Sign detection device and sign detection method

Abstract

To provide a sign detection device that detects a sign of failure in an elevator device.SOLUTION: A sign detection device according to an embodiment comprises a data acquisition unit and a sign detection unit. The data acquisition unit acquires time-series data measured during elevation operation of a car of an elevator device. The sign detection unit uses AI using supervised data to detect whether or not there is a sign of failure in an article configuring the elevator device on the basis of the acquired time-series data.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] Embodiments of the present invention relate to a sign detection device and a sign detection method.

Background Art

[0002] When an abnormality occurs in an elevator device, there is an abnormality detection device that identifies the defective component. Existing abnormality detection devices detect that an abnormality has occurred. Therefore, the abnormality cannot be detected until after a failure has occurred.

[0003] For example, when a door motor fails, the doors of the car cannot be opened or closed, and there is a possibility that passengers may be trapped inside the car. An abnormality detection device that detects that an abnormality has occurred cannot reduce the possibility of trapping passengers inside the car.

[0004] Also, when the hoist motor (winch) fails, the elevator device cannot be used for a long time for the work of replacing the hoist motor. If it is possible to detect a sign of an abnormality in a component, maintenance can be performed at night or the like before a failure occurs, and inconvenience to users can be reduced.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The present invention has been made in view of the above circumstances, and an object thereof is to detect a sign of a failure of an elevator device.

Means for Solving the Problems

[0007] The omen detection device according to the embodiment for solving the above problems includes a data acquisition unit and an omen detection unit. The data acquisition unit acquires time-series data measured during the elevating operation of the elevator car of the elevator device. The omen detection unit detects whether there is an omen of abnormality in the supplies constituting the elevator device based on the acquired time-series data by an AI using supervised data.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Mode for Carrying Out the Invention

[0009] Hereinafter, this embodiment will be described with reference to the drawings. For the description, an XYZ coordinate system composed of an X-axis, a Y-axis, and a Z-axis that are orthogonal to each other will be used as appropriate. The diagrams and flowcharts used in the description of this embodiment are merely examples.

[0010] (Embodiment 1) FIG. 1 is a perspective view of an elevator device 10 according to this embodiment. The elevator device 10 is disposed inside a hoistway 100 provided in a building such as a commercial facility or a residential facility. As shown in FIG. 1, the elevator device 10 includes a car 31, a counterweight 50, a hoisting motor 40, guide rails 21 to 24, and a control panel 70 (elevator control device).

[0011] Each of the guide rails 21 to 24 is a member with its longitudinal direction being the Z-axis direction. The guide rails 21 and 22 are a pair of members for guiding the car 31 to move up and down freely. Also, the guide rails 23 and 24 are a pair of members for guiding the counterweight 50 to move up and down freely. The guide rails 21 and 22 are arranged spaced apart in the Y-axis direction. Also, the guide rails 23 and 24 are similarly arranged spaced apart from each other in the Y-axis direction. In FIG. 1, the guide rails 23 and 24 of the counterweight 50 are arranged spaced apart from the guide rails 21 and 22 of the car 31 in the X-axis direction. Note that the arrangement of the guide rails 21 to 24 is not limited to the arrangement shown in FIG. 1.

[0012] The car 31 is a unit that accommodates users and moves up and down the hoistway 100. The car 31 is arranged between the guide rails 21 and 22 and is attached to the guide rails 21 and 22 so as to be movable in the vertical direction.

[0013] An opening 31a for entering and exiting the interior is formed on the +X side surface of the car 31. The opening 31a is closed or opened by a pair of doors 32 that move along the side surface of the car 31. The doors 32 are opened and closed by an opening and closing motor (not shown in FIG. 1).

[0014] The counterweight 50 is attached to the guide rails 23 and 24 so as to be movable in the vertical direction. The weight of the counterweight 50 is adjusted to a predetermined ratio with respect to the weight of the car 31.

[0015] The hoisting motor 40 is a motor for hoisting the car 31. The hoisting motor 40 is arranged at the upper part of the hoistway 100 such that its rotation axis is parallel to the Y-axis. A pulley 42 is fixed to the rotation axis of the hoisting motor 40.

[0016] A wire 43 is wound around the pulley 42 of the hoisting motor 40. One end of the wire 43 is fixed to the car 31 and the other end is fixed to the counterweight 50.

[0017] The control panel 70 is arranged in the hoistway 100. The control panel 70 houses a control device for controlling the hoisting motor 40 and the equipment provided in the car 31 and the like.

[0018] Figure 2 is a block diagram showing the control system of the elevator apparatus 10. The control system includes a control unit 80 and a drive unit 91 housed in the control panel 70, and an operation panel 36 provided in the car 31.

