JP7779220B2 - Failure prediction system and failure prediction method - Google Patents

Description

The present invention relates to technology for predicting equipment failures.

Vehicle production facilities use a large number of devices that operate repeatedly. As a result, failure of any of these devices can result in extended shutdowns of the production facilities. To manage equipment failures, for example, Patent Document 1 below discloses an equipment failure diagnosis method. This equipment failure diagnosis method is a technology that, when a failure is determined to have occurred based on the operating status of one cycle of the equipment, identifies and displays the device involved in the equipment failure. This equipment failure diagnosis method can reduce the effort required to discover failures.

Japanese Patent Application Publication No. 5-189026

Avoiding prolonged shutdowns of production equipment requires technology that can identify and address abnormalities before equipment failure occurs. However, the equipment failure diagnosis methods described above are technologies that identify equipment that has already failed, and are not able to determine abnormalities before the equipment fails, making them difficult to adopt as a solution to the problems described above. Furthermore, even if maintenance is performed on each piece of equipment in advance and an upper limit on the subsequent cumulative operating time is set to ensure operation, equipment failures can occur regardless of the period up until the cumulative operating time reaches the upper limit, so this does not fundamentally solve the problem of equipment management.

The present invention was made in consideration of these issues and aims to provide failure prediction technology with excellent accuracy in predicting equipment failures.

One aspect of the present invention is
A failure prediction system for predicting failures of equipment that repeats multiple operations,
an operation time detection unit that detects an operation time required from start to stop of the device for each operation;
a transition information deriving unit that derives transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation based on the operation time detected by the operation time detecting unit;
a failure prediction unit that predicts that a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived by the transition information derivation unit;
Equipped with
the transition information derivation unit is configured to derive, as the transition information, a scatter plot having a horizontal axis representing the number of times the device is operated and a vertical axis representing a cumulative operation time, which is a cumulative value of the operation time;
the failure prediction unit is configured to predict that a failure of the device is approaching when three points of data from the nth time to the (n+2)th time of the cumulative operating time are connected by an ascending curve in the scatter diagram derived by the transition information derivation unit.
is located.
Another aspect of the present invention is
A failure prediction system for predicting failures of equipment that repeats multiple operations,
an operation time detection unit that detects an operation time required from start to stop of the device for each operation;
a transition information deriving unit that derives transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation based on the operation time detected by the operation time detecting unit;
a failure prediction unit that predicts that a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived by the transition information derivation unit;
Equipped with
the transition information derivation unit is configured to derive, as the transition information, a scatter plot having a horizontal axis representing the number of times the device is operated and a vertical axis representing a cumulative operation time, which is a cumulative value of the operation time;
the failure prediction unit is configured to predict that a failure of the device is approaching when three points of data from the nth time to the (n+2)th time of the cumulative operating time are connected by an ascending curve in the scatter diagram derived by the transition information derivation unit, and when the (n+3)th time of the cumulative operating time, which is estimated to be on an extension of the ascending curve, exceeds a predetermined upper control limit value.
is located.

Another aspect of the present invention is a method for producing a semiconductor device comprising:
A failure prediction method for predicting failure of equipment that repeats multiple operations, comprising:
an operation time detection step of detecting an operation time required from start to stop of the device by an operation time detection unit every time;
a transition information deriving step of deriving transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation by a transition information deriving unit based on the operation time detected in the operation time detecting step;
a failure prediction step of predicting, by a failure prediction unit, that a time when a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived in the transition information derivation step;
and
In the transition information deriving step, a scatter plot is derived as the transition information, with the horizontal axis representing the number of times the device is operated and the vertical axis representing the cumulative operation time, which is the cumulative value of the operation time;
the failure prediction step predicts that a failure of the device is imminent when three points of data from the nth time to the (n+2)th time of the cumulative operating time are connected by an ascending curve in the scatter diagram derived in the transition information derivation step;
is located.
Another aspect of the present invention is a method for producing a semiconductor device comprising:
A failure prediction method for predicting failure of equipment that repeats multiple operations, comprising:
an operation time detection step of detecting an operation time required from start to stop of the device by an operation time detection unit every time;
a transition information deriving step of deriving transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation by a transition information deriving unit based on the operation time detected in the operation time detecting step;
a failure prediction step of predicting, by a failure prediction unit, that a time when a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived in the transition information derivation step;
and
In the transition information deriving step, a scatter plot is derived as the transition information, with the horizontal axis representing the number of times the device is operated and the vertical axis representing the cumulative operation time, which is the cumulative value of the operation time;
a failure prediction method in which, in the failure prediction step, three points of data from the nth time to the (n+2)th time of the cumulative operating time are connected by an ascending curve in the scatter diagram derived in the transition information derivation step, and it is predicted that a time when a failure of the device will occur is approaching when the cumulative operating time of the (n+3)th time, which is estimated to be on an extension of the ascending curve, exceeds a predetermined upper control limit value;
is located.

