Apparatus and method for diagnosing failure of rotating machine by using noise, vibration, and current signals

Abstract

The present invention relates to an apparatus for diagnosing a failure of a rotating machine by using a vibration signal and a current signal. The apparatus for diagnosing a failure of a rotating machine according to one embodiment of the present invention comprises a terminal device, the terminal device comprising: a single-phase or three-phase power and arc state monitoring unit connected to an electric line connected to a rotating machine using a single-phase or three-phase power source to measure a single-phase or three-phase voltage and current, analyze single-phase or three-phase power quality, and detect an arc; and a noise and vibration analysis unit which measures vibration generated from a rotary shaft included in the rotating machine and analyzes the degree of vibration. A failure of the rotating machine is diagnosed from information outputted by the single-phase or three-phase power and arc state monitoring unit and the noise and vibration analysis unit.

Description

Device and method for diagnosing failures of rotating machinery using noise, vibration, and current signals

[0001] The present invention relates to an apparatus and method for diagnosing whether a rotating machine is faulty by utilizing noise (sound waves, ultrasound), vibration, and current signals.

[0002]

[0003] Rotating machinery is a device in which components rotate around an axis to transmit or convert energy. Rotating machinery is used across various industries and can save energy and increase productivity by shifting the position of objects.

[0004] A rotating machine may include a shaft that transmits power generated in a driving body to a driven body, and a bearing that fixes the shaft and rotates it while supporting its load.

[0005] Rotating machinery can experience mechanical or electrical failures. Mechanical failures in rotating machinery can cause electrical failures, or electrical failures can cause mechanical failures, thereby affecting other types of failures.

[0006] To diagnose the failure of rotating machinery at an early stage, the diagnostic device includes a vibration sensor, a current sensor, and a temperature sensor, and after receiving the values ​​measured by the sensors, analyzes the vibration and rotation data to determine whether a failure has occurred.

[0007] However, since a configuration that diagnoses rotating machinery using only sensors has limitations in accurately diagnosing whether a fault exists, a method for diagnosing the fault of rotating machinery more accurately is required.

[0008]

[0009] The objective of the present invention is to provide an apparatus and method for diagnosing whether a rotating machine is faulty by utilizing noise (sound waves, ultrasound), vibration, and current signals.

[0010]

[0011] To achieve the above objectives, a fault diagnosis device for a rotating machine according to one embodiment of the present invention comprises: a single-phase or three-phase power and arc state monitoring unit connected to an electrical line connected to a rotating machine using a single-phase or three-phase power supply, which measures single-phase or three-phase voltage and current, analyzes single-phase or three-phase power quality, and detects an arc; and a noise and vibration analysis unit that measures vibrations occurring in a rotating shaft included in the rotating machine and analyzes the degree of vibration; and diagnoses whether the rotating machine is faulty from the information output by the single-phase or three-phase power and arc state monitoring unit and the noise and vibration analysis unit.

[0012] In one embodiment, the single-phase or three-phase power and arc state monitoring unit includes a main calculation unit, and the main calculation unit calculates arc state, channel, leakage current, and power state information from the measured voltage and current information.

[0013] In one embodiment, the single-phase or three-phase power and arc state monitoring unit includes a low-pass filter, a high-pass filter, and a plurality of band-pass filters, and inputs a current signal to the low-pass filter, the high-pass filter, and the plurality of band-pass filters to separate it by frequency and detect an arc.

[0014] In one embodiment, the single-phase or three-phase power and arc state monitoring unit comprises: a plurality of comparators connected to the band-pass filter or the high-pass filter, each having a different first reference value preset and comparing the output signal magnitude of the band-pass filter or the high-pass filter with the first reference value; a counter connected to the comparator, which increases the count by one if the output signal magnitude of the band-pass filter or the high-pass filter is greater than the first reference value set in the connected comparator; and a count analysis unit connected to the counter, which checks the count counted by the counter for a preset time and transmits a signal to a main calculation unit if it exceeds a preset second reference value.

[0015] In one embodiment, the system further includes a server computer having a first machine learning model trained to diagnose whether a rotating machine is faulty; and a database management system for storing data.

[0016] In one embodiment, the server computer transmits result data of learning the first machine learning model to a terminal device, and the terminal device includes a second machine learning model and sets the second machine learning model with parameters included in the result data.

[0017] In one embodiment, the terminal device inputs data output by the single-phase or three-phase power and arc state monitoring unit and the noise and vibration analysis unit into the second machine learning model, and the second machine learning model outputs information diagnosing whether the rotating machine is faulty in response to the input data.

[0018] A fault diagnosis method for a rotating machine according to one embodiment of the present invention comprises: a step in which a terminal device measures vibrations occurring in a rotating shaft included in the rotating machine; a step in which the terminal device measures voltage and current of the rotating machine; a step in which a server computer transmits result data learned from a first machine learning model to the terminal device, and the terminal device sets a second machine learning model with parameters included in the received data; a step in which the terminal device inputs the measured vibration information of the rotating machine into the second machine learning model, and the second machine learning model outputs vibration analysis information of the rotating machine; and a step in which the terminal device inputs the measured voltage and current information of the rotating machine into the second machine learning model, and the second machine learning model outputs power quality analysis information of the rotating machine.