[0019] The operation panel 36 is provided on the inner wall surface of the car 31. The operation panel 36 is an interface for receiving the destination floor and the like from the user of the car 31. The user can register the destination floor of the car 31 and open and close the door 32 by operating the operation panel 36. The operation panel 36 is connected to the control unit 80 housed in the control panel 70 via the cable 44 shown in FIG. 1.

[0020] The drive unit 91 shown in FIG. 2 drives the hoisting motor 40 and the opening / closing motor 41 by supplying power to the hoisting motor 40 and an opening / closing motor 41 (not shown in FIG. 1) that drives the door 32 of the car 31. The drive unit 91 drives the hoisting motor 40 based on an instruction from the control unit 80. Further, the drive unit 91 drives the opening / closing motor 41 based on an instruction from the control unit 80.

[0021] Figure 3 is a block diagram of the drive unit 91. The drive unit 91 has an inverter 13 and a measuring device 15. The inverter 13 is a power supply device that supplies power to the hoisting motor 40 and the opening / closing motor 41. The inverter 13 is composed of a switching regulator. When the hoisting motor 40 and the opening / closing motor 41 are formed of three-phase AC motors, the inverter 13 outputs a three-phase AC voltage.

[0022] The measurement device 15 includes a current measurement device 151 and a voltage measurement device 152. The current measurement device 151 is a device that measures the current supplied by the inverter 13 to the lifting motor 40 and the opening / closing motor 41 for each phase. The voltage measurement device 152 is a device that measures the output voltage of the inverter 13 for each phase.

[0023] In addition, the measurement device 15 includes a vibration sensor 154 and a torque sensor 155. The vibration sensor 154 is attached to the bottom of the car 31 and measures the vibration associated with the lifting operation of the car 31 and the opening / closing operation of the door 32. The torque sensor 155 measures the torque applied to the lifting motor 40 associated with the lifting operation of the car 31 and the torque applied to the opening / closing motor 41 associated with the opening / closing operation of the door 32.

[0024] Figure 4 is a physical block diagram of the control unit 80. The control unit 80 is a computer having a CPU (Central Processing Unit) 81, a main memory unit 82, an auxiliary storage unit 83, and an interface unit 84 that are interconnected via a bus 85. The CPU 81 executes the processes described later according to the programs stored in the auxiliary storage unit 83. The main memory unit 82 includes a RAM (Random Access Memory) and the like. The main memory unit 82 is used as a working area for the CPU 81. The auxiliary storage unit 83 includes non-volatile memories such as a ROM (Read Only Memory) and a semiconductor memory. The auxiliary storage unit 83 stores the programs executed by the CPU 81 and various parameters. In addition, the auxiliary storage unit 83 stores the supervised data used for anomaly detection by AI (Artificial Intelligence). The supervised data is time-series data with labels such as "no anomaly", "sign of anomaly", and "anomaly". Note that it may be semi-supervised data with partial labeling.

[0025] The interface unit 84 has a serial interface, a parallel interface, a wireless LAN interface, etc. The operation panel 36 and the drive unit 91 are connected to the CPU 81 via the interface unit 84. Also, an input / output device 93 composed of a keyboard, a display, etc. is connected to the interface unit 84.

[0026] Figure 5 is a functional block diagram of the control unit 80. The CPU 81 of the control unit 80 realizes the drive unit control section 71 and the sign detection device 72 by executing the programs stored in the auxiliary storage section 83.

[0027] The drive unit control section 71 controls the drive unit 91 based on the input from the operation panel 36 or the call panels on each floor. For example, when the drive unit control section 71 rotates the hoist motor 40 forward via the drive unit 91, the car 31 ascends and the counterweight 50 descends. When the drive unit control section 71 rotates the hoist motor 40 in reverse via the drive unit 91, the car 31 descends and the counterweight 50 ascends. Also, when the drive unit control section 71 rotates the opening / closing motor 41 forward via the drive unit 91, the door 32 of the car 31 and the doors provided at each floor landing are controlled to open, and when the opening / closing motor 41 is rotated in reverse, the door 32 of the car 31 and the doors provided at each floor landing are controlled to close.

[0028] The sign detection device 72 has an operation pattern setting section 73, a data acquisition section 74, a sign detection section 75, an extraction section 76, a Fourier transform section 77, a subtraction processing section 78, and an article identification section 79.

[0029] The driving pattern setting unit 73 sets a driving pattern that defines the moving speed and acceleration of the car body 31 in the drive unit control unit 71. As the driving pattern, there is a driving pattern during normal driving for transporting users and a driving pattern during sign detection for detecting whether there is a sign of abnormality in the supplies constituting the elevator device 10. The driving pattern is stored in the storage unit in advance. The driving pattern setting unit 73 selects the driving pattern from the storage unit and supplies it to the drive unit control unit 71.

