Diagnostic device and diagnostic method for electric motors

Abstract

A diagnostic device (100) for a motor (5) comprises a measurement circuit (7) for acquiring current and voltage signals, an FFT analysis unit (111) for frequency-analyzing the acquired current and voltage signals and outputting current analysis results and voltage analysis results, a noise determination unit (112), a noise removal unit (113), and a spectrum peak extraction unit (114), and diagnoses abnormalities in the motor (5) based on the extracted diagnostic spectrum peaks. The noise determination unit (112) performs noise determination by comparing a first spectrum appearing in the current analysis result with a second spectrum corresponding to the first spectrum within the voltage analysis result, the noise removal unit (113) removes noise components from the current analysis result, and the spectrum peak extraction unit (114) extracts diagnostic spectrum peaks based on the current analysis result after noise removal.

Description

Technology Field

[0001] The present invention relates to a diagnostic device and a diagnostic method for an electric motor. Background Technology

[0002] To diagnose an abnormality in a motor, a diagnostic support device, which is a conventional motor diagnostic device, comprises an amplitude calculation unit that acquires a q-axis current from a power converter controlling the motor and calculates the amplitude for each frequency related to the q-axis current from the acquired q-axis current, and a display unit that displays the calculation result of the amplitude calculation unit. The display unit displays the change over time of the frequency spectrum of the q-axis current, which is the calculation result of the amplitude calculation unit, and supports the diagnosis of the abnormality (e.g., Patent Document 1). Prior art literature

[0003] Patent Document 1: International Patent Publication WO2019 / 198563 The problem to be solved

[0004] In the prior art described in the above patent document 1, an abnormality diagnosis of a motor is performed using frequency analysis of the current, and DC components or noise components are removed in advance using a filter or the like. However, it is difficult to reliably remove various noise components, and it is difficult to reliably extract frequency components for diagnosis due to residual noise components.

[0005] The present invention discloses a technology for solving the above-mentioned problems, and aims to provide a diagnostic device and a diagnostic method for an electric motor that diagnose motor abnormalities with high reliability and precision by preventing the influence of noise. means of solving the problem

[0006] The diagnostic device for an electric motor disclosed herein comprises: a signal acquisition unit that detects a current flowing through the electric motor and a voltage applied to the electric motor to acquire a current signal and a voltage signal, respectively; a frequency analysis unit that performs frequency analysis on the acquired current signal and the voltage signal to output a current analysis result and a voltage analysis result, respectively; a noise determination unit that performs a noise determination by comparing a first spectrum, which is a spectrum component of a spectrum peak appearing in the current analysis result, with a second spectrum, which is a spectrum component corresponding to the first spectrum within the voltage analysis result; a noise removal unit that removes a noise component from the current analysis result based on the determination result of the noise determination unit; and a spectrum peak extraction unit that extracts a diagnostic spectrum peak based on the result of removing the noise component from the current analysis result, and diagnoses an abnormality of the electric motor based on the extracted diagnostic spectrum peak.

[0007] In addition, the diagnostic method for an electric motor disclosed herein comprises a signal acquisition step for acquiring a current signal and a voltage signal by detecting a current flowing through the electric motor and a voltage applied to the electric motor, respectively; a frequency analysis step for outputting a current analysis result and a voltage analysis result by frequency analyzing the current signal and the voltage signal, respectively; a noise determination step for performing a noise determination by comparing a first spectrum, which is a spectrum component of a spectrum peak appearing in the current analysis result, with a second spectrum, which is a spectrum component corresponding to the first spectrum within the voltage analysis result; a noise removal step for removing a noise component from the current analysis result based on the determination result by the noise determination step; and a spectrum peak extraction step for extracting a diagnostic spectrum peak based on the result of removing the noise component from the current analysis result by the noise removal step, and diagnosing an abnormality of the electric motor based on the extracted diagnostic spectrum peak. Effects of the invention

[0008] According to the electric motor diagnostic device and diagnostic method disclosed herein, it is possible to diagnose abnormalities in an electric motor with high reliability and precision by preventing the influence of noise. Brief explanation of the drawing

[0009] FIG. 1 is a schematic diagram showing a main circuit and a diagnostic device including a motor according to Embodiment 1. FIG. 2 is a block diagram showing the main parts of a diagnostic device according to Embodiment 1. FIG. 3 is a flowchart explaining the operation of the motor in its initial state in the diagnostic device according to Embodiment 1. FIG. 4 is a flowchart explaining the operation during abnormal diagnosis of a motor in a diagnostic device according to Embodiment 1. FIG. 5 is a spectrum waveform after frequency analysis, illustrating the operation of the motor in its initial state in the diagnostic device according to Embodiment 1. FIG. 6 is a spectrum waveform after frequency analysis, illustrating the operation during abnormal diagnosis of a motor in a diagnostic device according to Embodiment 1. FIG. 7 is a configuration diagram showing an example of hardware that realizes each function of the computation processing unit according to Embodiment 1. FIG. 8 is a configuration diagram showing another example of hardware that realizes each function of the operation processing unit according to Embodiment 1. FIG. 9 is a drawing illustrating a noise removal unit according to embodiment 2. Specific details for implementing the invention

[0010] Embodiment 1.

[0011] Hereinafter, embodiments will be described with reference to the drawings.

[0012] FIG. 1 is a schematic diagram showing a main circuit and a diagnostic device including a motor according to Embodiment 1. In this case, the wiring circuit that supplies power to the motor is referred to as the main circuit.

[0013] As shown in FIG. 1, a main circuit (1) introduced from a power system is provided with a circuit breaker (2), an electromagnetic contactor (3), and an instrument transformer (4). In the instrument transformer (4), the load current flowing from the main circuit (1) to the motor and the voltage applied to the motor (5) are detected. A motor (5), such as a three-phase induction motor, which is a load, is connected to the main circuit (1), and a mechanical equipment (6) is driven by the motor (5).