[0019] In one embodiment, the terminal device measures the voltage and current of a rotating machine in the step of: a sensor measuring current, voltage, and leakage current in a single-phase line of the rotating machine and transmitting the measured signals to a current signal processing unit, a voltage signal processing unit, and a leakage current signal processing unit, respectively; a voltage signal processing unit transmitting the voltage measurement signal to an analog-to-digital converter and converting it into a first digital signal; a voltage characteristic analysis unit classifying the voltage value in the first digital signal into an instantaneous value, an RMS value, and an average value, and analyzing the characteristics thereof; a leakage current signal processing unit transmitting the leakage current measurement signal to an analog-to-digital converter and converting it into a second digital signal; a leakage current analysis unit classifying the leakage current value in the second digital signal into an instantaneous value, an RMS value, and an average value, and analyzing the characteristics thereof; a current signal processing unit transmitting the current measurement signal to a low-pass filter, a band-pass filter, and a high-pass filter; and transmitting the signal output by the low-pass filter to an analog-to-digital converter and converting it into a third digital signal. The method includes: a step in which a frequency determination unit analyzes the quality and frequency characteristics of the current in a third digital signal; a step in which a comparator compares the signal output by a band-pass filter and a high-pass filter with a first reference value; a step in which a counter increases the count by one when the signal output by the band-pass filter and the high-pass filter is greater than the first reference value; and a step in which a count analysis unit checks the count counted by the counter for a preset time, and if the checked count exceeds a second reference value, transmits a signal to a main calculation unit.

[0020]

[0021] In the initial stages prior to a breakdown in rotating machinery, changes in vibration occur, which may lead to the sequential appearance of noise, heat, and smoke. By implementing the present invention, changes in vibration can be detected to predict whether a breakdown of the rotating machinery will occur. Furthermore, by performing maintenance to prevent breakdowns in advance based on the predicted information, costs can be reduced.

[0022] In addition, changes in driving current due to load fluctuations of rotating machinery can be simultaneously detected to improve the accuracy of fault prediction.

[0023] By detecting in advance arcs that cause fires originating from loads, power sources, and lines connected to the contacts of switchboards, distribution panels, and control panels, it is possible to cut off the power and take necessary safety management measures. Furthermore, by detecting in advance the risk of electrical fires caused by vibrations during an earthquake, fires can be prevented by cutting off the power supply to the switchboards, distribution panels, and control panels, as well as the power supplied to the loads.

[0024] Since the present invention non-contactly measures sound waves or ultrasound using a sensor, it is possible to inspect the condition in real time during operation without stopping the rotating machine. Furthermore, the sensor for noise diagnosis is relatively small and easy to attach, making it easy to install additionally on the rotating machine.

[0025] Early-stage defects, such as microscopic cracks in bearings or gears and wear caused by insufficient lubrication, are easily overlooked by conventional vibration sensors due to small signal changes. However, the present invention can detect abnormalities by increasing high-frequency signals of 20 kHz or higher at the early stages of defect progression, thereby enabling the detection of defects that are not detected by vibration or temperature changes.

[0026]

[0027] FIGS. 1 and FIGS. 2 show the configuration of a fault diagnosis device for a rotating machine utilizing vibration signals and current signals according to an embodiment of the present invention.

[0028] FIGS. 3 and FIGS. 4 illustrate each step of a fault diagnosis method for a rotating machine using vibration signals and current signals according to an embodiment of the present invention.

[0029] FIG. 5 shows the configuration of a single-phase power and arc state monitoring unit in one embodiment of the present invention.

[0030] FIG. 6 illustrates the operation method of a single-phase power and arc state monitoring unit in one embodiment of the present invention.

[0031] FIG. 7 shows the configuration of a three-phase power and arc state monitoring unit in one embodiment of the present invention.

[0032] FIG. 8 shows the frequency band that the filter passes through in one embodiment of the present invention.

[0033] FIGS. 9 to 11 show the waveform of an arc detection signal in one embodiment of the present invention.

[0034] FIG. 12 shows the configuration of a first or second machine learning model in one embodiment of the present invention.

[0035] FIG. 13 shows the configuration of a device for diagnosing a failure of a rotating machine using noise, vibration, and current signals according to one embodiment of the present invention.

[0036]

[0037] The present invention may be implemented with various modifications without departing from the spirit, and may have one or more embodiments. Furthermore, the embodiments described in the “specific details for implementing the invention” and “drawings,” etc., in the present invention are examples for specifically explaining the present invention and do not limit or restrict the scope of the rights of the present invention.

[0038] Accordingly, anything that a person skilled in the art to which the present invention pertains can easily deduce from the “specific details for carrying out the invention” and “drawings,” etc., of the present invention may be interpreted as falling within the scope of the present invention.

[0039] In addition, the size and shape of each component shown in the drawings may be exaggerated for the purpose of explaining the embodiments and do not limit the actual size and shape of the invention.

[0040] Unless specifically defined otherwise in the specification of the present invention, terms used therein may have the same meaning as generally understood by those skilled in the art to which the present invention pertains.

[0041] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0042]

[0043] In the present invention, the rotating device may be a device in which components rotate using electricity. For example, it may be any one of a pump, compressor, mixer, cooler fan, blower, motor, turbine, engine, or gearbox.

[0044] FIGS. 1 and FIGS. 2 show the configuration of a fault diagnosis device for a rotating machine utilizing vibration signals and current signals according to an embodiment of the present invention.

[0045] As illustrated in FIG. 1, a fault diagnosis device for a rotating machine (100, hereinafter “fault diagnosis device”) utilizing vibration signals and current signals may include a terminal device (110) and a server computer (120).

[0046] The terminal device (110) is connected to a rotating machine and can measure the vibration, voltage, and current of the rotating machine, and analyze and diagnose whether there is a fault. The terminal device (110) can analyze the power quality of a rotating machine using single-phase or three-phase power and detect arcs. It can also analyze the degree of vibration of a rotating machine rotating around three axes. In addition, it can diagnose whether there is a fault in the rotating machine from the measured data, store the measured data and diagnostic information, and display it using a display device or output it to an external device.

[0047] The terminal device (110) may each include a single-phase power and arc state monitoring unit (111), a noise and vibration analysis unit (112), a control unit (113), a storage unit (114), a display unit (115), and a communication unit (116).