In each of the above-described aspects, the operating time of a device that repeats multiple operations is detected for each operation. Based on the detected operating time, transition information for the device's operating time from any nth to (n+2)th operation is derived. Then, based on the derived transition information, it is determined that the operating time has increased three consecutive times from the nth to (n+2)th operation, and a prediction is made that the device is nearing failure. This makes it possible to predict the occurrence of an equipment failure by comparing the transitions in the device's operating time for the most recent three consecutive operations, enabling highly accurate failure predictions that do not rely on guaranteed upper limits in the device's specifications or guaranteed upper limits set during equipment maintenance. As a result, anomalies can be addressed before a failure occurs, reducing the occurrence of long-term shutdowns of production facilities equipped with multiple devices.

As described above, each of the above aspects makes it possible to provide failure prediction technology with excellent accuracy in predicting equipment failures.

FIG. 1 is a configuration diagram of a failure prediction system according to a first embodiment. 2 is a diagram for explaining the concept of failure prediction processing in the failure prediction unit in FIG. 1 ; FIG. 3 is a partially enlarged view of FIG. 2 . FIG. 2 is a flowchart showing a failure prediction method according to the first embodiment. FIG. 10 is a diagram corresponding to FIG. 2 according to a second embodiment.

Preferred embodiments of the above aspects are described below.

In the failure prediction system of the above aspect, the transition information derivation unit is preferably configured to derive, as the transition information, a scatter plot with the number of times the device is operated on the horizontal axis and the cumulative operating time, which is the cumulative value of the operation time, on the vertical axis, and the failure prediction unit is preferably configured to predict that a failure of the device is imminent when three points of data from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter plot derived by the transition information derivation unit.

This failure prediction system uses a scatter diagram derived from the equipment's operating time to predict equipment failure based on the relationship between three-point data and an ascending curve from the nth to (n+2)th cumulative operating time.

In the failure prediction system of the above aspect, the transition information derivation unit is preferably configured to derive, as the transition information, a scatter plot with the horizontal axis representing the number of times the device is operated and the vertical axis representing cumulative operation time, which is the cumulative value of the operation time; and the failure prediction unit is preferably configured to predict that a failure of the device is imminent when, in the scatter plot derived by the transition information derivation unit, three points of data for the cumulative operation time from the nth operation to the (n+2)th operation are connected by an ascending curve, and the cumulative operation time of the (n+3)th operation, which is estimated to be on an extension of the ascending curve, exceeds a predetermined upper control limit.

This failure prediction system can reflect the upper management limit in its failure predictions, in addition to the last three consecutive operating conditions of the equipment, thereby improving the accuracy of failure predictions compared to analysis based only on the last three consecutive operating conditions of the equipment.

In the failure prediction system of the above aspect, the ascending curve preferably constitutes a quadratic curve or an approximation curve approximated by a quadratic curve.

This failure prediction system simplifies the process of determining whether the three data points from the nth cumulative operating time to the (n+2)th cumulative operating time are connected by an ascending curve by using a relatively simple quadratic curve or an approximation thereof.

The failure prediction system of the above aspect preferably includes a notification output unit that notifies the user of the prediction results by the failure prediction unit.

This failure prediction system can quickly notify on-site users of production equipment that an equipment failure is imminent.

In the failure prediction method of the above aspect, in the transition information derivation step, a scatter plot with the number of times the device is operated on the horizontal axis and the cumulative operating time, which is the cumulative value of the operation time, on the vertical axis is derived as the transition information, and in the failure prediction step, it is preferable to predict that a failure of the device is imminent when three points of data from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter plot derived in the transition information derivation step.

