JP2021177294A - Failure prediction system and failure prediction method - Google Patents

Abstract

To provide a failure prediction system and a failure prediction method which allow failure of a base carrier of a working machine to be predicted with high reliability at a low cost.SOLUTION: The failure prediction system taking a working machine as an object comprises: detection means (pressure sensors 61L to 63R, a pump torque calculation unit, and a motor torque calculation unit) which detect, in an upper swivel 12, a prescribed state quantity relating to a hydraulic circuit 50 with travel motors 20L and 20R activated; a server device including a failure predictor which predicts failure of a base carrier 11 in accordance with a prescribed failure prediction model showing relationship between state quantity and failure, on the basis of the state quantity detected by the detection means; and a display device which outputs a prediction result of the failure predictor.SELECTED DRAWING: Figure 3

Description

The present invention relates to a failure prediction system and a failure prediction method for a work machine having a lower traveling body.

Patent Document 1 below discloses a method for estimating the expected useful life of mechanical parts for mechanical parts such as mobile cranes for ports. The method records the process data of a machine when performing work, transfers the recorded process data to a database, and estimates the expected useful life of the machine parts by analyzing the process data stored in the database. Process data includes speed, load, power, temperature, hydraulic conditions, etc. for individual machine parts.

JP-A-2018-14092

According to the method disclosed in Patent Document 1, process data that can be directly measured in each machine part is acquired for failure prediction of the machine part.

In a work machine such as a hydraulic excavator having a crawler type lower traveling body, crawler traveling is performed on a rough road with severe undulations, an unpaved road with many irregularities, a steep slope, etc. due to the characteristics of the work content. There are many. Since each part of the lower traveling body is severely damaged due to the stress during running of the crawler, it is recommended to perform maintenance such as parts replacement before the actual failure occurs.

Therefore, by diverting the method disclosed in Patent Document 1 to predict the failure of each part of the lower traveling body, a strain gauge is attached to each part of the lower traveling body, and the stress applied to each part during crawler running is applied. Can be considered as a method of directly measuring with the strain gauge.

However, since a large stress is applied to the lower traveling body when the crawler is running, if a sensor such as a strain gauge is placed in the lower traveling body, the sensor itself may frequently fail due to the stress. The prediction is unreliable. Further, it is necessary to pass the wiring for connecting the sensor in the lower traveling body and the controller in the upper rotating body through the joint portion between the lower traveling body and the upper rotating body. Therefore, the wiring may be damaged or broken during the turning operation of the upper swing body, and the reliability of failure prediction is low. Moreover, since it is necessary to additionally mount the sensor and the wiring for each component of the lower traveling body which is the target of failure prediction, the manufacturing cost of the work machine increases.

The present invention has been made in view of such circumstances, and obtains a failure prediction system and a failure prediction method capable of realizing failure prediction of a lower traveling body of a work machine with high reliability and low cost. The purpose is.

In the failure prediction system according to one aspect of the present invention, the traveling motor in the lower traveling body is driven by the upper rotating body, the lower traveling body, and the hydraulic oil supplied from the hydraulic pump in the upper rotating body via the control valve. A failure prediction system for a work machine having a hydraulic circuit, the detection means for detecting a predetermined state amount of the hydraulic circuit during operation of the traveling motor in the upper swing body, and the detection. Based on the state amount detected by the means, a prediction means for predicting a failure of the lower traveling body by a predetermined failure prediction model representing the relationship between the state amount and the failure, and an output for outputting the prediction result by the prediction means. It is provided with means.

According to this aspect, the detecting means detects a predetermined state quantity regarding the hydraulic circuit during operation of the traveling motor in the upper swing body, and the predicting means detects a predetermined failure based on the state quantity detected by the detecting means. The failure of the lower traveling body is predicted by the prediction model. Therefore, since it is not necessary to dispose a sensor such as a strain gauge in the lower traveling body where a large stress is applied during traveling, the sensor itself does not fail or the wiring is not broken due to the stress. As a result, it becomes possible to improve the reliability of the failure prediction system. Moreover, since the existing pressure sensor in the upper swing body can be used as the detection means, it is possible to realize the failure prediction system at low cost.

In the above embodiment, the state quantity includes at least one of the pressure or torque of the hydraulic pump, the drive control signal for driving the control valve, the pressure, torque, or the operating time of the traveling motor. Is desirable.

According to this aspect, at least one of the pressure or torque of the hydraulic pump, the drive control signal for driving the control valve, the pressure, torque of the traveling motor, or the operating time is used as the state quantity of the hydraulic circuit. As a result, it becomes possible to appropriately execute the failure prediction model creation process and the failure prediction process using the failure prediction model.

In the above aspect, an information processing device for creating the failure prediction model by machine learning is further provided by using the state quantity as an explanatory variable and the presence or absence of the failure as an objective variable and using the collected failure information of the lower traveling body as teacher data. It is desirable to prepare.