[0048] The single-phase power and arc state monitoring unit (111) is connected to an electrical line connected to a rotating machine using single-phase or three-phase power, and can measure the voltage and current of the single phase. From the measured voltage and current information, it can analyze the quality of the power and detect an arc. The main calculation unit can calculate the arc state, channel, leakage current, and power state information from the measured voltage and current information and process the data. The main calculation unit may be a component included in the single-phase power and arc state monitoring unit (111) or may be an independent component.

[0049] The noise and vibration analysis unit (112) can measure vibrations generated from a rotating shaft included in a rotating machine and analyze the degree of vibration. The rotating machine may include, for example, three rotating shafts. The noise and vibration analysis unit (112) can process the analyzed degree of vibration into data.

[0050] The control unit (113) can operate the operation of all components included in the terminal device (110), and can operate the transmission of electrical signals or the transmission and reception of binary data within the components, or the transmission and reception of binary data between the server computer (120) or an external device. The control unit (113) can interpret and process instructions that can be read by a computer. The control unit (113) may include a microprocessor.

[0051] The storage unit (114) can store electrical signals input to the terminal device (110) in a binary data format, or store data produced by processing the configuration of the terminal device (110) and data input to the terminal device (110). The storage unit (114) can store commands and data that can be read by a computer. The storage unit (114) may include flash memory or a solid state drive (SSD).

[0052] The display unit (115) can display electrical signals or data input to the terminal device (110) and data produced by processing the configuration of the terminal device (110), and show them to a person using the terminal device (110) (hereinafter, “user”).

[0053] The communication unit (116) can receive computer-readable commands and data from a server computer (120) or an external device, or transmit to a server computer (120) or an external device electrical signals that are input into a terminal device (110) and converted into binary data format, or data produced by processing the configuration of the terminal device (110).

[0054] The server computer (120) can receive and store data measured or produced by the terminal device (110). The server computer (120) includes a processor, a memory device, and an input / output device, and can read, interpret, and process instructions that can be read by a computer. The server computer (120) may be located adjacent to the terminal device (110) or located at a remote location, and is not restricted by its installation location.

[0055] A server computer (120) may include a first machine learning model (121) trained to diagnose whether a rotating machine is faulty and a database management system (122) that stores data. When the server computer (120) inputs data produced by a terminal device (110) into the first machine learning model (121), the first machine learning model (121) may output information diagnosing whether a rotating machine is faulty in response to the input data.

[0056] The server computer (120) can process information regarding the diagnosis of a malfunction into data and transmit the data to the terminal device (110). The terminal device (110) can display the received data so that the user can check whether the rotating device is malfunctioning.

[0057] The server computer (120) can transmit the result data of learning the first machine learning model (121) to the terminal device (110). The result data of learning the first machine learning model (121) may include parameters such as weights and bias information.

[0058] The terminal device (110) may further include a second machine learning model and may set the second machine learning model with parameters included in the data received from the server computer (120). When the terminal device (110) inputs the produced data into the second machine learning model without transmitting data to the server computer (120), the second machine learning model may output information diagnosing whether the rotating machine is faulty in response to the input data. The terminal device (110) may acquire the information diagnosing whether the rotating machine is faulty output by the second machine learning model and store it in the storage unit (114) or display it through the display unit (115).

[0059] The terminal device (110) can be used to diagnose whether there is a failure in a device or facility that uses DC power in addition to AC power. For example, it can be used to monitor the charge status of a battery or to shut down power facilities in the event of an earthquake.

[0060] Meanwhile, according to an embodiment of the present invention, mechanical defects can be diagnosed in a non-contact manner by utilizing noise (audible sound waves and ultrasound) generated during the operation of a rotating machine. When a rotating machine operates normally, it exhibits a unique acoustic spectrum; however, when abnormalities such as bearing defects, imbalance, misalignment, wear, or insufficient lubrication occur, abnormal signals different from the usual may appear in the audible range or ultrasonic band. These signals are detected in real time through noise sensors such as audible microphones, contact acoustic sensors, and ultrasonic sensors. Subsequently, the detected acoustic signals are analyzed using frequency analysis techniques such as FFT (Fast Fourier Transform), STFT (Short-Time Fourier Transform), and wavelet transform to capture characteristic frequency components or pattern changes compared to the normal state, thereby diagnosing signs of failure. By monitoring the condition of the rotating machine through spectrum analysis in the frequency domain in this manner, initial abnormal signals can be detected early, preventing progression to serious failure.

[0061] The noise and vibration analysis unit (112) can detect bearing defects in a rotating machine. If a defect occurs in the ball bearing or roller bearing of the rotating machine, or if the raceway is damaged, high-frequency ultrasonic waves in the range of about 20 to 60 kHz may be emitted. This is caused by minute shocks and elastic waves occurring inside the bearing, and although it is a very weak signal in the initial defect stage, it has a unique frequency component. The noise analysis unit (112) can detect frequencies such as BPFO, BPFI, BSF, and FTF, which are the natural frequencies of the bearing defect, through spectrum analysis. For example, if there is an outer ring defect (BPFO) or an inner ring defect (BPFI), the corresponding natural frequency and its integer multiple harmonic components appear in the acoustic spectrum, and the type of damage to a specific bearing can be determined by these characteristics.

[0062] The noise analysis unit (112) can detect poor lubrication of the rotating machine. When the lubricating oil in the rotating part is insufficient or deteriorated, friction increases, and the ultrasonic band noise level tends to rise. By comparing the acoustic spectrum and signal level before and after lubrication, the noise analysis unit (112) can determine the appropriate time for lubrication and detect problems caused by over-lubrication or insufficient lubrication in advance.

