Failure prediction device

Abstract

The purpose of the present invention is to accurately predict failure of a robot without erroneous detection even if the pattern of operation of the robot changes.　This failure prediction device is provided with: an evaluation data collection unit that collects evaluation data for at least a drive shaft of a robot working on the basis of a work program; and an erroneous detection determination unit that uses the evaluation data to derive an evaluation formula for evaluating the evaluation data, and determines whether the evaluation data is an evaluation data value attributed to a factor other than failure of the robot, or an evaluation data value attributed to a failure factor for the robot, on the basis of the evaluation formula and the evaluation data.

Description

Technical Field

[0001] The present invention relates to a failure prediction device.

Background Art

[0002] There is a need for a technology that can predict a robot's failure, that is, detect signs of a robot's failure and notify the user before the robot becomes unable to operate properly. Generally, it is known that the driving torque during operation increases when a robot fails. Therefore, it has been proposed to predict a failure by monitoring changes in the driving torque and the like. For example, Patent Document 1 describes a technique for performing failure diagnosis based on the magnitude relationship between the measured values of a group of sensors such as a torque sensor, a temperature sensor, and an acceleration sensor and a reference value. In addition, Patent Document 2 collects the torque values of the drive shafts of robots operating according to a given work program, derives an evaluation formula that approximates the time change of the most recent torque value from the collected torque values, and based on the evaluation formula and the time change of the torque value when the drive shaft has failed in the past, it is determined that a failure of the drive shaft has occurred ruto A failure threshold value, which is a torque value, is set, an estimated value of the torque value when a preset prediction time has elapsed in the evaluation formula is calculated, and it is determined whether a failure of the drive shaft is predicted within the prediction time by comparing the estimated value with the failure threshold value. The technique is described.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] The degree to which the drive torque increases before a robot fails, and the drive torque at the time of failure, vary depending on the robot's operating pattern. Therefore, the method of comparing the current measured value with a reference value, as described in Patent Document 1, may lead to the conclusion that a robot is nearing failure even if it still has a long lifespan remaining. Furthermore, in Patent Document 2, it is necessary to receive a signal from the robot control device that identifies the work program being executed or a signal that notifies of a change in the work program, reset the collected torque values ​​each time the work program is changed, and collect new torque values, which is time-consuming. Furthermore, every time factors other than failures change, such as changes in the robot's operating pattern due to changes in the work program, Patent Documents 1 and 2 may mistakenly detect a failure.

[0005] Therefore, even if the robot's operating pattern changes, it is desirable to accurately predict whether the evaluation data values ​​detected in conjunction with the change in the operating pattern are due to factors that would cause the robot to fail, without falsely detecting them as failures. [Means for solving the problem]

[0006] One embodiment of the failure prediction device of the present disclosure includes: an evaluation data collection unit that collects evaluation data for at least the drive axes of a robot performing work based on a work program; and a false detection determination unit that derives an evaluation formula for evaluating the evaluation data using the evaluation data, and determines, based on the evaluation formula and the evaluation data, whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot, or an evaluation data value due to a factor that causes the failure of the robot.

[0007] One embodiment of the failure prediction device of the present disclosure includes: an evaluation data collection unit that collects evaluation data of a robot performing work based on a work program; a false detection determination unit that determines whether the evaluation data is an evaluation data value due to factors other than the failure of the robot or an evaluation data value due to factors that cause the robot to fail; and a display control unit that displays on a display unit at least one of the prediction result of the failure of the robot predicted based on the evaluation data and information indicating the determination result of the false detection determination unit, wherein the display control unit, when it is determined that the evaluation data value is due to factors other than the failure of the robot, prioritizes displaying on the display unit information indicating that the evaluation data value is due to factors other than the failure of the robot. [Effects of the Invention]

[0008] According to one embodiment, even if the robot's motion pattern changes, it is possible to accurately predict whether the evaluation data value detected in conjunction with the change in motion pattern is due to a factor that would cause the robot to fail, without falsely detecting it as a failure. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows an example of the configuration of a robot system equipped with a fault prediction device according to one embodiment. [Figure 2] This figure shows an example of a formula for evaluating torque values. [Figure 3] This figure shows an example of an evaluation formula for false detection of torque values. [Figure 4] This figure shows an example of an evaluation formula for false detection of torque values. [Figure 5] This figure shows an example of a display controlled by the display control unit. [Figure 6] This is a flowchart explaining the failure prediction process of a failure prediction device. [Modes for carrying out the invention]

[0010] Hereinafter, one embodiment will be described with reference to the drawings. Figure 1 is a diagram showing an example of the configuration of a robot system equipped with a fault prediction device 1 according to one embodiment.

