Motor abnormality diagnosis device, abnormality diagnosis method, and abnormality diagnosis program

Abstract

To perform real-time diagnosis of normality and abnormality of a motor on the basis of the degree of divergence between a first feature quantity related to the motor state quantity such as a current and a voltage calculated by a motor model consisting of mathematical expressions and determinants and a second feature quantity extracted from the motor state quantity of the actual machine.SOLUTION: A motor abnormality diagnosis device includes a detection unit 10 as a motor state quantity generation unit that detects at least a voltage applied to a motor and a current and speed of the motor, a motor model calculation unit 20 that calculates, as a motor model, a mathematical formula or determinant containing the values detected by the detection unit 10 and a motor constant, a first feature quantity extraction unit 30 that extracts a first feature quantity from motor state quantities such as a current and a voltage estimated using the motor model, a second feature quantity extraction unit 40 that extracts a second feature quantity from the same type of motor state quantity as that used for extracting the first feature quantity, a comparative diagnosis unit 50 that diagnoses normality / abnormality of the motor on the basis of the degree of divergence of the second feature quantity from the first feature quantity, and an anomaly notification unit 60 that notifies an abnormality when the abnormality is diagnosed.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an abnormality diagnosis technology for motors driven by power conversion devices, and more particularly to an abnormality diagnosis technology for motors that serve as the power sources of cranes, elevators, belt conveyors, compressors, fans / blowers, robots, etc., used in various buildings, manufacturing / processing machines, transport machines, conveying / loading devices, etc. [Background technology]

[0002] In the various fields mentioned above, motors are not only operated independently, but are often used in combination with power conversion devices such as inverters, PLCs (Programmable Logic Controllers), motion controllers, various sensors, and the like. Power conversion devices and motors are becoming increasingly specialized depending on the field of application, and in systems having a plurality of power conversion devices and motors, the contents of fault diagnosis and operation verification are becoming more complex.

[0003] Currently, a test environment called Power HILS (Hardware In The Loop System or Hardware In The Loop Simulator) is being used to efficiently verify the operation of power conversion equipment. This Power HILS uses real-time simulation to mimic the operation of the main circuit and motor of the power conversion equipment.

[0004] Normally, motor load devices (also called MG sets or dynamos) have been widely used to verify the operation of power conversion devices. This device has the problem of being expensive and large in size because it needs to be prepared for each capacity of the power conversion device. In addition, because it uses an actual motor, which is a rotating body, it needs to be monitored by workers from a safety standpoint.

[0005] Therefore, there is a growing trend to replace the operation of actual machines with electronic devices without the need for an actual motor (hereinafter also referred to as actual machine).For example, the "motor simulator" described in Patent Document 1 is a motor emulator that can reproduce various load conditions in real time and with actual power using a simulated load inverter and software without using an actual machine, and can electronically perform load tests on a motor under conditions equivalent to the actual mechanical load.

[0006] There are two methods for controlling the motor emulator. The first method is the "ItoV" method, which detects the current and calculates the induced voltage of the motor from the voltage equation. This method is used by many motor simulators because the calculation structure is simple. However, the power conversion device that can be used is limited to an inverter that has a current control system and is vector controllable, so it is not very versatile.

[0007] In contrast, the second method, the "VtoI" method, detects the voltage and current and performs current feedback control. Although this method requires complex calculations including differential equations and sensor circuits, it has the advantage that it can be applied to voltage-controlled V / f control inverters used in many motor drive systems.

[0008] On the other hand, from the perspective of preventive maintenance, there is an increasing need for Condition Based Maintenance (CBM), which constantly diagnoses the condition of motors and other equipment installed in factories to detect abnormalities at an early stage and perform maintenance as needed based on the results. For example, in the "preventive maintenance device for electric motor" described in Patent Document 2, evaluation data showing the relationship between the operation amount of the motor and the state amount acquired by a sensor is compared with a correlation evaluation model. In the "preventive maintenance device for electric motor" described in Patent Document 3, a correlation model between the operation amount and state amount when the motor is normal is stored in advance, and the state amount calculated based on the correlation model in response to the operation amount during actual operation is compared with the state amount acquired during operation. Each of these preventive maintenance devices detects motor abnormalities at the pre-symptomatic stage based on the respective comparison results, and encourages prompt maintenance.