This failure prediction method makes it possible to predict equipment failures based on the relationship between three points of data and the ascending curve on a scatter plot from the nth cumulative operating time to the (n+2)th cumulative operating time.

In the failure prediction method of the above aspect, in the transition information derivation step, a scatter plot with the horizontal axis representing the number of times the device is operated and the vertical axis representing the cumulative operating time, which is the cumulative value of the operating time, is derived as the transition information, and in the failure prediction step, it is preferable to predict that a failure of the device is imminent when, in the scatter plot derived in the transition information derivation step, three points of data for the cumulative operating time from the nth operation to the (n+2)th operation are connected by an ascending curve, and the cumulative operating time of the (n+3)th operation, which is estimated to be on an extension of the ascending curve, exceeds a predetermined upper control limit.

This failure prediction method allows the failure prediction to reflect not only the most recent three consecutive operating conditions of the equipment, but also the upper management limit, thereby improving the accuracy of failure prediction compared to analysis based only on the most recent three consecutive operating conditions of the equipment.

In the failure prediction method of the above aspect, the ascending curve preferably constitutes a quadratic curve or an approximation curve approximated by a quadratic curve.

This failure prediction method simplifies the process of determining whether the three data points from the nth cumulative operating time to the (n+2)th cumulative operating time are connected by an ascending curve by using a relatively simple quadratic curve or an approximation thereof.

The failure prediction method of the above aspect preferably includes a notification output step in which the notification output unit notifies the user of the prediction result obtained by the failure prediction step.

This failure prediction method allows on-site users of production equipment to be quickly notified that an equipment failure is imminent.

Below, we will explain an embodiment of a technology that predicts failures in equipment that repeatedly operates multiple times, with reference to the drawings.

(Embodiment 1)
As shown in FIG. 1 , the failure prediction system 1 according to the first embodiment is installed in a vehicle production facility, and includes a control panel 11 , a data server 14 , and a facility management device 20 .

Each control panel 11 is electrically connected to the equipment 10 to be controlled. Typical examples of the equipment 10 include motors, cylinders, and other operating devices. The control panel 11 is equipped with multiple sensors 12 that detect the start and stop of the equipment 10. Data D from the multiple sensors 12 is measured by an input/output time measuring instrument 13 (hereinafter simply referred to as "measuring instrument 13"). The data D measured by the measuring instrument 13 is transmitted to and stored in a data server 14. This data D includes sensor information that can identify the actual start and stop timing of the equipment 10 (e.g., the output time of an input signal from the sensor 12 to the equipment 10, the detection time of an output signal from the equipment 10 to the sensor 12, the ON and OFF times of each sensor 12, etc.). The data D stored in the data server 14 is transmitted to the equipment management device 20 via a wireless or wired communication line NT.

The equipment management device 20 is a computer device having a known CPU (Central Processing Unit), ROM, RAM, an interface for input/output with external devices, etc. This equipment management device 20 is preferably configured as a desktop or notebook personal computer (PC), a tablet terminal, a mobile terminal, etc.

The equipment management device 20 performs the essential function of the failure prediction system 1, namely, the function of predicting failures of equipment 10 that repeatedly operates multiple times, based on the data D measured by the measuring instrument 13. This equipment management device 20 includes an operation time detection unit 21, a transition information derivation unit 22, a failure prediction unit 23, and an alarm output unit 24.

The measuring instrument 13 measures information regarding the operating time t required from start-up to shutdown of the device 10 for each operation. Here, "one time" can also be referred to as "one cycle." The operating time t at this time is detected by the operating time detection unit 21 performing calculations using the aforementioned data D. That is, the operating time detection unit 21 is configured to detect the operating time t required from start-up to shutdown of the device 10 for each operation in cooperation with the measuring instrument 13. The transition information derivation unit 22 is configured to derive transition information It (scatter plot S described below) of the operating time t. The failure prediction unit 23 is configured to predict that a failure of the device 10 is imminent based on a predetermined determination result described below. The notification output unit 24 is configured to output a notification of the prediction result by the failure prediction unit 23 to the user.