[0063] The noise analysis unit (112) can detect imbalance and misalignment of the rotating machine. If there is mass imbalance or shaft misalignment of the rotor or shaft, high-frequency noise resulting from mechanical shock may occur in combination with low-frequency vibration during rotation. The noise analysis unit (112) can synchronously collect and simultaneously analyze signals from vibration sensors, such as accelerometers, and signals from acoustic sensors. Since minute distortions or residual imbalances that are difficult to identify with vibration signals appear as unique abnormal patterns in acoustic signals, analyzing the information measured by the two sensors simultaneously can increase diagnostic accuracy.

[0064] The noise analysis unit (112) can detect defects in the gearbox of a rotating machine. If the gear teeth of a reduction gear or transmission are broken or worn, periodic impact noise may occur whenever it rotates. The rotational sound of a healthy gear appears as a constant gear mesh noise, but a damaged gear emits an irregular impact sound that is “ticking” whenever the teeth mesh, which can leave distinct peaks and harmonic patterns in the acoustic signal. The noise analysis unit (112) can analyze the harmonics and sidebands of the acoustic spectrum to capture the gear defect signal.

[0065] The noise analysis unit (112) may include portable equipment or sensors.

[0066] Portable equipment can be, for example, a portable ultrasonic scanner that allows an operator to directly measure at various points on rotating machinery in the field. Portable ultrasonic scanners can capture high-frequency sounds beyond the audible range, allowing users to listen to them through headphones and view the spectrum in real time. This enables the detection of abnormal sounds originating from components such as bearings or gearboxes in rotating machinery, and allows for the identification of defect locations.

[0067] The sensor is an acoustic or ultrasonic sensor that is fixedly installed on a rotating machine (motor, pump, turbine) and continuously collects acoustic data generated during operation. The noise analysis unit (112) can transmit the collected data to a server computer (120) or a SCADA system via wiring or wireless communication. The server computer (120) or the SCADA system can continuously compare the current data with a set reference spectrum and generate an alarm if an abnormal sign exceeding a threshold is detected.

[0068] The server computer (120) extracts various features, such as RMS, peak value, energy in a specific frequency band, and entropy, from the raw signal in the time domain. Additionally, the entire spectrum or time-frequency pattern is treated as a 2D image and used as input to a deep learning neural network. For example, a one-dimensional signal can be processed by a CNN (Convolutional Neural Network) to learn the features of anomalies, or the time-series pattern of the signal can be configured to be analyzed by an LSTM (Long Short-Term Memory) network. The first machine learning model (121) trained in this way can classify whether new sensor data is in a normal state or an abnormal state in real time. The first machine learning model (121) can also perform detailed diagnoses, such as classifying bearing defect types and predicting the degree of gear damage, and the diagnosis results can be provided in the form of a dashboard or linked with an equipment control system to be used for automatic action.

[0069] As illustrated in FIG. 2, the fault diagnosis device (200) may include a terminal device (210) and a server computer (220).

[0070] The terminal device (210) may each include a three-phase power and arc state monitoring unit (211), a noise and vibration analysis unit (212), a control unit (213), a storage unit (214), a display unit (215), and a communication unit (216). The remaining components, excluding the three-phase power and arc state monitoring unit (211), and the server computer (220) may be identical to the configuration described in FIG. 1.

[0071] The three-phase power and arc state monitoring unit (211) is connected to an electrical line connected to a rotating machine using three-phase power, and can measure the three-phase voltage and current. From the measured voltage and current information, it can analyze the quality of the power and detect an arc. The three-phase power and arc state monitoring unit (211) may include a plurality of single-phase power and arc state monitoring units (111). For example, it may include three single-phase power and arc state monitoring units (111). The three-phase power and arc state monitoring unit (211) includes a main calculation unit, and the main calculation unit can calculate arc state, channel, leakage current, and power state information from the measured voltage and current information and process the data.

[0072] FIGS. 3 and FIGS. 4 illustrate each step of a fault diagnosis method for a rotating machine using vibration signals and current signals according to an embodiment of the present invention.

[0073] Figure 3 is a method for diagnosing whether a rotating machine using single-phase power is faulty.

[0074] The noise and vibration analysis unit (112) measures the vibration generated from the rotating shaft included in the rotating machine. (Step S1010)

[0075] The single-phase power and arc state monitoring unit (111) measures the single-phase voltage of the rotating machine (step S1020) and measures the single-phase current of the rotating machine (step S1030).

[0076] Steps S1010 through S1030 may be executed in a different order.

[0077] The server computer (120) transmits the result data of the first machine learning model (121) to the terminal device (110). Then, the terminal device sets the second machine learning model using the parameters included in the received data. (Step S1040)

[0078] The control unit (113) or the main calculation unit inputs the measured vibration information of the rotating machine into the second machine learning model, and the second machine learning model outputs vibration analysis information of the rotating machine. (Step S1050) For example, information analyzing whether there is an abnormality in the rotating shaft or bearing configuration of the rotating machine can be obtained.

[0079] The control unit (113) or the main calculation unit inputs the measured single-phase voltage and single-phase current information of the rotating machine into the second machine learning model, and the second machine learning model outputs power quality analysis information of the rotating machine. (Step S1060) For example, information analyzing the voltage and current status of the rotating machine and whether an arc is detected can be obtained.

[0080] Steps S1050 and S1060 can be executed in a different order.

[0081] The terminal device (110) processes the measured information or the information obtained by inputting it into the second machine learning model into data, and then transmits it to the server computer (120). (Step S1070)

[0082] The fault diagnosis device (100) can repeat each step for the same or different rotating machine and obtain vibration analysis information and power quality analysis information of the rotating machine.

[0083] Figure 4 is a method for diagnosing whether a rotating machine using three-phase power is faulty.

[0084] The noise and vibration analysis unit (212) measures the vibrations generated from the three rotating shafts included in the rotating machine. (Step S2010)

[0085] The three-phase power and arc state monitoring unit (211) measures the three-phase voltage of the rotating machine (step S2020) and measures the three-phase current of the rotating machine (step S2030).

