A method for compensating time delay estimation error in current tracking control of robot joint motors

Abstract

The present invention discloses a method for compensating for delay estimation error of current tracking control of robot joint motor. The method comprises: implementing a delay control algorithm on the robot joint motor, adjusting the gain coefficient to make the motor run normally and stably; establishing a delay estimation error estimation model, inputting the estimated value of the delay estimation error at the previous moment, the actual current vector and the reference value into the model for processing, and outputting the estimated value of the current delay estimation error; constructing and implementing an improved delay control algorithm, improving the error dynamics equation of the robot joint motor, and finally obtaining the control voltage and controlling the motor through SVPWM to realize the current tracking control delay estimation error compensation. The present invention can scientifically and reasonably estimate the delay estimation error, give its specific discrete iterative expression from the perspective of gradient descent, and reduce the current tracking error while compensating for the delay estimation error by adding a compensation term in the control law, thereby realizing high-precision current tracking of the robot joint motor.

Description

Technical Field

[0001] The invention relates to a motor delay estimation error compensation method, and in particular to a robot joint motor current tracking control delay estimation error compensation method. Background Art

[0002] Robots play an increasingly important role in many fields such as manufacturing, medical treatment, and services. The precision of their trajectory tracking and the compliance of their force control have attracted widespread attention. For the current mainstream robots, especially collaborative robots, their joint motors usually only run a single torque / current loop, so the high-precision force and position control of the robot depends on accurate motor current control. However, most of the current research in the field of robotics focuses on upper-level motion control, and less attention is paid to the control of permanent magnet synchronous motors as core drive components under robot operating conditions. In addition, the change frequency of the torque / current reference value of the joint motor above 100Hz, coupled with the variable speed condition updated in real time, brings fast-changing uncertain disturbance terms to the joint motor control system, which will directly affect the accuracy of the time delay estimation TDE (Time Delay Estimation), thereby reducing the current tracking control performance of the time delay control algorithm TDC (Time Delay Control). Therefore, compensating for the delay estimation error is of great significance to improving the current tracking control accuracy under fast-changing conditions and improving the robot motion control performance.

[0003] At present, there are two improvement ideas and solutions for delay error compensation, namely variable gain coefficient and corrected uncertainty. Among them, the variable gain coefficient method constructs a function of the gain coefficient with respect to the tracking error to form an adaptive law or a sliding mode reaching law, so as to realize the automatic adjustment of the gain coefficient as the tracking error changes. In the motor current tracking control, this method dynamically adjusts the gain coefficient according to the actual current tracking error, effectively compensating for the problem of amplified delay estimation error under small gain coefficients and the problem of amplified noise interference under large gain coefficients, thereby improving the current tracking accuracy. In essence, the variable gain coefficient is an indirect compensation method for the delay estimation error. However, this method relies on the selection of adaptive law and reaching law. Suitable adaptive law and reaching law are usually difficult to find and there are few guiding methods. In addition, in order to improve the control performance, computationally complex terms such as sign function or even exponential function are generally added to the adaptive law and reaching law, which will increase the performance requirements for the digital controller and reduce the control efficiency of the controller.

[0004] The modified uncertainty rule reduces the error of the delay estimation by modifying the uncertainty of the delay estimation. When modifying the delay estimate, a prediction method is often used. The uncertainty of the delay estimation at the current moment is predicted and modified by combining the difference of the uncertainty of the previous two moments with the correction step length, and this part of the correction value is added to the control law to reduce the error of the estimated delay estimation. This method directly compensates for the delay error and usually corresponds to a certain control compensation structure, such as internal model control, which is more intuitive to understand and apply. However, the current research focus of this method is on the adaptive selection of the correction step length, while less attention is paid to the explanation and understanding of the scientificity and rationality of the correction term itself, which will lead to low analytical and interpretative properties of the phenomena and results in actual control, and poor extensibility and extension of the method. Summary of the invention

[0005] In order to solve the problems existing in the background technology, the present invention provides a method for compensating delay estimation errors in current tracking control of a robot joint motor.