According to this aspect, it is possible to appropriately create a failure prediction model by machine learning using the state quantity as an explanatory variable and the presence or absence of a failure as an objective variable and the collected failure information of the lower traveling body as teacher data. Become.

In the above aspect, the work machine further includes a work device connected to the upper swing body, and the state amount includes the traveling motor and the operating state of the work machine at the time of detecting the state amount. It is desirable to add operation state information indicating whether it is a combined operation state with the work device or an independent operation state of the traveling motor.

According to this aspect, a failure prediction model creation process and a failure prediction process using the failure prediction model are performed by distinguishing the state quantity according to whether the operation state of the work machine is a combined operation state or a single operation state. It is possible to improve the accuracy of.

In the above aspect, it is desirable that the state quantity is added with travel path information indicating whether the travel path of the work machine is flat or sloped at the time of detecting the state quantity.

According to this aspect, a failure prediction model is created and a failure using the failure prediction model is performed by distinguishing the state quantity according to whether the traveling path of the work machine is flat or sloping when the state quantity is detected. It is possible to improve the accuracy of the prediction process.

In the above aspect, it is desirable that the failure information includes information indicating the name of the failed component in the lower traveling body, and the predicting means further predicts the failed component in the lower traveling body.

According to this aspect, it is possible to create a failure prediction model capable of predicting a failed part of the lower traveling body by including information indicating the name of the failed part in the lower traveling body in the failure information. As a result, maintenance can be performed on a component-by-part basis.

In the above aspect, it is desirable that the server device for wireless communication with the work machine is further provided, and the server device has the prediction means.

According to this aspect, since the server device has a prediction means, when a large number of work machines are present, information is aggregated in the server device, so that the accuracy of the failure prediction model can be improved and the accuracy of the failure prediction model can be improved, and the accuracy of each work machine can be improved. Since it is possible to omit the installation of a high-performance information processing device, it is possible to reduce the system introduction cost.

In the above aspect, it is desirable that the working machine has the predictive means.

According to this aspect, since the work machine has the prediction means, an external server device or the like becomes unnecessary, and the system configuration can be simplified.

In the failure prediction method according to one aspect of the present invention, the traveling motor in the lower traveling body is driven by the upper rotating body, the lower traveling body, and the hydraulic oil supplied from the hydraulic pump in the upper rotating body via the control valve. It is a failure prediction method executed by a failure prediction system for a work machine having a hydraulic circuit, and detects a predetermined state amount related to the hydraulic circuit during operation of the traveling motor in the upper swing body. Based on the detected state amount, the failure of the lower traveling body is predicted by a predetermined failure prediction model representing the relationship between the state amount and the failure, and the prediction result is output.

According to this aspect, a predetermined state amount related to the hydraulic circuit during operation of the traveling motor is detected in the upper swing body, and a failure of the lower traveling body is predicted by a predetermined failure prediction model based on the detected state amount. .. Therefore, since it is not necessary to dispose a sensor such as a strain gauge in the lower traveling body where a large stress is applied during traveling, the sensor itself does not fail or the wiring is not broken due to the stress. As a result, it is possible to improve the reliability of the failure prediction system. Moreover, since the existing pressure sensor in the upper swing body can be used as the detection means, it is possible to realize the failure prediction system at low cost.

According to the present invention, it is possible to realize failure prediction of a lower traveling body of a work machine with high reliability and low cost.

It is a figure which shows typically the structure of the work machine which is the failure prediction target of the failure prediction system which concerns on embodiment of this invention. It is a figure which shows the whole structure of the failure prediction system which concerns on embodiment of this invention. It is a figure which shows by extracting the part which is related to the drive of a traveling motor in the hydraulic circuit in a work machine. It is a figure which shows the function which a controller has. It is a figure which shows an example of the input / output signal of a controller. It is a flowchart which shows the flow of the process which a controller executes about the creation and recording of the state quantity data. It is a flowchart which shows the flow of the process which a controller executes about the transmission of the state quantity data. It is a figure which shows the structure of a server apparatus. It is a flowchart which shows the flow of the process which a server device executes with respect to the failure prediction of a lower traveling body.

FIG. 1 is a diagram schematically showing a configuration of a work machine 1 which is a failure prediction target of the failure prediction system according to the embodiment of the present invention. In the example of this embodiment, the work machine 1 is a hydraulic excavator equipped with a bucket. However, the work machine 1 is not limited to a construction machine such as a hydraulic excavator, and may be another work machine such as an agricultural machine or an industrial machine.

As shown in FIG. 1, the work machine 1 includes a crawler-type lower traveling body 11, an upper rotating body 12 arranged on the lower traveling body 11, and a working device 13 connected to the upper rotating body 12. There is. The upper swivel body 12 can be swiveled with respect to the lower traveling body 11 by a swivel motor composed of a hydraulic motor.