[0086] Steps S2010 through S2030 may be executed in a different order.

[0087] The server computer (220) transmits the result data of the first machine learning model (221) to the terminal device (210). Then, the terminal device sets the second machine learning model using the parameters included in the received data. (Step S2040)

[0088] The control unit (213) or the main calculation unit inputs the measured vibration information of the rotating machine into the second machine learning model, and the second machine learning model outputs vibration analysis information of the rotating machine. (Step S2050) For example, information analyzing whether there is an abnormality in the rotating shaft or bearing configuration of the rotating machine can be obtained.

[0089] The control unit (213) or the main calculation unit inputs the measured three-phase voltage and three-phase current information of the rotating machine into the second machine learning model, and the second machine learning model outputs power quality analysis information of the rotating machine. (Step S2060) For example, information analyzing the voltage and current status of the rotating machine and whether an arc is detected can be obtained.

[0090] Steps S2050 and S2060 can be executed in a different order.

[0091] The terminal device (210) processes the measured information or the information obtained by inputting it into the second machine learning model into data, and then transmits it to the server computer (220). (Step S2070)

[0092] The fault diagnosis device (200) can repeat each step for the same or different rotating machine and obtain vibration analysis information and power quality analysis information of the rotating machine.

[0093] FIG. 5 shows the configuration of a single-phase power and arc state monitoring unit in one embodiment of the present invention.

[0094] The single-phase power and arc state monitoring unit (111) can analyze voltage and current, leakage current and arc current. The single-phase power and arc state monitoring unit (111) may include a current signal processing unit (11), a low-pass filter (12), a band-pass filter (13), a high-pass filter (14), an analog-to-digital converter (15, 22, 32), a frequency determination unit (16), a voltage signal processing unit (21), a voltage characteristic analysis unit (23), a leakage current signal processing unit (31), a leakage current analysis unit (33), a comparator (C), a counter (B), and a count analysis unit (D).

[0095] The current signal processing unit (11) can be connected to one of the pairs of electrical lines connected to the rotating machine. The current signal processing unit (11) can convert the current signal. For example, the current signal can be converted from the time domain to the frequency domain. Additionally, a current sensor can be included to measure the current signal input to the current signal processing unit (11).

[0096] The single-phase power and arc state monitoring unit (111) can separate current signals by frequency using a filter and detect arcs. The current signal passes through the current signal processing unit (11) and can be input to the low-pass filter (12), band-pass filter (13), and high-pass filter (14), respectively. Arcs occurring in the electric line can occur in all frequency bands, while switching noise or impulses can occur in specific frequency bands. Among these, the intensity of the arc may be high in the low-frequency band, and switching noise may occur frequently. The current signal processing unit (11) can convert the current signal input to the low-pass filter (12) to distinguish between arcs and non-arc impulses.

[0097] The low-pass filter (12) can pass a current signal having a frequency lower than or equal to a first reference frequency. The first reference frequency may be, for example, 2 kHz, and the low-pass filter (12) can pass a current signal having a frequency in the range of 0 Hz to 2 kHz.

[0098] The current signal passing through the low-pass filter (12) can be converted into a digital signal through an analog-to-digital converter (15). The converted digital signal can be input to a frequency determination unit (16). The frequency determination unit (16) can analyze the frequency band of the current signal passing through the low-pass filter (12) and then output the analyzed frequency band information.

[0099] A band-pass filter (13) can pass a current signal having a specific frequency band. For example, a band-pass filter (13) can pass a current signal with a frequency of 1.95 kHz to 100 kHz. A single-phase power and arc state monitoring unit (111) may include a plurality of band-pass filters (13), and each band-pass filter (13) can pass a current signal having a different frequency band. For example, among the first to n frequency bands divided into n parts between a first reference frequency and a second reference frequency, the first to n band-pass filters can pass a current signal having a frequency band corresponding to itself. (A first frequency band can be assigned to the first band-pass filter, ..., and an i-th frequency band can be assigned to the i-th band-pass filter.)

[0100] A plurality of comparators (C) can be connected in parallel for each band-pass filter (13). And the output signal of the band-pass filter (13) can be input to the plurality of comparators (C).

[0101] Different reference values ​​can be pre-set for each comparator (C). The comparator (C) can compare the output signal magnitude of the bandpass filter (13) with the set reference value.

[0102] A counter (B) can be connected in series to each comparator (C). The counter (B) can count when the output signal magnitude of the bandpass filter (13) is greater than a reference value set in the connected comparator (C). For example, the count can be increased by one.

[0103] The count analysis unit (D) can be connected in parallel with a plurality of counters (B). The counter analysis unit (D) checks the count of the counters (B) during a preset time period, and when the count exceeds a preset reference value, it can transmit a signal to the main calculation unit to notify it.

[0104] The high-pass filter (14) can pass a current signal having a frequency greater than or equal to a second reference frequency. The second reference frequency may be, for example, 95 kHz, and the high-pass filter (14) can pass a current signal greater than or equal to 95 kHz.

[0105] The configuration connected to the high-pass filter (14) may be the same as the configuration connected to the band-pass filter (13). The count analysis unit (D) provided in correspondence with the high-pass filter (14) can transmit a signal to the main calculation unit to notify when the number of output signal magnitudes of the high-pass filter (14) that are greater than the first reference value of the comparator (C) exceeds the second reference value during a set time.

[0106] The voltage signal processing unit (21) can be connected between electrical lines connected to the rotating machine. The voltage signal processing unit (21) can convert the voltage signal. For example, it can convert the voltage signal from the time domain to the frequency domain. Additionally, it can include a voltage sensor to measure the voltage signal input to the voltage signal processing unit (21).

[0107] The voltage signal processing unit (21) can be connected to an analog-to-digital converter (22), and the analog-to-digital converter (22) can be connected to a voltage characteristic analysis unit (23). The voltage characteristic analysis unit (29) can analyze the voltage signal to distinguish between an arc and a non-arc impulse.