[0009] As another diagnostic technique, there is known an "electric motor diagnostic device" as disclosed in Patent Document 4, which analyzes the power spectrum of the current flowing through the motor and detects an abnormality when the sideband wave of the averaged power spectrum exceeds a set value. Further, methods of diagnosing abnormalities include attaching a vibration sensor to the motor and performing spectral analysis of the sensor output signal to observe changes in specific frequency components, and comparing changes in the motor's power consumption with past data to diagnose abnormalities. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] Patent No. 4998693 [Patent Document 2] JP 2012-137386 A [Patent Document 3] JP 2013-223284 A [Patent Document 4] International Publication No. 2019 / 3389 Summary of the Invention [Problem to be solved by the invention]

[0011] The "motor simulator" described in Patent Document 1 simulates the behavior of a motor driven by an inverter using a separate inverter, and so the circuit configuration tends to become complicated. In the "electric motor preventive maintenance devices" described in Patent Documents 2 and 3, the correlation evaluation models and correlation models are constructed from approximations based on past data and experimental failure data, and there is a problem in that the abnormality factors that can be detected are limited. Furthermore, in Patent Document 4, the set value that is the standard for detecting an abnormality is also determined based on data from past motor failures, so if there are only a few failure cases, the accuracy of determining an abnormality is low. Some of the other conventional techniques are intended only for direct-start motors, and have the problem that they cannot be applied to power conversion devices such as inverters that are controlled by PWM (Pulse Width Modulation).

[0012] Therefore, an object of the present invention is to provide a motor abnormality diagnosis device, an abnormality diagnosis method, and an abnormality diagnosis program that can diagnose whether a motor is normal or abnormal in real time based on the degree of deviation between a first feature quantity related to a motor state quantity such as voltage or current calculated using a motor model consisting of an equation or a determinant, and a second feature quantity related to the motor state quantity obtained from an actual machine, with a motor driven by a power conversion device of various control methods as the diagnosis target. [Means for solving the problem]

[0013] In order to achieve the above object, the motor abnormality diagnosis device of the present invention includes a motor state quantity generation unit that generates as motor state quantities each value including at least an applied voltage equivalent value to a motor driven by a power conversion device, a current equivalent value flowing through the motor, and a motor speed equivalent value; a motor model calculation unit that calculates an equation or a determinant including the motor state quantities and a motor constant as a motor model; a first feature quantity extraction unit that extracts a first feature quantity from the motor state quantities such as current and voltage estimated and calculated using the motor model; a second feature quantity extraction unit that receives the motor state quantities generated by the motor state quantity generation unit without passing through the motor model calculation unit and extracts a second feature quantity based on a motor state quantity of the same type as the motor state quantity estimated and calculated by the first feature quantity extraction unit from the input motor state quantities; and a comparison diagnosis unit that diagnoses whether the motor is normal or abnormal based on the degree of deviation of the second feature quantity from the first feature quantity.

[0014] Here, the first feature extraction unit uses a motor model to estimate the current flowing through the motor from, for example, a detection value, command value, estimated value, etc. of the voltage applied to the motor among the motor state quantities, and extracts the first feature from a change in the estimated current. Alternatively, the first feature extraction unit uses a motor model to estimate the voltage applied to the motor from, for example, a detection value, command value, estimated value, etc. of the current flowing through the motor, and extracts the first feature from a change in the estimated voltage. The motor state quantities are input directly to the second feature quantity extraction unit without going through the motor model calculation unit. This second feature quantity extraction unit extracts the second feature quantity based on the same type of motor state quantity (current or voltage in the above example) as the motor state quantity estimated and calculated by the first feature quantity extraction unit, from the input motor state quantities.

[0015] Furthermore, the present invention also provides an abnormality diagnosis method comprising procedures executed by each unit of the abnormality diagnosis device described above, and a program for implementing the functions of each unit of the abnormality diagnosis device. Effect of the Invention

[0016] In the present invention, a first feature quantity is extracted using a motor model relating to the voltage, current, speed, etc. of the motor to be diagnosed, and a second feature quantity is extracted from the detection values ​​and command values ​​of the current, voltage, etc. of the motor during operation, and the deviation between these first and second feature quantities is used as a judgment criterion to diagnose whether the motor is normal or abnormal online. If the motor is diagnosed as abnormal, the current, voltage, and other values ​​and their changes can be analyzed in detail to identify the cause of the abnormality, and measures can be taken to quickly repair or remove the abnormal part, minimizing damage. In addition, by installing the abnormality diagnosis function according to the present invention inside a power conversion device or in an external PLC, cloud server, etc., it is also possible to apply it to abnormality diagnosis of mechanical equipment equipped with multiple motors. [Brief description of the drawings]

