Anomaly detection device and anomaly detection procedure

Abstract

Anomaly detection device, comprising: a model acceleration calculation unit (1) to calculate a model acceleration which is a predicted value of an engine acceleration; a motor acceleration calculation unit (3) to calculate the motor acceleration from a set of position and speed information of a motor (6); and an anomaly detection unit (4) to determine, based on the result of a comparison between the engine acceleration and the model acceleration, whether a power transmission mechanism (7) is abnormal, wherein the model acceleration calculation unit (1) includes a control system simulation unit (9) to calculate the model acceleration according to a position command sent to a single-shaft control system (5), wherein the control system simulation unit (9) includes a control target simulation unit (18) which is a model of a control target viewed by each shaft, the anomaly detection device further comprising a model correction unit (2) to calculate a friction parameter detection value of the engine (6) as a value identified as a friction parameter by performing an identification process, wherein the control target simulation unit (18) modifies an estimated friction associated with each shaft direction according to the value identified as the friction parameter calculated by the model correction unit (2), wherein The model acceleration calculation unit (1) further comprises a displacement correction unit (19) to calculate a speed command value and an acceleration command value as a position parameter value according to the position command, and the control target simulation unit (18) modifies the position parameter value according to the value identified as the friction parameter, which was calculated by the model correction unit (2).

Description

Area

[0001] The present invention relates to an anomaly detection device and an anomaly detection method for a power transmission mechanism, such as a reduction gear, an acceleration gear and a ball screw, which is used in a drive unit of a mechanical system, such as a robot. General state of the art

[0002] According to the relevant state of the art, power transmission mechanisms for transferring motor power to corresponding joints are frequently used in articulated industrial robots. In an industrial robot that has been in use for a long time, an increase in the loads acting upon it during operation can impair a power transmission mechanism, leading to a failure. If a robot cannot deliver its original performance, or if an industrial robot suddenly stops and work cannot be resumed due to a persistent fault, the production line must be stopped for repairs. Therefore, there was a need for a technique to determine whether power transmission mechanisms were impaired.

[0003] A method described in patent literature 1 discloses a technique for determining whether a power transmission mechanism is impaired by comparing a vibration pattern caused by impairment of the power transmission mechanism with a vibration pattern when the power transmission mechanism is driven normally. Additionally, the method described in patent literature 1 determines, on a scale derived from the disturbance torque acting on the power transmission mechanism, whether the power transmission mechanism is impaired.

[0004] Furthermore, DE69614367T2 describes a method for estimating disturbances in servomotors that, for example, drive feed shafts of machine tools or robot arms. The aim is to precisely estimate external loads affecting the motor in order to prevent damage to the mechanical system. For this purpose, the difference between the estimated and actual acceleration or velocity is used to determine the disturbance load acting on the motor.

[0005] US Patent 2004 / 0036621A1 discloses a method and device for controlling servo drives, particularly for machine tools. It employs an adaptive control system that takes into account parameters such as inertia and friction to compensate for deviations between setpoint and actual values.

[0006] DE102010036500A1 describes a control device for electric motors used in machine tools or industrial machines with the simultaneous determination of inertia, viscous friction and Coulomb friction using an inverse transfer function model instead of a Fourier transform.

[0007] The US 2012 / 0022690A1 combines the estimation of mechanical parameters such as inertia and friction with adaptive control that corrects deviations between the model and reality. This is intended to increase processing accuracy without requiring complex or memory-intensive calculation methods. List of prior art literature

[0008] Patent literature 1: JP 2008 - 32 477 A Brief description of the technical problem

[0009] According to the aforementioned techniques of the relevant prior art, it is possible to determine, based on the vibration pattern or the derived scale, whether a power transmission mechanism is impaired. However, the techniques of the relevant prior art are disadvantageous in that the influence of frictional torque, which changes depending on the temperature of the power transmission mechanism, cannot be eliminated.

[0010] The present invention was developed in light of the foregoing and one of its objectives is to provide an anomaly detection device which, using engine acceleration, which is less frequently affected by a change in friction, determines with high accuracy whether a power transmission mechanism is abnormal. Solution to the problem

[0011] The problem described above is solved by an anomaly detection device according to claim 1, and an anomaly detection method according to claim 8. Advantageous effects of the invention

[0012] The present invention produces an advantageous effect by determining with high accuracy, using engine acceleration, which is less frequently affected by a change in friction (a change in property), whether a power transmission mechanism is abnormal. Brief description of the drawings Fig. Figure 1 is a block diagram of an anomaly detection device of a first embodiment. Fig. Figure 2 is a block diagram illustrating the interior of a model acceleration calculation unit according to the first embodiment. Fig. Figure 3 is a block diagram illustrating the interior of a control system simulation unit according to the first embodiment. Fig. Figure 4 is a block diagram illustrating the interior of a feedback control unit according to the first embodiment. Fig. Figure 5 is a diagram illustrating the exemplary implementation of a CPU for control in an anomaly detection device according to the first embodiment. Fig. Figure 6 is a configuration diagram in a case where the definition of the range of the anomaly detection device is changed in the first embodiment. Fig. Figure 7 is a configuration diagram in a case where the definition of the range of the anomaly detection device is changed in the first embodiment. Fig. Figure 8 is a block diagram illustrating the interior of a control system simulation unit according to a third embodiment. Fig. Figure 9 is a block diagram illustrating the interior of a control system simulation unit according to a fourth embodiment. Fig. Figure 10 is a block diagram illustrating the interior of a model acceleration calculation unit according to a fifth embodiment. Fig. Figure 11 is a block diagram illustrating the interior of a model acceleration calculation unit according to a seventh embodiment. Fig. Figure 12 is a block diagram illustrating the interior of a model acceleration calculation unit according to an eighth embodiment. Fig. Figure 13 is a block diagram of an anomaly detection device according to a ninth embodiment. Fig. Figure 14 is a block diagram illustrating the interior of a control system simulation unit according to the ninth embodiment. Fig. Figure 15 is a block diagram illustrating the interior of a control system simulation unit according to the ninth embodiment. Fig. Figure 16 is a block diagram illustrating the interior of a control system simulation unit according to the ninth embodiment. Fig. Figure 17 is a diagram illustrating the exemplary implementation of the anomaly detection device according to the ninth embodiment. Fig. Figure 18 is a block diagram of an anomaly detection device according to a tenth embodiment. Fig. Figure 19 is a diagram illustrating the exemplary implementation of the anomaly detection device according to the tenth embodiment. Fig. Figure 20 is a flowchart of an anomaly detection procedure according to a fifteenth embodiment. Fig. Figure 21 is a flowchart of an anomaly detection procedure according to a sixteenth embodiment. Fig. Figure 22 is a block diagram illustrating a configuration of an anomaly detection device of a seventeenth embodiment. Fig. Figure 23 is a block diagram illustrating a configuration of an anomaly detection device of an eighteenth embodiment. Fig. Figure 24 is a block diagram illustrating a configuration of an anomaly detection device of a nineteenth embodiment. Fig. Figure 25 is a block diagram illustrating the interior of a control system simulation unit of the nineteenth embodiment. Fig. Figure 26 is a diagram illustrating the interior of a control system simulation unit according to a twentieth embodiment. Fig. Figure 27 is a hardware configuration diagram of the first to twentieth embodiments. Fig. Figure 28 is a hardware configuration diagram of the first to twentieth embodiments. Description of embodiments

[0013] Anomaly detection devices and anomaly detection methods according to embodiments of the present invention are described in detail below with reference to the drawings. It should be noted that the present invention is not limited to these embodiments. First embodiment.