[0011] The robot system comprises a robot R, a robot control device C that controls the robot R, a fault prediction device 1 that communicates with the robot control device C, and a monitor M which serves as a display unit whose display content is controlled by the fault prediction device 1. The robot R, the robot control device C, the fault prediction device 1, and the monitor M may be directly connected to each other via a connection interface (not shown). Alternatively, the robot R, the robot control device C, the fault prediction device 1, and the monitor M may be interconnected via a network (not shown), such as a LAN (Local Area Network) or the Internet. In this case, the robot R, the robot control device C, the fault prediction device 1, and the monitor M are equipped with a communication unit (not shown) for communicating with each other via such connection.

[0012] Robot R has multiple drive axes J1, J2, J3, J4, J5, and J6, and is equipped with a hand H capable of holding a workpiece at its tip. Robot R can perform desired tasks by moving the hand H by driving the drive axes J1 to J6. Robot R can be a vertical articulated robot as shown in the figure, but is not limited to this, and may be a Cartesian coordinate robot, a SCARA robot, a parallel link robot, etc. Robot R is configured to have a torque detection unit (not shown) that detects the torque value of each drive axis J1 to J6 (for example, the torque and current value of the motors that drive the drive axes).

[0013] The robot control device C operates the robot R according to a given work program. Specifically, the robot control device C can have a known configuration that calculates the position or speed of each drive axis J1 to J6 of the robot R at each time required for the operation according to the work program, and applies the necessary current to each drive axis J1 to J6 of the robot R. Further, the robot control device C receives a feedback signal of the drive torque of each drive axis J1 to J6 from the robot R, and transmits the torque values of each drive axis J1 to J6 to the failure prediction device 1.

[0014] The monitor M can be constituted by a display device such as a liquid crystal display panel. The monitor M may be attached to the failure prediction device 1, may be attached to the robot control device C, or may be provided at a location separated from the failure prediction device 1 and the robot control device C, for example, at a position where it can be visually recognized from many places in the factory where the robot R is installed.

[0015] The failure prediction device 1 is, for example, a computer device having a CPU, a memory, a communication interface, etc. The CPU is a processor that controls the failure prediction device 1 as a whole. The CPU reads out the system program and the application program stored in the memory via the bus, and controls the entire failure prediction device 1 according to the system program and the application program. Thereby, as shown in FIG. 1, the failure prediction device 1 is configured to realize the functions of the evaluation data collection unit 10, the evaluation formula derivation unit 20, the false detection determination unit 30, the prediction determination unit 40, and the display control unit 50.

[0016] The evaluation data collection unit 10 collects evaluation data of at least the drive axes of the robot R that works based on the work program. Specifically, the evaluation data collection unit 10 collects, for example, the time-series torque values of the drive axes J1 to J6 of the robot R operating according to the work program from the robot control device C as evaluation data. That is, the evaluation data collection unit 10 stores the value of each drive torque of the drive axes J1 to J6 at each time.

[0017] The evaluation formula derivation unit 20 derives a first evaluation formula and a second evaluation formula for each drive axis J1 to J6, which approximate the time change of the torque values ​​(evaluation data) of the drive axes J1 to J6 of the robot R operating according to the most recent work program (within a certain time range up to the present) from the torque values ​​(evaluation data) collected by the evaluation data acquisition unit 10. Specifically, the evaluation formula derivation unit 20 performs a normalization process that converts the torque values ​​Tq(t) (evaluation data) of each of the drive axes J1 to J6 of the robot R to values ​​between 0 and 1, for example, using formula (1). <Tq(t)> =Tq(t)-Tq min / (Tq max -Tq min ) × (1-0) (1) Here, Tq(t) represents the torque value at the most recent time t.<Tq(t)> This indicates that Tq(t) is a normalized value between 0 and 1. max This represents the most recent maximum torque value, which becomes "1" after normalization using equation (1). Tq min This represents the smallest torque value in the most recent period, and becomes "0" after normalization using equation (1). As shown in Figure 2, the evaluation formula derivation unit 20 represents the normalized torque value shown by the solid line.<Tq(t)> The first evaluation formula, shown by the dashed line, is derived as a linear function Y(t) = at + b (where a and b are constants) by linear regression of the time change of the value. In other words, the evaluation formula derivation unit 20 derives the torque value obtained by normalizing the evaluation formula Y(t) = at + b through linear regression.<Tq(t)> Calculate the values ​​of the constants a and b in the expression that represents this. In Figure 2, for example, the torque values ​​Tq(t) for the most recent 10 points are shown as evaluation data.