[0014] The diagnostic device (100) comprises a measurement circuit (7) as a signal acquisition unit that acquires current and voltage signals from the output of an instrument transformer (4), and a computation processing unit (8) that performs the diagnosis of an abnormality of the motor (5) using the current and voltage signals input from the measurement circuit (7). Additionally, the abnormality of the motor (5) is diagnosed including an abnormality of the mechanical equipment (6).

[0015] Additionally, the diagnostic device (100) is provided with a rating information setting circuit (10) for pre-setting necessary rating information, such as the power frequency of the motor (5), the rated output of the motor (5), the rated current, the number of poles, and the rated rotational speed, and a setting information memory circuit (9) for storing the rating information set by the rating information setting circuit (10). The rating information is information that can be easily obtained from the manufacturer's catalog of the motor (5) or from a nameplate mounted on the motor (5).

[0016] When diagnosing an abnormality of the motor (5) using the current signal and voltage signal input from the measurement circuit (7), the calculation processing unit (8) performs calculation processing by referring to the rated information in the setting information memory circuit (9).

[0017] Additionally, the motor (5) to be diagnosed may be multiple units, in which case the rated information of the multiple motors (5) is set in advance and maintained in the setting information memory circuit (9). The measurement circuit (7) acquires current signals and voltage signals from the instrument transformer (4) of each motor (5) and outputs them to the calculation processing unit (8). For each motor (5), the calculation processing unit (8) refers to the rated information of each motor (5) within the setting information memory circuit (9) and performs an abnormality diagnosis process of the motor (5) using the current signals and voltage signals.

[0018] In the following description, the diagnosis of an abnormality in one motor (5) will be explained.

[0019] The diagnostic device (100) further comprises a display unit (11), a driving circuit (12), an output circuit (13), and a communication circuit (14), each connected to a computation processing unit (8).

[0020] The display unit (11) displays the abnormal state or alarm of the motor (5), which is the result output of the abnormal diagnosis processing by the calculation processing unit (8), and also displays detected physical quantities such as current signals and voltage signals. The driving circuit (12) outputs a control signal to open and close the electronic contactor (3) based on the result calculated by the calculation processing unit (8). The output circuit (13) outputs the abnormal state or alarm of the motor (5), which is the result output of the abnormal diagnosis processing by the calculation processing unit (8), to the outside.

[0021] The external monitoring device (200) is configured as a PC (personal computer), etc., and is connected to one or more diagnostic devices (100). It appropriately receives information from the computation processing unit (8) through the communication circuit (14) and monitors the operating status of the diagnostic device (100). The connection between the external monitoring device (200) and the communication circuit (14) of the diagnostic device (100) may be made using a cable or wirelessly. A network may be established between the multiple diagnostic devices (100) and the connection may be made via the Internet.

[0022] Additionally, the diagnostic device (100) may have parts other than the main parts, such as the computational processing unit (8) and the measurement circuit (7), configured as external components. Furthermore, by providing a signal acquisition unit within the computational processing unit (8) and configuring it to acquire a signal from the measurement circuit (7), the diagnostic device (100) can be configured with parts other than the computational processing unit (8) configured as external components.

[0023] FIG. 2 is a block diagram showing the main parts of a diagnostic device (100).

[0024] As shown in FIG. 2, the computation processing unit (8) comprises a load rate calculation unit (110) that calculates the load rate of the motor (5), an FFT analysis unit (111) that performs frequency analysis of the current signal and voltage signal input from the measurement circuit (7) by FFT (Fast Fourier Transform), a noise determination unit (112) that performs noise determination based on the analysis result, and a noise removal unit (113) that removes noise components. Additionally, the computation processing unit (8) comprises a spectrum peak extraction unit (114) for extracting a diagnostic spectrum peak, a correction processing unit (115) for correcting a diagnostic spectrum peak, a reference value storage unit (116) for storing a diagnostic spectrum peak after correction in the initial state of the motor (5) in normal condition as a reference value, a diagnostic value storage unit (117) for storing a diagnostic spectrum peak after correction in the case of abnormal diagnosis of the motor (5) as a diagnostic value, and a judgment unit (118) for making a good or bad judgment by comparing the diagnostic value with the reference value.

[0025] In addition, in this embodiment, the diagnostic device (100) collects information (reference value) of the motor (5) in its initial state as a normal motor (5) and maintains it in the reference value storage unit (116). Then, when diagnosing an abnormality of the motor (5), it collects information (diagnostic value) of the motor (5) and diagnoses the motor (5) by comparing this diagnostic value with the maintained reference value.

[0026] For this reason, the load rate calculation unit (110), FFT analysis unit (111), noise determination unit (112), noise removal unit (113), spectrum peak extraction unit (114), and correction processing unit (115) within the calculation processing unit (8) each operate when collecting information on the initial state of the motor (5) and when collecting information on the abnormality diagnosis of the motor (5).

[0027] Current and voltage signals from the measurement circuit (7) are input to the operation processing unit (8).

[0028] The load rate calculation unit (110) calculates the load rate of the motor (5) based on the current signal and the voltage signal.

[0029] The FFT analysis unit (111) performs FFT analysis on the current signal and voltage signal, respectively, to extract spectrum peaks and outputs the current analysis result and voltage analysis result.

[0030] The noise determination unit (112) compares the current analysis result and the voltage analysis result, which are the outputs of the FFT analysis unit (111), and performs noise determination for each spectrum.

[0031] The noise removal unit (113) removes noise components from the current analysis result based on the determination result of the noise determination unit (112).

[0032] The spectrum peak extraction unit (114) extracts a diagnostic spectrum peak related to the diagnosis of an abnormality of the motor (5) from the spectrum peak in the current analysis result after removing the noise component in the noise removal unit (113).

[0033] The correction processing unit (115) corrects the diagnostic spectrum peak extracted by the spectrum peak extraction unit (114) according to the load rate calculated by the load rate calculation unit (110).