[0014] Fig. Figure 1 is a block diagram illustrating an anomaly detection device according to a first embodiment of the present invention. The anomaly detection device according to the present embodiment comprises a model acceleration calculation unit 1, a motor acceleration calculation unit 3, and an anomaly determination unit 4. The model acceleration calculation unit 1 performs a process to calculate a predicted value in order to calculate a model acceleration of a motor 6 connected to a power transmission mechanism 7, which is included in a system as a model acceleration. The motor acceleration calculation unit 3 calculates motor acceleration from position information or rotational speed information of the motor 6.The anomaly detection unit 4 determines, based on the result of the comparison between the motor acceleration and the model acceleration, whether the power transmission mechanism 7 for a drive shaft is abnormal. It should be noted that, according to the present embodiment, the power transmission mechanism 7 functions as a reduction gear, an accelerator, or a geared motor.

[0015] The system is an industrial machine, such as an industrial robot (not illustrated), a processing machine, a forming machine, or a conveyor belt, and a drive mechanism of the system has one or more drive shafts. The present embodiment is an embodiment that can be applied to any system; however, a case is described below as an example in which an industrial robot having six drive shafts is to be controlled. In the case of an industrial robot having six drive shafts, six drive motors 6 and six power transmission mechanisms 7 are provided.

[0016] In the event that the anomaly detection device has the exemplary configuration shown in Fig. As illustrated in Figure 1, a position command for the motor 6, which drives a corresponding shaft, is generated in a control device included in the system. A single-shaft control system 5 controls the motor 6 according to the position command given by the system.

[0017] The model acceleration calculation unit 1 receives the position command and calculates a model acceleration, which is the model acceleration of the motor 6 according to the position command, by performing the procedure for calculating the predicted value. Fig. Figure 2 is a block diagram illustrating the interior of the model acceleration calculation unit 1 according to the first embodiment. In the model acceleration calculation unit 1, the position command is input into a control system simulation unit 9, and a time distinction is performed twice to output an estimated displacement of the control system simulation unit 9 in a process to calculate a predicted value, which is carried out by a distinction unit 10 to calculate the model acceleration.

[0018] Fig. Figure 3 is a block diagram illustrating the internal structure of the control system simulation unit 9 according to the first embodiment. The control system simulation unit 9 includes a feedforward control unit 11, a feedback control unit 12, and a simplified single-wave model 13. The feedforward control unit 11 consists of a single- or multi-stage filter unit, performs feedforward control for the position command input from an external device, and outputs a control variable that is controlled by the feedforward control. Fig. Figure 4 is a block diagram illustrating the internal structure of the feedback control unit 12 according to the first embodiment. The feedback control unit 12 comprises a proportional control unit 16, a proportional-integral control unit 17, and a differential processing unit 21 of a control system used for feedback control of the plant. The proportional control unit 16 performs position-proportional and speed-proportional-integral control on an input control variable, which is an output of the feedforward control.

[0019] The simplified single-shaft model 13 is an approximate model of a control target as viewed by each shaft in an inertial frame of reference for inertia and friction. The output of the estimated displacement from the simplified single-shaft model 13 in the control system simulation unit 9 is an estimated value of the engine displacement, taking into account the characteristics of the pre-feedback and post-feedback control. It should be noted that, in the present embodiment, the control system simulation unit 9 calculates each of the estimated acceleration and the estimated rotational speed in the model acceleration calculation unit 1. Specifically, the control system simulation unit 9 can calculate the estimated displacement by first calculating the estimated acceleration, then integrating the calculated acceleration over time to obtain the estimated rotational speed, and finally integrating the estimated rotational speed over time.

[0020] Alternatively, the control system simulation unit 9 can output the estimated rotational speed to the model acceleration calculation unit 1, perform the time differentiation on the output rotational speed once, and then output the calculated value as the model acceleration. The control system simulation unit 9 controls each of the six shafts of a six-shaft industrial robot. It should be noted that, while a model that also incorporates the properties of the control target is used as the control system simulation unit 9 in the present embodiment, the feedforward control unit 11 can also be used as a simple model of the control system simulation unit 9.

[0021] A friction model included in the simplified single-shaft model 13 calculates the frictional torque using a friction parameter calculated by a model correction unit 2. It can be assumed that the frictional force fi of the i-th shaft of the robot can be modeled by the following formula (1). fi=k1i*sgn(vi)+k2i*vi

[0022] In formula (1), vi represents the rotational speed of the i-th shaft, and sgn() is a function that returns 1 if the result is positive, -1 if the result is negative, and 0 if the result is 0. In the present embodiment, the frictional force of each shaft is expressed by a sum of Coulomb friction, which depends on the direction of motion, and viscous friction, which is proportional to the rotational speed. It should be noted that viscous friction is not necessarily proportional to the rotational speed increased to a power of 1, but it can be assumed to be proportional to the rotational speed increased to a power other than 1, e.g., proportional to the rotational speed increased to a power of 0.5. This document provides an exemplary description of the case where viscous friction is assumed to be proportional to the rotational speed increased to a power of 1.The model acceleration calculation unit 1 calculates the parameters k1i and k2i in formula (1) by performing a successive identification process.

[0023] In the present embodiment, the model acceleration calculation unit 1 uses the motor torque τm, calculated by multiplying the measured motor current value by a torque constant, or the motor torque τm detected by a torque sensor. It should be noted that the motor displacement pm is measured by a motor displacement measurement unit, such as an encoder. The model correction unit 2 calculates the time differential vm from the measured motor displacement pm and the time differential vm. The drive torque τ1, which is different from the frictional force, is then calculated from the motor displacement pm, the time differential vm, and the time differential am using the following formula (2). It should be noted that the single-shaft control system 5 in the anomaly detection device according to the present embodiment includes a motor current measurement unit.The motor current measuring unit can calculate the motor torque τm by multiplying the measured motor current by the torque constant. τl=M(pm) am+h(pm,vm)+g(pm)

[0024] The values ​​of yi, Ri and ri are calculated by the following formulas (3) to (5), and kpi[k] is calculated by the following formula (6): yi[k]=sgn(vmi[k],vmi[k] Ri[k]=Ri[k−1]+st*(−si*Ri[k−1]+yi[k]Tyi[k] ri[k]=ri[k−1]+st*(−si*ri[k−1]+(τmi[k]−τli[k])*yi[k]T) kpi[k]=kpi[k−1]−st*Gi(Ri[k]kpi[k−1]−ri[k]) where the i-th shaft components of the drive torque τl, the motor torque τm and the time differential vm are represented by τli, τmi and vmi respectively, and the value of a k-th identification period is represented by [k].

[0025] In this process, a first element of kpi[k] is an identified value k1i[k] of k1i in the k-th identification period, and a second element of kpi[k] is k2i[k], which is an identified value of k2i in the k-th identification period. Specifically, kpi, which is calculated by formula (6), is obtained as a vector with two rows and one column. An element in the first row and first column of kpi is the first element. An element in the second row and second column of kpi is the second element. Note that st represents a period of successive identification, and si and Gi represent preset increases. The model correction unit 2 outputs the values ​​k1i[k] and k2i[k], identified as friction parameters and representing identification results for each wave, as friction parameter identification results to a control target simulation unit 18 in the model acceleration calculation unit 1.The control target simulation unit 18 is a model of the control target as viewed from each shaft. It should be noted that an identified increase is a constant and consists of a matrix of constants. The frictional force is calculated in the control target simulation unit 18 based on formula (1).

[0026] In the first embodiment, the control target simulation unit 18 can function as the simplified single-shaft model 13. It should be noted that the control target simulation unit 18 can modify an estimated friction associated with each shaft direction of the industrial robot, according to the value identified as the friction parameter, which was calculated by the model correction unit 2.