[0018] Furthermore, the evaluation formula derivation unit 20, for example, as shown in Figure 3, uses a normalized torque value.<Tq(t)> A second evaluation formula (hereinafter also called the "false detection evaluation formula") S(t) (shown by the dashed line) is derived by binarizing the time change of the torque value. Note that in the binarization process, the torque value is used in the second evaluation formula.<Tq(t)> If the evaluation data is less than or equal to the average value of (maximum value "1" + minimum value "0") (i.e., "0.5" shown by the dashed line), the torque value will be set to "0".<Tq(t)> Evaluation data that exceeds the average value of (maximum value "1" + minimum value "0") (i.e., "0.5" shown by the dashed line) is set to "1". Figure 4 shows the torque value.<Tq(t)> This figure shows an example of the torque value misdetection evaluation formula S(t) when the time change of takes a different pattern than that shown in Figure 3. As shown in Figure 4, when the evaluation data fluctuates around "0.5", the false detection evaluation formula S(t) changes to "0" or "1" in accordance with that fluctuation.

[0019] The false detection determination unit 30 determines, based on the evaluation formula derivation unit 20, the first evaluation formula and the second evaluation formula (false detection evaluation formula) which are evaluation formulas for evaluating evaluation data using, for example, the torque values ​​(evaluation data) for each drive shaft J1 to J6, and the evaluation data, whether the evaluation data for each drive shaft is an evaluation data value due to a change in the work program or an evaluation data value due to a factor that causes a failure of the robot R.

[0020] The false detection determination unit 30 determines the value calculated by the first evaluation formula Y(t)=at+b derived by the evaluation formula derivation unit 20 using formulas (2) and (3), and the normalized torque value.<Tq(t)> The first threshold value r1 is determined by the difference with the evaluation data, and the value calculated by the second evaluation formula (false detection evaluation formula) S(t) is normalized to the torque value.<Tq(t)> A second threshold, r2, is calculated based on the difference with the evaluation data. r1=sqrt(Σ(<Tq(t)> -Y(t)) 2 ) (2) r2=sqrt(Σ(<Tq(t)> -S(t)) 2 ) (3) The false detection determination unit 30 determines that the evaluation data is a normal value, including cases where the evaluation data value is due to factors other than the failure of the robot R, when threshold r1 ≥ threshold r2. Hereinafter, for simplicity, this determination will also be referred to as "determining that the evaluation data is a false detection." Conversely, the false detection determination unit 30 compares the calculated threshold r1 and threshold r2, and if threshold r1 < threshold r2, it determines that the evaluation data value is due to some factor that causes a failure of the robot R. Hereafter, for simplicity, this determination will also be referred to as "determining that the evaluation data is not a false detection." Furthermore, using thresholds r1 and r2 in this manner, the evaluation data Ta It has been confirmed from previously accumulated robot production operation data that it is possible to determine whether the evaluation data values ​​include those from normal conditions, such as those caused by factors other than robot failure, or whether they include evaluation data values ​​caused by factors that lead to robot failure.

[0021] For example, if the false detection determination unit 30 determines that the evaluation data is not a false detection, the prediction determination unit 40 determines whether the slope a of the first evaluation formula Y(t) derived by the evaluation formula derivation unit 20 exceeds a preset threshold for the drive axis of the robot R that was determined not to be a false detection. If the slope a of the first evaluation formula Y(t) exceeds a preset threshold, the prediction determination unit 40 determines that a failure of the drive axis is predicted. Furthermore, if the false detection determination unit 30 determines that the evaluation data is a false detection, the prediction determination unit 40 will skip predicting a drive shaft failure.