[0034] The load rate calculation unit (110), FFT analysis unit (111), noise determination unit (112), noise removal unit (113), spectrum peak extraction unit (114), and correction processing unit (115) described above operate in both the initial state information collection of the motor (5) and the abnormality diagnosis information collection of the motor (5), as described above. Accordingly, the output of the correction processing unit (115) becomes a reference value when collecting information on the initial state of the motor (5), and becomes a diagnosis value when collecting information on the abnormality diagnosis of the motor (5).

[0035] The reference value storage unit (116) stores the diagnostic spectrum peak (reference value) after correction in the initial state of the motor (5), and the diagnostic value storage unit (117) stores the diagnostic spectrum peak (diagnostic value) after correction during the diagnosis of an abnormality of the motor (5).

[0036] The judgment unit (118) diagnoses an abnormality of the motor (5) by making a judgment of good or bad based on the diagnostic value generated during the diagnosis of an abnormality of the motor (5) by referring to the reference value in the reference value storage unit (116).

[0037] The operation of the diagnostic device (100) configured as described above is explained below based on the flowchart and spectrum diagram.

[0038] FIG. 3 is a flowchart explaining the operation of the diagnostic device (100) in the initial state of the electric motor (5).

[0039] FIG. 4 is a flowchart explaining the operation of the diagnostic device (100) during the diagnosis of an abnormality of the electric motor (5).

[0040] FIG. 5 is a spectrum waveform after frequency analysis, illustrating the operation of the diagnostic device (100) in the initial state of the electric motor (5).

[0041] FIG. 6 is a spectrum waveform after frequency analysis, which explains the operation of the diagnostic device (100) when diagnosing an abnormality of the electric motor (5).

[0042] First, the operation of the diagnostic device (100) in the initial state of the electric motor (5) will be explained based on FIG. 3 and FIG. 5.

[0043] The diagnostic device (100) collects information on the initial state of the motor (5), that is, the initial output flow shown in FIG. 3, at predetermined intervals.

[0044] When the initial output flow is initiated, the diagnostic device (100) detects that the motor (5) is in an operating state (step S101), and the measurement circuit (7) acquires a current signal due to the load current flowing through the motor and a voltage signal due to the voltage applied to the motor (5). For example, if the motor (5) is a three-phase induction motor, at least one phase current signal and voltage signal are acquired (step S102).

[0045] Next, the load rate calculation unit (110) calculates the load rate of the motor (5) based on the current signal and the voltage signal (step S103).

[0046] Next, the FFT analysis unit (111) performs frequency analysis by FFT analyzing the current signal and the voltage signal, respectively, to derive current analysis results and voltage analysis results including spectrum waveforms (step S104), and normalizes the obtained current analysis results and voltage analysis results (step S105). Subsequently, the FFT analysis unit (111) detects all spectrum peaks from each of the current analysis results and voltage analysis results after normalization.

[0047] By this, current analysis results and voltage analysis results including the spectrum waveform and spectrum peaks after normalization processing are generated (step S106).

[0048] Spectral waveforms based on current analysis results and spectral waveforms based on voltage analysis results are shown in the left and center diagrams of FIG. 5, respectively. The dotted lines in the diagrams represent noise components. Additionally, the right diagram of FIG. 5 will be described later.

[0049] In the current analysis results shown in the left side of Fig. 5, the following spectrum having spectral peaks appears. Spectrum A1 of the power frequency f0 and harmonic spectrum A2 of integer multiples other than 1 of the power frequency f0 appear. Additionally, at frequencies (f0±fr) shifted by the rotation frequency fr on both sides of spectrum A1, sideband spectra B1 and B2 of spectrum A1 appear, and noise spectra N1, N2, N3, and N4, which are noise components, also appear. In this case, since the noise spectrum N2 is superimposed on the sideband spectrum B1, the superimposed spectrum BN appears, and the noise spectrum N2 and the sideband spectrum B1 are not distinguishable.

[0050] The sideband spectra B1 and B2 are used for diagnosing abnormalities because they increase when there is an abnormal sign in the motor (5), and k1 and k2 appear as positive integers with the frequency |k1·f0±k2·fr|. In this case, for simplicity, the harmonic spectrum A2 only shows the frequency 2f0, which is twice the power frequency f0, and for the sideband spectra B1 and B2, only the frequency (f0±fr) is shown.

[0051] In the voltage analysis results shown in the center of Fig. 5, the following spectra with spectral peaks appear. Spectrum A1a of the power frequency f0 and harmonic spectrum A2a of integer multiples (in this case, 2 times) of the power frequency f0 other than 1 appear. Additionally, noise components such as noise spectra N1a, N2a, N3a, and N4a appear. The noise spectra N1a, N2a, N3a, and N4a appear in correspondence with the noise spectra N1, N2, N3, and N4 obtained from the analysis results by current FFT.

[0052] In the spectrum waveform diagram based on the voltage analysis results, the sideband spectrum associated with the rotation frequency fr does not appear.

[0053] If the spectrum resulting from the current analysis is set as the first spectrum and the spectrum resulting from the voltage analysis is set as the second spectrum, it can be seen that the second spectrum A1a, A2a, N1a, N2a, N3a, and N4a appear corresponding to the first spectrum A1, A2, N1, BN, N3, and N4, respectively.

[0054] We will return to the initial output flow shown in Figure 3 and continue the explanation.

[0055] After the current analysis result and voltage analysis result including the spectrum waveform and spectrum peak after normalization processing are generated, the noise determination unit (112) compares the current analysis result and voltage analysis result to determine the noise of each spectrum.

[0056] The operation of the noise determination unit (112) is described in detail with reference to FIG. 5.

[0057] The noise determination unit (112) compares the first spectrum A1, A2, N1, BN, N3, N4 based on the current analysis result with the second spectrum A1a, A2a, N1a, N2a, N3a, N4a based on the voltage analysis result. Then, it determines whether the corresponding first spectrum A1, A2, N1, BN, N3, N4 and the second spectrum A1a, A2a, N1a, N2a, N3a, N4a commonly contain a noise component (hereinafter, common noise component).