The lower traveling body 11 has components such as a traveling motor 20, a frame 21, a roller 22, an idler 23, a crawler 24, and a shoe 25. The traveling motor 20 is composed of a hydraulic motor. The lower traveling body 11 has a pair of left and right traveling motors 20L and a right traveling motor 20R that can be independently driven as traveling motors 20. In FIG. 1, only the left traveling motor 20L appears. Inside the upper swing body 12, a controller 31 that controls the entire work machine 1 is arranged. The working device 13 has a boom 41, an arm 42, and a bucket 43. However, other attachments such as a breaker or a nibbler may be attached instead of the bucket 43. Further, the working device 13 has a boom cylinder 44, an arm cylinder 45, and a bucket cylinder 46 as actuators for driving the boom 41, the arm 42, and the bucket 43, respectively.

FIG. 2 is a diagram showing an overall configuration of a failure prediction system according to the present embodiment. As shown in FIG. 2, the failure prediction system includes a plurality of work machines 1A to 1C (three machines in the example of FIG. 2), a server device 2, a management device 3, a communication network 4, and a display device 5. Each work machine 1A to 1C corresponds to the work machine 1 shown in FIG. The communication network 4 is an arbitrary wired or wireless communication network such as an Internet communication network or a public telephone line network. The server device 2 can communicate with the work machines 1A to 1C via the communication network 4. The management device 3 can communicate with the work machines 1A to 1C and the server device 2 via the communication network 4. The display device 5 is a liquid crystal display, an organic EL display, or the like, and is connected to the management device 3. The management device 3 and the display device 5 are installed in the operation center where the administrator of the failure prediction system is stationed.

FIG. 3 is a diagram showing an extracted portion of the hydraulic circuit 50 in the work machine 1 that is related to the drive of the traveling motor 20. As shown in FIG. 3, the hydraulic circuit 50 includes a hydraulic pump 51 (left hydraulic pump 51L and right hydraulic pump 51R), a prime mover 52, a hydraulic oil tank 53, a relief valve 54 (left relief valve 54L and a right relief valve 54R), and traveling. It includes a motor control valve 55 (left traveling motor control valve 55L and right traveling motor control valve 55R) and a traveling motor 20 (left traveling motor 20L and right traveling motor 20R). By driving the control valve 55 for the traveling motor by the pressure of the pilot signal in response to the operator's lever operation or pedal operation, the availability, direction and flow rate of hydraulic oil supplied from the hydraulic pump 51 to the traveling motor 20 can be determined. Be controlled.

The operation lever or operation pedal may be a so-called electric lever or electric pedal that outputs an electric signal according to the amount of operation by the operator. In this case, the control system of the traction motor control valve 55 is configured to include the operation lever or operation pedal, a predetermined controller, and a proportional control valve. The electric signal output from the operation lever or the operation pedal is input to the controller, and the controller inputs the command current value corresponding to the electric signal to the proportional valve. The proportional valve outputs the pilot signal pressure according to the command current value. By driving the traction motor control valve 55 by the pilot signal pressure, the availability of hydraulic oil supply from the hydraulic pump 51 to the traction motor 20, the supply direction, and the flow rate are controlled.

Among the hydraulic circuits 50 shown in FIG. 3, the traveling motor 20 is arranged in the lower traveling body 11, and the other components are arranged in the upper rotating body 12. A swivel joint 56 is arranged at a connection portion of the circuit between the lower traveling body 11 and the upper rotating body 12. In the example of this embodiment, a so-called two-pump type hydraulic circuit is adopted. For example, the left hydraulic pump 51L is shared for driving the left traveling motor 20L, the swivel motor, and the arm cylinder 45, and the right hydraulic pump 51R. Is shared for driving the right traveling motor 20R, the boom cylinder 44, and the bucket cylinder 46.

Further, the hydraulic circuit 50 has pressure sensors 61 to 63 (left pressure sensors 61L to 63L and right pressure sensors 61R to 63R). The pressure sensors 61 to 63 are arranged in the upper swing body 12. The pressure sensor 61 measures the pressure of the hydraulic pump 51, that is, the pressure of the hydraulic oil at the outlet point of the hydraulic pump 51, and outputs signals S1 (S1L and S1R) indicating the measured values. The pressure sensor 62 measures the pressure of the hydraulic oil at the first port (hydraulic oil outlet port during forward traveling) on the traveling motor 20 side of the traveling motor control valve 55, and signals S6 (S6L and S6L) indicating the measured values. S6R) is output. This pressure is substantially equal to the pressure of the hydraulic oil at the first port (hydraulic oil inlet port during forward traveling) of the traveling motor 20. The pressure sensor 63 measures the pressure of hydraulic oil at the second port (hydraulic oil inlet port during forward traveling) on the traveling motor 20 side of the traveling motor control valve 55, and signals S7 (S7L and S7R) indicating the measured values. ) Is output. This pressure is substantially equal to the pressure of the hydraulic oil at the second port (hydraulic oil outlet port during forward traveling) of the traveling motor 20. The pressure of the traveling motor 20 is obtained as the difference between the pressure measured by the pressure sensor 62 (signal S6) and the pressure measured by the pressure sensor 63 (signal S7).