[0030] FIG. 6 is an example of a driving pattern for identifying supplies with signs of abnormality. FIG. 6(A) shows a driving pattern in which the speed of the car body 31 is changed with an acceleration lower than that during normal driving. By changing the speed with a low acceleration, it becomes easier to detect noise caused by vibrations and operating cycles of supplies that appear at a specific speed. FIG. 6(B) shows a driving pattern in which the car body 31 is moved while maintaining a constant speed. FIG. 6(C) shows a driving pattern in which the speed of the car body 31 is increased with an acceleration higher than that during normal driving and then the ascending / descending speed of the car body 31 is decreased with a low acceleration. FIG. 6(D) shows a driving pattern in which the acceleration control of FIG. 6(A) is performed step by step. Note that the driving patterns shown in FIG. 6 are merely examples and are not limited thereto. The driving pattern for identifying supplies with signs of abnormality is set not only for the ascending / descending control of the car body 31 but also for the opening / closing control of the door 32.

[0031] Returning to FIG. 5, the data acquisition unit 74 acquires time-series data measured during the ascending / descending operation of the car body 31. For example, the data acquisition unit 74 acquires time-series data indicating the waveform of the voltage or current supplied by the inverter 13, which supplies power to the hoist motor 40 that drives the ascending / descending of the car body 31 or the opening / closing motor 41 that opens and closes the door 32 of the car body 31, as measured by the measuring device 15, time-series data of the vibration waveform during the ascending / descending operation of the car body 31 measured by the vibration sensor 154, and time-series data of the torque waveform of the opening / closing motor 41 during the opening and closing of the door 32 measured by the torque sensor 155.

[0032] In Embodiment 1, the case where the data acquisition unit 74 acquires time-series data indicating the waveforms of voltage or current from the measuring device 15 will be described. The time-series data measured by the measuring device 15 is temporarily stored in the storage unit. The data acquisition unit 74 acquires the time-series data from the storage unit. The time length of the time-series data acquired in one operation pattern is, for example, 10 seconds, 1 minute, or 10 minutes. FIG. 7 is an example showing a part of the time-series data of the acquired voltage waveform. The time-series data includes various frequency components of noise caused by vibrations and operating cycles of supplies, in addition to the frequency components corresponding to voltage changes associated with motor control.

[0033] Incidentally, the supplies constituting the elevator device 10 each have their own operating cycles. This operating cycle is also related to the operation pattern. For example, the rotation cycle of the sheave related to the cable connecting the car 31 and the counterweight 50 is determined by the diameter of the sheave and the lifting speed of the car 31. Also, the cycle of vibrations of the guide rails 21 to 24, etc., is determined by the lifting speed and acceleration of the car 31 and the moving distance. Also, the rotation cycle of the sheave used for opening and closing the door 32 is determined by the diameter of the sheave and the opening and closing speed of the door 32. FIG. 8 shows an example of the respective specific operating cycles of the supplies constituting the elevator device 10. The information shown in FIG. 8 is stored in the storage unit for each operating speed of the operation pattern.

[0034] Returning to FIG. 5, the sign detection unit 75 detects whether there is a sign of abnormality in the supplies constituting the elevator device 10 based on the acquired time-series data by an AI using supervised data. The sign detection unit 75 is composed of an AI having a learning function. The AI is configured using, for example, a neural network or a support vector machine.

[0035] The supervised data used for the AI included in the sign detection unit 75 is created based on the time-series data acquired by the data acquisition unit 74 during the elevating operation of the car 31 from when there is no abnormality in the supplies constituting the elevator device 10 until an abnormality occurs. The time-series data is measured, for example, during regular inspections every January. For the supervised data, a label of "abnormality present" is assigned to the time-series data when there is an abnormality in the supplies, and a label of "no abnormality" is assigned to the time-series data when there is no abnormality in the supplies. If an abnormality occurs in the supplies constituting the elevator device 10 within a predetermined period (for example, one month) after the time-series data is acquired, a label of "sign present" is assigned instead of the "no abnormality" label of the corresponding time-series data.

[0036] The extraction unit 76 extracts a part of the time-series data by specifying the time length from the time-series data acquired by the data acquisition unit 74. FIG. 9 is a diagram for explaining the time length of the time-series data. The time length is set based on the operation cycle specific to the supplies shown in FIG. 8. For example, the time length Wm for detecting a sign of abnormality in the sheave is a time length (for example, about 0.1 second) that is several times (for example, 2 times or 4 times) of 1 / fm, and the time length Wn for detecting a sign of abnormality in the guide rail is a time length that is several times of 1 / fn (for example, about 100 seconds). When inspecting the supplies related to the door 32, the time length may be set to about 2 seconds, for example.

