Method for diagnosing the state of an asynchronous motor

Abstract

The invention relates to a method for diagnosing the state of an asynchronous motor (M) to which a three-phase voltage UR,S,T(t) with a supply frequency (fU) is applied, having the following steps: a) detecting a time series (Ut) of the voltage UR,S,T(t) and a time series (It) of a three-phase motor current IR,S,T(t) which flows through the asynchronous motor (M) as a result of the voltage UR,S,T(t) applied to the asynchronous motor (M); b) calculating, using an FFT, a current spectrum I(f,t) and a voltage spectrum U(f,t) on the basis of the detected time series (It, Ut); c) calculating one or more of the following time-dependent spectral variables 1-6: 1) current components Ia(f,t), Ib(f,t), wherein Ia(f,t) and Ib(f,t) are formed using a Park transformation of IR,S,T; 2) total current amplitude AI(f,t) = √((Ia(f,t))2 + (Ib(f,t))2); 3) phase angle between current components φab(f,t) = arctan(Ia / Ib); 4) phase angle between current and voltage components φUI(f,t) = arctan(U(f,t) / I(f,t)); 5) instantaneous spectral output P(f,t) = U(f,t)*I(f,t); and 6) admittance X(f,t) = I(f,t) / U(f,t); d) analyzing the calculated spectral variables for an anomaly; and if an anomaly is ascertained, reporting the anomaly.

Description

[0001]202223006 1 Description Method for condition diagnosis of an asynchronous motor The present invention relates to a method for condition diagnosis of an asynchronous motor. Due to their cost-effective and robust design, asynchronous motors (= ASM) are very widespread 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 MW range. Asynchronous motors are therefore among the largest consumers of the 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. A well-known method for detecting faults in electric motors is so-calledMotor Current Signature Analysis (MCSA), see e.g. EP3961230A1 (Siemens AG) March 2, 2022 and Kumar, K. Vinoth, et al.: A Review of Voltage and Current Signature Diagnosis in Industrial Drives. International Journal of Power Electronics and Drive Systems (IJPEDS) Volume 1, Number 1, September 2011, pages 75 – 82, ISSN 2088-8694. MCSA is based on the recording of the electrical phases, usually three (L1, L2, L3), of an electric motor. In addition to MCSA, the spectral components of the voltage are used to operate CM (MVSA). Measuring devices such as For example, the current / voltage measurement module of the SIMOCODE motor management system from Siemens AG measures not only the current I but also the voltage U of the electrical phases. The measurements are performed with a high sampling frequency and as synchronously as possible. These measured values ​​are then analyzed.For example, the Fast Fourier Transformation (FFT) is used to detect or quantify faults and operating conditions of the motor in the frequency domain. The 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 for the time-resolved analysis of processes. ASM work 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, meaning that a change in slip changes the frequency position of the fault frequencies. Based on the change in the amplitudes of these characteristic frequencies in the frequency spectrum, motor faults can be detected using MCSA. One task of MCSA is therefore to determine the slip s of asynchronous machines as accurately as possible. One way to determine the slip precisely is to determine the frequency of the so-called principal slot harmonics (PSH), i.e., those frequencies that originate in the structure of the rotor, see the following equation: f. PSH = [k•(R / p)•(1-s) ± ν] • f supply Eq. (1) where f PSH Frequency of the Principal Slot Harmonics f supply Frequency of the supply voltage, e.g. 50 Hz k Natural number 202223006 3 p Number of pole pairs = Number of pole pairs R Number of rotor bars = Number of rotor bars Slip Odd numbers 1, 3, 5, 7, ... = Order of the harmonics of fsupply For this purpose, the PSH frequencies f PSH which are a function f of the number of rotor bars R (e.g. 28, 44, 150, ...), the number of pole pairs p and the slip s (usually between 0 and 0.05 (0-5 %)): f PSH = f(R,p,s), evaluated based on the respective current signal of one or more electrical phases (L1, L2, L3). The evaluation can also be carried out in a spectrum of the reduced current magnitudes D and Q or in a spectrum of the total amplitude of the phase currents. The reduced current magnitudes D and Q are the current magnitudes along the axes d and q of the two-axis coordinate system into which the three-phase quantities along the axes U, V, W are converted by the Park transformation. One problem with this is