FIG. 4 is a diagram showing a function of the controller 31, and FIG. 5 is a diagram showing an example of an input / output signal of the controller 31. As shown in FIG. 4, the controller 31 includes an operating state determination unit 71, a traveling path determination unit 72, a counter 73, a pump torque calculation unit 74, and a motor torque calculation unit 75. These functions may be realized as software by executing a predetermined program by a processor such as a CPU, or may be realized as hardware by using a dedicated circuit such as FPGA. Further, the controller 31 is connected to a non-volatile storage unit 32 such as a flash memory and a communication unit 33 for wireless communication with the server device 2 and the management device 3 via the communication network 4.

With reference to FIGS. 4 and 5, signals S1 (S1L and S1R) indicating the pressure of the hydraulic pumps 51 (51L and 51R) are input to the controller 31 from the pressure sensors 61 (61L and 61R). The controller 31 inputs the pump current for controlling the tilt angle of the cylinder block (not shown) to the hydraulic pump 51 as signals S2 (S2L and S2R) for controlling the discharge amount of hydraulic oil. Signals S3 (S3L and S3R) indicating the pressure of the pilot signal for driving the traction motor control valves 55 (55L and 55R) are input to the controller 31. A signal S4 indicating the pressure of a pilot signal for driving a boom cylinder control valve, an arm cylinder control valve, a bucket cylinder control valve, or a swivel motor control valve (all not shown) is input to the controller 31. Will be done. The controller 31 inputs a signal S5 (S5L and S5R) for setting the capacitance of the variable capacitance type traveling motor 20 (20L and 20R) to the traveling motor 20 via a tilt switching valve (not shown). Signals S6 (S6L and S6R) indicating the pressure of the first port of the traction motor control valve 55 are input to the controller 31 from the pressure sensors 62 (62L and 62R). Signals S7 (S7L and S7R) indicating the pressure of the second port of the traction motor control valve 55 are input to the controller 31 from the pressure sensors 63 (63L and 63R).

As described above, the left hydraulic pump 51L is shared for driving the left traveling motor 20L, the swivel motor, and the arm cylinder 45, and the right hydraulic pump 51R is shared for driving the right traveling motor 20R, the boom cylinder 44, and the bucket cylinder 46. NS. That is, the hydraulic pump 51 is shared for driving the traveling motor 20 and the working device 13 (including the swivel motor). The operation state determination unit 71 indicates that the operation state of the work machine 1 is a state in which the traveling motor 20 and the working device 13 are driven at the same time (combined operation state), or a state in which the traveling motor 20 is driven independently. (Independent operation state) is determined. Therefore, first, the operation state determination unit 71 continuously inputs a pilot signal pressure having a magnitude of a certain value or more for a certain period of time or more with respect to the pilot signal pressure (signal S3) shown in FIGS. It is determined that the work machine 1 is running, that is, the running motor 20 is being driven during the period T1, T2, T3) of 5. Next, the operating state determination unit 71 compares the pressure value of the pilot signal pressure (signal S4) shown in FIG. 4 with a predetermined threshold value for each of the periods T1, T2, and T3, and determines the pilot signal pressure. When the pressure value is equal to or higher than the threshold value, it is determined to be in the combined operation state, and when the pressure value of the pilot signal pressure is less than the threshold value, it is determined to be in the independent operation state.

The travel path determination unit 72 determines whether the travel path of the work machine 1 is flat or sloped when the work machine 1 is traveling in a state where the travel motor 20 is operating independently. Assuming that the torque of the hydraulic pump 51 is T, the pressure is P, and the tilting capacity (discharge capacity per rotation of the pump) is q, the relationship of T = (P × q) / (2π) is established. When the work machine 1 is traveling on level ground at the maximum speed by inputting the maximum travel pilot signal pressure, the pressure P is smaller than the relief pressure, so the tilt capacity q is the maximum tilt capacity qmax. Become. Assuming that the slope of the traveling path gradually increases from that state, the pressure P gradually increases while maintaining the maximum tilting capacity qmax until the slope reaches a predetermined value, but the slope is relevant. When the value exceeds a predetermined value, a larger pressure P is required, so that the tilting capacity q decreases from the maximum tilting capacity qmax as the pressure P increases. That is, the traveling speed of the work machine 1 is reduced. In the travel path determination unit 72, when the work machine 1 is traveling in the independent operation state of the travel motor 20, the tilt capacity q maintains the maximum tilt capacity qmax (pressure P is a predetermined value P0). In the following cases), it is determined that the running path is flat, and when the tilting capacity q is lower than the maximum tilting capacity qmax (when the pressure P exceeds the predetermined value P0), the running path is sloped. Is determined. However, the traveling path determination unit 72 uses the detection value of the attitude sensor such as the acceleration sensor or the gyro sensor instead of using the tilting capacity q or the pressure P of the hydraulic pump 51, so that the traveling path of the work machine 1 is on a flat surface. It may be determined whether the land is a slope or a slope.