[0027] Motor displacement, which indicates the position of motor 6, is measured by a displacement measurement unit, such as an encoder, or rotational speed information, which indicates the rotational speed of motor 6, is input to the motor acceleration calculation unit 3. The motor acceleration calculation unit 3 performs time differentiation on the motor displacement twice or on the rotational speed information once to calculate the motor acceleration. Note that the motor displacement indicates a motor angle. The motor acceleration calculation unit 3 subtracts the motor acceleration calculated by the motor acceleration calculation unit 3 from the model acceleration calculated by the model acceleration calculation unit 1. The result of this subtraction is input to the anomaly detection unit 4.While the subtraction is performed after model acceleration and the motor acceleration is calculated in the present embodiment, the model correction unit 2 can subtract the motor displacement from the estimated displacement within the model acceleration calculation unit 1 and perform time differentiation on the estimated displacement twice. Alternatively, the model correction unit 2 can perform time differentiation on the estimated displacement within the model acceleration calculation unit 1 once to output the model speed, subtract the motor speed from the output model speed, and then perform time differentiation on the calculated speed once. Alternatively, the two performances of time differentiation can be replaced by a process that uses a high-pass filter with appropriate properties.The high-pass filter has a property that allows only signal components with frequencies higher than the preset frequency to pass through. It should be noted that the anomaly detection unit 4 can determine whether the force transmission mechanism 7 is abnormal based on the signal components that have passed through the high-pass filter.

[0028] The anomaly detection unit 4 determines, based on the comparison between the engine acceleration and the model acceleration, whether the power transmission mechanism 7 for the drive shaft is abnormal. Specifically, the anomaly detection unit 4 calculates the maximum absolute value of the result of subtracting the engine acceleration from the model acceleration. If the calculated maximum absolute value is equal to or greater than a reference value, the power transmission mechanism 7 is determined to be abnormal. Fig. Figure 5 illustrates the exemplary implementation of a CPU for control in an anomaly detection device 101 according to the first embodiment. In the example shown in Fig. As illustrated in Figure 5, in the anomaly detection device 101, a target machine control 28 is implemented by a central processing unit (CPU). The CPU performs the corresponding functions of the model acceleration calculation unit 1, the model correction unit 2, the motor acceleration calculation unit 3, an instruction generation unit 29, and the anomaly detection unit 4 on a single-shaft controller 25 or a control target machine 27 as a control target. It should be noted that the CPU can be attached to an external personal computer (PC) (not illustrated) to control the single-shaft controller 25 or the control target machine 27.

[0029] The Fig. 6 and Fig. Figure 7 shows configuration diagrams for cases where the definition of the range of the anomaly detection device is changed in the first embodiment. While the anomaly detection device, which in Fig. As illustrated in Figure 5, the control target machine 27 includes the motor, the power transmission mechanism, a load unit 8, and the like. Since the information on the motor 6 is used to detect an anomaly, the anomaly detection device can be configured to include the model acceleration calculation unit 1, the model correction unit 2, the motor acceleration calculation unit 3, and the anomaly detection device 4, but not the single-shaft control system and the control target machine 27, as shown in Figure 5. Fig. Figure 6 illustrates this. Alternatively, the anomaly detection device can be implemented by an external computer, as shown in Figure 6. Fig. 7 illustrates.

[0030] It is known that as a reduction gear deteriorates, the vibrations it causes can become increasingly pronounced. When detecting an anomaly based on the vibration level of a reduction gear, it is assumed that the accuracy of the detection will be improved by using data from processes with a small torque component that is not a vibration. Additionally, using data from a section where the rotational speed is constant during the measurement process aids the analysis, as the vibrations caused by the reduction gear are related to the rotational speed.

[0031] In a case where the reduction gear is impaired, the properties may not only be such that the vibrations caused by the reduction gear are increased, but other changes in its characteristics may also occur. For example, if a shaft reduction gear is impaired and increasingly wears, the rigidity of the reduction gear decreases. In a ball screw mechanism to which a preload is applied, the preload is released and the rigidity of the ball screw mechanism decreases as the ball screw mechanism wears. When the rigidity of the reduction gear is reduced in this way, vibration can be excited during acceleration or deceleration, or vibration can be excited immediately after a stop.Therefore, to determine a decrease in rigidity due to current or torque, it is necessary to determine the current in a section where acceleration or deceleration is performed, or the current immediately after stopping. Using a high-pass filter for differences or similar factors is effective in removing vibrations due to the decrease in rigidity of a current waveform; however, it is necessary to eliminate the influence of acceleration / deceleration torque and frictional torque, which naturally occur during acceleration / deceleration. The influence of acceleration / deceleration torque can be eliminated by calculation if parameters such as the weight and center of gravity of the system, such as the robot that represents the control target, are known.However, the frictional torque cannot be eliminated beforehand, as the frictional torque varies depending on the temperature of the joint, even in the same machine, which leads to the problem that a change in the rigidity of a machine cannot be determined with great accuracy.

[0032] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Second embodiment.

[0033] A second embodiment differs from the first embodiment in the interior of the model correction unit 2 and the interior of the anomaly detection unit 4. In the first embodiment, the model correction unit 2 performs successive identification of the friction parameter to determine the friction parameter. In the present embodiment, a segment of a period during which the anomaly detection is to be performed, or a period specifying the period during which the anomaly detection is to be performed, is preset, and the model correction unit 2 performs a process of identifying a friction parameter by the least squares method of drive torque, engine displacement, engine speed, and engine acceleration within the segment or period during which the anomaly detection is performed.In the present embodiment, the model acceleration calculation unit 1 uses the motor torque τm, calculated by multiplying the measured motor current value by a torque constant, or the motor torque τm detected by a torque sensor. Note that motor displacement measured by a motor displacement measurement unit, such as an encoder, is represented by pm. The model correction unit 2 calculates a difference τdi between the drive torque τl and the motor torque τm of an i-th shaft using the following formula (7). Note that the drive torque τl is calculated by substituting the time differential vm of the motor displacement pm and the time differential am of the time differential vm into formula (2). τdi=Tmi−τli

[0034] In this case, the frictional force is identified using data from the first period to the nth period of the identification period, where a vector of n rows, in which the element in the mth row is τdi[m], represented by Yti, a matrix of n rows and two columns, in which the element in the mth row and first column is sgn(vmi[m]) and the element in the mth row and second column is vmi[m], represented by Ai, and a pseudo-inverse matrix of Ai by Ai + As shown, the first element of P, calculated by the following formula (8), is an estimate of the Coulomb friction coefficient k1i, and the second element of P is an estimate of the viscous friction coefficient k2i. Note that n represents the number of samples in each period. m represents each row of Yti and Ai. n and m are natural numbers that satisfy 1 ≤ m ≤ n. P=Ai+Yti

[0035] The identified or estimated friction parameter is a fixed value in each segment of the anomaly detection. The control target simulation unit 18 in the model acceleration calculation unit 1 calculates the frictional torque using the identified or estimated friction parameter. The anomaly detection unit 4 calculates the maximum absolute value of the result of subtracting the engine acceleration from the model acceleration in a segment during which the anomaly detection is performed. If the calculated maximum absolute value is equal to or greater than a reference value, the anomaly detection unit 4 determines that the power transmission mechanism 7 is abnormal.

[0036] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Third embodiment.

[0037] A third embodiment differs from the first embodiment in the interior of the control system simulation unit 9. Therefore, only a description of the interior of the control system simulation unit 9 will be given. Fig. Figure 8 is a block diagram illustrating the interior of the control system simulation unit 9 according to the third embodiment. In the first embodiment, the simplified single-shaft model 13 is used as the control target simulation unit 18 in the control system simulation unit 9. Fig. Figure 8 illustrates the control system simulation unit 9 according to the third embodiment. A model of a rigid body 14, which is in Fig. Figure 8 illustrates the acceleration a using the following formula (9), calculates the rotational speed v by time integration of the acceleration a, and calculates the displacement p by time integration of the rotational speed v. In formula (9), τ represents an output of the feedback control unit 12, and a, v, and p represent the acceleration, rotational speed, and displacement respectively, calculated by the rigid body model 14. a=M(p)−1(τ−h(p,v)−g(p)−f(v))

[0038] M(p) -1represents a reversal matrix of M(p). M(p) represents an inertia matrix. h(p,v) represent the centrifugal force and the Coriolis force, g(p) represents the gravitational force, and f(v) represents the frictional force. The control system simulation unit 9 outputs the calculated displacement p as the estimated displacement to a motion formula calculation unit. Additionally, a parameter identified by the model acceleration calculation unit 1 is used in the calculation of the frictional force f(v) of formula (9). The displacement p calculated by the rigid body model 14 is input as the estimated displacement into the feedback control unit 12.