[0017] [Figure 1] 1 is a block diagram of an abnormality diagnosis device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a block diagram showing a case where the embodiment of FIG. 1 is applied to a sensorless vector control system for a PM motor. [Diagram 3] FIG. 2 is a block diagram showing a case where the embodiment of FIG. 1 is applied to a motor emulator. [Figure 4] FIG. 1 is a diagram showing typical failure modes of a motor. [Diagram 5] 2 is a flowchart showing an abnormality diagnosis operation by the abnormality diagnosis device of FIG. 1. [Figure 6] 2 is a configuration diagram showing an example of a power conversion device incorporating the abnormality diagnostic device of FIG. 1. [Figure 7] FIG. 1 is a configuration diagram showing an application example of an abnormality diagnosis device according to the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. 1 is a block diagram of a motor abnormality diagnosis device according to an embodiment of the present invention. The abnormality diagnosis device is composed of a detection unit 10, a motor model calculation unit 20, a first feature amount extraction unit 30, a second feature amount extraction unit 40, a comparison diagnosis unit 50, and an abnormality notification unit 60.

[0019] In the above configuration, the functions of each unit other than the detection unit 10 are realized, for example, by a program implemented in a processor in a power conversion device that drives a three-phase motor as a diagnosis target. As the processor, for example, a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), or a microcomputer capable of high-speed calculations can be used. It should be noted that a diagnostic device may be provided independently outside the power conversion device, and the above program may be implemented in a DSP or FPGA built into the diagnostic device to realize the functions of each unit other than the detection unit 10. Next, the function of each part in FIG. 1 will be described in order.

[0020] (1) Detection unit 10 The detection unit 10 detects at least motor state quantities such as current, voltage, speed, etc., for a three-phase motor that is an actual machine. If necessary, the temperature may be detected as a motor state quantity in order to correct the winding resistance of the motor. The detection unit 10 has a function of detecting a motor state quantity as an example of a motor state quantity generation unit, but the motor state quantity generation unit of the present invention may be provided with a function of generating a command value of the motor state quantity, or a function of generating an estimated value of the motor state quantity. That is, in the case where the detection unit 10 is not provided and, for example, the voltage applied to the motor (output voltage of the power conversion device) cannot be detected, an output voltage command value or an estimated value obtained from the control circuit of the power conversion device may be used instead of the voltage detection value. Furthermore, in the case where the speed of the motor cannot be detected, a speed command value or an estimated value obtained from the control circuit may be used instead of the speed detection value. That is, in the present invention, it is sufficient that the motor state quantity generating unit obtains the detected values, command values, or estimated values ​​of the current, voltage, speed, etc. as the motor state quantities related to the motor to be diagnosed. Note that the motor state quantities are not limited to the detected values, command values, or estimated values ​​of the current, voltage, speed, etc., but may also be "unit values" indicating the ratios to the values ​​(100%) during rated operation. In the embodiment shown in FIG. 1, the detection unit 10 is assumed to obtain detection values ​​of the current, voltage, speed, etc. of the motor.

[0021] When driving a motor by PWM control of a power conversion device, the pulsed PWM voltage applied to the motor is detected. This detected voltage may be a line voltage or a phase voltage, but when a line voltage that is easy to detect is obtained, it is desirable to convert the detected line voltage to a phase voltage. To obtain a pulsed PWM voltage, for example, a ΔΣ modulator can be used to convert it into a digital value, or a low-pass filter can be used in consideration of the control system. If the power conversion device outputs a sine wave voltage, the detected voltage can be used as is. The current flowing through the motor is detected using a hall sensor or shunt resistor, and converted into a digital value using an A / D converter. The motor speed is detected using an encoder that outputs A-phase and B-phase pulses, a resolver, etc. The speed detection method may be either optical or magnetic. The detection values ​​of current, voltage, speed, etc., obtained by the detection unit 10 described above are input to a motor model calculation unit 20 and a second feature amount extraction unit 40.

[0022] (2) Motor model calculation unit 20 The motor model calculation unit 20 calculates a motor model as a mathematical expression or a determinant using each detection value input from the detection unit 10 and motor constants. Here, the motor constants are the motor impedance (winding resistance and inductance), flux linkage, etc. These motor constants may use known values, or may use values ​​obtained by offline or online tuning.