that the amplitudes A(f PSH ) of the PSH frequencies f PSH outside the mains supply frequency f supply compared to the amplitude A(f supply ) of the supply current are very small: A(f PSH ) < 10 -2 A(f supply). Faults such as broken rotor bars (BB), eccentricities (ECC), bearing faults, the influence of coupled machines, and even the load state of the machine can be detected in the frequency spectrum of the motor current by a change in amplitude at characteristic frequencies. In the MCSA, the amplitudes are evaluated for a wide variety of fault cases: i) Broken rotor bars (BB = Broken Bars): f BB1 = (1 ± 2k•s) • f supply Eq. (2) ii) Static / Dynamic Rotor Eccentricity (= ECC; 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 202223006 4 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 supplyFrequency 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 d Index depending on the eccentricity type (static: n d = 0; dynamic: n d = 1, 2, 3, 4, ...) N b Number of rolling elements D b / D pCharacteristic diameters on the rolling bearing ß Contact angle For some types of faults, knowledge of the number of rotor bars R is necessary. This can be known from the design or estimated using known methods. In reality, however, not only the above-mentioned theoretical fault frequencies often occur, but drive train-specific vibrations can lead to new, mostly unknown frequencies in the current spectrum. This is particularly the case for imbalances, alignment errors, and similar "mechanical" faults, which lead to air gap asymmetries that manifest themselves as harmonics of the rotation frequency and, in theory, therefore, lead to a modulation on the supply frequency. Real drives can exhibit even more complex vibration modes due to natural and resonant vibrations. There are known fault cases in which the theoretical frequencies are insufficient to evaluate the faults and their severity in the current spectrum.An object of the present invention is therefore improved fault detection based on MCSA. This object is achieved according to the invention by a method having the features specified in claim 1. The method according to the invention serves to diagnose the condition of an asynchronous motor to which a three-phase voltage U is applied. R,S,T (t) with a supply frequency. One step of the method comprises recording a voltage time series U t the voltage U R,S,T (t) and a current time series I t a three-phase motor current I R,S,T (t), which is due to the voltage U applied to the asynchronous motor R,S,T (t) flows through the asynchronous motor. A further step involves 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). A further step involves calculating 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 between current components: φ ab (f,t) = arctan(I a / I b ); 4) Phase angle φ UI 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); and 6) admittance X(f,t) = I(f,t) / U(f,t). A further step comprises analyzing the calculated spectral variables for an anomaly. This step “analyzing the calculated spectral variables” can comprise the following steps: forming a feature vector from one or more of the time-dependent spectral variables; and using the formed feature vector as input values ​​of an ML algorithm, ie as training data or as test data. A further step of the method according to the invention comprises reporting the anomaly if an anomaly has been detected. 202223006 6 The invention utilizes known machine learning algorithms to detect the fault condition of an electrical machine based on the frequencies according to the invention. One approach is to use an "unsupervised" anomaly detector.A motor is trained in the "good" state for a specific period of time, i.e., the algorithm in the training phase only sees the "good" state but is not aware of the "bad" state. After this training phase, anomaly detection is activated, i.e., if changes are detected after this time, they are reported to the user. Such an AI requires input values, the so-called "features." These must be obtained from the current samples. In principle, one could view the current samples as "features," but this does not produce good results, as explained in the following points a) to c): a) Very large "feature" vectors: With typical sampling rates of 2 to 20 kSps (Sps = samples per second), a recording duration of approximately 1 second results in feature vectors with 2,000 to 20,000 values. The "information content" of a single value is low. Therefore, the aim here is to "condense" the data, orto increase the entropy. b) Phase position: In the time domain, the problem of "phase position" arises. Depending on the point