[0039] It should be noted that, although the output of the control system simulation unit 9 is the estimated displacement, in the present embodiment the estimated acceleration and the estimated rotational speed are calculated in the control system simulation unit 9. Thus, the estimated acceleration in the control system simulation unit 9 can be used as an output of the model acceleration calculation unit 1. Alternatively, the control system simulation unit 9 can output the calculated estimated rotational speed to the model acceleration calculation unit 1, perform the time differentiation of the output estimated rotational speed once, and then output the calculated value as the model acceleration.

[0040] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Fourth embodiment.

[0041] A fourth embodiment differs from the third embodiment in the interior of the control system simulation unit 9. Therefore, only a description of the interior of the control system simulation unit 9 will be given. Fig. Figure 9 is a block diagram illustrating the interior of the control system simulation unit 9 according to the fourth embodiment. While the rigid body model 14 is used as the model of the control target simulation unit 18 in the third embodiment, a joint flexibility consideration model 15, as shown in Figure 9, is used in the fourth embodiment. Fig. Figure 9 illustrates this. A joint coupled by a linkage mechanism is provided between the motor 6 and the power transmission mechanism 7, and a model to be used takes into account the flexibility of the joint, which has the properties of a linear spring with a constant spring constant. The joint flexibility consideration model 15 is a model that drives the load unit 8 side using a value obtained by multiplying the difference between the displacement on the load unit side and the displacement on the motor side by the spring constant.

[0042] In the joint flexibility consideration model 15, am, representing a motor acceleration vector, vm, representing a motor speed vector, and pm, representing a motor displacement vector from each shaft, are used for the model to drive the side of the load unit 8. In the joint flexibility consideration model 15, al, representing a displacement vector, vl, representing a speed vector, and pl, representing a displacement vector on the side of the load unit 8, which is an output shaft of the power transmission mechanism 7, τm, representing a motor torque vector, and τ1, representing a power transmission mechanism 7 output torque vector, are used for the model to drive the side of the load unit 8.Additionally, in the joint flexibility consideration model 15 lm, which represents a diagonal matrix consisting of motor moments of inertia, f, which represents a vector consisting of the frictional torque, M, which represents an inertia matrix, h, which represents a vector consisting of the centrifugal force and the Coriolis force, g, which represents a vector consisting of the gravitational force, and Kb, which represents a diagonal matrix consisting of the spring constant of the joint of each shaft, are replaced in the following formulas (10) to (12), which enables the calculation of the motor acceleration am, the output torque τ1 of the power transmission mechanism 7, and the acceleration al on the side of the load unit 8. am=Im−1(xm−il−f(vm)) il=Kp(pl−pm) a1=M(pl)−1(il−h(pl,vl)−g(pl))

[0043] The motor speed vm is calculated by the time integration of the elements of the motor acceleration am.

[0044] The engine displacement pm is calculated by the time integration of the elements of the engine speed vm. The rotational speed vl on the side of load unit 8 is calculated by the time integration of the elements of the acceleration al on the side of load unit 8. The displacement pl on the side of load unit 8 is calculated by the time integration of the elements of the rotational speed vl on the side of load unit 8. In systems with n drive shafts, the engine acceleration am, the engine speed vm, the acceleration al, and the rotational speed vl can be expressed as a vector with n rows and one column. Specifically, a component in the first row and first column of elements is the first element. A component in the second row and first column is the second element. A component in the third row and first column is the third element. Thus, a component in the nth row and first column is the nth element.It should be noted that Im. -1 and M(pl) -1 The reversal matrices of Im and M(pl) are represented. While the output of the control system simulation unit 9 is the estimated engine displacement in the present embodiment, the estimated engine acceleration in the control system simulation unit 9 can be used as an output of the model acceleration calculation unit 1, since the estimated engine acceleration and the estimated engine speed are also calculated in the control system simulation unit 9. Alternatively, the control system simulation unit 9 can output the estimated engine speed to the model acceleration calculation unit 1, perform the time derivative of the output engine speed once, and then output the calculated value as the model acceleration.

[0045] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Fifth embodiment.

[0046] A fifth embodiment differs from the first embodiment in the internal structure of the model correction unit 2 and the internal structure of the model acceleration calculation unit 1. Therefore, only a description of the internal structure of the model correction unit 2 and the internal structure of the model acceleration calculation unit 1 will be provided. In the model correction unit 2, successive identification of the friction parameter is performed by the same unit as in the first embodiment. Furthermore, the control system simulation unit 9 in the model acceleration calculation unit 1 includes the feedforward control unit 11, the feedback control unit 12, the control target simulation unit 18, and a two-stage differentiation unit. The model correction unit 2 calculates the model acceleration using a friction parameter value on which successive identification is performed.Additionally, the model correction unit 2 also calculates the engine acceleration from the engine displacement, which is measured by an encoder. The model correction unit 2 identifies a parameter of a correction function such that ea, which represents the result of subtracting the model acceleration a from the engine acceleration am, becomes as small as possible.

[0047] In the fifth embodiment, it is assumed that two values, namely the speed command value vd, obtained by differentiating a position command, and an acceleration command value ad, obtained by further differentiating a position command, are input into the correction function, and the correction function can be expressed by a polynomial that includes the acceleration command value and the speed command value, which are the position parameter values. Specifically, the correction function can be expressed by the following formula (13) as a weighted sum of the acceleration command value raised to the power of 1 and the acceleration command value raised to the power of 2, and the speed command value raised to the power of 1 and the speed command value raised to the power of 2.The model correction unit 2 also recognizes the weight coefficients kali, ka2i, ka3i, and ka4i in the subsequent processes. Specifically, eai is calculated using the following formula (13). The suffix i indicates a value on an i-th wave. eai=ka1i*adi+ka2i*adi2+ka3i*vdi+ka4i*vdi2

[0048] In this case, yai, Rai, and rai are calculated using formulas (14) to (16), and kpai[k] is calculated using formula (17). Note that k represents a value in the k-th period of the identification period. yai[k]=[adi[k],adi[k]2,vdi[k],vdi[k]2] Rai[k]=Rai[k−1]+st*(−si*Rai[k−1]+yai[k]Tyai[k]) rai[k]=rai[k−1]+st*(−si*rai[k−1]+(ami[k]−ai[k])*yai[k]T) kpai[k]=kpai[k−1]−st*Gi(Rai[k]kpai[k−1]−rai[k])

[0049] Model Correction Unit 2 sends the calculated identified values ​​to Model Acceleration Computing Unit 1. The identified values ​​sent to Model Acceleration Computing Unit 1 are the first element of kpai[k], which is an identified value ka1i[k] of ka1i in the k-th identification period; the second element of kpai[k], which is an identified value ka2i[k] of ka2i in the k-th identification period; the third element of kpai[k], which is an identified value ka3i[k] of ka3i in the k-th identification period; and the fourth element of kpai[k], which is an identified value ka4i[k] of ka4i in the k-th identification period. Specifically, kpai, calculated by formula (17), is obtained as a vector with four rows and one column. An element in the first row and first column of kpai is the first element. An element in the second row and first column of kpai is the second element.An element in the third row and first column of kpai is the third element. An element in the fourth row and first column of kpai is the fourth element. Model Correction Unit 2 also sends the values ​​of kpai[k] to Model Acceleration Computing Unit 1.