[0058] The noise determination unit (112) collects the spectral peaks of both the first spectrum A1, A2, N1, BN, N3, N4 and the second spectrum A1a, A2a, N1a, N2a, N3a, N4a in a time series. Then, it determines whether the variation trend of the dispersion of the spectral peaks matches the corresponding first spectrum A1, A2, N1, BN, N3, N4 and second spectrum A1a, A2a, N1a, N2a, N3a, N4a. Then, for the matching first spectrum N1, BN, N3, N4 and second spectrum N1a, N2a, N3a, N4a, it determines that they contain a common noise component.

[0059] In this case, the standard deviation of the spectrum peaks is observed as a dispersion. The noise determination unit (112) observes the standard deviation σ1 of the spectrum peaks of each first spectrum A1, A2, N1, BN, N3, N4 and the standard deviation σ2 of the spectrum peaks of each second spectrum A1a, A2a, N1a, N2a, N3a, N4a, respectively. Then, if the fluctuation trends of the standard deviations σ1 and σ2 of both the corresponding first spectrum and second spectrum match, it is determined that there is a common noise component. Here, the fluctuation of both standard deviations σ1 and σ2 exceeds the set value, for example, a rapid increase, and the fluctuation trends of the dispersion of the spectrum peaks match.

[0060] However, the spectral peaks in spectra that are integer multiples of the power frequency f0, such as spectra A1 and A1a and harmonic spectra A2 and A2a, hardly fluctuate. Furthermore, the spectral peaks of sideband spectra B1 and B2 also do not fluctuate significantly. In contrast, the noise spectra N1–N4 and N1a–N4a, which appear due to the influence of power noise, etc., show large fluctuations in their spectral peaks.

[0061] Furthermore, the noise spectra N1, N2, N3, and N4 in the current analysis results and the noise spectra N1a, N2a, N3a, and N4a in the voltage analysis results appear correspondingly, and the spectral peaks of the corresponding noise spectra fluctuate in the same way. In addition, in this case, the fluctuation of the spectral peak of noise spectrum N2 does not appear directly, but appears as a fluctuation of spectrum BN.

[0062] As described above, the noise determination unit (112) observes the standard deviation σ1 of the spectrum peaks of each first spectrum A1, A2, N1, BN, N3, N4 and the standard deviation σ2 of the spectrum peaks of each second spectrum A1a, A2a, N1a, N2a, N3a, N4a, respectively. By doing so, combinations of the first and second spectra (N1, N1a), (BN, N2a), (N3, N3a), and (N4, N4a) in which the fluctuation trends of the dispersion of the spectrum peaks match are determined to have a common noise component (step S107).

[0063] Next, the noise removal unit (113) removes the noise components included in the first spectrum N1, BN, N3, N4 within the current analysis result that is determined to have common noise components, i.e., the components caused by the noise spectrum N1, N2, N3, N4. The noise removal unit (113) maintains, for example, past performance information, and removes the noise components by referring to the maintained information (step S108).

[0064] Next, the FFT analysis unit (111) determines the spectrum waveform and spectrum peaks as noise-free current analysis results for the spectra A1, A2, B1, and B2 after the noise component (noise spectrum N1, N2, N3, N4) is removed by the noise removal unit (113) from all spectra A1, A2, BN, B2, N1, N3, and N4 in the current analysis results.

[0065] The spectrum waveform resulting from this noise-free current analysis is the right-hand diagram of Fig. 5.

[0066] As shown in FIG. 5, based on the current analysis results and voltage analysis results shown in the left and center diagrams, respectively, noise components within the current analysis results are removed, and a noise-free current analysis result shown in the right diagram is obtained (step S109).

[0067] Next, the spectrum peak extraction unit (114) extracts spectrum peaks B1 and B2 of the sideband spectrum related to the rotational frequency fr from the spectrum peaks in the noise-free current analysis result as diagnostic spectrum peaks related to the diagnosis of abnormalities in the motor (5) (step S110).

[0068] Next, the correction processing unit (115) corrects the diagnostic spectrum peak extracted by the spectrum peak extraction unit (114) according to the load rate calculated by the load rate calculation unit (110), and generates a reference value which is the diagnostic spectrum peak after correction. This reference value represents the initial state of the motor (5) during normal operation, that is, the reference state of the motor (5).

[0069] However, the sideband spectrum B1 and B2, which increase due to an abnormality in the motor (5), have spectrum peaks that fluctuate according to the change in the load rate of the motor (5). Therefore, by correcting the diagnostic spectrum peaks according to the calculated load rate, the bias caused by the load rate is eliminated and averaged. In addition, when correcting, correction according to the frequency of each spectrum may be performed together with the load rate (step S111).

[0070] Next, the correction processing unit (115) calculates a judgment threshold α, which is used in judgment during abnormal diagnosis, for each diagnostic spectrum peak (reference value) after correction, with a value greater than the diagnostic spectrum peak (reference value) (step S112), and stores the diagnostic spectrum peak (reference value) and the judgment threshold α in the reference value storage unit (116) (step S113).

[0071] As described above, the diagnostic device (100) collects information on the initial state of the motor (5) by the initial output flow shown in steps S101 to S113, and the diagnostic spectrum peak (reference value) and judgment threshold α of the initial state are stored in the reference value storage unit (116).

[0072] Next, the operation of the diagnostic device (100) during the diagnosis of an abnormality of the electric motor (5) will be explained based on FIGS. 4 and FIGS. 6.

[0073] The diagnostic device (100) collects information when diagnosing an abnormality of the motor (5), that is, executes the abnormality diagnosis flow shown in FIG. 4 at predetermined intervals. In addition, steps S201 to S211 shown below are described in an appropriately omitted manner because they are operations similar to steps S101 to S111 described above.

[0074] When the abnormal diagnosis flow is initiated, the diagnosis device (100) detects that the motor (5) is in an operating state (step S201), and the measurement circuit (7) acquires a current signal due to the load current flowing through the motor and a voltage signal due to the voltage applied to the motor (5) (step S202).

[0075] Next, the load rate calculation unit (110) calculates the load rate of the motor (5) based on the current signal and the voltage signal (step S203).