[0037] Returning to FIG. 5, the Fourier transform unit 77 generates a spectrum distribution by Fourier-transforming the time-series data extracted by the extraction unit 76. FIG. 10 is an example of the spectrum distribution generated by the Fourier transform unit 77. The spectral values of the frequencies corresponding to the operation cycles of the respective supplies shown in FIG. 8 are large values.

[0038] The subtraction processing unit 78 performs a process of subtracting the value of the spectral distribution resulting from the operation of the supplies for which no sign is determined from the value of the spectral distribution generated by the Fourier transform unit 77. For example, when it is determined that there is no sign of abnormality in the roller guide shown in FIG. 8, the value of the spectral distribution resulting from the operation when there is no sign in the roller guide is subtracted from the value of the spectral distribution shown in FIG. 10. Regarding the time-series data acquired in the operation pattern shown in FIG. 6, the data of the spectral distribution when there is no sign for each supply corresponding to the time length specified by the extraction unit 76 is stored in the storage unit in advance. By this subtraction process, the spectrum indicated by the dotted line shown in FIG. 11 is deleted, or the value of the spectrum indicated by the dotted line becomes smaller.

[0039] The supply identification unit 79 identifies the supplies with a sign of abnormality based on the generated spectral distribution by AI using the supervised data. The supply identification unit 79 is composed of an AI having a learning function. The AI is configured by using, for example, a neural network or a support vector machine. The AI constituting the supply identification unit 79 detects whether there is a sign of abnormality in the supplies constituting the elevator device 10 by comparing the feature amount of the spectral distribution output by the subtraction processing unit 78 with the feature amount of the supervised data.

[0040] The supervised data used for the AI included in the supply identification unit 79 is created based on the time-series data indicating the voltage or current waveform acquired by the data acquisition unit 74. The supervised data is data in which the label "abnormality present" is assigned to the data of the spectral distribution in the case of abnormality, and the label "no abnormality" is assigned to the data of the spectral distribution in the case of no abnormality. Also, among the data to which the label "no abnormality" is assigned, if an abnormality occurs within a subsequent predetermined period (for example, one month), the corresponding data is assigned the label "sign present" instead of the label "no abnormality". And it is created as supervised data associated with the data identifying the supplies with a sign of abnormality (specific names such as motors and sheaves).

[0041] For example, the spectrum distribution when the sheave is deformed may be labeled as "Abnormal: Sheave is deformed", and the spectrum distribution when the sheave has a scratch may be labeled as "Abnormal: Sheave has a scratch". Also, if an abnormality occurs in the guide rail within, for example, one month after the time series data is acquired, labels such as "Sign: Looseness in the mounting screw of the guide rail" or "Sign: Decrease in contact resistance between the sheave and the wire" may be attached.

[0042] This supervised data is created in large numbers for each time length set by the extraction unit 76 and for each operation pattern shown in FIG. 6, and is stored in the storage unit. This supervised data is also created in large numbers for each time length set by the extraction unit 76 and for each operation pattern shown in FIG. 6 even when there is no sign of abnormality in all supplies, and is stored in the storage unit with labels such as "No abnormality" or "No sign".

[0043] Next, with reference to the flowchart shown in FIG. 12, a sign detection process (sign detection method) for detecting a sign of abnormality in the supplies constituting the elevator device 10 will be described. The following control is performed based on a program stored in the auxiliary storage unit 83, and the control entity is the control unit 80 (CPU 81). The measuring device 15 measures time series data of the output current and output voltage of the inverter 13 and notifies the control unit 80 of the measured data.

[0044] The sign detection device 72 detects whether or not there has been a hoisting operation of the car 31 (step S11). If there has been no hoisting operation of the car 31 (step S11: No), the sign detection device 72 continues to monitor for the hoisting operation of the car 31. If there has been a hoisting operation of the car 31 (step S11: Yes), the sign detection device 72 acquires the time series data of the output current and output voltage of the inverter 13 measured by the data acquisition unit 74 (step S12). Step S12 is a data acquisition step.

[0045] Next, the sign detection device 72 determines whether there is a sign of abnormality in the supplies constituting the elevator device 10 based on the acquired time-series data by using the supervised data that constitutes the sign detection unit 75 (step S13). Step S13 is the sign detection step.