[0050] Fig. Figure 10 is a block diagram illustrating the interior of the model acceleration calculation unit 1 according to the fifth embodiment. Since the processing in the feedforward control unit 11 and the feedback control unit 12 according to the fifth embodiment is the same as in the first embodiment, its description will not be repeated, and a description of the interior of a displacement correction unit 19 will be given. The displacement correction unit 19 receives a position command and performs a time differentiation of the position command to calculate a speed command and an acceleration command. Subsequently, the control target simulation unit 18 calculates an acceleration correction value eai based on formula (13) using a correction parameter kpai[k], which is input by the model correction unit 2.The control target simulation unit 18 integrates the calculated eai twice, adds the integration result as a displacement correction value to the estimated displacement, and outputs the sum. Alternatively, in the present embodiment, instead of performing the correction in the control system simulation unit 9, the control target simulation unit 18 can output the estimated displacement and the acceleration correction value eai, and output the result of adding the acceleration correction value eai to the estimated acceleration, which was calculated by performing the time differentiation on the estimated displacement twice, as the model acceleration.

[0051] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Sixth embodiment.

[0052] A sixth embodiment differs from the fifth embodiment in the internal structure of the model correction unit 2 and the anomaly detection unit 4, and therefore only a description of the model correction unit 2 and the anomaly detection unit 4 will be provided. In the sixth embodiment, similar to the second embodiment, a segment of a period during which the anomaly detection is to be performed, or a period defining the period during which the anomaly detection is to be performed, is preset, and the model correction unit 2 detects a friction parameter by the least-squares method of drive torque τmi, engine displacement pmi, engine speed vmi, and engine acceleration ami within the segment or period during which the anomaly detection is performed.The model correction unit 2 then identifies a parameter of an acceleration correction value by the least squares method, expressed by formula (13), of eai, which is the result of subtracting the model acceleration from the engine acceleration, the acceleration command adi, and the speed command vdi. The model correction unit 2 outputs the identified friction parameter and the identified acceleration correction value parameter as identical values ​​in the segment during which the anomaly determination is performed to the model acceleration calculation unit 1. The anomaly determination unit 4 calculates the maximum absolute value of the result of subtracting the engine acceleration from the model acceleration in a segment during which the anomaly determination is performed.If the calculated maximum value of the absolute values ​​is equal to or greater than a reference value, the anomaly detection unit 4 determines that the power transmission mechanism 7 is abnormal.

[0053] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Seventh embodiment.

[0054] A seventh embodiment differs from the fifth embodiment by the model correction unit 2 and the model acceleration calculation unit 1; therefore, only a description of the model correction unit 2 and the model acceleration calculation unit 1 will be given. While the acceleration command and the speed command are inputs for the acceleration correction value ea in the fifth embodiment, the time difference of an output from the feedforward control unit 11 and a further time difference therefrom are the inputs vd and ad, respectively, in the seventh embodiment. The model correction unit 2 performs successive identification of the output of the feedforward control unit 11 using formulas (14) to (17) in the same way as the fifth embodiment.

[0055] Fig. Figure 11 is a block diagram illustrating the interior of the model acceleration calculation unit 1 according to the seventh embodiment. The displacement correction unit 19 receives the output from the feedforward control unit 11, performs two time differentiations on an estimated displacement, which is an output from the control system simulation unit 9, and calculates an acceleration correction value ea based on formula (13) using the correction parameter kpai[k] input from the model correction unit 2. The control target simulation unit 18 integrates the calculated ea twice, adds the integration result as a displacement correction value to the estimated displacement, and outputs the sum.Alternatively, in the present embodiment, instead of performing the correction in the control system simulation unit 9, the control target simulation unit 18 can output the estimated displacement and the acceleration correction value ea and output the result of adding the acceleration correction value ea to the estimated acceleration, which was calculated by performing the temporal differentiation on the estimated displacement twice, as the model acceleration.

[0056] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Eighth embodiment.

[0057] An eighth embodiment differs from the fifth embodiment by the model correction unit 2 and the model acceleration calculation unit 1; therefore, only a description of the model correction unit 2 and the model acceleration calculation unit 1 will be given. While the acceleration command and the speed command are inputs for the acceleration correction value ea in the fifth embodiment, the time difference of an estimated engine displacement, which is an output of the control target simulation unit 18, and a further time difference thereof are the inputs vd and ad, respectively, in the eighth embodiment. The model correction unit 2 performs successive identification of the output of the control target simulation unit 18 using formulas (14) to (17) in the same way as the fifth embodiment.

[0058] Fig. Figure 12 is a block diagram illustrating the interior of the model acceleration calculation unit 1 according to the eighth embodiment. The displacement correction unit 19 receives the output from the control target simulation unit 18, performs two time differentiations, and calculates an acceleration correction value ea based on formula (13) using the correction parameter kpai[k] input from the model correction unit 2. The control target simulation unit 18 integrates the calculated ea twice, adds the integration result as a displacement correction value to the estimated displacement, and outputs the sum.Alternatively, in the present embodiment, instead of performing the correction in the control system simulation unit 9, the control target simulation unit 18 can output the estimated displacement and the acceleration correction value ea and output the result of adding the acceleration correction value ea to the estimated acceleration, which was calculated by performing the temporal differentiation on the estimated displacement twice, as the model acceleration.

[0059] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Ninth embodiment.

[0060] Fig. Figure 13 is a block diagram of an anomaly detection device according to a ninth embodiment. The ninth embodiment differs from the first embodiment in that the model correction unit 2 is not included. In the ninth embodiment, the control target simulation unit 18 in the model acceleration calculation unit 1 uses a predetermined fixed value as a friction parameter. Fig. Figure 14 is a block diagram illustrating the interior of the control system simulation unit 9 according to the present embodiment. As shown in Fig. As illustrated in Figure 14, the control system simulation unit 9 uses the simplified single-shaft model 13 as the control target simulation unit 18.

[0061] Additionally illustrated Fig. 15 a block diagram of the interior of the control system simulation unit 9 in a case where the model of a rigid body 14 is used as the control target simulation unit 18, and Fig. Figure 16 illustrates a block diagram of the interior of the control system simulation unit 9 in a case where the joint flexibility consideration model 15 is used as the control target simulation unit 18. The control system simulation unit 9 can be used as the feedforward control unit 11. Fig. Figure 17 illustrates the exemplary implementation of the anomaly detection device of the ninth embodiment. In the anomaly detection device according to the present embodiment, the model acceleration calculation unit 1, the motor acceleration calculation unit 3, the anomaly determination unit 4, and the single-shaft control system 5 are all implemented by a CPU of the single-shaft control 25, which controls the motor 6 of the control target.

[0062] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Tenth embodiment.

[0063] Fig. Figure 18 illustrates a block diagram of an anomaly detection device according to a tenth embodiment. In the tenth embodiment, the determination of whether an anomaly occurs is based on data from repeated identical operations. The anomaly detection device 101 according to the present embodiment includes a recording unit that records operational data, that is, operational information on equipment when operations are normal. The recording unit functions as a source data recording unit 34. Examples of operational data when operation is normal include engine torque and engine displacement. The source data recording unit 34 inputs recorded data into the model acceleration calculation unit 1.The model acceleration calculation unit 1 calculates a friction parameter from the operating data acquired as source data and also acquires the calculated friction parameter as a source friction parameter in the source data acquisition unit 34. The interior of the model correction unit 2 is similar to that in the second embodiment, and the friction parameters are identified in a batch from the operating data acquired in the source data acquisition unit 34 when operations are normal.