[0006] The technical solution adopted by the present invention is:

[0007] The robot joint motor current tracking control delay estimation error compensation method of the present invention comprises:

[0008] Step 1) Implement a traditional time delay control algorithm on the robot joint motor, and adjust the gain coefficient of the time delay control algorithm to make the robot joint motor run normally and stably.

[0009] Step 2) Establish a delay estimation error estimation model for the normal and stable operation of the robot joint motor, input the estimated value of the current tracking control delay estimation error of the robot joint motor at the previous moment, the actual current vector and the reference current vector at the previous two moments into the delay estimation error estimation model, and after processing, the delay estimation error estimation model outputs the estimated value of the current tracking control delay estimation error of the robot joint motor at the current moment.

[0010] Step 3) An improved time delay control algorithm is constructed according to the estimated value of the current tracking control delay estimation error of the robot joint motor at the current moment, and the improved time delay control algorithm is implemented on the robot joint motor, thereby improving the error dynamics equation of the robot joint motor under the action of delay control, and finally obtaining the control voltage of the robot joint motor.

[0011] Step 4) The three-phase drive voltage of the robot joint motor is obtained according to the control voltage of the robot joint motor, so as to control the robot joint motor to perform current tracking control delay estimation error compensation of the robot joint motor and realize high-precision current tracking of the robot joint motor.

[0012] In the step 1), the robot joint motor is first controlled by a traditional time-delay control algorithm under a preset initial gain coefficient, and the preset initial gain coefficient is a given smaller gain coefficient close to 0 to ensure that the system can be stably controlled; then the gain coefficient is gradually increased so that the current tracking accuracy of the system is gradually improved; until the current tracking error of the robot joint motor no longer decreases, it is determined that the traditional time-delay control algorithm under the gain coefficient at this time is used to finally control the robot joint motor, so that the robot joint motor can operate normally and stably.

[0013] In the step 2), the delay estimation error estimation model when the robot joint motor is running normally and stably is as follows:

[0014] ε ^( kT s )= ε ^(( k -1) T s )+ K c [ I ref (( k -1) T s )- I (( k -1) T s )]- K c ( I 2×2 - T s K p )[ I ref (( k -2) T s )- I (( k -2) T s )]

[0015] in, ε ^( ) represents the delay estimation error of the current tracking control of the robot joint motor ε ( ) k and T s Respectively represent the moment and control period in discrete time domain; K c and K pdenote the diagonal constant coefficient matrix and the positive definite diagonal constant value matrix respectively, K c To correct the step size, K c =diag( K c1 , K c2 ), K c1 and K c2 Denote the diagonal constant coefficient matrices K c The d-axis and q-axis correction step lengths in , K p =diag( K p1 , K p2 ), K p1 and K p2 denote positive definite diagonal constant matrices K p The d-axis and q-axis current tracking error convergence parameters are used to adjust the d-axis and q-axis current tracking error convergence speed; I ( )and I ref ( ) represent the actual current vector and reference current vector of the robot joint motor respectively; I 2×2 represents the 2nd-order identity matrix.

[0016] ε ^(( k -1) T s )and ε ^( kT s ) are respectively the previous ( k -1) T s Moment and Present kT s The estimated value of the time delay estimation error of the current tracking control of the robot joint motor at time.

[0017] In the step 3), the improved delay control algorithm is as follows:

[0018] V ( kT s )= A 2 u (( k- 1) Ts )+ N 2^( kT s )+ A 2 ε ^( kT s )

[0019] in, V ( ) represents the stator voltage vector to be applied to the robot joint motor; k and T s Respectively represent the moment and control period in discrete time domain; A 2 represents the gain coefficient, which is a diagonal constant coefficient matrix, A 2=diag( A 21 , A 22 ), A 21 and A 22 Denote the diagonal constant coefficient matrices A The gain coefficients of the d-axis and q-axis in 2; u ( ) represents the control component of the control law of the traditional time-delay control algorithm, which is usually constructed in the form of error dynamics equation; N 2^( ) represents the lumped uncertainty of the robot joint motor N 2( )’s estimated delay; ε ^( ) represents the delay estimation error of the current tracking control of the robot joint motor ε ( ) is an estimated value.