[0022] As shown in Figure 5, the display control unit 50 displays the most recent torque values ​​(evaluation data) as a graph on the monitor M screen, and if the prediction judgment unit 40 determines that a failure of drive shafts J1 to J6 is predicted, it displays this fact using a marker or the like. This allows the user to be notified that a failure is predicted. The method of displaying the marker indicating a predicted failure may be, for example, as shown in Figure 5, by displaying a straight line L on the torque value graph that indicates the time of the last torque value when the prediction judgment unit 40 determined that a failure of drive shafts J1 to J6 was predicted. It is preferable to display this straight line L in a color that is more conspicuous than the line plotting the torque values ​​(for example, red). Furthermore, the display control unit 50 may, when the false detection determination unit 30 determines that the evaluation data is a false detection, prioritize displaying information indicating that the evaluation data is a false detection on the monitor M.

[0023] <Fault prediction processing of fault prediction device 1> Next, referring to Figure 6, we will explain the flow of the failure prediction process of the failure prediction device 1. Figure 6 is a flowchart illustrating the fault prediction process of the fault prediction device 1. The flow shown here is executed repeatedly while the robot R is operating based on the work program.

[0024] In step S1, the evaluation data acquisition unit 10 collects time-series torque values ​​of the drive axes J1 to J6 of the robot R operating according to the work program from the robot control device C as evaluation data.

[0025] In step S2, the evaluation formula derivation unit 20 performs a normalization process on the torque values ​​Tq(t) of the most recent drive shafts J1 to J6 among the torque values ​​collected in step S1, and the normalized torque value<Tq(t)> The time evolution of is linearly regressed to derive the first evaluation equation Y(t) for each drive shaft J1 to J6.

[0026] In step S3, the evaluation formula derivation unit 20 uses the torque value normalized in step S2.<Tq(t)> The time evolution is binarized to derive the second evaluation formula, the false detection evaluation formula S(t).

[0027] In step S4, the false detection determination unit 30 uses the first evaluation formula Y(t) derived in step S2 and the normalized torque value<Tq(t)> The threshold r1 is based on the difference with the evaluation data, the second evaluation formula is the false detection evaluation formula S(t), and the normalized torque value.<Tq(t)> A threshold r2, based on the difference with the evaluation data, is compared to determine whether the evaluation data for each drive shaft J1 to J6 is a false detection.

[0028] In step S5, the prediction and judgment unit 40 determines whether the slope a of the first evaluation formula Y(t) derived in step S2 exceeds a preset threshold for the drive axis of the robot R for which the evaluation data was determined not to be a false detection in step S4, that is, whether a failure of the drive axis is predicted.

[0029] In step S6, the display control unit 50 displays the most recent torque value as a graph on the screen of the monitor M.

[0030] In step S7, if the display control unit 50 determines in step S5 that a failure of the drive shafts J1 to J6 is predicted, it displays the time of the endpoint of the torque value that formed the basis of that determination as a straight line L on the graph.

[0031] As described above, the fault prediction device 1 according to one embodiment normalizes the torque values ​​Tq(t) of the drive shafts J1 to J6 due to the operation of the robot R according to the currently executing work program, and the normalized torque value<Tq(t)> The first evaluation formula Y(t) is derived, which shows the change in the torque value, and the second evaluation formula, the false detection evaluation formula S(t), is also derived. The fault prediction device 1 uses the first evaluation formula Y(t) and the normalized torque value.<Tq(t)> The threshold r1 is based on the difference with the evaluation data, the second evaluation formula is the false detection evaluation formula S(t), and the normalized torque value.<Tq(t)> By comparing a threshold r2 based on the difference with the evaluation data, it is determined whether the evaluation data for each drive axis J1 to J6 is a false detection, and it is determined whether a failure is predicted for the drive axis of robot R for which the evaluation data is determined not to be a false detection. As a result, even if the operating pattern of the robot R changes, the failure prediction device 1 can accurately predict whether the evaluation data value detected in conjunction with the change in the operating pattern is due to a factor that will cause the robot R to fail, without mistakenly detecting it as a failure.

[0032] Although one embodiment has been described above, the failure prediction device 1 is not limited to the embodiment described above, and may include modifications, improvements, etc., to the extent that the objective can be achieved.