[0023] Although each detection value and motor constant by the detection unit 10 is obtained as a value of a stationary coordinate system such as U, V, W phase, etc., it is common to use it by replacing it with the well-known d, q axis coordinate system (the d axis is the control axis in the magnetic flux direction of the motor, and the q axis is the control axis perpendicular to the d axis), which is a rotating coordinate system. For this reason, in this embodiment, the current, voltage, inductance, etc. in the formulas and determinants are expressed as values ​​in the d, q axis coordinate system. Of course, there is no problem in expressing the current, voltage, inductance, etc. as values ​​in the stationary coordinate system.

[0024] When the motor to be diagnosed is a PM motor that is vector-controlled by a power conversion device, the voltage equation in the d, q coordinate system is expressed by, for example, Equation 1.

number

[0025] As a differential equation for obtaining current from the voltage equation of Equation 1, by transforming the input signal into a voltage form, we obtain, for example, Equation 2.

number

[0026] On the other hand, when considering a squirrel-cage induction motor as a representative example of a motor that does not use permanent magnets, the voltage equation in the d, q axis coordinate system is expressed as Equation 3.

number

[0027] A differential equation for obtaining the d-axis and q-axis components of the primary current i1 and the secondary current i2 from the voltage equation of Equation 3 is expressed by Equation 4.

number

[0028] As described above, the current flowing through the motor can be calculated by applying the motor voltage, angular velocity, etc. detected by the detection unit 10 together with the motor constants to the differential equations of Equation 2 and Equation 4 and solving them. Note that the differential equation may use magnetic flux instead of current. Differential equations, i.e. f(x,y)=dy / dx or f(x n ,y n )=dy n / dx n To solve this problem, the well-known Euler method, Runge-Kutta method, or the like may be used. By the way, the Euler method is expressed as Equation 5, and the Runge-Kutta method is expressed as Equation 6.

number

number

[0029] By solving differential equations quickly using these methods, it is possible to estimate the current flowing through the motor. In order to estimate the motor current in real time, it is desirable to use the aforementioned DSP or FPGA, or a microcomputer capable of high-speed calculations. In this way, the motor model calculation unit 20 obtains a current estimate value in real time using the motor model, and outputs the calculation result including this current estimate value to the first feature extraction unit 30.

[0030] When the power conversion device performs current control such as vector control, the voltage applied to the motor can be estimated by directly using the above-mentioned voltage equation for a PM motor (Formula 1) or the voltage equation for an induction motor (Formula 3) depending on the type of motor, and the calculation result including this voltage estimate is output to the first feature extraction unit 30.

[0031] Here, FIG. 2 is a block diagram in which this embodiment is applied to a sensorless vector control system for a PM motor. In this control system, a real voltage is generated by controlling a main circuit 100 of the power conversion device according to a voltage command value calculated via a speed regulator 92 and a current regulator 93, and an adding / subtracting means 94 calculates the deviation between an actual current when the real voltage is applied to a motor M of a real machine and a current estimate value calculated by a motor model MM in a control circuit of the power conversion device according to the above-mentioned real voltage. The above-mentioned motor model MM is expressed, for example, by the following formula 7.

number

[0032] The motor model MM in FIG. 2 calculates the voltage command value (v d ,v q ), motor constants, and speed estimate ω r ∧etc., current i d ,i q Estimate. The current deviation calculated by the addition / subtraction means 94 is input to a position / speed estimator 95 to obtain the estimated speed value ω r ∧ , and this speed estimate ω r ∧ The rotor pole position is estimated by integrating the speed command value and the speed estimate value ω r ∧ The deviation from this is calculated by an adding / subtracting means 91, and a speed regulator 92 calculates a current command value so as to eliminate this deviation.

[0033] Next, a method for obtaining the mechanical load (load torque) driven by the motor will be described. For example, the torque of a PM motor is expressed by Equation 8. [Number 8] T=p o {φi q +(L d -L q )i d i q} where T is the torque, p o is the number of pole pairs, and φ is the flux linkage. Moreover, the relationship between speed and power is expressed by Equation 9. [Number 9] P=2π N T Here, P is the power, N is the rotational speed (rpm) [r / min], and T is the torque [N / m].

[0034] Furthermore, the angular velocity calculated using the inertia, the generated torque, and the load torque is given by Equation 10. [Number 10] ω m =(1 / J)∫(T M -T L ) Here, ω m is the angular velocity, J is the inertia, T M is the generated torque, T L is the load torque.