[0046] FIG. 13 is a diagram schematically showing the time-series data attached to the supervised data when there is no sign of abnormality. The time-series data when there is no sign is created by attaching the label "no sign" to the time-series data of the current shown in FIG. 13. FIG. 14 is a diagram schematically showing the time-series data of the current attached to the supervised data when there is a sign of abnormality in the bearing. The supervised data when there is a sign is created by attaching the label "sign present" to the time-series data of the current shown in FIG. 14. The sign detection device 72 compares the feature amounts of the supervised data as shown in FIGS. 13 and 14 with the feature amounts of the time-series data of the output current and output voltage acquired by the data acquisition unit 74 to determine whether there is a sign of abnormality in the supplies constituting the elevator device 10.

[0047] When there is no sign of abnormality in the supplies (step S14: No), the sign detection device 72 transfers the process to step S16. On the other hand, when there is a sign of abnormality in the supplies (step S14: Yes), the sign detection device 72 sets a flag indicating that there is a sign (step S15) and transfers the process to step S16.

[0048] Next, the sign detection device 72 determines whether the sign detection for all supplies has been completed (step S16). When the sign detection for all supplies has not been completed (step S16: No), the sign detection device 72 repeats the processes from step S11 to step S16. When the sign detection for all supplies has been completed (step S16: Yes), the process ends. When the flag indicating that there is a sign is set in step S15, the sign detection device 72 outputs to the input / output device 93 that there is a sign of abnormality and performs a process of specifying the supply that caused the sign of abnormality.

[0049] Next, with reference to the flowchart shown in FIG. 15, the measurement process of data for identifying the supplies that caused the sign of abnormality will be described. The following control is performed based on the program stored in the auxiliary storage unit 83, and the control entity is the control unit 80 (CPU 81). The measuring device 15 measures the time-series data of the output current and output voltage of the inverter 13 and notifies the control unit 80 of the measured time-series data. Here, the case of measuring data by moving the car 31 up and down in the four driving patterns shown in FIG. 6 will be described.

[0050] The driving pattern setting unit 73 first sets the driving pattern shown in FIG. 6(A) to the drive unit control unit 71 (step S31). The measuring device 15 measures the time-series data of the voltage and current when the car 31 is moving under control or the door 32 is opening / closing under the corresponding driving pattern (step S32) and stores it in the storage unit.

[0051] The driving pattern setting unit 73 determines whether all the driving patterns have been executed (step S33). If not all the driving patterns have been executed (step S33: No), the driving pattern setting unit 73 returns to step S31 and sets the unexecuted driving pattern. For example, in the second step S31, the driving pattern setting unit 73 sets the driving pattern shown in FIG. 6(B) to the drive unit control unit 71. Also, in the third step S31, the driving pattern setting unit 73 sets the driving pattern shown in FIG. 6(C) to the drive unit control unit 71. On the other hand, if all the driving patterns have been executed (step S33: Yes), the sign detection device 72 ends the data measurement process.

[0052] Next, with reference to the flowchart shown in FIG. 16, the identification process for identifying the supplies with signs of abnormality will be described. It is assumed that the data acquisition process described with reference to FIG. 15 has ended, and the measured time-series data is stored in the storage unit. The supervised data corresponding to the driving pattern and the time duration set by the driving pattern and extraction unit 76 is stored in the storage unit in advance.

[0053] First, the data acquisition unit 74 extracts time-series data from the storage unit for each driving pattern (step S51). For example, the data acquisition unit 74 first extracts the time-series data measured in the driving pattern shown in FIG. 6(A) from the storage unit.

[0054] Next, the extraction unit 76 extracts a part of the time-series data by specifying the time length from the time-series data acquired by the data acquisition unit 74 (step S52). The time length is set in advance for each article to be inspected according to the operation cycle of the article to be inspected. The extraction unit 76 first sets the shortest time length. For example, when the roller guide shown in FIG. 8 is the inspection target, the extraction unit 76 sets the time length Wi shown in FIG. 9 to a time length that is several times (for example, 2 times or 4 times) of 1 / fi. Then, the extraction unit 76 supplies the time-series data of the time length Wi to the Fourier transform unit 77.

[0055] Next, the Fourier transform unit 77 generates a spectrum distribution obtained by Fourier-transforming the time-series data of the time length Wi (step S53). The spectrum distribution obtained by Fourier-transforming the time-series data limited to the time length Wi does not include noise caused by articles with an operation cycle longer than fi, such as the sheaves and guide rails shown in FIG. 8, or the ratio of the power of the spectrum caused by articles with an operation cycle longer than fi in the converted spectrum distribution is low.

[0056] Next, the subtraction processing unit 78 performs a process of subtracting the value of the spectrum distribution caused by the operation of the article for which it is determined that there is no sign of abnormality from the value of the spectrum distribution generated by the Fourier transform unit 77 (step S54). In the process of the first step S54, since there is no article for which it is determined that there is no sign of abnormality, the process of step S54 is skipped.