The counter 73 counts the number of clocks of a predetermined clock signal whose clock period is known. The controller 31 detects the drive start and drive stop of the traveling motor 20 by the pilot signal pressure represented by the signal S3, and is based on the count value of the counter 73 from the detection of the drive start to the detection of the drive stop. Then, the operating time of the traveling motor 20 is calculated.

The pump torque calculation unit 74 calculates the tilt capacity q of the hydraulic pump 51 from the pump current represented by the signal S2 by using a known relational expression between the pump current of the hydraulic pump and the tilt capacity. Further, the pump torque calculation unit 74 uses the relational expression T = (P × q) / (2π) [Nm] to reduce the oil pressure based on the pressure P (signal S1) of the hydraulic pump 51 and the tilt capacitance q. The torque T of the pump 51 is calculated.

The motor torque calculation unit 75 calculates the pressure of the traveling motor 20 as the difference between the pressure measured by the pressure sensor 62 (signal S6) and the pressure measured by the pressure sensor 63 (signal S7). Further, the motor torque calculation unit 75 uses the known relational expression of the torque, pressure and capacity of the traveling motor to obtain the pressure of the traveling motor 20 calculated above and the capacity of the traveling motor 20 represented by the signal S5. The torque of the traveling motor 20 is calculated based on the above.

The controller 31 travels including the pressure and torque of the hydraulic pump 51, the pressure, torque, and operating time of the traveling motor 20 detected as described above, and the pilot signal pressure of the traveling motor control valve 55. State quantity data S8 representing a predetermined state quantity relating to the hydraulic circuit during operation of the motor 20 is created. The state quantity data S8 may include all of these plurality of state quantities, or may include only a part thereof. Further, instead of the pilot signal pressure, other drive control signals for driving the control valve 55 for the traveling motor, such as a hydraulic or electric signal output from the operation lever or the operation pedal, may be included. Further, the state quantity data S8 further includes operating state information indicating whether the operating state of the work machine 1 at the time of detecting the state quantity is a combined operating state or a single operating state. Further, the state quantity data S8 further includes travel path information indicating whether the travel path of the work machine 1 at the time of detecting the state quantity is a flat ground or a slope. Further, the state quantity data S8 further includes detection time information indicating the detection time (including the date) of the state quantity and unique identification information of the work machine 1. Other information may be further included in the state quantity data S8. The controller 31 records the created state quantity data S8 in the storage unit 32. Further, the controller 31 transmits the state quantity data S8 read from the storage unit 32 from the communication unit 33 to the server device 2 via the communication network 4.

FIG. 6 is a flowchart showing a flow of processing executed by the controller 31 regarding the creation and recording of the state quantity data S8. When the power of the work machine 1 is turned on, the controller 31 starts executing this process.

First, in step SP101, the operating state determination unit 71 determines whether or not the work machine 1 is running, depending on whether or not a pilot signal pressure (signal S3) having a magnitude equal to or greater than a certain value is continuously input for a certain period of time or longer. That is, it is determined whether or not the traveling motor 20 is driven.

When the work machine 1 is not running (step SP101: NO), the operation state determination unit 71 repeatedly executes the process of step SP101.

When the work machine 1 is running (step SP101: YES), then in step SP102, the operation state determination unit 71 determines whether or not the pressure value of the pilot signal pressure (signal S4) is equal to or higher than a predetermined threshold value. It is determined whether or not the working device 13 (including the swivel motor) is driven.

When the work device 13 is not driven (step SP102: NO), that is, when the traveling motor 20 is in an independent operating state, then in step SP103, the traveling path determination unit 72 has a tilting capacity q of the hydraulic pump 51. Whether or not the traveling path of the work machine 1 is flat is determined by whether or not the maximum tilting capacity qmax is maintained or whether or not the pressure P of the hydraulic pump 51 is equal to or less than a predetermined value P0.

When the traveling path of the work machine 1 is on a flat ground (step SP103: YES), then in step SP104, the operation state determination unit 71 indicates a working state indicating an independent operating state and a flat ground traveling state as the operating state of the work machine 1. Set K1.