[0035] If a load torque observer is designed using the above, the generated torque T can be calculated from the transition of the motor speed and the power applied to the motor. M Calculate the load torque T L can be estimated. In addition, if the motor is equipped with a speed sensor, both sides of formula 10 are differentiated to obtain formula 11, which is then transformed to formula 12, and the angular velocity ω detected by the speed sensor is expressed as m Enter this into equation 12 to obtain the load torque T L It is also possible to calculate [Number 11] (d / dt)ω m =(1 / J)(T M -T L ) [Number 12] T L =T M -J(d / dt)ω m

[0036] Load torque observers that estimate load torque have been proposed for a long time and are mainly used to improve the accuracy of motor control. By incorporating this load torque observer into the control circuit 210 of the motor emulator shown in Fig. 3, for example, it is possible to estimate the inertia and load torque from the torque command and the current rotation speed, and reproduce the behavior of the motor that drives the mechanical load (load torque).

[0037] The motor emulator shown in FIG. 3 controls the power conversion device 200 and the power conversion device 230 under test, which are connected via a coupling network 220, so as to provide the power conversion device 230 with a motor driven by a three-phase AC power source G as a simulated load. That is, the control circuit 210 is set with a motor constant of the motor to be simulated by the operation of the power conversion device 200. The control circuit 210 uses the current detection value by the current detector 222 and the motor constant to calculate simulated speed / position detection signals so that the voltage detection value by the voltage detector 221 becomes equal to the voltage calculated from the voltage equation of the motor to be simulated, and provides the simulated speed / position detection signals to the power conversion device under test 230. The power conversion device under test 230 performs feedback control using the speed / position detection signals, and creates a state similar to that when the motor is actually rotated.

[0038] (3) First feature extraction unit 30 and second feature extraction unit 40 1, the first feature extraction unit 30 extracts the current estimation value itself input from the motor model calculation unit 20, or the change (transition) tendency of the current estimation value obtained by various methods described below, the state of pulsation, etc. as a first feature. The second feature detection unit 40 extracts the current detection value itself from each detection value input from the detection unit 10, or the change tendency of the current detection value, the state of pulsation, etc. as a second feature. The first and second feature amounts are not limited to the current information described above, but may be the motor voltage, speed or load torque itself, or the change (transition) tendency of these values, the state of pulsation, or the like.

[0039] When the feature is, for example, current information, the method of extracting the feature may involve extracting the amount of change, rate of change, average value, etc. along the time axis, or removing the time axis before performing the analysis. Methods for removing the time axis may include any of the following: a method using a Lissajous figure, a waveform counting method such as the rainflow method or the range pair method, frequency spectrum analysis using FFT (Fast Fourier Transform), a method using a moving average or a notch filter that extracts signals of specific frequencies, multivariate analysis, etc.

[0040] As a specific example, when the first and second feature quantities are current information, the three-phase currents of U, V, and W phases are transformed into a stationary coordinate system and expressed as i u ,i v ,iw Then, converting these to Id, Iq in the d, q axis coordinate system, which is a rotating coordinate system, a Lissajous figure is created. The Lissajous figure is then treated as a point cloud and subjected to multivariate analysis to determine the size and center of gravity of the point cloud and the slope component of the principal component of the point cloud using independent principal component analysis, and the size and slope of the slope component are then determined for each of the first and second feature quantities. When using the waveform counting method, the number of cycles of the amplitude is calculated for each of the first and second feature amounts for each motor speed using the waveform counting method, and a frequency distribution is created. The transition of this frequency distribution may be calculated for each of the first and second feature amounts, and used for diagnosis in the comparison diagnosis unit 50 described below.

[0041] (4) Comparison and diagnosis unit 50 The comparative diagnosis unit 50 constantly compares the first and second feature amounts to diagnose whether the motor is normal or abnormal. That is, if the first and second feature amounts match (if the deviation of the second feature amount from the first feature amount is zero), the comparative diagnosis unit 50 diagnoses the motor as normal, and if the first and second feature amounts do not match (if the deviation of the second feature amount from the first feature amount is equal to or greater than a predetermined threshold), the comparative diagnosis unit 50 diagnoses the motor as abnormal. Figure 4 shows typical failure modes as causes of motor abnormalities. These are merely examples, and it goes without saying that other failure modes also exist.

[0042] Possible criteria for the comparative diagnosis unit 50 to diagnose an abnormality include when the deviation becomes equal to or exceeds a threshold value as described above, and when the rate of increase in the difference between the first feature amount and the second feature amount becomes equal to or exceeds a predetermined value. In addition, the second feature amount may be input to a trained neural network that uses the first feature amount as training data, and the degree of deviation of the output value at that time from the training data may be evaluated to determine an abnormality.