[0033] <Example 1> In one embodiment, the evaluation formula derivation unit 20 is a normalized torque value<Tq(t)> The time change of was binarized to derive the false detection evaluation formula S(t), which is the second evaluation formula, but is not limited to this. For example, the evaluation formula derivation unit 20 uses the normalized torque value<Tq(t)> Alternatively, the time evolution of the function can be approximated to derive the false detection evaluation formula S(t), which is the second evaluation formula for the step function.

[0034] <Modification 2> Furthermore, for example, in the above embodiment, the evaluation formula derivation unit 20 uses the normalized torque value<Tq(t)> The first evaluation formula Y(t) was derived by linear regression of the time change of the torque value Tq(t), but the method is not limited to this. For example, the evaluation formula derivation unit 20 may derive the first evaluation formula Y(t) by linear regression of the time change of the torque value Tq(t).

[0035] <Variation 3> Furthermore, in the above-described embodiment, the evaluation formula derivation unit 20 was configured as a different functional unit from the false detection determination unit 30, but it may also be included in the false detection determination unit 30.

[0036] <Modification 4> Furthermore, in the above-described embodiment, for example, the false detection determination unit 30 determined whether the evaluation data was a false detection or not, and the prediction determination unit 40 determined whether the drive shaft of the robot R, for which the evaluation data was determined not to be a false detection, was faulty or not. However, the system is not limited to this. For example, the prediction determination unit 40 may predict a drive shaft failure based on the evaluation data, and the false detection determination unit 30 may determine whether the evaluation data is a false detection or not after the prediction determination unit 40 has predicted a drive shaft failure.

[0037] In one embodiment, each function included in the fault prediction device 1 can be implemented by hardware, software, or a combination thereof. Here, implementation by software means that it is implemented by a computer reading and executing a program.

[0038] Programs can be stored and supplied to a computer using various types of non-transitory computer-readable medium. Non-transitory computer-readable mediums include various types of tangible storage mediums. Examples of non-transitory computer-readable mediums include magnetic storage media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical storage media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memory (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, RAMs). Programs may also be supplied to a computer using various types of transient computer-readable mediums. Examples of transient computer-readable mediums include electrical signals, optical signals, and electromagnetic waves. Transitory computer-readable mediums can be supplied to a computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0039] Furthermore, the step of writing the program to be recorded on the recording medium includes not only processes that are performed chronologically in that order, but also processes that are not necessarily performed chronologically, but are executed in parallel or individually.

[0040] In other words, the failure prediction device of this disclosure can take various forms having the following configurations.

[0041] (1) The failure prediction device 1 of the present disclosure includes an evaluation data collection unit 10 that collects evaluation data for at least the drive axes of a robot R that performs work based on a work program, and a false detection determination unit 30 that derives an evaluation formula for evaluating the evaluation data using the evaluation data, and determines, based on the evaluation formula and the evaluation data, whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot R, or an evaluation data value due to a factor that causes the failure of the robot R. According to this failure prediction device 1, even if the operation pattern of robot R changes, it is possible to accurately predict whether the evaluation data value detected in conjunction with the change in the operation pattern is due to a factor that will cause robot R to fail, without falsely detecting it as a failure.

[0042] (2) In the fault prediction device 1 described in (1), the false detection determination unit 30 may derive an evaluation formula for each evaluation data.

[0043] (3) In the failure prediction device 1 described in (1) or (2), if the false detection determination unit 30 determines that the evaluation data is an evaluation data value due to a factor causing failure of the robot R, the device may also include a prediction determination unit 40 that predicts a failure of the drive shaft based on the evaluation data.

[0044] (4) In the failure prediction device 1 described in any of (1) to (3), the false detection determination unit 30 may derive a threshold value based on the difference between the evaluation data and the value of the evaluation formula, and determine, based on the evaluation data and the threshold value, whether the evaluation data is an evaluation data value due to factors other than the failure of the robot R, or an evaluation data value due to factors that cause the failure of the robot R.

[0045] (5) In the fault prediction device 1 described in (4), the evaluation formula includes a first evaluation formula obtained by linear regression of the evaluation data and a second evaluation formula obtained by applying a step function or binarization to the evaluation data, and the threshold may be derived based on the difference between the evaluation data and the value of the first evaluation formula or the value of the second evaluation formula.