[0057] Next, the supply item identification unit 79 detects whether there is a sign of abnormality in the supply items constituting the elevator device 10 by comparing the feature amount of the spectral distribution output by the subtraction processing unit 78 with the feature amount of the supervised data (step S55). In the spectral distribution obtained by Fourier-transforming the time-series data limited to the time length Wi, noise caused by supply items with an operation period longer than fi, such as the sheaves and guide rails shown in FIG. 8, is not included, or the ratio of the power of the spectrum caused by supply items with an operation period longer than fi in the transformed spectral distribution is low. Therefore, in the spectral distribution obtained by Fourier-transforming the time-series data limited to the time length Wi, the ratio occupied by the feature amount of the noise caused by supply items with an operation period shorter than fi becomes high. By making the feature amount of the supply item of interest prominent, the supply item identification unit 79 can accurately detect whether there is a sign of abnormality in the supply item of interest (supply item with an operation period shorter than fi).

[0058] The sign detection device 72 determines whether analysis has been performed for all supply items (step S56). Specifically, it is determined whether all the settings of the time lengths shown in FIG. 9 corresponding to the supply items shown in FIG. 8 to be inspected have been executed. If the analysis for all supply items has not been completed (step S56: No), the sign detection device 72 returns to step S52 and repeats the processing from step S52 to step S56 with a different time length.

[0059] In the subtraction process of the second step S54, for example, if it is determined in the process of the first step S55 that there is no sign of abnormality in the roller guide, the value of the spectral distribution caused by the operation of the roller guide when there is no sign of abnormality in the roller guide is subtracted from the value of the spectral distribution generated by the Fourier transform unit 77 in the second step S53. Explaining with reference to FIG. 11, since the spectral component indicated by the dotted line is removed, the spectral component in the frequency band lower than the frequency fi is emphasized.

[0060] Also, in the subtraction process of the third step S54, for example, when it is determined in the process of the first step S55 that there is no sign of abnormality in the roller guide, and it is determined in the process of the second step S55 that there is no sign of abnormality in the motor and the sheave, the value of the spectral distribution due to the operation of the roller guide when there is no sign of abnormality in the roller guide, the motor, and the sheave is subtracted from the value of the spectral distribution generated by the Fourier transform unit 77 in the third step S53. Explaining with reference to FIG. 11, since the spectral components indicated by the dotted line and the spectral components indicated by the dashed-dotted line are removed, the spectral components in the frequency band lower than the frequency fm are emphasized.

[0061] By subtracting the value of the spectral distribution due to the operation of the supplies for which it has been determined that there is no sign of abnormality from the value of the spectral distribution generated by the Fourier transform unit 77, the proportion occupied by the spectral components of the noise caused by the supplies for which the presence or absence of a sign of abnormality in the spectral distribution has not been determined increases. Since the characteristic amount of the supplies for which the presence or absence of a sign of abnormality has not been determined becomes prominent, the supplies specifying unit 79 can accurately detect whether there is a sign of abnormality in the supplies for which the presence or absence of a sign of abnormality has not been determined.

[0062] Next, the sign detection device 72 determines whether the analysis has been completed for all the operation patterns (step S57). If the analysis has not been completed for all the operation patterns (step S57: No), the sign detection device 72 returns to step S51 and repeats the processing from step S51 to step S57. In the processing of the second step S51, for example, the time-series data measured in the operation pattern shown in FIG. 6(B) is extracted from the storage unit. The processing from step S52 to step S57 is the same as that of the first time. In the processing of the third step S51, for example, the time-series data measured in the operation pattern shown in FIG. 6(C) is extracted from the storage unit. The processing from step S52 to step S57 is the same as that of the first time.

[0063] On the other hand, when the analysis of all the operation patterns has been completed (step S57: Yes), the sign detection device 72 ends the specific process of the supplies with signs. When detecting a sign that becomes abnormal in the process of step S55, the sign detection device 72 outputs the supplies with the abnormal sign to the input / output device 93.

[0064] Next, with reference to the flowchart shown in FIG. 17, the creation process of the supervised data used for the AI of the sign detection unit 75 will be described.

[0065] For example, during the regular inspection every month, the time-series data of the voltage and current of the inverter 13 during the lifting operation of the car 31 is acquired (step S71). When a failure is found in the elevator device 10 during the regular inspection (step S72: Yes), it is stored in the storage unit as supervised data with the label "abnormal" attached to the acquired time-series data (step S73). On the other hand, when no failure is found in the elevator device 10 during the regular inspection (step S72: No), it is stored in the storage unit as supervised data with the label "no abnormality" attached to the acquired time-series data (step S74).