[0076] Next, the FFT analysis unit (111) performs FFT analysis on the current signal and the voltage signal, respectively, to derive the current analysis result and the voltage analysis result (step S204), normalizes the obtained current analysis result and the voltage analysis result (step S205), and detects all spectrum peaks. By doing so, the spectrum waveform after normalization and the current analysis result and the voltage analysis result including the spectrum peaks are generated (step S206).

[0077] Spectral waveforms based on current analysis results and spectral waveforms based on voltage analysis results are shown in the left and center diagrams of FIG. 6, respectively. The dotted lines in the diagrams represent noise components. Additionally, the right diagram of FIG. 6 will be described later.

[0078] In the current analysis results shown in the left side of Fig. 6, the following spectrum having spectral peaks appears. Spectrum A1 of the power frequency f0 and harmonic spectrum A2 of integer multiples other than 1 of the power frequency f0 appear. Additionally, at frequencies (f0±fr) shifted by the rotation frequency fr on both sides of spectrum A1, sideband spectra B3 and B4 of spectrum A1 appear, and noise spectra N1, N2, N3, and N4, which are noise components, also appear. In this case, since noise spectrum N2 is superimposed on sideband spectrum B3, spectrum BNa after superimposition appears, and noise spectrum N2 and sideband spectrum B3 are not distinguishable.

[0079] In the voltage analysis results shown in the center of Fig. 6, the following spectra with spectral peaks appear. Spectrum A1a of the power frequency f0 and harmonic spectrum A2a of integer multiples (in this case, 2 times) of the power frequency f0 other than 1 appear. Additionally, noise components such as noise spectra N1a, N2a, N3a, and N4a appear. The noise spectra N1a, N2a, N3a, and N4a appear in correspondence with the noise spectra N1, N2, N3, and N4 obtained from the analysis results by current FFT.

[0080] In the spectrum waveform diagram based on the voltage analysis results, the sideband spectrum associated with the rotation frequency fr does not appear. Additionally, in Figure 6, for convenience, the same symbols as those in Figure 4, which shows the spectrum diagram of the initial state, were used for spectra B3, B4, and BNa associated with sideband spectra B3 and B4.

[0081] If the spectrum resulting from the current analysis is called the first spectrum and the spectrum resulting from the voltage analysis is called the second spectrum, it can be seen that the second spectrum A1a, A2a, N1a, N2a, N3a, and N4a appear corresponding to the first spectrum A1, A2, N1, BNa, N3, and N4, respectively.

[0082] We will return to the abnormal diagnosis flow shown in Figure 4 and continue the explanation.

[0083] After the current analysis result and voltage analysis result including the spectrum waveform and spectrum peak after normalization processing are generated, the noise determination unit (112) compares the current analysis result and the voltage analysis result and performs a noise determination of each spectrum, similar to step S107.

[0084] The operation of the noise determination unit (112) will be explained with reference to FIG. 6.

[0085] The noise determination unit (112) compares the first spectrum A1, A2, N1, BNa, N3, N4 based on the current analysis result with the second spectrum A1a, A2a, N1a, N2a, N3a, N4a based on the voltage analysis result. Then, it determines whether the variation trend of the dispersion of the spectrum peaks matches in the corresponding first and second spectra, and for the matching first spectrum N1, BNa, N3, N4 and second spectrum N1a, N2a, N3a, N4a, it determines that they contain a common noise component (step S207).

[0086] Next, the noise removal unit (113) removes the noise components included in the first spectrum N1, BNa, N3, N4 within the current analysis result that is determined to have common noise components, i.e., the components caused by the noise spectrum N1, N2, N3, N4 (step S208).

[0087] Next, the FFT analysis unit (111) determines the spectrum waveform and spectrum peaks as noise-free current analysis results for the spectra A1, A2, B3, and B4 after the noise component (noise spectrum N1, N2, N3, N4) has been removed by the noise removal unit (113) from all spectra A1, A2, BNa, B4, N1, N3, and N4 in the current analysis results.

[0088] The spectrum waveform resulting from this noise-free current analysis is the right-hand side of Fig. 6.

[0089] As shown in FIG. 6, based on the current analysis results and voltage analysis results shown in the left and center diagrams, respectively, noise components within the current analysis results are removed, and a noise-free current analysis result shown in the right diagram is obtained (step S209).

[0090] Next, the spectrum peak extraction unit (114) extracts spectrum peaks B3 and B4 of the sideband spectrum related to the rotational frequency fr from the spectrum peaks in the noise-free current analysis result as diagnostic spectrum peaks related to the diagnosis of abnormalities in the motor (5) (step S210).

[0091] Next, the correction processing unit (115) corrects the diagnostic spectrum peak extracted by the spectrum peak extraction unit (114) according to the load rate calculated by the load rate calculation unit (110), and generates a diagnostic value which is the diagnostic spectrum peak after correction. This diagnostic value represents the diagnostic state of the motor (5) when an abnormality is diagnosed (step S211).

[0092] Next, the correction processing unit (115) stores each diagnostic spectrum peak (diagnostic value) after correction in the diagnostic value storage unit (117) (step S212).

[0093] Next, the judgment unit (118) performs a pass / fail judgment for each diagnostic spectrum peak (diagnostic value) by referring to the reference value and judgment threshold α in the reference value storage unit (116) (step S213). In this case, the corrected spectrum peaks (diagnostic values) of the sideband spectrums B3 and B4 at the time of abnormal diagnosis are larger than the corrected spectrum peaks (reference values) of the sideband spectrums B1 and B2 in the initial state and are also larger than the judgment threshold α, so a defect in the diagnostic value is detected (step S214).

[0094] This diagnoses the abnormality of the motor (5).