[0064] During anomaly detection, when the same operation as an operation corresponding to the operating data recorded in the original data acquisition unit 34 is to be carried out, the model correction unit 2 calculates a current friction parameter and outputs the calculated current friction parameter to the model acceleration calculation unit 1. In the model acceleration calculation unit 1, the time differentiation is performed on the engine displacement recorded in the original data acquisition unit 34 in order to calculate the engine acceleration of the operating data recorded in the original data acquisition unit 34 as time series data data1. After that, in the

[0065] Model acceleration calculation unit 1 inputs a position command and a friction parameter, acquired in the original data acquisition unit 34, into the control system simulation unit 9, and performs time differentiation at the output of the control system simulation unit 9. Using the same data acquired in the original data acquisition unit 34, model acceleration calculation unit 1 calculates the time series data data2 of the model acceleration, which includes the friction parameter calculated from the operating data at the time of the original data acquisition.Additionally, in the model acceleration calculation unit 1, the position command and the current friction parameter, which was output by the model correction unit 2, are entered into the control system simulation unit 9 and a time differentiation is performed on the output of the control system simulation unit 9 in order to calculate the time series data data3 of the model acceleration.

[0066] The model acceleration calculation unit 1 outputs the engine acceleration at the time of the original data acquisition, with data corrected for the influence of a change in friction as time series data, i.e., data1 - data2 + data3. The model acceleration calculation unit 1 inputs the result of subtracting the engine acceleration from the calculated model acceleration at the same time into the anomaly detection unit 4. The anomaly detection unit 4 calculates the maximum value of absolute values ​​or the average value of absolute values ​​of the input for the target operation and determines that an anomaly occurs if the maximum value or the average value is equal to or greater than a reference value. The exemplary implementation of the anomaly detection device 101 of the tenth embodiment is described in Fig. Figure 19 illustrates this. The model correction unit 2, the model acceleration calculation unit 1, the motor acceleration calculation unit 3, and the anomaly detection unit 4 are implemented by a CPU of a machine controller, which controls a control target, such as a robot. The original data acquisition unit 34 is implemented by a memory 32, which is included in the machine controller.

[0067] The anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal because the influence of an unmodeled factor of a control objective can be reduced. Eleventh version.

[0068] An anomaly detection device according to an eleventh embodiment has a configuration similar to that shown in the block diagram in Fig. Figure 1 illustrates this. Since the difference between the present embodiment and the first embodiment lies within the anomaly detection unit 4, only a description of the interior of the anomaly detection unit 4 will be given. In the eleventh embodiment, the average of the absolute values ​​of inputs is used for anomaly detection. The average can be calculated as an average value within a separately defined segment or as an average value over each predetermined period. The anomaly detection unit 4 determines that an anomaly occurs when the calculated average value is greater than a reference value.

[0069] The anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal because the influence of an unmodeled factor of a control objective can be reduced. Twelfth embodiment.

[0070] An anomaly detection device according to a twelfth embodiment has a configuration similar to that shown in the block diagram in Fig. Figure 1 illustrates this. Since the difference between the present embodiment and the first embodiment lies within the anomaly detection unit 4, only a description of the interior of the anomaly detection unit 4 will be given. In the twelfth embodiment, the anomaly detection unit 4 includes a dimensionless symptom parameter calculation unit that calculates a dimensionless symptom parameter. The dimensionless symptom parameter is used to determine whether the force transmission mechanism 7 is abnormal. Examples of the dimensionless symptom parameter include the form factor, impact index, crest factor, skew, and bulge. The anomaly detection unit 4 calculates at least one of these dimensionless symptom parameters in a separately defined segment or at each predetermined time interval.The anomaly detection unit 4 had a predetermined threshold and can determine whether the force transmission mechanism 7 is abnormal by comparing the calculated dimensionless symptom parameter with the threshold.

[0071] The shape factor, impact index, vertex factor, slant, and kurtosis are calculated using statistics, such as the mean, distribution, and standard deviation of n time series data points used in the tenth embodiment, and are used as dimensionless symptom parameters for anomaly determination. It should be noted that the statistics are calculated and obtained by the Model Acceleration Computing Unit 1. Specific procedures for calculating the shape factor, impact index, vertex factor, slant, and kurtosis using the calculated statistics, such as the mean, distribution, and standard deviation, are described below. The shape factor can be calculated by dividing the standard deviation calculated above by the mean of the absolute values.The impact index can be calculated by dividing a peak value by the standard deviation above. Note that the peak value is an average of absolute values ​​from 10 time series data points taken in descending order from the n time series data points. The crest factor can be calculated by dividing the peak value used above by the average of the absolute values. The slope is a value representing the degree of deformation in the positive and negative directions from the average value of the vibration waveform of vibration in the force transmission mechanism and can be calculated by the following formula (18), where the average of the absolute values ​​of the time series data points is X. i by X a is represented and the standard deviation is divided by X rmsis shown. Note that N represents the number of time series data points. The kurtosis is a value that represents how impulsive a waveform is and can be calculated using the following formula (19). [Formula 1] Oblique={∑i=1N(|Xi|−Xa)3} / (N−1) / Xrms3 [Formula 2] Curvature={∑i=1N(|Xi|−Xa)4} / (N−1) / Xrms4

[0072] The anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device according to the present embodiment can determine with high accuracy whether the force transmission mechanism 7 is abnormal because the influence of an unmodeled factor of a control objective can be reduced. Thirteenth embodiment.

[0073] An anomaly detection device according to a thirteenth embodiment has a configuration similar to that shown in the block diagram in Fig. Figure 1 illustrates this. Since the difference from the first embodiment lies within the anomaly detection unit 4, only a description of the interior of the anomaly detection unit 4 will be given. In the thirteenth embodiment, frequency analysis is used for anomaly detection. The anomaly detection unit 4 determines whether an anomaly occurs in the transition of the peak frequency in the frequency analysis result. Specifically, in a case where 16 Hz is set as a reference value, the anomaly detection unit 4 identifies the power transmission mechanism 7 as abnormal if the peak frequency, which is 20 Hz in a normal state immediately after the power transmission mechanism 7 starts operating, has progressively decreased to a value equal to or less than the reference value.

[0074] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device 101 according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal because it can more clearly detect a change in the rigidity of the power transmission mechanism 7. Fourteenth embodiment.

[0075] An anomaly detection device according to a fourteenth embodiment has a configuration similar to that shown in the block diagram in Fig. Figure 1 illustrates this. Since the difference from the first embodiment lies within the anomaly detection unit 4, only a description of the interior of the anomaly detection unit 4 will be given. In the fourteenth embodiment, an output from a bandpass filter is used for anomaly detection. The bandpass filter has a property that allows only signal components of a preset frequency to pass through. It should be noted that the anomaly detection unit 4 can determine that the power transmission mechanism 7 is abnormal if a signal component that has passed through the bandpass filter is detected.Specifically, in a case where the force transmission mechanism 7 is determined to be abnormal if the vibration frequency has become equal to or less than 16 Hz due to impairment of the force transmission mechanism 7, the anomaly detection unit 4 determines that the force transmission mechanism 7 is abnormal if the average value of absolute values ​​of the result of the vibration frequency of the force transmission mechanism 7, which has passed through the bandpass filter of 16 Hz, has become equal to or greater than the reference value.

[0076] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device 101 according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal because it can more clearly detect a change in the rigidity of the power transmission mechanism 7. Fifteenth embodiment.

[0077] Fig. Figure 20 illustrates a flowchart of an anomaly detection procedure of a fifteenth embodiment. The model acceleration calculation unit 1 begins a count, i.e., resets N to 1, and identifies a friction parameter (steps S1 and S2). A unit for identifying the friction parameters is similar to the model correction unit 2 in the first embodiment. The model acceleration calculation unit 1 calculates the model acceleration (step S3). The model acceleration is calculated by the same unit as the model acceleration calculation unit 1 in the first embodiment. The model acceleration calculation unit 1 subtracts the engine acceleration from the model acceleration (step S4). In the anomaly detection procedure according to the fifteenth embodiment, steps S2 to S4 are repeated unless a predetermined reference value has not been reached.The anomaly detection unit 4 determines whether the predetermined reference value has been reached and calculates the maximum absolute value of the result of subtracting the engine acceleration from the model acceleration when the predetermined reference value has been reached (steps S5 and S6). The anomaly detection unit 4 determines whether the maximum absolute value of the result of subtracting the engine acceleration from the model acceleration is equal to or greater than the reference value and issues an alarm, indicating that the power transmission mechanism 7 is abnormal if the maximum value is equal to or greater than the reference value (step S8).