[0043] If any of the typical failure modes shown in FIG. 4 are expected, a method may be adopted in which the failure mode is temporarily simulated to calculate a motor model, and the gain and vibration components are adjusted while being fed back so as to obtain the same waveform as that of the actual motor, and then whether or not there is an abnormality is determined. In addition, there is a certain relationship between changes in load torque and speed and changes in current, and normally the current increases in response to changes in load. Therefore, it is also possible to generate the first and second feature quantities by combining multiple types of signals and waveforms such as load torque, current, and speed, and then diagnose whether the motor is normal or abnormal by comparing these first and second feature quantities. Furthermore, there are rules and conditions for the current change pattern depending on the type of motor and the control method. If a situation occurs that does not follow these rules or conditions, it can be assumed that some kind of abnormality or its symptoms have occurred. Taking these into consideration, it is advisable to compare the first feature value based on the motor model and the second feature value based on the actual machine in real time.

[0044] (5) Abnormality notification section 60 The abnormality notification unit 60 has a function of notifying the occurrence of an abnormality recognized from the diagnosis result by the comparison and diagnosis unit 50. As a specific method of abnormality notification, an abnormality occurrence signal may be transmitted to the outside using a communication standard such as RS485 or Ethernet, or may be displayed and output on a display screen of a power conversion device or the like equipped with a control circuit. Also, the abnormality occurrence signal may be output as a voltage signal or current signal from a digital output terminal or the like of the device.

[0045] As described above, in this embodiment, the functions of the detection unit 10, the motor model calculation unit 20, the first feature extraction unit 30, the second feature extraction unit 40, the comparison diagnosis unit 50, and the abnormality notification unit 60 make it possible to detect and notify motor abnormalities.

[0046] FIG. 5 is a flowchart showing a series of abnormality diagnosis processes performed by the above-mentioned components. First, the abnormality diagnosis device is started to start diagnosis (step S1), and each sensor of the detection unit 10 in Fig. 1 detects current, voltage, speed, etc. (step S2). Next, the motor model calculation unit 20 calculates a motor model consisting of the motor voltage equation and its differential equation (step S3). Next, the first feature extraction unit 30 extracts a first feature based on the motor model, and the second feature extraction unit 40 extracts a second feature based on each detection value input from the detection unit 10 (step S4).Then, the comparison and diagnosis unit 50 compares the first and second feature and diagnoses whether the motor is normal or abnormal based on the degree of discrepancy between the two (step S5), and the abnormality notification unit 60 notifies the result. Furthermore, the diagnosis result is stored in memory to end one cycle (step S6), and thereafter, the processes from step S2 onwards are repeated. This series of processes (steps S2 to S6) is repeatedly executed at a fixed cycle.

[0047] Next, FIG. 6 is a configuration diagram showing an example of a power conversion device equipped with the abnormality diagnostic device of this embodiment. In Fig. 6, 100 is a main circuit of a power conversion device consisting of a rectifier circuit, an inverter section, etc. A voltage sensor 120 and a current sensor 130 are connected between the main circuit 100 and a motor M, and a speed sensor 140 is attached to the motor M. These sensors 120, 130, and 140 constitute the detection section 10 in Fig. 1.

[0048] The voltage of the DC intermediate circuit of the main circuit 100 is detected by a voltage sensor 110 and input to the control circuit 70, and the detection values ​​by the voltage sensor 120, the current sensor 130, and the speed sensor 140 are also input to the control circuit 70. The control circuit 70 generates a PWM command for the switching elements of the inverter unit of the main circuit 100 using each of the input detection values, and sends a drive signal to each switching element via a gate drive circuit 80.

[0049] On the other hand, the configuration of the abnormality diagnosis device connected to the control circuit 70 is substantially the same as that in Fig. 1. In Fig. 6, the feature extraction unit is shown as a single block (30, 40), but this block has both the function of the first feature extraction unit 30 and the function of the second feature extraction unit 40 in Fig. 1.