[0020] The control voltage of the robot joint motor is the stator voltage vector to be applied to the robot joint motor V ( ). Under the control law of the improved time-delay control algorithm, the right side of the error dynamics equation is made as close to 0 as possible. According to the properties of the first-order differential equation, the current tracking error will also tend to 0.

[0021] The control component of the conventional delay control algorithm u ( )as follows:

[0022] u (( k- 1) T s )= I ´(kT s )+ ε ( kT s )

[0023] ε ( kT s )= A 2 -1 [ N 2( kT s )- N 2^( kT s )]

[0024] in, I ´( ) represents the actual current vector of the robot joint motor I ( ) is the first-order differential value of .

[0025] The lumped uncertainty of the robot joint motor is N 2( )as follows:

[0026] N 2( kT s )=[ N 2d ( kT s ), N 2q ( kT s )] T

[0027] N 2d ( kT s )=( L d - A 21 ) i d ´+ Ri d - w e L d i d

[0028] N 2q ( kT s )=( Lq - A 22 ) i q ´+ Ri q - w e ( L d i d + ψ f )

[0029] in, N 2d ( )and N 2q ( ) represent the uncertainties of the d-axis and q-axis of the robot joint motors respectively; L d and L q They represent the d-axis inductance and q-axis inductance of the robot joint motor respectively; i d and i d ´represent the d-axis current of the robot joint motor and its first-order differential value, i q and i q ´represent the q-axis current and its first-order differential value of the robot joint motor respectively; R , w e and ψ f They represent the resistance, electrical angular velocity and stator flux of the robot joint motor respectively.

[0030] In the step 3), the error dynamics equation of the improved robot joint motor under the time delay control is as follows:

[0031] e ´( kT s )+ K p e (( k -1) T s )= ε ( kT s )+△( kT s )- ε ^( kT s ):=ε ~( kT s )

[0032] △( kT s )= I ref ´( kT s )- I ref ´(( k -1) T s )

[0033] e ( kT s )= I ref ( kT s )- I ( kT s )

[0034] in, e ( )and e ´( )represent the current tracking error and its first-order differential value of the robot joint motor respectively; k and T s Respectively represent the moment and control period in discrete time domain; K p represents a positive definite diagonal constant matrix; ε ( )and ε ^( ) represent the time delay estimation error and its estimated value of the current tracking control of the robot joint motor respectively; △( ) represents the difference in the rate of change of the reference current at adjacent moments; ε ~( ) represents the time delay estimation error of the current tracking control of the robot joint motor after compensation; I ( ) represents the actual current vector of the robot joint motor, I ref ( )and I ref ´( ) represent the reference current vector and its first-order differential value of the robot joint motor respectively.

[0035] In the step 4), the control voltage of the robot joint motor includes the d-axis voltage and the q-axis voltage to be applied to the robot joint motor. The d-axis voltage and the q-axis voltage to be applied to the robot joint motor are converted into the α-axis voltage and the β-axis voltage after dq-αβ coordinate transformation, and then the switching devices of the inverter of the robot joint motor are controlled to be turned on and off after space vector pulse width modulation SVPWM (Space Vector PulseWidth Modulation), so as to obtain the three-phase drive voltage of the robot joint motor, and finally control the operation of the robot joint motor.

[0036] The electronic device of the present invention comprises: a memory and a processor coupled to each other, wherein the memory stores program data, and the processor calls the program data to execute the method as described above.

[0037] The computer-readable storage medium of the present invention stores program data thereon, and when the program data is executed by a processor, the method described above is implemented.

[0038] The beneficial effects of the present invention are:

[0039] 1) Based on the analysis of the influence of traditional time delay estimation error on current tracking accuracy, the present invention estimates and compensates the time delay estimation error, and gives a discrete iterative expression of the error estimation term. By adding the compensation term to the control law of the new time delay control algorithm, a more accurate current tracking effect is obtained.

[0040] 2) This invention explains the scientific nature of the discrete iterative expression corresponding to the given error estimation term from the perspective of the gradient descent algorithm, as well as the rationality of the control law of the new time-delay control algorithm with compensation terms, providing theoretical guarantee for the application of this algorithm in actual robot joint motors.