Next, the specific processing performed by the equipment management device 20 will be described with reference to Figures 2 to 4. The failure prediction method of embodiment 1 is a method for predicting failures in equipment 10, and is made possible by sequentially executing the steps from the first step S101 to the fourth step S104 in Figure 4. In this embodiment, it is preferable that all of these steps be executed by the equipment management device 20.

The operation time detection unit 21 detects the operation time t of the device 10 in a first step S101 (operation time detection step) in FIG. 4. The transition information derivation unit 22 then derives transition information It of any three consecutive operation times t based on the operation time t detected by the operation time detection unit 21 in a second step S102 (transition information derivation step) in FIG. 4. The transition information derivation unit 22 derives a scatter diagram S (see FIG. 2) as the transition information It of any nth to (n+2)th operation of the device 10. Then, the failure prediction unit 23 performs failure prediction processing using the scatter diagram S in a third step S103 (failure prediction step) in FIG. 4.

Now, we will explain the scatter diagram S in Figure 2. This scatter diagram S has the number of activations N of the device 10 on the horizontal axis and the cumulative activation time T, which is the cumulative value of the activation time t, on the vertical axis. In this scatter diagram S, the intervals between the number of activations N on the horizontal axis are constant. In this embodiment, a scatter diagram S is shown that shows the correlation between the number of activations N and the cumulative activation time T for the number of activations N from the (n-1)th to the (n+3)th activation of the device 10.

In this scatter plot S, when the device 10 is operating normally and without any abnormalities, the cumulative operating time T for each operation, excluding the initial state until the moving parts (not shown) of the device 10 are adjusted, forms a straight, upward-sloping reference line Lr. In other words, the operating time t of the device 10 remains roughly constant for all operations. However, when an abnormality begins to occur in the device 10, the increase in the cumulative operating time T until the next operation (i.e., the actual operating time t required) increases. In this case, the time difference between the previous and next cumulative operating times T exceeds the normal time difference ΔT. In Figure 2, the time difference A between the cumulative operating time T(n+1) and the cumulative operating time T(n), and the time difference B between the cumulative operating time T(n+2) and the cumulative operating time T(n+1) both exceed the normal time difference ΔT.

The failure prediction process by the failure prediction unit 23 is a process that predicts that a failure of the device 10 is imminent when it is determined that the operating time t has increased three consecutive times, from the nth to the (n+2)th time, based on the scatter diagram S in Figure 2. A consecutive increase in the operating time t is determined, specifically, by whether or not three points of data from the nth to the (n+2)th time of the cumulative operating time T are connected by an ascending curve L in the scatter diagram S. If these three points of data are connected by an ascending curve L, it is predicted that a failure of the device 10 is imminent; if not, it is predicted that the device 10 will not fail immediately.

The "ascending curve L" referred to here is a curve that is substantially convex downward in the range from the nth to the (n+2)th plot and whose differential coefficient (the slope of the tangent at each plot point) is always positive. In this case, the differential coefficient may or may not be constant. This ascending curve L is typically preferably a quadratic curve with a positive coefficient or an approximation curve that is approximated by this quadratic curve. By making the ascending curve L a relatively simple quadratic curve or its approximation curve, the process of determining whether the above three data points are connected by the ascending curve L can be simplified. However, the ascending curve L is not limited to quadratic curves or their approximation curves, and other curves may be applied as needed.

When the above three data points are connected by an ascending curve L, it is estimated that the (n+3)th cumulative operating time T(n+3) is plotted on an extension of this ascending curve L (i.e., on the curve indicated by the two-dot chain line in Figure 2) (see the square plot in Figure 2). Therefore, there is a high possibility that the cumulative operating time T will increase significantly in the (n+3)th operation and exceed the predetermined upper limit control value UL. The upper limit control value UL is typically the guaranteed upper limit value in the specifications of the device 10, or the guaranteed upper limit value set during maintenance of the device 10. Therefore, if the cumulative operating time T exceeds the upper limit control value UL, there is a high possibility that the device 10 will actually malfunction.

In this embodiment, therefore, a failure prediction for the device 10 is made based on the relationship between the above three data points and the rising curve L, and when the above three data points are connected by the rising curve L, it is predicted that a failure of the device 10 is imminent. The cumulative operating time T of the most recent three consecutive times is used as one set for continued prediction. This allows the operating status of the device 10 for the most recent three consecutive times to be constantly analyzed without using the upper control limit value UL, and the results of this analysis to be reflected in the failure prediction for the device 10.