[0050] That is, the first feature extraction unit 30 extracts first features from the current estimates, voltage estimates, etc. calculated by the motor model calculation unit 20 using the motor model, and outputs them to the comparative diagnosis unit 50, while the second feature extraction unit 40 extracts second features from the current detection values, voltage detection values, and speed detection values ​​received via the control circuit 70, and outputs them to the comparative diagnosis unit 50. As described above, the comparative diagnosis unit 50 diagnoses whether the motor M is normal or abnormal based on the degree of deviation between the first characteristic amount and the second characteristic amount, and outputs the result to the abnormality notification unit 60. An abnormality occurrence signal is input from the abnormality notification unit 60 to the control circuit 70, and the control circuit 70 transmits the abnormality information to the outside and displays the abnormality on a display screen.

[0051] Fig. 7 is a block diagram showing an application example of the abnormality diagnostic device of the present invention. In Fig. 7, M1 and M2 are motors to be diagnosed, and 240 and 250 are power conversion devices that drive the motors M1 and M2, respectively. An external diagnostic device 310 is provided between the power conversion device 240 and the motor M1, and an external diagnostic device 320 is provided between the power conversion device 250 and the motor M2. The external diagnostic devices 310 and 320 are capable of communicating with each other via an industrial high-speed communication path 330. Numeral 400 denotes a controller that comprehensively controls the entire system.

[0052] The external diagnostic devices 310, 320 have the function of an abnormality diagnostic device as shown in Fig. 1, and diagnose the normality / abnormality of the motors M1, M2 in real time. The controller 400 constantly collects these diagnostic results, making it possible to comprehensively monitor the conditions of the motors M1, M2. According to this application example, in a factory or plant equipped with a plurality of motor drive systems each consisting of a power conversion device and a motor, it is possible to grasp the states of a plurality of motors online and in real time. [Explanation of symbols]

[0053] 10: Detection section 20: Motor model calculation section 30: First feature extraction unit 40: Second feature extraction unit 50: Comparative diagnosis section 60: Abnormality notification section 70: Control circuit 80: Gate drive circuit 91,94: Addition and subtraction means 92: Speed ​​regulator 93: Current regulator 95:Position / velocity estimator 100: Main circuit 110,120: Voltage sensor 130: Current sensor 140: Speed ​​sensor 200: Power conversion device 210: Control circuit 220: Coupling network 221: Voltage sensor 222: Current sensor 230: Power conversion device under test 240,250: Power conversion device 310, 320: External diagnostic device 330: Industrial high-speed communication channels 400: Controller G: Three-phase AC power supply M, M1, M2: Motor MM: Motor model

Claims

1. a motor state quantity generating unit that generates, as motor state quantities, values ​​including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; a motor model calculation unit that calculates a mathematical expression or a determinant including the motor state quantity and a constant of the motor as a motor model; a first feature amount extraction unit that extracts a first feature amount based on a motor state amount estimated and calculated by the motor model calculation unit using the motor model; a second feature quantity extraction unit which receives the motor state quantity generated by the motor state quantity generation unit without passing through the motor model calculation unit, and extracts a second feature quantity from the input motor state quantities based on a motor state quantity of the same type as the motor state quantity estimated and calculated by the first feature quantity extraction unit; a comparative diagnosis unit that calculates a degree of deviation of the second feature amount from the first feature amount and diagnoses whether the motor is normal or abnormal based on the degree of deviation; Equipped with When the first feature amount and the second feature amount are current information, A motor abnormality diagnosis device that uses multivariate analysis to determine the slope component of the principal component of the point group of a Lissajous figure created based on the current information, and determines the magnitude or slope of the slope component for each of the first feature amount and the second feature amount.

2. a motor state quantity generating unit that generates, as motor state quantities, values ​​including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; a motor model calculation unit that calculates a mathematical expression or a determinant including the motor state quantity and a constant of the motor as a motor model; a first feature amount extraction unit that extracts a first feature amount based on a motor state amount estimated and calculated by the motor model calculation unit using the motor model; a second feature quantity extraction unit which receives the motor state quantity generated by the motor state quantity generation unit without passing through the motor model calculation unit, and extracts a second feature quantity from the input motor state quantities based on a motor state quantity of the same type as the motor state quantity estimated and calculated by the first feature quantity extraction unit; a comparative diagnosis unit that calculates a degree of deviation of the second feature amount from the first feature amount and diagnoses whether the motor is normal or abnormal based on the degree of deviation; Equipped with a frequency distribution of the number of cycles of amplitude is created for each speed of the motor using a waveform counting method, and a transition of this frequency distribution is obtained for each of the first characteristic amount and the second characteristic amount.

3. In the motor abnormality diagnosis device according to claim 1 or 2, 2. A motor abnormality diagnosis device, wherein the motor model is a differential equation for determining a current flowing through the motor.