If the travel path of the work machine 1 is not on a flat ground (step SP103: NO), then in step SP108, the operation state determination unit 71 indicates the operation state of the work machine 1 as a single operation state and a slope travel state K2. To set.

When the work device 13 is being driven (step SP102: YES), then in step SP112, the operation state determination unit 71 indicates the combined operation state of the traveling motor 20 and the work device 13 as the operation state of the work machine 1. Set the working state K3.

In the example of the present embodiment, the operation state determination unit 71 classifies the operation state of the work machine 1 into three types of work states K1 to K3, but the operation state may be further classified.

Following step SP104, in step SP105, the controller 31 detects the pressures of the hydraulic pump 51 and the traveling motor 20 and the pilot signal pressure of the traveling motor control valve 55. Next, in step SP106, the pump torque calculation unit 74 and the motor torque calculation unit 75 calculate the torques of the hydraulic pump 51 and the traveling motor 20, respectively. Next, in step SP107, the controller 31 calculates the operating time of the traveling motor 20 in the working state K1. The controller 31 adds work state information (information including operating state information and travel path information) indicating the work state K1, detection time information, and unique identification information of the work machine 1 to these state quantities. The state quantity data S8 is created.

Following step SP108, in step SP109, the controller 31 detects the pressures of the hydraulic pump 51 and the traveling motor 20 and the pilot signal pressure of the traveling motor control valve 55. Next, in step SP110, the pump torque calculation unit 74 and the motor torque calculation unit 75 calculate the torques of the hydraulic pump 51 and the traveling motor 20, respectively. Next, in step SP111, the controller 31 calculates the operating time of the traveling motor 20 in the working state K2. The controller 31 creates the state quantity data S8 by adding the work status information indicating the work status K2, the detection time information, and the unique identification information of the work machine 1 to these state quantities.

Following step SP112, in step SP113, the controller 31 detects the pressures of the hydraulic pump 51 and the traveling motor 20 and the pilot signal pressure of the traveling motor control valve 55. Next, in step SP114, the pump torque calculation unit 74 and the motor torque calculation unit 75 calculate the torques of the hydraulic pump 51 and the traveling motor 20, respectively. Next, in step SP115, the controller 31 calculates the operating time of the traveling motor 20 in the working state K3. The controller 31 creates the state quantity data S8 by adding the work status information indicating the work status K3, the detection time information, and the unique identification information of the work machine 1 to these state quantities.

Following steps SP107, SP111, and SP115, in step SP116, the controller 31 records the created state quantity data S8 in the storage unit 32.

The controller 31 repeatedly executes the generation and recording process of the state quantity data S8 shown in FIG. 6 at each control cycle in which the controller 31 outputs a control signal or at regular time intervals. As a result, a plurality of state quantity data S8 are accumulated in the storage unit 32. The controller 31 records in the storage unit 32 the detected or calculated values of the pressure and torque of the hydraulic pump 51 and the traveling motor 20 and the pilot signal pressure of the traveling motor control valve 55 among the above-mentioned state quantities. Alternatively, the average value for each predetermined time (for example, 1 minute) may be calculated and the average value may be recorded in the storage unit 32.

FIG. 7 is a flowchart showing a flow of processing executed by the controller 31 with respect to the transmission of the state quantity data S8. When the power of the work machine 1 is turned on, the controller 31 starts executing this process.

First, in step SP201, the controller 31 determines whether or not a predetermined time has elapsed since the last transmission of the state quantity data S8. This predetermined time is a preset fixed transmission time interval, for example, one hour. However, the time interval is not limited to one hour, and may be any time interval such as half a day (12 hours) or one day (24 hours).

If the predetermined time has not elapsed (step SP201: NO), the controller 31 repeatedly executes the process of step SP201.

When the predetermined time has elapsed (step SP201: YES), the controller 31 then reads the untransmitted state quantity data S8 from the storage unit 32 in step SP202. The communication unit 33 transmits the state quantity data S8 read from the storage unit 32 to the server device 2 via the communication network 4.