[0108] The leakage current signal processing unit (31) can be connected between electrical lines connected to the rotating machine. The leakage current signal processing unit (31) can measure the current input to and output to the rotating machine. The leakage current signal processing unit (31) can convert the signal of the measured leakage current. For example, the leakage current signal can be converted from the time domain to the frequency domain. Additionally, a leakage current sensor can be included to measure the leakage current signal input to the leakage current signal processing unit (31).

[0109] The leakage current signal processing unit (31) can be connected to an analog-to-digital converter (32), and the analog-to-digital converter (32) can be connected to a leakage current analysis unit (33). The leakage current analysis unit (33) can analyze the frequency of the converted leakage current signal to determine whether leakage current has occurred in an electrical line or rotating machine.

[0110] The main component receives frequency analysis information of a current signal that has passed through a low-pass filter (12), information on whether the number of output signals of a band-pass filter (13) or a high-pass filter (14) that are greater than a first reference value has exceeded a second reference value during a set time, analysis information of a voltage signal, and analysis information of a leakage current signal, calculates arc state, channel, leakage current, and power state information, and then processes the calculated information.

[0111] The main component can detect whether an arc has occurred by comparing the output signal magnitude of a filter with a reference value set in one of the multiple comparators (C), and evaluate the quality of power by comparing the output signal magnitude of a filter with a reference value set in the remaining comparators (C). The reference value set in the comparator (C) used to detect whether an arc has occurred may be higher than the reference value set in the remaining comparators (C).

[0112] The main component can evaluate the quality of power by analyzing the number of changes in the output signal magnitude of the band-pass filter (13) or high-pass filter (14) that is greater than the first reference value of the comparator (C) during a set time.

[0113] The storage unit (114) stores information calculated by the main calculation unit, and the display unit (115) can display information calculated by the main calculation unit. The communication unit (115) can transmit information calculated by the main calculation unit to a server computer (120) or an external device.

[0114] FIG. 6 illustrates the operation method of a single-phase power and arc state monitoring unit in one embodiment of the present invention.

[0115] The sensor measures the current, voltage, and leakage current in the electrical line connected to the rotating machine, and transmits the measured signals to the current signal processing unit (11), voltage signal processing unit (21), and leakage current signal processing unit (31), respectively. (Step S3010)

[0116] The voltage signal processing unit (21) transmits the voltage measurement signal to the analog-to-digital converter (22) and converts it into a first digital signal. (Step S3020)

[0117] The voltage characteristic analysis unit (23) classifies the voltage value in the first digital signal into instantaneous value, RMS value, and average value, and analyzes the characteristics. (Step S3030)

[0118] The leakage current signal processing unit (31) transmits the leakage current measurement signal to the analog-to-digital converter (32) and converts it into a second digital signal. (Step S3040)

[0119] The leakage current analysis unit (33) classifies the leakage current value in the second digital signal into instantaneous value, RMS value, and average value, and analyzes the characteristics thereof. (Step S3050)

[0120] The current signal processing unit (11) transmits the current measurement signal to a low-pass filter (12), a band-pass filter (13), and a high-pass filter (14). (Step S3060)

[0121] The signal output by the low-pass filter (12) is transmitted to the analog-to-digital converter (15) and converted into a third digital signal. (Step S3070)

[0122] The frequency determination unit (16) analyzes the quality and frequency characteristics of the current in the third digital signal. (Step S3080)

[0123] For the signals output by the band-pass filter (13) and the high-pass filter (14), the comparator (C) compares them with a first reference value. (Step S3090)

[0124] When the signal output by the band-pass filter (13) and the high-pass filter (14) is greater than the first reference value, the counter (B) increases the count by one. (Step S3100)

[0125] The count analysis unit (D) checks the count counted by the counter (B) during a preset time, and if the checked count exceeds a second reference value, it transmits a signal to the main calculation unit. (Step S3110)

[0126] Using the information collected from the main production unit, the failure of the rotating machine is determined, and the determination result is transmitted to the server computer (120). (Step S3120) The main production unit inputs the collected information into a second machine learning model, and the second machine learning model can calculate and output information diagnosing whether the rotating machine is faulty.

[0127] The first set step consisting of steps S3020 to S3030, the second set step consisting of steps S3040 to S3050, and the third set step consisting of steps S3060 to S3110 may be executed in a different order. And all steps may be executed repeatedly.

[0128] FIG. 7 shows the configuration of a three-phase power and arc state monitoring unit in one embodiment of the present invention.

[0129] The three-phase power and arc state monitoring unit (211) may include three single-phase power and arc state monitoring units (111), a three-phase voltage and current measurement circuit unit (41), and a leakage current sensor (42). The single-phase power and arc state monitoring unit (111) may be identical to the configuration shown in FIG. 5.

[0130] The three-phase voltage and current measurement circuit (41) is connected to the three-phase AC power line (L) of the rotating machine and can measure the current flowing through the line (L) or the voltage between the lines (L). For example, the first single-phase power and arc state monitoring unit can measure the voltage between line A and the neutral line (not shown) of the rotating machine and the current of line A or line B. The second single-phase power and arc state monitoring unit can measure the voltage between line B and the neutral line of the rotating machine and the current of line B or line C. The third single-phase power and arc state monitoring unit can measure the voltage between line C and the neutral line of the rotating machine and the current of line C or line A.

[0131] Alternatively, the first single-phase power and arc state monitoring unit may measure the voltage between the ends of line A and line B of the rotating machine (inter-phase voltage between phase A and phase B), the second single-phase power and arc state monitoring unit may measure the voltage between the ends of line B and line C of the rotating machine (inter-phase voltage between phase B and phase C), and the third single-phase power and arc state monitoring unit may measure the voltage between the ends of line C and line A of the rotating machine (inter-phase voltage between phase C and phase A).