[0095] As described above, the diagnostic device (100) of the motor (5) according to this embodiment comprises a measurement circuit (7) for acquiring a current signal and a voltage signal, an FFT analysis unit (111) for frequency-analyzing the acquired current signal and voltage signal and outputting a current analysis result and a voltage analysis result, a noise determination unit (112) for performing noise determination, a noise removal unit (113), and a spectrum peak extraction unit (114). The noise determination unit (112) performs noise determination by comparing a first spectrum, which is a spectrum component of a spectrum peak appearing in the current analysis result, with a second spectrum, which is a spectrum component corresponding to the first spectrum within the voltage analysis result. The noise removal unit (113) removes a noise component from the current analysis result based on the determination result of the noise determination unit (112). The spectrum peak extraction unit (114) extracts a diagnostic spectrum peak based on the result of removing the noise component from the current analysis result.

[0096] And, the diagnostic device (100) diagnoses abnormalities in the motor (5) based on the extracted diagnostic spectrum peaks.

[0097] By this, the diagnostic device (100) can diagnose abnormalities in the motor (5) with high reliability and precision while preventing the influence of noise. In addition, it is not limited to a driving method based on vector control as in Patent Document 1, but can be applied to motors (5) based on a wide range of driving methods.

[0098] In addition, in this embodiment, the diagnostic device (100) generates a reference value representing the reference state of the motor (5) based on a diagnostic spectrum peak generated by the sequential operation of the FFT analysis unit (111), the noise determination unit (112), the noise removal unit (113), and the spectrum peak extraction unit (114) for the current signal and voltage signal acquired during normal operation of the motor (5). Then, the diagnostic device (100) generates a diagnostic value representing the diagnostic state of the motor (5) based on a diagnostic spectrum peak generated by the sequential operation of the FFT analysis unit (111), the noise determination unit (112), the noise removal unit (113), and the spectrum peak extraction unit (114) for the current signal and voltage signal acquired during abnormal diagnosis of the motor (5), and diagnoses the abnormality of the motor (5) by comparing this diagnostic value with the reference value.

[0099] By this, both the reference value representing the reference state of the motor (5) and the diagnostic value representing the diagnostic state of the motor (5) can be generated as highly reliable, high-precision values ​​with noise components removed, and by comparing the diagnostic value and the reference value, the precision and reliability of the abnormal diagnosis of the motor (5) are further improved.

[0100] In addition, the diagnostic method of the motor (5) according to this embodiment comprises a signal acquisition step (step S102, step S202) for acquiring a current signal and a voltage signal, a frequency analysis step (step S104, step S204) for frequency analyzing the current signal and the voltage signal, respectively, and outputting a current analysis result and a voltage analysis result, a noise judgment step (step S107, step S207) for performing a noise judgment, a noise removal step (step S108, step S208), and a spectrum peak extraction step (step S110, step S210).

[0101] The noise determination step (step S107, step S207) performs a noise determination by comparing a first spectrum, which is a spectral component of a spectral peak appearing in the current analysis result, with a second spectrum, which is a spectral component corresponding to the first spectrum within the voltage analysis result. The noise removal step (step S108, step S208) removes noise components from the current analysis result based on the determination result by the noise determination step. The spectral peak extraction step (step S110, step S210) extracts diagnostic spectral peaks based on the result of removing noise components from the current analysis result by the noise removal step.

[0102] And, the above diagnostic method diagnoses an abnormality in the motor (5) based on the diagnostic spectrum peak extracted in the spectrum peak extraction step (step S110, step S210).

[0103] By this, it is possible to diagnose abnormalities in the motor (5) with high reliability and precision by preventing the influence of noise. In addition, it is not limited to a driving method based on vector control as in Patent Document 1, but can be applied to motors (5) based on a wide range of driving methods.

[0104] In addition, in the diagnostic method of the motor (5) according to this embodiment, the series of five steps, including the signal acquisition step (step S102, step S202), the frequency analysis step (step S104, step S204), the noise determination step (step S107, step S207), the noise removal step (step S108, step S208), and the spectrum peak extraction step (step S110, step S210), are operated in both the normal state of the motor and the abnormal state of the motor. Then, when the motor (5) is normal, a reference value representing the reference state of the motor (5) is generated based on the extracted diagnostic spectrum peak, and when the motor (5) is abnormal, a diagnostic value representing the diagnostic state of the motor (5) is generated based on the extracted diagnostic spectrum peak, and the abnormal state of the motor (5) is diagnosed by comparing the diagnostic value with the reference value.

[0105] By this, both the reference value representing the reference state of the motor (5) and the diagnostic value representing the diagnostic state of the motor (5) can be generated as highly reliable, high-precision values ​​with noise components removed, and the precision and reliability of the abnormal diagnosis of the motor (5) are further improved by comparing the diagnostic value and the reference value.

[0106] In addition, in the above embodiment 1, the diagnostic device (100) is shown to generate a reference value indicating the standard state of the motor (5) when the motor (5) is normal, and to generate a diagnostic value indicating the diagnostic state of the motor (5) when the motor (5) is abnormal, and to compare it with the reference value. However, the diagnostic device (100) may be a device that generates a diagnostic spectrum peak that becomes the reference value or the diagnostic value. For example, when the motor (5) is abnormal, a diagnostic value indicating the diagnostic state of the motor (5) may be generated, and if it exceeds a preset threshold value, it may be diagnosed as an abnormality of the motor (5). In that case as well, it is possible to diagnose the abnormality of the motor (5) with high reliability and high precision while preventing the influence of noise.

[0107] However, the function of the operation processing unit (8) of the above embodiment 1 is realized by a processing circuit.

[0108] FIG. 7 is a configuration diagram showing an example of hardware that realizes each function of the operation processing unit (8). In this case, the operation processing unit (8) is configured with a processing circuit (130A), which is dedicated hardware.

[0109] Additionally, the processing circuit (130A) corresponds to, for example, a single circuit, a complex circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof.

[0110] Additionally, FIG. 8 is a configuration diagram showing another example of hardware that realizes each function of the operation processing unit (8). In this case, the processing circuit (130B) is equipped with a processor (131) and a memory (132).

[0111] In the processing circuit (130B), the function of the computation processing unit (8) is realized by software, firmware, or a combination of software and firmware. The software and firmware are described as programs and stored in memory (132). The processor (131) realizes each function by reading and executing the program stored in memory (132).