FIG. 8 is a diagram showing the configuration of the server device 2. As shown in FIG. 8, the server device 2 includes a communication unit 81, a data analysis unit 82, a data storage unit 83, and a failure predictor 84. The failure predictor 84 predicts the failure time of the lower traveling body 11 of the work machine 1 by a failure prediction model showing the relationship between the state quantity and the failure time. As shown in FIG. 2, the server device 2 can communicate with a plurality of work machines 1A to 1C via the communication network 4. The server device 2 stores the state quantity data S8 received from the plurality of work machines 1A to 1C in the data storage unit 83. In addition, when a failure occurs in the lower traveling body 11 in any of the work machines 1, the unique identification information of the work machine 1, the date and time when the failure occurred, the name of the part in which the failure occurred, the content of the failure (symptom, degree, etc.) The failure information including the above is transmitted from the work machine 1 to the server device 2. However, the management device 3 may collect the failure information from the work machine 1, and the management device 3 may transmit the failure information to the server device 2. The server device 2 stores the received failure information in the data storage unit 83. The failure predictor 84 uses a plurality of state quantities stored in the data storage unit 83 as explanatory variables, the presence or absence of failure of each component of the lower traveling body 11 as an objective variable, and a plurality of states stored in the data storage unit 83. A failure prediction model is created by machine learning such as deep learning using a neural network using the failure information of. Here, as the explanatory variable, an average value or an integrated value of a plurality of state quantities for each work machine may be used. As the objective variable, information indicating whether or not a failure has occurred at an arbitrary predetermined cumulative operating time (cumulative operating days, cumulative operating months, etc.) is used. As the teacher data, information such as the cumulative operating time until the failure, which is extracted based on the collected failure information, can be used, and further, the above-mentioned predetermined accumulation also for the work machine in which the failure has not occurred. Data on a work machine that is the same as or close to the operating time may be used as the teacher data. The failure predictor 84 sequentially updates the failure prediction model every time the server device 2 receives new state quantity data S8 and new failure information even after the failure prediction model has been created.

FIG. 9 is a flowchart showing a flow of processing executed by the server device 2 regarding the failure prediction of the lower traveling body 11.

First, in step SP301, the communication unit 81 receives the state quantity data S8 from the work machine 1.

Next, in step SP302, the data analysis unit 82 confirms whether or not the operating time of the traveling motor 20 included in the received state quantity data S8 is "0", so that the traveling data is added to the state quantity data S8. Is included or not.

If the travel data is not included (step SP302: NO), the server device 2 ends the process.

When the traveling data is included (step SP302: YES), the data analysis unit 82 then confirms the unique identification information of the work machine 1 included in the received state quantity data S8 in step SP303. , It is determined whether the work machine 1 that is the source of the state quantity data S8 is a machine that has already been registered in the server device 2 or a machine that has not been registered.

If the work machine 1 is a registered machine (step SP303: YES), then in step SP304, the data analysis unit 82 is assigned to the work machine 1 in the total storage space of the data storage unit 83. The state amount data S8 received in step SP301 is stored in the storage area.

If the work machine 1 is an unregistered machine (step SP303: NO), then in step SP305, the data analysis unit 82 allocates a storage area related to the work machine 1 in the entire storage space of the data storage unit 83. .. Next, in step SP306, the data analysis unit 82 stores the state quantity data S8 received in step SP301 in the storage area allocated in step SP305.

Following steps SP304 and SP306, in step SP307, the server device 2 reads the state quantity data S8 stored in the data storage unit 83 in steps SP304 and SP306 from the data storage unit 83, and reads the state quantity data S8 from the failure predictor 84. Enter in. The failure predictor 84 predicts the failure time of each component of the lower traveling body 11 of the work machine 1 by the failure prediction model based on the state quantity data S8. The failure predictor 84 outputs a failure flag of "1" when there is a part whose predicted failure time is close (for example, within one month), and "0" when there is no part whose predicted failure time is close. "Failure flag is output.

When the failure predictor 84 outputs the failure flag of "1" (step SP307: YES), then in step SP308, the communication unit 81 uses the unique identification information of the work machine 1 and the lower traveling body 11 in which a failure is predicted. A failure alarm including the part name and the predicted failure time is transmitted to the management device 3 via the communication network 4. The management device 3 notifies the administrator of the failure prediction information by displaying the received failure alarm on the display device 5.

When the failure predictor 84 outputs the failure flag of "0" (step SP307: NO), the server device 2 ends the process.

According to the failure prediction system according to the present embodiment, the pressure sensors 61 to 63, the pump torque calculation unit 74, and the motor torque calculation unit 75 (detection means) are in a predetermined state regarding the hydraulic circuit during operation of the traveling motor 20. The amount is detected in the upper swing body 12. The failure predictor 84 (prediction means) predicts the failure of the lower traveling body 11 by a predetermined failure prediction model based on the state quantity detected by the detection means. Therefore, since it is not necessary to dispose a sensor such as a strain gauge in the lower traveling body 11 to which a large stress is applied during traveling, the sensor itself does not fail or the wiring is not broken due to the stress. As a result, it becomes possible to improve the reliability of the failure prediction system. Moreover, since the existing pressure sensor 61 and the like in the upper swing body 12 can also be used as the detection means, the failure prediction system can be realized at low cost.

Further, as the state quantity of the hydraulic circuit, the pressure or torque of the hydraulic pump 51, the drive control signal (pilot signal pressure, etc.) for driving the control valve 55 for the traveling motor, the pressure, torque, or the operating time of the traveling motor 20. By using at least one of them, it is possible to appropriately execute the failure prediction model creation process and the failure prediction process using the failure prediction model.