Alternatively, the three data points are connected by an ascending curve L, and when the time difference between the (n+3)th cumulative operating time T(n+3) and the (n+2)th cumulative operating time T(n+2), which are estimated to be on an extension of the ascending curve L, is equal to or greater than twice the time difference B (see Figure 2), it may be possible to predict that a failure of the device 10 is imminent. This is effective in improving the accuracy of failure predictions for the device 10.

Alternatively, when the three data points are connected by an ascending curve L and the (n+3) cumulative operating time T(n+3), which is estimated to be on an extension of the ascending curve L, exceeds the upper limit control value UL, it may be predicted that the device 10 is nearing a failure. This allows the upper limit control value UL to be reflected in the failure prediction in addition to the operating status of the device 10 for the most recent three consecutive times. In this case, the accuracy of the failure prediction can be improved compared to when analysis is based only on the operating status of the device 10 for the most recent three consecutive times.

In contrast, as shown in Figure 3, which is part of Figure 2, examples in which the above three data points are not connected by an ascending curve L include a first comparative example in which they are connected by a straight line La, a second comparative example in which they are connected by a curve Lb, and a third comparative example in which they are connected by a curve Lc.

In the first comparative example, although the cumulative operating time T increases three consecutive times from the nth to the (n+2)th operation, the three data points are connected by a straight line La, which is essentially different from the increasing curve L of this embodiment. The operating time T remains constant from the nth to the (n+2)th operation. In this case, the rate of increase of the cumulative operating time T changes around the nth operation, and although it increases after the nth operation, the rate of increase remains constant from the nth to the (n+2)th operation, so it cannot be said that the device 10 is unambiguously in an abnormal state.

In the second comparative example, curve Lb is an upwardly convex curve and does not correspond to the ascending curve L of this embodiment. Furthermore, in the third comparative example, curve Lc is a downwardly convex curve like the ascending curve L, but the differential coefficient of curve Lc (the slope of the tangent at each plot point) is negative in the range from the nth to the (n+1)th time, and therefore it is differentiated from the ascending curve L of this embodiment.

Although not specifically shown, it is preferable that the determination of whether the above three data points are connected by an ascending curve L in the scatter diagram S be performed automatically using a known program pre-installed in the equipment management device 20 or an AI model (trained model) based on artificial intelligence. Alternatively or additionally, the user may visually view the scatter diagram S and make the determination directly.

The notification output unit 24 outputs the prediction result by the failure prediction unit 23 as a notification in the fourth step S104 (notification output step) in Figure 4. The term "notification output" here broadly encompasses various output formats, such as screen display output, audio output, alarm output, and printout output. This allows on-site users of the production equipment to be promptly notified that a failure of the equipment 10 is imminent.

Next, we will explain the effects of the above-mentioned embodiment 1.

In embodiment 1, the operation time t of the device 10, which repeats multiple operations, is detected for each operation. Furthermore, based on the detected operation time t, transition information It of the operation time t of the device 10 from any nth operation to the (n+2)th operation is derived. Then, based on the derived transition information It (scatter plot S), it is determined that the operation time t has increased three consecutive times from the nth operation to the (n+2)th operation, and it is predicted that the device 10 is about to malfunction. This allows the occurrence of a malfunction in the device 10 to be predicted by comparing the transitions of the operation time t of the device 10 for the most recent three consecutive operations, enabling highly accurate malfunction predictions that do not rely on the guaranteed upper limit UL in the device 10 specifications or the guaranteed upper limit UL set during device maintenance. As a result, an abnormality can be addressed before a malfunction occurs in the device 10, and long-term shutdowns of production facilities equipped with multiple devices 10 can be reduced.

Furthermore, if the trends in the operating time t for two consecutive times are compared, it is difficult to determine whether this is an abnormality or an error fluctuation, and if the trends in the operating time t for four or more consecutive times are compared, the cumulative operating time T will have risen significantly, making it highly likely that the device 10 has already reached a breakdown. For this reason, neither of these measures is effective in detecting an abnormality before the device 10 breaks down.

As described above, embodiment 1 provides a failure prediction system 1 and a failure prediction method that have excellent accuracy in predicting equipment failures.