[0112] The program stored in the memory (132) can also be said to execute the procedures or methods of each part described above on the computer. Here, the memory (132) is a non-volatile or volatile semiconductor memory, such as, for example, RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable and Programmable Read Only Memory). In addition, magnetic disks, flexible disks, optical disks, compact disks, mini disks, DVDs, etc. also correspond to memory (202).

[0113] In addition, regarding the functions of the above-described operation processing unit (8), some parts may be implemented in dedicated hardware, and some parts may be implemented in software or firmware.

[0114] In this way, the processing circuit can realize the function of the above-described computation processing unit (8) by means of hardware, software, firmware, or a combination thereof.

[0115] Embodiment 2.

[0116] In this embodiment 2, the noise removal unit (113) is equipped with an information processing device that processes information using a neural network and removes noise components, namely components caused by noise spectrum N1, N2, N3, N4.

[0117] The rest of the configuration is the same as in the above embodiment 1, and the configuration and operation of the noise removal unit (113), which is different from embodiment 1, will be explained below.

[0118] As described above, the noise removal unit (113) removes noise components included in the first spectrum N1, BN(BNa), N3, and N4 within the current analysis result determined to have common noise components, that is, components caused by the noise spectrum N1, N2, N3, and N4. Here, noise removal for generating reference values ​​of the initial state of the motor (5) is described. That is, as shown in FIG. 5, noise components included in the first spectrum N1, BN, N3, and N4 are removed.

[0119] FIG. 9 is a drawing illustrating a noise removal unit according to embodiment 2.

[0120] As shown in FIG. 9, an information processing device (125) using a neural network (120) processes input signals (126), which are noisy spectrum peaks PA, PB, and PC, in the neural network (120) and outputs output signals (127), which are noisy spectrum peaks PA1, PB1, and PC1.

[0121] All spectral peaks of the first spectrum N1, BN, N3, and N4 containing noise components are used as one input signal (126), and each spectral peak PA, PB, and PC correspond to one input signal (126). That is, the spectral peaks PA, PB, and PC of each input signal (126) are each signals that integrate the data of all spectral peaks of the first spectrum N1, BN, N3, and N4, and are input to the neural network (120) in a time series.

[0122] Likewise, the spectrum peaks PA1, PB1, and PC1 of each output signal (127) are signals that integrate data of all spectrum peaks after removing noise components from the first spectrum N1, BN, N3, and N4, respectively, and are output from the neural network (120) in a time series corresponding to the input signal (126). In this case, all spectrum peaks after removing noise components from the first spectrum N1, BN, N3, and N4 become 0, the spectrum peak of B1, 0, and 0.

[0123] That is, the noise removal unit (113) takes the noisy spectrum peaks PA, PB, and PC as inputs and derives the noise-free spectrum peaks PA1, PB1, and PC1, and based on this result, the FFT analysis unit (111) determines the noise-free current analysis result.

[0124] The neural network (120) is equipped with a learning process (121) that performs learning based on an input signal (126), and an inference process (122) that performs inference based on the input signal (126) based on information obtained from the learning process (121), and derives an output signal (127) based on the inference process (122).

[0125] For the neural network (120), a Recurrent Neural Network (RNN), suitable for processing time series data, is used. In addition, it is not limited to an RNN, and any recursive processing such as retraining is possible.

[0126] The neural network (120) learns the trends of the noise-free spectrum peaks PA1, PB1, and PC1 from the past, that is, the previous values, and performs the process of estimating the noise-free spectrum peaks PA1, PB1, and PC1 from the current noise-free spectrum peaks PA, PB, and PC.

[0127] There is a possibility that the spectrum peaks may gradually increase due to the aging degradation of the motor (5). In this embodiment, the noise removal unit (113) can distinguish between the increase in spectrum peaks due to the degradation of the motor (5) and the temporary and steep increase in spectrum peaks due to noise components, because the neural network (120) learns the trends of the noise-free spectrum peaks PA1, PB1, and PC1 up to the previous value. In this way, the neural network (120) frequently feeds back the data of the noise-free spectrum peaks PA1, PB1, and PC1 for further learning, and estimates the current noise-free spectrum peaks PA1, PB1, and PC1 by considering the increase in each spectrum peak due to the degradation of the motor (5).

[0128] By this, the removal of noise components can be realized at high speed and with high precision.

[0129] Although the present invention describes various exemplary embodiments and examples, the various features, modes, and functions described in one or more embodiments are not limited to the application of a specific embodiment but can be applied to the embodiments alone or in various combinations.

[0130] Accordingly, countless variations not exemplified herein are conceived within the scope of the technology disclosed herein. For example, this includes cases where at least one component is modified, added, or omitted, and furthermore, cases where at least one component is extracted and combined with components of other embodiments. Explanation of the symbols

[0131] 5: Electric motor 7: Measurement circuit 110: Load Rate Calculation Unit 111: FFT Analysis Unit 112: Noise determination unit 113: Noise removal unit 114: Spectrum peak extraction unit 115: Correction processing unit 116: Reference value containment section 118: Judgment section 120: Neural Network A1, A2, N1, BN, BNa, N3, N4: First spectrum A1a, A2a, N1a, N2a, N3a, N4a: Second spectrum B1, B2: Sideband spectrum PA, PB, PC: Noisy spectral peaks PA1, PB1, PC1: Noise-free spectrum peaks α: Judgment threshold σ1, σ2: Standard deviation