[0066] The sign detection device 72 determines whether a failure has occurred in any of the supplies constituting the elevator device 10 before a predetermined period (for example, one month) has elapsed since the acquisition of the time-series data (step S75). When a failure has occurred in any of the supplies constituting the elevator device 10 before a predetermined period (for example, one month) has elapsed since the acquisition of the time-series data (step S75: Yes), the sign detection device 72 changes the label of the corresponding time-series data from "no abnormality" to "sign present", and stores it in the storage unit as supervised data with the label "sign present" attached (step S76).

[0067] On the other hand, if no failure occurs in any of the supplies that make up the elevator device 10 before a predetermined period (e.g., one month) has elapsed since the acquisition of the time-series data (step S75: No), the sign detection device 72 keeps the label of the corresponding time-series data as "no abnormality" and stores it in the storage unit as teacher data with the label of "no abnormality" (step S77).

[0068] As described above, the sign detection device 72 according to the first embodiment can detect whether there is a sign of abnormality in the supplies that make up the elevator device 10 based on the acquired time-series data by using AI with teacher data that constitutes the sign detection unit 75. Thereby, maintenance can be performed before a failure occurs, so that, for example, the possibility of confining passengers in the car can be reduced.

[0069] Further, the sign detection device 72 according to the first embodiment includes an extraction unit 76 that extracts time-series data of a specified time length from the time-series data acquired by the data acquisition unit 74, a Fourier transform unit 77 that generates a spectrum distribution by performing a Fourier transform on the time-series data extracted by the extraction unit 76, and an article identification unit 79 that identifies an article having a sign of abnormality based on the spectrum distribution generated by AI using teacher data. By performing the extraction process by the extraction unit 76, in the spectrum distribution obtained by performing a Fourier transform on the time-series data limited to an arbitrary time length Wx, the ratio occupied by the feature amount of the noise caused by the supplies having a period shorter than the operation period fx (Wx is several times 1 / fx) increases. By making the feature amount of the article of interest prominent, the article identification unit 79 can accurately detect whether there is a sign of abnormality in the article of interest (the article having a period shorter than fx). Therefore, maintenance such as replacement of the article can be performed before the article fails. Thereby, the possibility of confining passengers in the car can be reduced. Further, since a spare article can be prepared before the article fails, the maintenance time of the elevator device 10 can be shortened.

[0070] In addition, the sign detection device 72 according to Embodiment 1 includes a subtraction processing unit 78 that subtracts the value of the spectral distribution resulting from the operation of the supplies for which it has been determined that there is no sign of abnormality from the value of the spectral distribution generated by the Fourier transform unit 77. By subtracting the value of the spectral distribution resulting from the operation of the supplies for which it has been determined that there is no sign of abnormality from the value of the spectral distribution generated by the Fourier transform unit 77, the proportion occupied by the spectral components of the noise caused by the supplies for which the presence or absence of a sign of abnormality in the spectral distribution has not been determined increases. Since the feature amount of the supplies for which the presence or absence of a sign of abnormality has not been determined becomes prominent, the supply identification unit 79 can accurately identify the supplies with a sign of abnormality.

[0071] In addition, as described with reference to the flowchart shown in FIG. 17, the sign detection device 72 according to Embodiment 1 performs the creation process of the supervised data. Since the accuracy of the determination by the AI is improved as the supervised data increases in the sign detection device 72 according to Embodiment 1, the detection accuracy of the sign of abnormality in the elevator device 10 can be improved.

[0072] As described above, the embodiments of the present invention have been described, but the present invention is not limited to the above embodiments. For example, in the above description, the case of determining the presence or absence of a sign of abnormality based on the time-series data measured during the lifting operation at the time of regular maintenance has been described. As another embodiment, the presence or absence of a sign of abnormality may be determined based on the time-series data measured during the lifting operation during normal operation. Further, when determining the presence or absence of a sign of abnormality based on the time-series data measured during normal lifting operation, in order to remove unspecified vibrations, the measurement data when there are no passengers may be used. The determination as to whether there are passengers in the car 31 may be made using a camera that captures the inside of the car, or a load sensor that measures the weight of the car 31.

[0073] In addition, the driving pattern at the time of abnormality detection may include not only the example shown in FIG. 6 but also the driving pattern during normal operation.

[0074] In the above description, the case of performing the additional process of the supervised data has been described. However, when a sufficient amount of supervised data has been stored in advance and the target anomaly detection accuracy is satisfied, the additional process of the supervised data may be omitted.