[0078] The anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection device according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal even if the friction of the power transmission mechanism 7 changes depending on conditions such as temperature. Sixteenth embodiment.

[0079] Fig. Figure 21 illustrates a flowchart of an anomaly detection procedure of a sixteenth embodiment. The model acceleration calculation unit 1 begins a count, i.e., resets N to 1, and identifies a friction parameter (steps S1 and S2). A unit for identifying the friction parameters is similar to the model correction unit 2 in the first embodiment. The model acceleration calculation unit 1 calculates the model acceleration (step S3). The model acceleration is calculated by the same unit as the model acceleration calculation unit 1 in the first embodiment. The model acceleration calculation unit 1 subtracts the engine acceleration from the model acceleration (step S4). In the anomaly detection procedure, it is determined whether a predetermined reference value has been reached, and steps S2 to S4 are repeated unless the reference value has been reached.When the predetermined reference value is reached, the anomaly detection unit 4 calculates the curvature of the subtraction result (steps S5 and S6). The anomaly detection unit 4 determines whether the calculated curvature is equal to or greater than a reference value (step S8). If the calculated curvature is determined to be equal to or greater than the reference value, the anomaly detection unit 4 issues an alarm and determines that the power transmission mechanism 7 is abnormal.

[0080] The anomaly detection method according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal by reducing the influence of a change in friction, which represents a change in properties. Additionally, the anomaly detection method according to the present embodiment can determine with high accuracy whether the power transmission mechanism 7 is abnormal even if the friction of the power transmission mechanism 7 changes depending on conditions such as temperature. Seventeenth embodiment.

[0081] Fig. Figure 22 is a block diagram illustrating a configuration of an anomaly detection device of a seventeenth embodiment. Since components other than a program generation unit 35 and a threshold change unit 36 ​​are the same as those in the first embodiment, their description is not repeated. The program generation unit 35 generates programs to cause a control target, such as an industrial robot or a machine tool, to operate. The instruction generation unit 29 generates an immediate instruction to cause the control target to operate, i.e., a position instruction for each shaft based on operating instructions described in the programs generated by the program generation unit 35. The immediate instruction is fed into the single-shaft control system 5 and also into the threshold change unit 36.The threshold change unit 36 ​​determines a threshold value based on the entered immediate command.

[0082] For example, the threshold change unit 36 ​​stores two types of thresholds and calculates a speed command and an acceleration command for each shaft from a position command for the corresponding shaft. These thresholds are thresholds for anomaly detection. The threshold change unit 36 ​​selects a smaller value as a threshold to output if the speed command and the acceleration command are equal to or less than corresponding concrete values, or it selects and outputs a larger value if the condition that the speed command and the acceleration command are equal to or less than corresponding concrete values ​​is not met.

[0083] The threshold change can involve switching between three or more values ​​instead of just two. The changed thresholds can be continuous values. Additionally, the threshold can be switched based on whether the position is close to a specific position, rather than based on rotational speed and acceleration. The threshold change unit 36 ​​can change the threshold in the anomaly detection unit 4. Eighteenth embodiment.

[0084] Fig. Figure 23 is a block diagram illustrating a configuration of an anomaly detection device of an eighteenth embodiment. Since components other than the program generation unit 35 and the threshold change unit 36 ​​are the same as those in the first embodiment, their description is not repeated. The program generation unit 35 generates programs to cause a control target, such as an industrial robot or a machine tool, to operate. In a program generated by the program generation unit 35, the operating instructions for causing a control target, such as an industrial robot or a machine tool, to operate are described, and a threshold or sensitivity setting for anomaly detection in the anomaly detection unit 4 is described, which is used when operation is carried out according to the operating instructions.

[0085] The command generation unit 29 generates an immediate command to initiate operation of the control target, i.e., a position command for each shaft based on operating instructions described in the programs generated by the program generation unit 35. The immediate command is input into the single-shaft control system 5. The threshold change unit 36 ​​modifies the threshold to be used for anomaly detection by the anomaly detection unit 4, based on the threshold or sensitivity of the anomaly detection described in the program generated by the program generation unit 35. In cases where the threshold is described in the program, the threshold is switched to the specific value defined in the program.In a case where the sensitivity is set in the program, the threshold contained in the threshold change unit 36 ​​is changed depending on the set sensitivity, and the changed threshold is output to the anomaly detection unit 4. The threshold change unit 36 ​​can change the threshold in the anomaly detection unit 4. Nineteenth version.

[0086] Fig. Figure 24 is a block diagram illustrating a configuration of an anomaly detection device of a nineteenth embodiment. Fig. Figure 25 is a block diagram illustrating the interior of a control system simulation unit of the nineteenth embodiment. The anomaly detection device of the nineteenth embodiment does not include the model correction unit, unlike the first embodiment. The interior of the model acceleration calculation unit 1 is similar to that shown in Figure 25. Fig. 2 illustrated. The control target simulation unit 18 in Fig. 25 incorporates the simplified single-shaft model 13 in a similar manner to the first embodiment, but the friction parameter used by the simplified single-shaft model 13 is not an output of the model correction unit 2 but is calculated using a friction parameter previously stored in the simplified single-shaft model 13. In the present embodiment, although the influence of a change in friction on the model acceleration is not considered, it is less likely that the motor acceleration will be affected by a change in friction than the motor torque, since the motor is subject to feedback control for each shaft, which creates the effect that anomaly detection can be carried out with high accuracy even if a change in friction occurs. Twentieth embodiment.

[0087] Fig. Figure 26 is a diagram illustrating the interior of a control system simulation unit of a twentieth embodiment. The configuration of the anomaly detection device of the twentieth embodiment is similar to the anomaly detection device described in Figure 26. Fig. 24 is illustrated.

[0088] The difference from the nineteenth embodiment is that the interior of the control system simulation unit 9 is as in Fig. Figure 26 illustrates the design. In the present embodiment, although the influence of a change in friction is not taken into account in the model acceleration, it is less likely that the motor acceleration will be affected by a change in friction than the motor torque, since the motor is subject to feedback control for each shaft, which creates the effect that anomaly detection can be carried out with high accuracy even if a change in friction occurs.

[0089] The effects produced by the anomaly detection devices and anomaly detection methods according to the embodiments described above are described below. With the anomaly detection device or anomaly detection method according to the embodiments, anomaly detection is performed based on motor acceleration, in which the influence of a change in friction caused by a change in temperature of a joint coupling through a connection mechanism between the motor 6 and the power transmission mechanism 7 is reduced compared to the motor current; therefore, anomaly detection can be performed with high accuracy, even in a case where the influence of a change in friction caused by a change in temperature is present.In addition to the anomaly detection device or method according to the embodiments, a change in friction caused by a change in the joint's temperature is identified and corrected; therefore, anomaly detection can be performed with greater accuracy than in a case where the change in friction is not corrected, even if the influence of the change in friction caused by a change in temperature is present. The anomaly detection device or method according to the embodiments corrects the influence of a change in friction before comparison with the acceleration data during normal operation; therefore, anomaly detection can be performed with high accuracy without detailed modeling of a control target.

[0090] Each embodiment is achieved by implementing the target machine control 28 using a CPU for control. Alternatively, each embodiment can be achieved by installing software, which functions as the corresponding components of the anomaly detection device, on a PC located outside the target machine control 28. A hardware configuration of the external PC running the software is now described with reference to the Fig. 27 and Fig. 28 described. The external PC according to the embodiments can be implemented by the hardware 100, which is described in Fig. Figure 27 illustrates this, i.e., a processing circuit 102. Specifically, the anomaly detection device includes a processing device for calculating the model acceleration, calculating the motor acceleration from the position information or the rotational speed information of the motor 6, and determining, based on the result of comparing the calculated motor acceleration with the calculated model acceleration, whether the power transmission mechanism 7 for a drive shaft is abnormal. The processing circuit 102 can be dedicated hardware or a central processing unit (CPU; also known as a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, a processor, or a digital signal processor (DSP)) that executes programs stored in memory.