Other embodiments related to the above-described first embodiment will be described below with reference to the drawings. In these embodiments, elements that are the same as those in the first embodiment will be assigned the same reference numerals, and descriptions of these same elements will be omitted.

(Embodiment 2)
The failure prediction system and the failure prediction method of the second embodiment are basically the same as those of the first embodiment. However, as shown in Fig. 5 , the scatter diagram S' derived by the transition information derivation unit 22 as the transition information It (see Fig. 1 ) is different from the scatter diagram S (see Fig. 2 ) of the first embodiment.

The scatter diagram S' has the number of activations N of the device 10 on the horizontal axis and the activation time t on the vertical axis. In this scatter diagram S', the intervals between the number of activations N on the horizontal axis are constant. In this embodiment, the scatter diagram S' shows the correlation between the number of activations N and the cumulative activations t for the number of activations N of the device 10 from the (n-1)th to the (n+3)th activation.

In this scatter plot S', when the device 10 is operating normally and without any abnormalities, the operating time t of the device 10 is roughly constant for all times, and connecting the operating times t of each time forms a horizontal reference line Mr. However, when abnormalities begin to occur in the device 10, the actual operating time t required increases. In this case, the time difference between the previous and next operating times t becomes a positive value. In Figure 2, the time difference A between operating times t(n+1) and t(n), and the time difference B between operating times t(n+2) and t(n+1) are both positive values.

The failure prediction process by the failure prediction unit 23 (see Figure 1) is a process that predicts that a failure of the device 10 is imminent when it is determined that the operating time t has increased three consecutive times, from the nth to the (n+2)th time, based on the scatter diagram S' in Figure 5. A consecutive increase in the operating time t is determined, specifically, by whether or not an ascending curve M connects three data points from the nth to the (n+2)th time of the operating time t in the scatter diagram S'. If these three data points are connected by an ascending curve M, it is predicted that a failure of the device 10 is imminent; if not, it is predicted that the device 10 will not fail immediately.

The "rising curve M" referred to here is a curve that, like the rising curve L in embodiment 1, is substantially convex downward in the range from the nth to the (n+2)th operation and whose differential coefficient (the slope of the tangent at each point) is always positive. When the above three data points are connected by such a rising curve M, it is estimated that the (n+3)th operation time t(n+3) is plotted on an extension of this rising curve M (i.e., on the curve indicated by the two-dot chain line in Figure 5) (see the square plot in Figure 5).

In this embodiment, as in the first embodiment, when the three data points are connected by an ascending curve M, it is predicted that the device 10 is about to fail. Alternatively, it may be predicted that the device 10 is about to fail when the three data points are connected by an ascending curve M and the time difference between the (n+3)th cumulative operating time T(n+3) and the (n+2)th cumulative operating time T(n+2), which are estimated to be on an extension of the ascending curve M, is equal to or greater than twice the time difference B (see FIG. 5). Alternatively, it may be predicted that the device 10 is about to fail when the three data points are connected by an ascending curve M and the (n+3)th cumulative operating time T(n+3), which is estimated to be on an extension of the ascending curve M, exceeds the upper limit management value UL.

Other configurations and methods are the same as in embodiment 1.

According to embodiment 2, a scatter plot S' derived from the operating time of the device 10 can be used to predict failures of the device 10 based on the relationship between the three-point data and the ascending curve M from the nth to the (n+2)th operating time t.

Other than that, it has the same effects as embodiment 1.

The present invention is not limited to the above-described embodiments, and various applications and modifications are possible without departing from the purpose of the present invention. For example, the following embodiments can be implemented by applying the above-described embodiments.

In the above embodiment, an example was given in which the prediction result by the failure prediction unit 23 (failure prediction step) is output to the user by the notification output unit 24 (notification output step), but instead, the prediction result may simply be output as data.

In the above-described embodiment, an example was given of a failure prediction system 1 and failure prediction used in vehicle production facilities, but the type of production facilities is not particularly limited, and it goes without saying that these technologies can also be applied to equipment failure prediction technologies in production facilities other than vehicles.