[0075] (Embodiment 2) In Embodiment 1, the case of detecting whether there is a sign of abnormality in the supplies constituting the elevator apparatus 10 based on the time-series data of the voltage or current of the inverter 13 has been described. However, whether there is a sign of abnormality in the supplies can also be detected by other methods. For example, based on the time-series data measured by the vibration sensor 154 or the torque sensor 155, it is also possible to detect whether there is a sign of abnormality in the supplies.

[0076] The sign detection device 72 according to Embodiment 2 uses the time-series data measured by the vibration sensor 154 or the torque sensor 155 shown in FIG. 3. The vibration sensor 154 measures the time-series data of the vibration during the ascending and descending operations of the car 31. The torque sensor 155 measures the time-series data of the torque applied to the opening / closing motor 41 during opening and closing. The data acquisition unit 74 acquires the time-series data measured by the vibration sensor 154 or the torque sensor 155.

[0077] The supervised data is created for the time-series data of the vibration waveform and the time-series data of the torque waveform. Further, the supervised data is created corresponding to the operation pattern shown in FIG. 6 and the normal operation pattern. When using the time-series data measured by the vibration sensor 154, it is preferable to perform sign detection based on the data when there is no passenger in the car 31 in order to remove unspecified vibrations due to the movement of the passengers.

[0078] Although some embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are included in the invention described in the claims and its equivalent scope.

Explanation of Signs

[0079] 10…Elevator device 13…Inverter 15…Measuring device 151…Current measuring device 152…Voltage measuring device 154…Vibration sensor 155…Torque sensor 21~24…Guide rail 31…Car 31a…Opening 32…Door 36…Operation panel 40…Lifting motor 41…Opening / closing motor 42…Pulley 43…Wire 44…Cable 50…Counterweight 70…Control panel 71…Drive unit control section 72…Foresight detection device 73…Operation pattern setting section 74…Data acquisition section 75…Foresight detection section 76…Extraction section 77…Fourier transform section 78…Subtraction processing section 79…Supplies identification section 80…Control unit 81…CPU 82…Main memory section 83…Auxiliary storage section 84…Interface section 85…Bus 91…Drive unit 93…Input / output device 100…Elevator shaft

Claims

1. A data acquisition unit that acquires time-series data measured during the elevating operation of the elevator car; A sign detection unit that uses AI with labeled data to detect whether there is a sign of abnormality in the supplies that make up the elevator device based on the acquired time-series data; A sign detection device having the above.

2. The labeled data used for the AI included in the sign detection unit is created based on the time-series data acquired by the data acquisition unit in the elevating operation of the car from when there is no abnormality in the supplies that make up the elevator device until an abnormality occurs, When an abnormality occurs in the supplies that make up the elevator device within a predetermined period after acquiring the time-series data, it is created as labeled data with a label indicating that there is a sign in the corresponding time-series data, The sign detection device according to claim 1.

3. The time-series data acquired by the data acquisition unit is time-series data indicating the waveform of the voltage or current measured by a measuring device that measures the waveform of the voltage or current supplied by a power source that drives the elevation of the car or opens and closes the car door to the motor, time-series data of the torque waveform measured by a torque sensor that measures the torque of the motor when the door opens and closes, time-series data of the vibration waveform measured by a vibration sensor that measures the vibration during the elevating operation of the car, The sign detection device according to claim 1 or 2, including the above.

4. An extraction unit that extracts time-series data of a specified time length from the time-series data acquired by the data acquisition unit; A Fourier transform unit that generates a spectrum distribution by performing a Fourier transform on the time-series data extracted by the extraction unit; An article identification unit that uses AI with labeled data to identify articles with a sign of abnormality based on the generated spectrum distribution; The sign detection device according to claim 1 or 2, having the above.

5. The time length is set based on the operating cycle of each supply that makes up the elevator device, The sign detection device according to claim 4.

6. It has a subtraction processing unit that subtracts the value of the spectrum distribution caused by the operation of the supplies determined to have no sign of abnormality from the value of the spectrum distribution generated by the Fourier transform unit, The article identification unit detects that there is a sign of abnormality in the supplies that make up the elevator device based on the spectrum distribution after the subtraction process. The omen detection device according to claim 4.

7. A data acquisition step of acquiring time-series data measured during the elevating operation of the elevator car of the elevator device, An omen detection step of detecting, by an AI using supervised data, whether there is an omen of abnormality in the supplies constituting the elevator device based on the acquired time-series data, An omen detection method including the above.

8. In the data acquisition step, Time-series data measured during the elevating operation of the car from when there is no abnormality in the supplies constituting the elevator device to when an abnormality occurs is acquired, When an abnormality occurs in the supplies constituting the elevator device within a predetermined period after acquiring the time-series data, supervised data with a label indicating that there is an omen in the corresponding time-series data is created, The omen detection method according to claim 7.

Images