4. In the motor abnormality diagnosis device according to claim 1 or 2, 2. A motor abnormality diagnosis device, wherein the motor model is a voltage equation for determining a voltage applied to the motor.

5. The motor abnormality diagnosis device according to any one of claims 1 to 3, the first feature extraction unit uses the motor model to estimate a current flowing through the motor from a voltage equivalent value generated by the motor state quantity generation unit, and extracts the first feature based on the estimated current.

6. In the motor abnormality diagnosis device according to any one of claims 1 to 4, the first feature extraction unit uses the motor model to estimate a voltage applied to the motor from a current equivalent value generated by the motor state quantity generation unit, and extracts the first feature based on the estimated voltage.

7. In the motor abnormality diagnosis device according to claim 1 or 2, the first feature quantity extraction unit extracts the first feature quantity based on a combination of a plurality of motor state quantities generated by the motor state quantity generation unit, using the motor model; The motor abnormality diagnosis device according to claim 1, wherein the second feature amount extraction unit extracts the second feature amount based on a combination of a plurality of motor state quantities of the same type as the combination of the plurality of motor state quantities.

8. generating motor state quantities including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; A step of calculating an equation or a determinant including the motor state quantity and a constant of the motor as a motor model; extracting a first feature quantity based on a motor state quantity estimated and calculated using the motor model; extracting a second feature quantity based on a motor state quantity of the same type as the motor state quantity used to extract the first feature quantity, without using the motor model; a step of determining a degree of deviation of the second feature amount from the first feature amount, and diagnosing whether the motor is normal or abnormal based on the degree of deviation; is executed repeatedly at a fixed cycle, When the first feature amount and the second feature amount are current information, A method for diagnosing an abnormality in a motor, comprising: determining, by multivariate analysis, a gradient component of a principal component of a group of points of a Lissajous figure created based on the current information; and determining the magnitude or gradient of the gradient component for each of the first feature amount and the second feature amount.

9. generating motor state quantities including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; A step of calculating an equation or a determinant including the motor state quantity and a constant of the motor as a motor model; extracting a first feature quantity based on a motor state quantity estimated and calculated using the motor model; extracting a second feature quantity based on a motor state quantity of the same type as the motor state quantity used to extract the first feature quantity, without using the motor model; a step of determining a degree of deviation of the second feature amount from the first feature amount, and diagnosing whether the motor is normal or abnormal based on the degree of deviation; is executed repeatedly at a fixed cycle, A motor abnormality diagnosis method comprising: creating a frequency distribution of the number of cycles of amplitude for each speed of the motor using a waveform counting method; and determining a transition of this frequency distribution for each of the first characteristic amount and the second characteristic amount.

10. A function of generating motor state quantities including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; A function of calculating an equation or a determinant including the motor state quantity and a constant of the motor as a motor model; a function of extracting a first characteristic quantity based on a motor state quantity estimated and calculated using the motor model; a function of extracting a second feature quantity based on a motor state quantity of the same type as the motor state quantity used to extract the first feature quantity, without using the motor model; a function of determining a degree of deviation of the second characteristic amount from the first characteristic amount and diagnosing whether the motor is normal or abnormal based on the degree of deviation; When the first feature amount and the second feature amount are current information, A function of obtaining a gradient component of a principal component of a group of points of a Lissajous figure created based on the current information by multivariate analysis, and obtaining a magnitude or gradient of the gradient component for each of the first feature amount and the second feature amount; A motor abnormality diagnosis program that causes a processor to repeatedly execute the above at a constant interval.

11. A function of generating motor state quantities including at least a value equivalent to a voltage applied to a motor driven by a power conversion device, a value equivalent to a current flowing through the motor, and a value equivalent to a speed of the motor; A function of calculating an equation or a determinant including the motor state quantity and a constant of the motor as a motor model; a function of extracting a first characteristic quantity based on a motor state quantity estimated and calculated using the motor model; a function of extracting a second feature quantity based on a motor state quantity of the same type as the motor state quantity used to extract the first feature quantity, without using the motor model; a function of determining a degree of deviation of the second characteristic amount from the first characteristic amount and diagnosing whether the motor is normal or abnormal based on the degree of deviation; a function of creating a frequency distribution of the number of cycles of amplitude for each speed of the motor using a waveform counting method, and determining a transition of this frequency distribution for each of the first feature amount and the second feature amount; A motor abnormality diagnosis program that causes a processor to repeatedly execute the above at a constant interval.

Images