in time at which the "recording" of the current values ​​begins, a random phase position results. In principle, it is possible to start the "recording" at a specific value, but this results in a "blur" of the phase position. A PLL could also be used, but this is also technically very complex. c) Quantity: The current values ​​are quantitatively dependent on the load, ie, a higher motor current flows at a higher load. However, anomaly detection should not classify a load change as an anomaly, but only as a motor anomaly. The current values ​​must therefore be "normalized."202223006 7 Instead, according to the invention, those features of a spectral amplitude distribution in one or more spectral types are used as features which: i) are as load-independent as possible, but have been found in laboratory tests to be correlated to the error class - e.g. triple winding harmonics in the phase angle between current components φ. ab(f). Alternative features include, for example, ii) error frequencies and their spectral amplitudes / phases from one or more spectral types that can be classically calculated using MCSA formulas. iii) spectral amplitudes / phases of the spectral types that can be determined from a spectrum of selected frequency resolution. The invention circumvents problems of "classical" MCSA, which are explained in the following points a) to d): a) Resolution vs. data length / recording time: On the one hand, good frequency resolution is desired for MCSA. On the other hand, this leads to long recording times during which the system must remain "stationary," and thus to large feature vectors, which quickly leads to the "curse of dimensionality." b) The error frequencies of classical MCSA are insufficient for detecting an existing error in general applications. c) The current amplitudes could be disturbed by voltage artifacts.d) The current and MCSA spectral types are not optimally suited for detecting all faults; sometimes faults can be detected more clearly in other spectral types. The invention uses spectral types, in particular phase relationships, and combinations of slip-independent WH-based features with features calculated according to MCSA formulas (WH = Winding Harmonics). 202223006 8 The inventive approach a) avoids an excessively high frequency resolution for the features (WH are slip-independent) and b) minimizes amplitude influences by using phase relationships, e.g., phase angle φ. ab (f,t) between current components: if the current components I a and I ban artifact of the voltage, this is divided out. A typical time resolution according to the invention is 1 s; therefore the feature spectra have a frequency resolution of 1 Hz. A constant speed must be present within this period; sufficiently steady-state operation exists if at least one of the following three conditions (i) to (iii), which can be read from the recorded data, is met: (i) provided that the efficiency factor cos φ of the motor does not change too significantly during this time; e.g., this condition is considered met if the efficiency factor cos φ of the motor changes by less than 3 times the standard deviation of the efficiency factor. ii) provided that the slip lines, i.e. spectral lines with slip such as PSH, ECC, BB, have a narrow width; e.g., this condition is considered met if the width of the slip lines is less than 3 times the frequency resolution.iii) provided that the slip line variance lies within df = 1 Hz. The variance is a measure of dispersion that characterizes the distribution of values ​​around the mean. It is the square of the standard deviation. The variance is calculated by dividing the sum of the squared deviations of all measured values ​​from the arithmetic mean by the number of measured values. For anomaly detector features according to i), the frequency resolution can be greater, since the WH frequency position does not change with the slip, i.e., more dynamic drives can be qualified and a significantly reduced computational effort is necessary. The evaluation of only the WH frequencies can also be advantageously applied in one or more other spectral types. The evaluation according to i), i.e., the features of an algorithm, e.g.An anomaly detector or a classification according to WH frequencies (in any spectrum types or combinations thereof) is based on the knowledge that errors which, according to classic MCSA, manifest themselves in air gap variations, depending on slip, due to the rotor passing frequency inducing a stator current, also manifest themselves in a superimposed manner in the mains frequency multiples (WH frequencies). Advantageous embodiments and further