Claims

Claim 1 A diagnostic device for a motor comprising: a signal acquisition unit that detects a current flowing through the motor and a voltage applied to the motor to acquire a current signal and a voltage signal, respectively; a frequency analysis unit that performs frequency analysis on the acquired current signal and the voltage signal to output a current analysis result and a voltage analysis result, respectively; a noise determination unit that performs a noise determination by comparing a first spectrum, which is a spectrum component of a spectrum peak appearing in the current analysis result, with a second spectrum, which is a spectrum component corresponding to the first spectrum within the voltage analysis result; a noise removal unit that removes a noise component from the current analysis result based on the determination result of the noise determination unit; and a spectrum peak extraction unit that extracts a diagnostic spectrum peak based on the result of removing the noise component from the current analysis result, and diagnoses an abnormality of the motor based on the extracted diagnostic spectrum peak, wherein the noise determination unit collects both spectrum peaks in a time series for the first spectrum and the second spectrum, and performs the noise determination based on the first spectrum and the second spectrum including the spectrum peaks collected in a time series. Claim 2 A diagnostic device for an electric motor according to claim 1, wherein the noise determination unit detects that the variation trend of the dispersion of the collected spectrum peaks on both sides matches in the first spectrum and the second spectrum, and determines that the first spectrum and the second spectrum contain a common noise component, and the noise removal unit removes the common noise component included in the first spectrum as the noise component. Claim 3 A diagnostic device for an electric motor according to claim 2, wherein the noise determination unit observes the standard deviation at each spectrum peak of the first spectrum and the second spectrum as the dispersion, and detects the consistency of the variation trend of the dispersion when both standard deviations vary beyond a set value. Claim 4 A diagnostic device for an electric motor according to claim 2, wherein the noise removal unit comprises an information processing device that processes information using a neural network, and performs learning by inputting a noisy peak, which is a spectrum peak when the first spectrum includes the noise component, and a noise-free peak, which is a spectrum peak when the noise component is removed from the first spectrum, and performs inference using the noisy peak as input based on the information obtained from the learning to derive the noise-free peak. Claim 5 A diagnostic device for an electric motor according to claim 3, wherein the noise removal unit comprises an information processing device that processes information using a neural network, and performs learning by inputting a noisy peak, which is a spectrum peak when the first spectrum includes the noise component, and a noise-free peak, which is a spectrum peak when the noise component is removed from the first spectrum, and performs inference using the noisy peak as input based on the information obtained from the learning to derive the noise-free peak. Claim 6 A diagnostic device for a motor according to any one of claims 1 to 5, wherein, for the current signal and voltage signal acquired by the signal acquisition unit during normal operation of the motor, a reference value representing a reference state of the motor is generated based on a first diagnostic spectrum peak, which is a diagnostic spectrum peak generated by the frequency analysis unit, the noise determination unit, the noise removal unit, and the spectrum peak extraction unit operating in sequence, and for the current signal and voltage signal acquired by the signal acquisition unit during abnormal diagnosis of the motor, a diagnostic value representing a diagnostic state of the motor is generated based on a second diagnostic spectrum peak, which is a diagnostic spectrum peak generated by the frequency analysis unit, the noise determination unit, the noise removal unit, and the spectrum peak extraction unit operating in sequence, and the diagnostic value is compared with the reference value to diagnose an abnormality of the motor. Claim 7 A diagnostic device for a motor according to claim 6, comprising a storage unit for storing the reference value of the motor generated during normal operation of the motor, and a judgment unit for determining the quality of the diagnostic value of the motor generated during abnormal diagnosis of the motor by comparing it with the reference value in the storage unit. Claim 8 A diagnostic device for a motor according to claim 7, wherein a judgment threshold is generated in correspondence with the reference value of the motor, the storage unit stores the judgment threshold together with the reference value, and the judgment unit performs a judgment of the quality of the diagnostic value using the judgment threshold. Claim 9 A diagnostic device for a motor according to claim 6, comprising a correction processing unit that corrects the diagnostic spectrum peak according to the load rate of the motor, wherein the correction processing unit corrects the first diagnostic spectrum peak to generate the reference value of the motor when the motor is normal, and corrects the second diagnostic spectrum peak to generate the diagnostic value of the motor when the motor is abnormal. Claim 10 A diagnostic device for a motor according to claim 7, comprising a correction processing unit that corrects the diagnostic spectrum peak according to the load rate of the motor, wherein the correction processing unit corrects the first diagnostic spectrum peak to generate the reference value of the motor when the motor is normal, and corrects the second diagnostic spectrum peak to generate the diagnostic value of the motor when the motor is abnormal. Claim 11 A diagnostic device for a motor according to claim 8, comprising a correction processing unit that corrects the diagnostic spectrum peak according to the load rate of the motor, wherein the correction processing unit corrects the first diagnostic spectrum peak to generate the reference value of the motor when the motor is normal, and corrects the second diagnostic spectrum peak to generate the diagnostic value of the motor when the motor is abnormal. Claim 12 Diagnosis of a motor comprising: a signal acquisition step for acquiring a current signal and a voltage signal by detecting, respectively, a current flowing to the motor and a voltage applied to the motor from a main circuit connected between a power system and the motor; a frequency analysis step for performing frequency analysis on the current signal and the voltage signal, respectively, and outputting a current analysis result and a voltage analysis result; a noise determination step for performing a noise determination by comparing a first spectrum, which is a spectral component of a spectrum peak appearing in the current analysis result, with a second spectrum, which is a spectral component corresponding to the first spectrum within the voltage analysis result; a noise removal step for removing a noise component from the current analysis result based on the determination result by the noise determination step; and a spectrum peak extraction step for extracting a diagnostic spectrum peak based on the result of removing the noise component from the current analysis result by the noise removal step; and diagnosing an abnormality of the motor based on the extracted diagnostic spectrum peak, wherein the noise determination step collects both spectrum peaks in a time series for the first spectrum and the second spectrum, and performs the noise determination based on the first spectrum and the second spectrum including the spectrum peaks collected in a time series. method. Claim 13 A method for diagnosing a motor according to claim 12, wherein the series of five steps, the signal acquisition step, the frequency analysis step, the noise determination step, the noise removal step, and the spectrum peak extraction step, operates in both the normal state of the motor and the abnormal state of the motor, and in the normal state of the motor, generates a reference value representing the reference state of the motor based on the extracted diagnostic spectrum peak, and in the abnormal state of the motor, generates a diagnostic value representing the diagnostic state of the motor based on the extracted diagnostic spectrum peak, and diagnoses the abnormal state of the motor by comparing the diagnostic value with the reference value.

Images