[0091] In a case where the processing circuit 102 is dedicated hardware, the processing circuit 102 is, for example, a single circuit, a compound circuit, a programmed processor, a parallel programmed processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination thereof. The model acceleration calculation unit 1, the motor acceleration calculation unit 3, and the anomaly detection unit 4 can each be implemented by the processing circuit 102, or the functions of the corresponding components can be implemented together by a single processing circuit.

[0092] In a case where the processing circuit 102 is a CPU, the anomaly detection device can be implemented by the hardware 100a, which is in Fig.Figure 28 illustrates this, namely a processor 103 and a memory 104 connected to the anomaly detection device. In this case, the model acceleration calculation unit 1, the motor acceleration calculation unit 3, and the anomaly detection unit 4 are implemented by software, firmware, or a combination of both. The software and firmware are described in the form of programs and stored in memory 104. The processor 103 implements the functions of the corresponding components by reading and executing the programs stored in memory 104.Specifically, the anomaly detection device includes a memory for storing programs which, when executed by the processing circuit 102, result in the execution of a step to calculate the model acceleration, a step to calculate the motor acceleration from the position information or the rotational speed information of the motor 6, and a step to determine whether the power transmission mechanism 7 is abnormal. In other words, these programs cause a computer to execute the processes and procedures of the model acceleration calculation unit 1, the motor acceleration calculation unit 3, and the anomaly detection unit 4.It should be noted that the memory can be, for example, a non-volatile or volatile semiconductor memory, such as random access memory (RAM), read-only memory (ROM), flash memory, erasable programmable read-only memory (EPROM) or electrically erasable programmable read-only memory (EEPROM), a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk or a digital versatile disc (DVD).

[0093] The configurations shown in the foregoing embodiments are examples of the present invention and can be combined with other known techniques or can be partially omitted or modified without departing from the scope of the present invention. Reference symbol list

[0094] 1 Model Acceleration Calculation Unit; 2 Model Correction Unit; 3 Motor Acceleration Calculation Unit; 4 Anomaly Detection Unit; 5 Single-Shaft Control System; 6 Motor; 7 Power Transmission Mechanism; 8 Load Unit; 9 Control System Simulation Unit; 10 Differentiation Unit; 11 Feedforward Control Unit; 12 Feedback Control Unit; 13 Simplified Single-Shaft Model; 14 Rigid Body Model; 15 Joint Flexibility Consideration Model; 16 Dependent Control Unit; 17 Dependent Integrated Control Unit; 18 Control Target Simulation Unit; 19 Displacement Correction Unit; 21 Difference Processing Unit; 25 Single-Shaft Control; 27 Control Target Machine; 28 Target Machine Control; 29 Instruction Generation Unit; 32 Memory in Control; 34 Origin Data Acquisition Unit; 35 Program Generation Unit; 36 Threshold Change Unit; 100, 100a Hardware; 102 Processing circuit; 103 Processor; 104 Memory.

Claims

[1] Anomaly detection device comprising: a model acceleration calculation unit (1) to calculate a model acceleration which is a predicted value of an engine acceleration; a motor acceleration calculation unit (3) to calculate the motor acceleration from a set of position and speed information of a motor (6); and an anomaly detection unit (4) to determine, based on the result of a comparison between the engine acceleration and the model acceleration, whether a power transmission mechanism (7) is abnormal, wherein the model acceleration calculation unit (1) includes a control system simulation unit (9) to calculate the model acceleration according to a position command sent to a single-shaft control system (5), wherein the control system simulation unit (9) includes a control target simulation unit (18) which is a model of a control target viewed by each shaft, the anomaly detection device further comprising a model correction unit (2) to calculate a friction parameter detection value of the engine (6) as a value identified as a friction parameter by performing an identification process, wherein the control target simulation unit (18) modifies an estimated friction associated with each shaft direction according to the value identified as the friction parameter calculated by the model correction unit (2), wherein The model acceleration calculation unit (1) further comprises a displacement correction unit (19) to calculate a speed command value and an acceleration command value as a position parameter value according to the position command, and the control target simulation unit (18) modifies the position parameter value according to the value identified as the friction parameter, which was calculated by the model correction unit (2). [2] Anomaly detection device according to claim 1, further comprising: a storage unit (34) to store operating information of the motor (6) when operation is normal; and the model correction unit (2) to calculate the operating information as the value identified as the original friction parameter by performing the identification process and to store the calculated value identified as the original friction parameter in the storage unit (34), wherein the model acceleration calculation unit (1) the model acceleration of the force transmission mechanism (7) by performing a calculation process of a predicted value according to the friction parameter identified as the original friction parameter Value calculated by the model correction unit (2) and the operating information stored in the memory unit (34). [3] Anomaly detection device according to one of claims 1 to 2, wherein the anomaly detection unit (4) includes a high-pass filter to allow a signal component with a frequency higher than a preset frequency to pass from signal components of the engine acceleration, wherein When a signal component passing through the high-pass filter is detected, the anomaly detection unit (4) determines whether the force transmission mechanism (7) is abnormal according to the detected signal component. [4] Anomaly detection device according to one of claims 1 to 2, wherein the anomaly detection unit (4) includes a frequency analysis unit to perform frequency analysis of vibrations in the power transmission mechanism (7), and The anomaly detection unit (4) determines, based on a transition from a peak frequency obtained by the frequency analysis unit, whether the power transmission mechanism (7) is abnormal. [5] Anomaly detection device according to one of claims 1 to 2, wherein the anomaly detection unit (4) includes a dimensionless symptom parameter calculation unit to calculate a dimensionless symptom parameter using statistics obtained from the model acceleration calculation unit (1), and the anomaly detection unit (4) determines, based on a result of the comparison between the dimensionless symptom parameter and a predetermined threshold, whether the force transmission mechanism (7) is abnormal. [6] Anomaly detection device according to claim 1, comprising: a threshold change unit (36) to change a threshold of the anomaly detection unit (4), wherein the threshold is used to determine whether the anomaly occurs, wherein the threshold to be output by the threshold change unit (36) is determined on the basis of an instruction generated by an instruction generation unit (29). [7] Anomaly detection device according to claim 1, comprising: a threshold change unit (36) to change a threshold in the anomaly detection unit (4), wherein the threshold is used to determine whether the anomaly occurs, wherein the threshold is set by a program generated by a program generation unit (35). [8] Anomaly detection procedures, including: a model acceleration calculation step of calculating a model acceleration, which is a predicted value of an engine acceleration; a motor acceleration calculation step of calculating the motor acceleration from one of position information and speed information of a motor (6); and an anomaly detection step of determining, based on a result of comparing the engine acceleration and the model acceleration, whether a power transmission mechanism (7) is abnormal, wherein the model acceleration calculation step includes a control system simulation step to calculate the model acceleration according to a position command sent to a single-shaft control system (5), wherein the control system simulation step includes a control target simulation step, which is a model of a control target viewed by each shaft, the anomaly detection procedure further comprising a model correction step to calculate a friction parameter detection value of the engine (6) as a value identified as a friction parameter by performing an identification process, wherein The control target simulation step modifies an estimated friction associated with each shaft direction, according to the value identified as the friction parameter, which was calculated by the model correction step, wherein The model acceleration calculation step further includes a displacement correction step to calculate a speed command value and an acceleration command value as a position parameter value according to the position command, and the control target simulation step modifies the position parameter value according to the value identified as the friction parameter, which was calculated by the model correction step.

Images