developments of the invention are specified in the dependent claims. According to a preferred embodiment of the invention, the amplitudes of the calculated spectral variables are normalized to the amplitude at the supply frequency. According to a preferred embodiment of the invention, the spectral variables are determined with a frequency resolution in the range of up to approximately 0.01 to 5 Hz, preferably 1 Hz, and a time resolution of approximately 1 s.According to a preferred embodiment of the invention, the current amplitudes and / or spectral quantities at the theoretical frequencies are determined after determining the slip s from PSH frequencies, knowing the number of rotor bars R, and using the determined amplitudes as a feature set. This type of feature is referred to as an MCSA feature set—of a specific spectral type. According to a preferred embodiment of the invention, the current amplitudes and / or spectral quantities at all frequencies are used as a feature set. The term "all frequencies" can include a so-called "MAX binning" to achieve a reduction in the number of frequencies. MAX binning with a binning length N. bin , where in N bin Samples of an FFT with N FFT frequencies the maximum as the new amplitude of the N bin-th frequency bins. This can also be carried out with an overlap of the bins. The inventive approach significantly reduces the dimensionality of the feature space through binning. According to a preferred embodiment of the invention, only those features with the measured amplitude are displayed which stand out significantly from the respective noise background. Using what is known as significance analysis, the environment of a maximum in a frequency band can be examined and, for example, the standard deviation s of the measured values ​​around the detected peak can be determined. The determined peak value must be more than N xs greater than the environment to be considered a significant peak; where N is an integer in the value range [1, ..., 6]. The environment can thus be set to a fixed smallest value, oriented towards the noise limit of the sensor signal chain, e.g., -100 dB.As a result, changed noise values ​​at non-significant frequencies do not result as "anomaly" values. New peaks that emerge from the noise, however, appear as new frequency peaks and are therefore used for error detection. The inventive approach significantly reduces the dimensionality of the feature space through significance analysis. In particular, the significance analysis avoids false alarms in anomaly detection, which could result from noise values ​​of non-significant frequencies that have changed compared to the learned state. According to a preferred embodiment of the invention, the amplitudes of the winding harmonics, i.e. the harmonics of the supply frequency f, are determined. supply, as a set of features. These can be calculated individually or as triples (3n), positive (3n+1) or negative (3n-1), with harmonics of the fundamental frequency. Alternatively, a distortion factor of the harmonics can be determined in relation to the fundamental frequency of the supply voltage. These WH-202223006 11 amplitudes, normalized to the fundamental frequency, have the advantage that their frequency position is independent of slip, meaning that error detection or anomaly deviation can be carried out largely independently of the operating point on the basis of these features. The invention is explained below with the aid of the accompanying drawings. They show schematically and not to scale: Fig. 1 shows a first embodiment of an inventive device for condition monitoring of a drive train; Fig. 2 shows a second embodiment of an inventive device for condition monitoring of a drive train; Fig.3 shows a flow diagram of an embodiment of the method according to the invention for the condition diagnosis of an asynchronous motor; Fig. 4 shows a typical MCSA-based amplitude spectrum; Fig. 5 shows the distortion factor as a load-dependent feature; Fig. 6 shows triple winding harmonics as a function of the torque; Fig. 7 shows a plot of the positive, negative and triple WH amplitudes of the current for different load levels; Figs. 8A and 8B show exemplary evaluations of two different asynchronous motors; Figs. 9A, 9B and 9C show confidence matrices of an exemplary supervised classifier; 202223006 12 Fig. 10A shows the spectral type of current amplitude for different set shaft alignment states; Fig. 10B shows the spectral type of admittance for different set shaft alignment states; Fig. 10C spectral type phase angles for different wave alignment conditions; Fig. 11 exemplary anomaly detector results