Further, the state quantity is used as an explanatory variable, the presence or absence of a failure of the lower traveling body 11 is used as an objective variable, and the failure information of the lower traveling body 11 collected from a plurality of working machines 1A to 1C is used as teacher data for machine learning to cause a failure. It becomes possible to appropriately create a prediction model.

Further, by distinguishing the state quantity according to whether the operating state of the work machine 1 is a combined operating state or a single operating state, the accuracy of the failure prediction model creation process and the failure prediction process using the failure prediction model can be improved. It becomes possible to improve.

Further, by distinguishing the state quantity according to whether the traveling path of the work machine 1 is flat or sloped at the time of detecting the state quantity, a failure prediction model creation process and a failure prediction process using the failure prediction model can be performed. It is possible to improve the accuracy.

Further, by including the information indicating the name of the faulty part in the lower traveling body 11 in the fault information, it is possible to create a fault prediction model capable of predicting the faulty part of the lower traveling body 11. As a result, maintenance can be performed on a component-by-part basis.

Further, since the server device 2 has the failure predictor 84, when a large number of work machines 1 exist, the information is aggregated in the server device 2, so that the accuracy of the failure prediction model can be improved and each work machine 1 can be improved. Since it is possible to omit the installation of a high-performance information processing device in the system, it is possible to reduce the system introduction cost.

<Modification example>
In the above embodiment, the server device 2 outside the work machine 1 has the failure predictor 84, but the work machine 1 itself may have the failure predictor 84. The failure predictor 84 predicts the failure time of each component of the lower traveling body 11 by the failure prediction model based on the state quantity data S8. However, instead of the failure prediction model, failure prediction may be performed by a simple process of comparing the cumulative operating time of each component with a predetermined allowable upper limit value. The failure predictor 84 displays a failure alarm on the display installed in the cockpit of the upper swing body 12 when there is a component whose predicted failure time is near.

According to this modification, since the work machine 1 itself has the failure predictor 84, the external server device 2 and the like are not required, and the system configuration can be simplified.

1 Work machine 2 Server device 11 Lower traveling body 12 Upper rotating body 13 Working device 20 Traveling motor 50 Hydraulic circuit 51 Hydraulic pump 55 Control valve for traveling motor 61-63 Pressure sensor 84 Failure predictor

Claims (9)

Targeting a work machine having an upper swing body, a lower traveling body, and a hydraulic circuit for driving a traveling motor in the lower traveling body by hydraulic oil supplied from a hydraulic pump in the upper rotating body via a control valve. It is a failure prediction system that
A detection means for detecting a predetermined state quantity of the hydraulic circuit during operation of the traveling motor in the upper swing body, and
Based on the state amount detected by the detection means, a prediction means for predicting a failure of the lower traveling body by a predetermined failure prediction model representing the relationship between the state amount and the failure.
An output means for outputting the prediction result by the prediction means and
A failure prediction system. The state quantity includes at least one of the pressure or torque of the hydraulic pump, the drive control signal for driving the control valve, the pressure, torque, or the operating time of the traveling motor, according to claim 1. Described failure prediction system. The claim further comprises an information processing device that creates the failure prediction model by machine learning by using the state quantity as an explanatory variable and the presence or absence of the failure as an objective variable and using the collected failure information of the lower traveling body as teacher data. The failure prediction system according to 1 or 2. The work machine further comprises a work device connected to the upper swing body.
The state amount is an operation indicating whether the operating state of the work machine at the time of detecting the state amount is a combined operation state of the traveling motor and the working device or an independent operating state of the traveling motor. The failure prediction system according to claim 3, to which state information is added. The failure prediction system according to claim 3 or 4, wherein travel path information indicating whether the travel path of the work machine is flat or sloped at the time of detecting the state quantity is added to the state quantity. The failure information includes information indicating the name of the failed part in the lower traveling body.
The failure prediction system according to any one of claims 3 to 5, wherein the prediction means further predicts a failure component of the lower traveling body. Further equipped with a server device for wireless communication with the work machine,
The failure prediction system according to any one of claims 1 to 6, wherein the server device has the prediction means. The failure prediction system according to any one of claims 1 to 6, wherein the work machine has the prediction means. For a work machine having an upper swing body, a lower traveling body, and a hydraulic circuit for driving a traveling motor in the lower traveling body by hydraulic oil supplied from a hydraulic pump in the upper rotating body via a control valve. It is a failure prediction method executed by the failure prediction system.
A predetermined amount of state related to the hydraulic circuit during operation of the traveling motor is detected in the upper swing body, and the state amount is detected.
Based on the detected state quantity, the failure of the lower traveling body is predicted by a predetermined failure prediction model representing the relationship between the state amount and the failure.
Output the prediction result,
Failure prediction method.

Images