[0046] (6) In the failure prediction device 1 described in (3), the prediction judgment unit 40 may skip predicting a drive shaft failure if the false detection judgment unit 30 determines that the evaluation data is an evaluation data value due to a factor other than a failure of the robot R.

[0047] (7) The failure prediction device 1 described in (1) includes a prediction judgment unit 40 that predicts a failure of the drive shaft based on evaluation data, and the false detection judgment unit 30 may, after the prediction judgment unit 40 has predicted a failure of the drive shaft, determine whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot R, or an evaluation data value due to a factor that will cause the failure of the robot R.

[0048] (8) In the failure prediction device 1 described in (5), the threshold includes a first threshold derived based on the difference between the evaluation data and the value of the first evaluation formula, and a second threshold derived based on the difference between the evaluation data and the value of the second evaluation formula, and the false detection determination unit 30 may determine that the evaluation data is an evaluation data value due to a factor other than a failure of the robot R when the first threshold is greater than or equal to the second threshold, and may determine that the evaluation data is an evaluation data value due to a factor that causes a failure of the robot R when the first threshold is less than the second threshold.

[0049] (9) The failure prediction device 1 of the present disclosure includes an evaluation data collection unit 10 that collects evaluation data of a robot R that performs work based on a work program, a false detection determination unit 30 that determines whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot R, or an evaluation data value due to a factor that will cause the robot R to fail, and a display control unit 50 that displays on a monitor M at least one of the prediction result of the failure of the robot R predicted based on the evaluation data and information indicating the determination result of the false detection determination unit 30, wherein the display control unit 50 determines that the evaluation data value is due to a factor other than the failure of the robot R, and prioritizes displaying on the monitor M information indicating that the evaluation data value is due to a factor other than the failure of the robot R. This fault prediction device 1 can achieve the same effect as (1). [Explanation of symbols]

[0050] 1. Fault prediction device 10 Evaluation Data Collection Department 20 Evaluation formula derivation section 30 False detection determination unit 40 Prediction and Judgment Department 50 Display Control Unit R Robot J1, J2, J3, J4, J5, J6 drive shafts

Claims

1. An evaluation data collection unit that collects evaluation data for at least the drive axes of a robot performing work based on a work program, An evaluation formula derivation unit that derives an evaluation formula for evaluating the evaluation data using the aforementioned evaluation data, which includes a first evaluation formula obtained by linear regression of the evaluation data and a second evaluation formula obtained by applying a step function or binarization process to the evaluation data, A false detection determination unit that determines, based on the first and second evaluation formulas and the evaluation data, whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot, or an evaluation data value due to a factor that causes the failure of the robot, Equipped with, The aforementioned false detection determination unit is A first threshold is derived based on the difference between the evaluation data and the value of the first evaluation formula, and a second threshold is derived based on the difference between the evaluation data and the value of the second evaluation formula. If the first threshold is greater than or equal to the second threshold, it is determined that the evaluation data is an evaluation data value due to a factor other than the failure of the robot; if the first threshold is less than the second threshold, it is determined that the evaluation data is an evaluation data value due to a factor that causes the failure of the robot. Fault prediction device.

2. The failure prediction device according to claim 1, further comprising a prediction determination unit that predicts the failure of the drive shaft based on the evaluation data when the false detection determination unit determines that the evaluation data is an evaluation data value due to a factor causing the failure of the robot.

3. The failure prediction device according to claim 2, wherein the prediction judgment unit skips predicting a failure of the drive shaft when the false detection judgment unit determines that the evaluation data is an evaluation data value due to a factor other than the failure of the robot.

4. The system includes a predictive determination unit that predicts the failure of the drive shaft based on the aforementioned evaluation data, The failure prediction device according to claim 1, wherein the false detection determination unit determines, after the prediction determination unit has predicted a failure of the drive shaft, whether the evaluation data is an evaluation data value due to a factor other than the failure of the robot, or an evaluation data value due to a factor that will cause the failure of the robot.

5. The system includes a display control unit that displays on a display unit at least one of the predicted failure result of the robot, based on the evaluation data, and the judgment result of the false detection determination unit. When the display control unit determines that the evaluation data value is due to a factor other than a robot failure, it prioritizes displaying information on the display unit indicating that the evaluation data value is due to a factor other than a robot failure. The fault prediction device according to claim 1.

Images