for the detection of misalignments; and Fig.12 the spectrum type phase angle between current components of the triple WH. Fig. 1 shows an arrangement with an asynchronous motor M, which is connected via supply lines 11 to a voltage source 10, which has a supply voltage U. supply with a voltage frequency f supply supplies, is electrically connected. The current I flowing through the supply lines 11 is measured by a current sensor 14, e.g., a shunt, a current transformer, or a Hall sensor. The voltage U applied to the supply lines 11 is measured by a voltage sensor 15, e.g., a shunt. Current and voltage values ​​are measured synchronously with a sampling frequency f A [1 / T] for a duration T A [T] is recorded. The number of samples N A is therefore N A = f A *T A For example, let f A = 3200 Hz and T A = 1.28 seconds, resulting in N A =4096 = 2 12Sampled values, one for current and one for voltage. The current and voltage measured values ​​recorded by the sensors 14, 15 are sent from the sensors 14, 15 to a computing unit 16. The method steps for calculating, using an FFT, a current spectrum, calculating one or more time-dependent spectral quantities, analyzing the calculated spectral quantities, and generating a message informing about an anomaly are 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 carrying out the method steps. This software is executed by the processor 18. Input values ​​can be transferred to the computing unit 16 by an input / output unit 22 connected to the computing unit, e.g., a PC.After detecting an anomaly, the computing unit 16 sends a corresponding message to the input / output unit 22. Fig. 2 shows an alternative arrangement in which the asynchronous motor M is designed as a drive machine of a drive train D, which also has a mechanical gearbox G and a work machine W. A torque provided by the asynchronous motor M is transmitted via the gearbox G to the work machine W. The work machine can be, for example, a conveyor belt, a roller or a cylinder. In addition to the asynchronous motor M of the drive train D, further electrical loads 24 are connected to the power grid N. Fig. 3 shows a flow diagram of an embodiment of the method according to the invention for diagnosing the condition of an asynchronous motor to which a three-phase voltage U. R,S,T (t) with a supply frequency f U T current values ​​I tof an electric current I flowing through the electric motor M, which flows through the asynchronous motor M due to the supply voltage U applied to the asynchronous motor M. This gives a time series I(1...T) of T current values ​​I t , while the asynchronous motor M is in good condition. During the recording phase, current values ​​I t recorded, which describe the "good condition" of the motor. These current values ​​I t are subsequently used for training the anomaly detector. The following aspects are very important: 202223006 14 In a first step 310, a voltage time series U t the voltage U R,S,T (t) and a current time series I t a three-phase motor current I R,S,T (t) which is determined by the voltage U applied to the asynchronous motor M R,S,T(t) flows through the asynchronous motor M. In a second step 320, a current spectrum I(f,t) and a voltage spectrum U(f,t) are calculated using an FFT based on the recorded time series I t , U t In a third step 330, one or more of the following time-dependent spectral quantities 330.1 to 330.6 are calculated: 330.1) The current components I a (f,t), I b (f,t) is calculated, where I a (f,t) and I b (f,t) by Park transformation from I R,S,T are formed; 330.2) The total current amplitude A I (f,t) = √((I a (f,t)) 2 + (I b (f,t)) 2 )calculated. 330.3) The phase angle φ ab between the current components: φ ab (f,t) = arctan(I a / I b ). 330.4) The phase angle φ UI between the current and voltage components: φ UI(f,t) = arctan(U(f,t) / I(f,t)). 330.5) The instantaneous spectral power P(f,t) = U(f,t)*I(f,t) is calculated. 330.6) The admittance X(f,t) = I(f,t) / U(f,t) is calculated. In a fourth step 340, the calculated spectral variables are analyzed for anomalies. This fourth step 340, “Analyzing the calculated spectral variables,” may comprise the following steps: forming a feature vector from one or more of the time-dependent spectral variables; and using the formed feature vector as input values ​​for an ML algorithm, i.e., as training data or as test data. In a fifth step 350, if an anomaly is detected, the anomaly is reported. Fig. 4 shows a typical MCSA-based amplitude spectrum (amplitude as a function of frequency) of a current signal, before (crosses) and after significance analysis (solid line).To reduce the