1 Failure prediction system 10 Equipment 21 Operation time detection unit 22 Transition information derivation unit 23 Failure prediction unit 24 Notification output unit t, t(n-1), t(n), t(n+1), t(n+2), t(n+3) Operation time It Transition information of operation time L Rise curve N Number of operation S, S' Scatter diagram (transition information)
S101 Operation time detection step S102 Transition information derivation step S103 Failure prediction step S104 Notification output step T, T(n-1), T(n), T(n+1), T(n+2), T(n+3) Cumulative operation time UL Upper limit of control

Claims (8)

A failure prediction system for predicting failures of equipment that repeats multiple operations,
an operation time detection unit that detects an operation time required from start to stop of the device for each operation;
a transition information deriving unit that derives transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation based on the operation time detected by the operation time detecting unit;
a failure prediction unit that predicts that a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived by the transition information derivation unit;
Equipped with
the transition information derivation unit is configured to derive, as the transition information, a scatter plot having a horizontal axis representing the number of times the device is operated and a vertical axis representing a cumulative operation time, which is a cumulative value of the operation time;
the failure prediction unit is configured to predict that a failure of the equipment is imminent when three points of data from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter diagram derived by the transition information derivation unit. A failure prediction system for predicting failures of equipment that repeats multiple operations,
an operation time detection unit that detects an operation time required from start to stop of the device for each operation;
a transition information deriving unit that derives transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation based on the operation time detected by the operation time detecting unit;
a failure prediction unit that predicts that a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived by the transition information derivation unit;
Equipped with
the transition information derivation unit is configured to derive, as the transition information, a scatter plot having a horizontal axis representing the number of times the device is operated and a vertical axis representing a cumulative operation time, which is a cumulative value of the operation time;
the failure prediction unit is configured to predict that a failure of the equipment is approaching when three points of data from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter diagram derived by the transition information derivation unit, and when the (n+3)th cumulative operating time, which is estimated to be on an extension of the ascending curve , exceeds a predetermined upper control limit value. 3. The failure prediction system according to claim 1 , wherein the ascending curve is a quadratic curve or an approximation curve that is approximated by a quadratic curve. 3. The failure prediction system according to claim 1 , further comprising a notification output unit that notifies a user of the prediction result by said failure prediction unit. A failure prediction method for predicting failure of equipment that repeats multiple operations, comprising:
an operation time detection step of detecting an operation time required from start to stop of the device by an operation time detection unit every time;
a transition information deriving step of deriving transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation by a transition information deriving unit based on the operation time detected in the operation time detecting step;
a failure prediction step of predicting, by a failure prediction unit, that a time when a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived in the transition information derivation step;
and
In the transition information deriving step, a scatter plot is derived as the transition information, with the horizontal axis representing the number of times the device is operated and the vertical axis representing the cumulative operation time, which is the cumulative value of the operation time;
In the failure prediction step, when three data points from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter diagram derived in the transition information derivation step, it is predicted that a failure of the equipment is imminent . A failure prediction method for predicting failure of equipment that repeats multiple operations, comprising:
an operation time detection step of detecting an operation time required from start to stop of the device by an operation time detection unit every time;
a transition information deriving step of deriving transition information of the operation time of the device from an arbitrary nth operation to an (n+2)th operation by a transition information deriving unit based on the operation time detected in the operation time detecting step;
a failure prediction step of predicting, by a failure prediction unit, that a time when a failure of the device is imminent when it is determined that the operating time has increased continuously from the nth time to the (n+2)th time based on the transition information derived in the transition information derivation step;
and
In the transition information deriving step, a scatter plot is derived as the transition information, with the horizontal axis representing the number of times the device is operated and the vertical axis representing the cumulative operation time, which is the cumulative value of the operation time;
In the failure prediction step, when three points of data from the nth to the (n+2)th cumulative operating time are connected by an ascending curve in the scatter diagram derived in the transition information derivation step, and the cumulative operating time of the (n+3)th cumulative operating time, which is estimated to be on an extension of the ascending curve, exceeds a predetermined upper control limit value , it is predicted that a failure of the equipment is imminent. 7. The failure prediction method according to claim 5 , wherein the ascending curve is a quadratic curve or an approximation curve that is approximated by a quadratic curve. 7. The failure prediction method according to claim 5 , further comprising a notification output step of outputting a notification of the prediction result from said failure prediction step to a user by a notification output unit.