[0132] The leakage current sensor (42) can be positioned between the lines (L) and can measure the current input to and output to the rotating machine. The leakage current sensor (42) can determine whether leakage current has occurred by calculating the difference in magnitude of the measured current.

[0133] The single-phase power and arc state monitoring unit (111) can analyze the quality of power and detect arcs in response to a single-phase power source combined by two lines.

[0134] FIG. 8 shows the frequency band that the filter passes through in one embodiment of the present invention.

[0135] As shown in (a), the frequency bands of electrical signals passed by the low-pass filter, band-pass filter, and high-pass filter may overlap. For example, if the frequency band of the low-pass filter is set to 0 to 2 kHz, the frequency band of the band-pass filter to 1.95 kHz to 100 kHz, and the frequency band of the high-pass filter to 95 kHz or higher, the 1.95 kHz to 2 kHz and 95 kHz to 100 kHz bands may overlap. By setting the frequency band of the band-pass filter to 1.5 kHz to 125 kHz to expand the area where the frequency bands overlap, the detection rate of the arc can be increased.

[0136] When an arc occurs, pulses can be detected across all frequency bands. As the current magnitude decreases, pulses can be detected primarily in electrical signals within the high-frequency band. Conversely, as the current magnitude increases, pulses can be detected across all frequency bands.

[0137] As shown in (b), the frequency bands of the electrical signals passed by the low-pass filter, band-pass filter, and high-pass filter can be set so that they do not overlap. Each of the multiple band-pass filters can divide the frequency band between the low-frequency band and the high-frequency band to pass electrical signals having their respective set frequency bands.

[0138] At this time, noise generated by a load operating in a specific frequency band can be distinguished from an arc.

[0139] FIGS. 9 to 11 show the waveform of an arc detection signal in one embodiment of the present invention.

[0140] The waveform shown in FIG. 9 represents an arc detection signal when a series arc occurs when a rotating machine uses AC power. The first waveform (W1) located at the top represents the voltage across the device where the series arc occurred, allowing the point in time when the series arc occurred to be identified. For example, it can be determined that a series arc occurred at the point when the voltage rises rapidly.

[0141] The second waveform (W2) and the third waveform (W3), located at the second and third positions, are waveforms representing the output of the filter to detect a series arc. It can be seen that the magnitude of the third waveform (W3) increases after the series arc occurs and the interruption time has elapsed.

[0142] The fourth waveform (W4) located at the bottom is a waveform representing the output of the serial arc detection signal, and it can be seen that the first serial arc occurs and after the interruption time has passed, the interruption trigger signal is generated.

[0143] The waveforms shown in FIG. 10 represent the output signals of filters for each frequency band when a series arc occurs when a rotating machine uses AC power. The fifth waveform (W5) located at the bottom represents the voltage across the device where the series arc occurred, and the point in time when the series arc occurred can be identified by observing the rapid change in its magnitude.

[0144] The sixth waveform (W6) located at the top is a waveform representing the signal output from a low-pass filter. The seventh waveform (W7) located below the sixth waveform (W6) is a waveform representing the signal output from a band-pass filter. The eighth waveform (W8) located below the seventh waveform (W7) is a waveform representing the signal output from a high-pass filter. The circuit breaker can operate at the point where the waveform of the series arc generation is interrupted.

[0145] The waveform shown in FIG. 11 represents an arc detection signal when a series arc occurs when a rotating machine uses a DC power source. The ninth waveform (W9) located at the top represents the voltage across the device where the series arc occurred, allowing the point in time when the series arc occurred to be identified. For example, it can be determined that a series arc occurred at the point when the voltage drops rapidly.

[0146] The 10th waveform (W10) and the 11th waveform (W11), located at the second and third positions, represent the output of the filter for detecting a series arc. The 12th waveform (W12), located at the bottom, represents the cutoff signal after a series arc is detected. It can be seen that the cutoff trigger signal is generated after the first series arc occurs and the cutoff time has elapsed.

[0147] FIG. 12 shows the configuration of a first or second machine learning model in one embodiment of the present invention.

[0148] The first or second machine learning model receives information analyzed by the single-phase power and arc state monitoring unit (111) or the three-phase power and arc state monitoring unit (211) and vibration information measured by the noise and vibration analysis unit (112, 212), and can calculate and output whether the rotating machine is faulty.

[0149] Multiple nodes (n) included in the input layer (I) i Each of the nodes can store information input to a first or second machine learning model. For example, each of the nodes can store frequency analysis information of a current signal that has passed through a low-pass filter (12), a second node can store information on whether the number of output signals of a band-pass filter (13) or a high-pass filter (14) that are greater than a first reference value has exceeded a second reference value during a set time period and the number of such values, analysis information of a voltage signal, analysis information of a leakage current signal, and vibration information measured by a noise and vibration analysis unit.

[0150] One node (n) in the input layer (I) i Each node included in the ) and the intermediate layer (M) is multiplied by a weight, the multiplied values ​​are summed, and the result of inputting the summed value into an activation function is obtained, and the node (n) of the input layer (I) i Node (n) of the intermediate layer (M) corresponding to ) m You can set the value of ).

[0151] The first hidden layer (H (1) In ) one node (n h(1) Corresponding to ), each node included in the intermediate layer (M) is multiplied by a weight, the multiplied values ​​are summed, and the result of inputting the summed value into an activation function is obtained, thereby obtaining the first hidden layer (H (1) that node (n) of ) h(1) You can set the value of ).

[0152] The value of that node in the i-th hidden layer can be set by multiplying each node in the i-1-th hidden layer by a weight corresponding to one node in the i-th hidden layer, summing the multiplied values, and then inputting the summed value into an activation function to obtain the result.

