following the first step 31, at least one associated anomaly detector, which is adapted to detect a fault condition of the asynchronous motor in the respective operating range, is assigned to each of the defined operating ranges. In a third step 33 following the second step 32, a current operating range of the asynchronous motor is determined.In a fourth step 34 following the third step 33, at least one anomaly detector assigned to the current operating range is activated, and the state of the asynchronous motor is monitored by the active anomaly detector. And in a fifth step 35 following the fourth step 34, if an anomaly is detected by the active anomaly detector, a corresponding signal is generated. Fig. 4 shows the definition of the different operating ranges of the asynchronous motor. In steps a), first, a slip value range S = [S. min ; S max ] of values ​​s i of the slip s that occur during operation of the asynchronous motor AM is determined. A frequency value range F = [f min ; f max ] of values ​​f ithe supply frequency f of the voltage that occur during operation of the asynchronous motor AM. In steps b), the determined slip value range S is divided into Q slip subranges S q The determined frequency value range is also divided into R frequency subranges F r In the steps under c), Q x R operating ranges B(q, r) are defined, defined by one of the said slip subranges S q 202218760 23 and one of the said frequency sub-ranges F rFor this purpose, Q x R anomaly detectors A(q, r) are provided, with one anomaly detector A(q, r) being assigned to a corresponding operating area B(q, r) for anomaly detection. Fig. 5 shows the configuration in which several anomaly detectors are responsible for one operating area. In a detection phase, after determining the current operating range B(q, r) of the asynchronous motor AM, activating the anomaly detector A(q, r) associated with the current operating range B(q, r), as well as one or more anomaly detectors A(q-1, r-1), A(q-1, r), A(q-1, r+1), A(q, r-1), A(q, r+1), A(q+1, r-1), A(q+1, r), A(q+1, r+1), the operating ranges B(q-1, r-1), B(q-1, r), B(q-1, r+1), B(q, r-1), B(q, r+1), B(q+1, r-1), adjacent to the current operating range B(q, r). B(q+1, r), B(q+1, r+1) of the electric motor AM.Subsequently, monitoring of the state of the asynchronous motor AM by the active anomaly detectors A(q, r), A(q-1, r-1), A(q-1, r), A(q-1, r+1), A(q, r-1), A(q, r+1), A(q+1, r-1), A(q+1, r), A(q+1, r+1).

Claims

202218760 24 claims 1. Method for monitoring the condition of an asynchronous motor (AM) with a slip (s) to which a supply voltage (U) with a supply frequency (f) is applied, comprising the following steps: - defining two or more different operating ranges (B) of the asynchronous motor (AM) as a function of the slip (s) and as a function of the supply frequency (f); - assigning to each of the defined operating ranges (B), at least one associated anomaly detector (A i-1 , A i , A i+1), which is adapted to detect a fault condition (E) of the asynchronous motor (AM) in the respective operating range (B); - determining a current operating range (B) of the asynchronous motor (AM); - activating the at least one anomaly detector (A) assigned to the current operating range (B), and monitoring the condition of the asynchronous motor (AM) by the active anomaly detector (A); and - in the event of detection of an anomaly by the active anomaly detector (A), generating a corresponding signal.

2. Method according to claim 1, comprising the following steps: - determining a slip value range (S) of values ​​(s i ) of the slip (s) that occur during operation of the asynchronous motor (AM); - Determining a frequency range (F) of values ​​(f i ) of the supply frequency (f) that occur during operation of the asynchronous motor (AM); - subdividing the determined slip value range (S) into Q slip subranges (Sq ); - subdividing the determined frequency value range into R frequency subranges (F r ); - Defining Q x R operating ranges (B(q, r)), defined by one of said slip subranges (S q ) and one of the said frequency sub-ranges (F r ); 202218760 25 - Providing Q x R anomaly detectors (A(q, r)), wherein an anomaly detector (A(q, r)) for anomaly detection is assigned to a corresponding operating range (B(q, r)); - In a training phase, training the anomaly detectors (A(q, r)); - In a detection phase, determining a current operating range (B(q, r)) of the asynchronous motor, activating the anomaly detector (A(q, r)) assigned to the current operating range (B(q, r)), and monitoring the state of the asynchronous motor (AM) by the active anomaly detector (A(q, r)); and - In case of detection of an anomaly by the active anomaly detector (A(q, r)), generating a signal indicating the anomaly by the active anomaly detector (A(q, r)). 3.Method according to claim 2, comprising the following steps: - In a detection phase, after determining the current operating range (B(q, r)) of the asynchronous motor (AM), activating the anomaly detector (A(q, r)) associated with the current operating range (B(q, r)) and one or more anomaly detectors (A(q-1, r-1), A(q-1, r), A(q-1, r+1), A(q, r-1), A(q, r+1), A(q+1, r-1), A(q+1, r), A(q+1, r+1)) associated with operating ranges of the electric motor adjacent to the current operating range (B(q, r)); and - monitoring the state of the asynchronous motor (AM) by the active anomaly detectors (A(q, r), A(q-1, r-1), A(q-1, r), A(q-1, r+1), A(q, r-1), A(q, r+1), A(q+1, r-1), A(q+1, r), A(q+1, r+1)).

4. The method according to claim 1, comprising: - assigning, to each of the defined operating ranges (B. i ), several anomaly detectors (A i-1 , A i , A i+1), whereby detection results of the assigned anomaly detectors (A i-1 , A i , A i+1 ) as a result vector; 202218760 26 - Applying a majority decision to the result vector in order to decide on an anomaly.

5. Method according to one of the preceding claims, wherein determining the current operating range (B) of the asynchronous motor (AM) comprises the following steps: - Determining values ​​(s i ) of the slip (s) and values ​​(f i ) of the supply frequency (f) during a time sequence T; - Based on the determined values ​​of slip (s) and supply frequency (f), determining the current operating range (B) of the asynchronous motor (AM).

6. Method according to claim 5, comprising the following steps: - subdividing the time sequence T into ν subunits of shorter duration T ν , where in each subunit a measure L s,νfor the slip; - activating an anomaly detector (A) if the differences in the dimensions L s,ν not exceed a predetermined difference value, and otherwise refrain from activating an anomaly detector (A).

Images