noise of the FFT features, the amplitudes of the non-significant frequencies are set to 10. -12 Furthermore, the amplitudes of the so-called winding harmonics, i.e. the harmonics of the supply frequency f supply, can be determined as features. These can be calculated individually or as triple WH (3n), positive WH (3n+1) or negative WH (3n-1), as harmonics of the fundamental frequency. Alternatively, a distortion factor of the harmonics can be determined in relation to the fundamental frequency of the supply voltage. These WH amplitudes, normalized to the fundamental frequency, have the advantage that their frequency position is independent of slip, meaning that fault detection or anomaly deviation can be carried out on the basis of these features largely independently of the operating point. Fig. 5 shows the distortion factor (THD = Total Harmonic Distortion) as a load-dependent feature for a well-aligned drive train "Healthy" and two increasingly severe misalignments "Misalign1", "Misalign2" as a function of torque. Distortion factor and THD represent the ratio of unwanted harmonic / nonlinear distortion to the original signal.The THD of the states are separable because the scatter widths or variances for each torque are smaller than the distance between neighboring states, but they are load-dependent; rising separation lines would be inscribed. Fig. 6 shows triple winding harmonics as a function of torque. At the phase angle φ. abBetween current components, the triple winding harmonics can be seen as a largely load-independent feature from approximately 10 Nm (only minor changes are observed with the torque (load)) for a well-aligned drive train "Healthy" and two increasingly severe misalignments "Misalign1" and "Misalign2" as a function of the torque. In particular, the values ​​of the severe misalignment "Misalign2" are above the other two states from approximately 10 Nm 202223006 16, and this without a noticeable increase with the load level. The triple winding harmonics feature can therefore serve as a distinguishing feature regardless of the exact load. Fig. 7 shows a plot of the positive, negative, and triple WH amplitudes of the current for 7 different load levels, with the load initially increasing gradually to the next higher level and then decreasing gradually to the next lower level.Curve A0 indicates the amplitude at a mains frequency of 50 Hz. The plot shows that the triple WH amplitudes in particular are largely load-independent. Consequently, the quantities derived from the triple WH amplitudes also show only a slight load dependence, so that if the exact load state has not been determined beforehand, they can be used preferentially as features for detecting faults. Fig. 8A and 8B show example evaluations of two different asynchronous motors connected to loads via shaft couplings. Fig. 8A relates to an asynchronous motor with a number of pole pairs p=1 and a number of rotor bars R=20. Fig. 8B relates to an asynchronous motor with a number of pole pairs p=2 and a number of rotor bars R=28. In each case, a good state (OK) with ideal alignment is considered in comparison to two states with increasingly severe misalignments (ALGN1, ALGN2).The respective current spectra (samples over frequency) as well as the identifiable confidence matrices (predicted label vs. true label) are shown using the label (supervised learning) for a) current amplitudes (“Currents”) and b) voltage amplitudes (“Voltage”). The confidence matrices using the voltage spectral type show almost perfect classifiability. Figs. 9A, 9B, and 9C show confidence matrices of an exemplary supervised classifier for ball bearings with varying degrees of damage (true label and predicted label, each ranging from 0 = OK = “good condition” to 9 = “end of life”) for various spectrum types. Fig. 9A, with the spectrum type "FFT Current Spectrum I_Spec," shows a confidence matrix relating to the current I(f,t). Fig. 9B, with the spectrum type "FFT Phase Iα, Iβ," shows a confidence matrix relating to the phase angle φ. ab(f,t) between current components. Fig. 9C with spectrum type "Best Accuracy Spec" shows a confidence matrix that uses "all" available spectral types 1) to 6), which are listed below: 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 between current components: φ ab (f,t) = arctan(I a / I b ) 4) Phase angle φ UI 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) From 9A to 9C, increasingly better separabilities are shown, and the overall classification accuracy increases from 85.1% to 93.6% and 95.9% respectively. Fig. 10A, 10B and 10C show spectral types for differently set wave alignment states (OK = good wave alignment = “good state”; ALGN = misalignment). Fig. 10A shows the spectral type current amplitude I(f,t). Fig. 10A shows the spectral type admittance X(f,t). Fig. 10A shows the spectral type phase angle φ ab(f,t). Within each state, the load increases in steps between 0% and 120% of the nominal load. The main fault frequency is approximately 175 Hz (slip-dependent) - see the outlined area. In the OK1 state, a voltage disturbance of 10 Hz is superimposed. The voltage disturbance is suppressed in the X and phase spectrum types and is barely measurable or visible. A particular advantage is that for small loads (in each state block on the left), better separation is achieved. For larger loads, the 175 Hz disturbance line deviates increasingly from the 202223006 18 slip-affected fault frequency anyway. Advantage: The frequency resolution for the X and phase spectrum types can be selected to be larger, and an anomaly detector would not falsely learn the disturbance lines as a characteristic of the good case, since they are not measurably present. Fig.Figure 11 shows exemplary anomaly detector results for the detection of misalignments (top: DataScenario = 2 with misalignment states, see Figure 8A; bottom: DataScenario = 3 with misalignment states, see Figure 8B), whereby only the WH current amplitudes (“WHonly”) are used in the learning and detection phases. The advantage here: no slip dependence of the spectral lines to be evaluated. The frequency resolution can be greater, more dynamic drives can be qualified, and only a significantly reduced computational effort is required. The evaluation of only the WH frequencies can also be advantageously applied in one or more other spectral types. Figure 12 shows the spectrum type phase angle between current components φ. ab(f,t) of the triple WH at 450 Hz for a broken-bar scenario for the states OK, BB1, BB1.5, and BB2 (OK = good state; BB = broken bar; BB1 = one motor rotor bar separated; BB1.5 = one motor rotor bar separated, a second motor rotor bar partially separated; BB2 = two motor rotor bars separated). Within each state, the motor load increases from left to right. The purely electrically altered conditions of the rotor, by separating one or two rotor bars, result in a significantly changed phase position of the triple WH from approximately 50° to 25°.

Claims

202223006 19 claims 1. Method for diagnosing the condition of an asynchronous motor (M) to which a three-phase voltage U R,S,T (t) with a supply frequency (f U ) is applied, with the following steps: a) Recording a voltage time series (U t ) of the voltage U R,S,T (t) and a current time series (I t ) of a three-phase motor current I R,S,T (t), which is due to the voltage U applied to the asynchronous motor (M) R,S,T (t) flows through the asynchronous motor (M); 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) Calculating 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) = √((Ia (f,t)) 2 + (I b (f,t)) 2 ) 3) Phase angle between current components φ ab (f,t) = arctan(I a / I b ) 4) Phase angle 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) d) analyzing the calculated spectral quantities for an anomaly; and e) upon detection of an anomaly, reporting the anomaly.

2. The method according to claim 1, comprising the following step: normalizing the amplitudes of the calculated spectral quantities to the amplitude at the supply frequency.

3. The method according to any one of the preceding claims, comprising the following step: 202223006 20 Determining the spectral quantities with a frequency resolution in the range of up to approximately 0.01 to 5 Hz, preferably 1 Hz, and a time resolution of approximately 1 s.

4. Method according to one of the preceding claims, comprising the following steps: Determining the amplitudes of the current and / or the spectral quantities at the theoretical frequencies after determining the slip s from PSH frequencies with knowledge of the rotor bar number R, and using the determined amplitudes as a feature set.

5. Method according to one of the preceding claims, comprising the following step: Using the amplitudes of the current and / or the spectral quantities at all frequencies as a feature set.

6. Method according to one of the preceding claims, comprising the following step: Displaying only those features with the measured amplitude that stand out significantly from the respective noise background. 7.Method according to one of the preceding claims, comprising the following step: determining the amplitudes of the winding harmonics, i.e. the harmonics of the supply frequency f. supply , as a set of features.

Images