[0153] In the output layer (O), one node (n o Corresponding to ), the last hidden layer (H (k) Each node included in ) is multiplied by a weight, the multiplied values ​​are summed, and the result of inputting the summed value into an activation function is obtained, and that node (n) of the output layer (O) o You can set the value of ).

[0154] Each node of the output layer (O) can represent information on the probability of occurrence according to the type of failure. For example, each node can numerically represent the degree of arc generation, the degree of vibration generation, the degree of noise generation, the degree of heat and smoke generation, and the degree of twisting in a rotating machine.

[0155] The terminal device (110) or server computer (120) can calculate a value corresponding to a mechanical or electrical failure of the rotating machine from a value stored in a node of the output layer (O). The calculated value can be stored in a storage unit (114) or a database management system (122), or displayed through a display unit (115).

[0156]

[0157] Although embodiments of the present invention have been described above, the present invention is not limited to the above embodiments. Various modifications may be made within the scope of the detailed description of the invention and the attached drawings, provided that such modifications do not depart from the spirit of the invention and do not impair its effects. Furthermore, it is obvious that such embodiments fall within the scope of the present invention.

Claims

1. A single-phase or three-phase power and arc state monitoring unit connected to an electrical line connected to a rotating machine using single-phase or three-phase power, measuring single-phase or three-phase voltage and current, analyzing single-phase or three-phase power quality, and detecting arcs; and A terminal device comprising a noise and vibration analysis unit that measures vibrations generated in a rotating shaft included in a rotating machine and analyzes the degree of vibration, and Diagnosing whether a rotating machine is faulty from the information output by the single-phase or three-phase power and arc state monitoring unit and the noise and vibration analysis unit. Fault diagnosis device for rotating machinery.

2. In Claim 1, The above single-phase or three-phase power and arc state monitoring unit includes a main output unit, and The above main calculation unit calculates arc state, channel, leakage current, and power state information from measured voltage and current information. Fault diagnosis device for rotating machinery.

3. In Claim 2, The above single-phase or three-phase power and arc state monitoring unit includes a low-pass filter, a high-pass filter, and a plurality of band-pass filters, and A current signal is input to the low-pass filter, the high-pass filter, and the plurality of band-pass filters to separate it by frequency and detect an arc. Fault diagnosis device for rotating machinery.

4. In Claim 3, The above single-phase or three-phase power and arc state monitoring unit is, A plurality of comparators connected to the band-pass filter or the high-pass filter, each having a different first reference value preset, and comparing the output signal magnitude of the band-pass filter or the high-pass filter with the first reference value; A counter connected to the above comparator, which increases the count by one when the output signal magnitude of the band-pass filter or the high-pass filter is greater than a first reference value set in the connected comparator; and A count analysis unit connected to the above counter, checking the number counted by the counter for a preset time, and transmitting a signal to the main calculation unit if it exceeds a preset second reference value; Fault diagnosis device for rotating machinery.

5. In Claim 1, A first machine learning model trained to diagnose whether a rotating machine is faulty; A server computer further comprising a database management system for storing data. Fault diagnosis device for rotating machinery.

6. In Claim 5, The above server computer transmits the result data of training the above first machine learning model to a terminal device, and The terminal device includes a second machine learning model and sets the second machine learning model using parameters included in the result data. Fault diagnosis device for rotating machinery.

7. In Claim 6, The terminal device inputs data output by the single-phase or three-phase power and arc state monitoring unit and the noise and vibration analysis unit into the second machine learning model, and The above second machine learning model outputs information diagnosing whether the rotating machine is faulty in response to input data. Fault diagnosis device for rotating machinery.

8. A step in which a terminal device measures vibrations occurring in a rotating shaft included in a rotating machine; A step in which a terminal device measures the voltage and current of a rotating machine; A step of transmitting result data obtained by a server computer training a first machine learning model to a terminal device, and setting a second machine learning model using parameters included in the data received by the terminal device; A step in which a terminal device inputs vibration information of a rotating machine measured into a second machine learning model, and the second machine learning model outputs vibration analysis information of the rotating machine; and A step comprising: a terminal device inputting voltage and current information of a rotating machine measured into a second machine learning model, and the second machine learning model outputting power quality analysis information of the rotating machine. Method for diagnosing faults in rotating machinery.

9. In Claim 8, In the step where the above terminal device measures the voltage and current of the rotating machine, A step in which a sensor measures current, voltage, and leakage current in a single-phase line of a rotating machine, and transmits the measured signals to a current signal processing unit, a voltage signal processing unit, and a leakage current signal processing unit, respectively; A step in which a voltage signal processing unit transmits a voltage measurement signal to an analog-to-digital converter and converts it into a first digital signal; A step in which a voltage characteristic analysis unit classifies voltage values ​​in a first digital signal into instantaneous values, RMS values, and average values, and analyzes their characteristics; A step in which a leakage current signal processing unit transmits a leakage current measurement signal to an analog-to-digital converter and converts it into a second digital signal; A step in which a leakage current analysis unit classifies leakage current values ​​from a second digital signal into instantaneous values, RMS values, and average values, and analyzes their characteristics; A step in which a current signal processing unit transmits a current measurement signal to a low-pass filter, a band-pass filter, and a high-pass filter; A step of transmitting the signal output by a low-pass filter to an analog-to-digital converter and converting it into a third digital signal; A step in which a frequency determination unit analyzes the quality and frequency characteristics of the current in the third digital signal; A step in which a comparator compares the signals output by the band-pass filter and the high-pass filter with a first reference value; A step in which, when the signal output by the band-pass filter and the high-pass filter is greater than the first reference value, the counter increases the count by one; and A step comprising: a count analysis unit checking the number counted by the counter during a preset time, and transmitting a signal to the main calculation unit if the checked number exceeds a second reference value. Method for diagnosing faults in rotating machinery.

Images