[0041] 3) By introducing the delay estimation error compensation term into the control law, the present invention can adapt to the actual robot joint motor operating conditions with rapid changes in current reference and speed, and achieve better current tracking performance.

[0042] 4) The control law used in the present invention is expressed in vector form, which can not only realize high-precision current tracking control of a single motor, but also be extended to multi-motor occasions to realize high-precision current tracking control of multiple motors.

[0043] In summary, the present invention can estimate the delay estimation error, specifically give the corresponding discrete iterative expression, and compensate for the adverse effect of the delay estimation error on the current tracking error by adding a compensation term in the control law, so that the current tracking error is reduced, thereby achieving high-precision current tracking of the robot joint motor. The scientificity and rationality of this method are explained and demonstrated from the perspective of gradient descent. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the overall control block diagram of the present invention;

[0045] Figure 2 is a flow chart of the method of the present invention;

[0046] Figure 3 The current tracking experiment results of the conventional time delay control algorithm and the improved time delay control algorithm of the present invention under the condition of constant speed and variable reference current are shown in FIG. Figure 3 (a) is the experimental result diagram under the control of the traditional time delay control algorithm. Figure 3 (b) is a graph showing the experimental results when the algorithm of the present invention is used;

[0047] Figure 4 The current tracking experiment results of the conventional time delay control algorithm and the improved time delay control algorithm of the present invention under the condition of variable speed and constant reference current are shown in FIG. Figure 4 (a) is the experimental result diagram under the control of the traditional time delay control algorithm. Figure 4 (b) is a graph showing the experimental results when the algorithm of the present invention is used;

[0048] Figure 5 The current tracking experiment results of the conventional time delay control algorithm and the improved time delay control algorithm of the present invention under the condition of simultaneous change of speed and reference current are shown in FIG. Figure 5 (a) is the experimental result diagram under the control of the traditional time delay control algorithm. Figure 5 (b) is a diagram of experimental results when the algorithm of the present invention is used. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] like Figure 1 As shown, it is an overall block diagram of the high-precision current tracking control of the robot joint motor using the delay estimation error compensation method of the present invention, which specifically includes key links such as the ideal error dynamic equation construction link, the delay estimation link, and the delay estimation error compensation link, among which the delay estimation error compensation link is the key link that is different from the traditional delay control algorithm; Figure 1 In the process of constructing the ideal error dynamics equation, in ( k -1) T s At this moment, the actual current vector of the robot joint motor is I (( k -1) T s ), reference current vector I ref (( k-1) T s ) and its first-order differential value I ref ´(( k -1) T s ) Input the ideal error dynamic equation building block and output the control component u (( k -1) T s ), the specific generation formula is expressed as:

[0051] u (( k -1) T s )= I ref ´(( k -1) T s )+ K p I ref (( k -1) T s )- K p I (( k -1) T s )

[0052] For the delay estimation link, the robot joint motor kT s Lumped uncertainty at time N 2( kT s ) N 2^( kT s )Depend on( k -1) T s Control voltage at the moment V (( k -1) T s )、( k -1) T s The first-order differential value of the actual current vector at time I ´(( k -1) T s ) is obtained through delay estimation, and the delay estimation equation is:

[0053] N 2^(kT s )= N 2(( k -1) T s )= V (( k -1) T s )- A 2 I ´(( k -1) T s )

[0054] The control law of the traditional time delay control algorithm can be obtained by constructing the ideal error dynamic equation link and the time delay estimation link as follows:

[0055] V ( kT s )= A 2 u (( k -1) T s )+ N 2^( kT s )

[0056] The control law of the above conventional time-delay control algorithm is used to control the robot joint motor, and the experimental results of its current tracking are used to compare with the time-delay control algorithm after experimental error estimation compensation of the present invention.

[0057] Then, for the delay estimation error compensation link, kT s The estimated value of the time delay estimation error of the robot joint motor current tracking control at time ε ^( kT s )Depend on( k -1) T s The actual current vector of the robot joint motor at this moment I (( k -1) T s ), reference current vector I ref (( k -1) T s ), the estimated value of the delay estimation error ε ^(( k -1) T s )as well as( k -2) Ts The actual current vector of the robot joint motor at this moment I (( k -2) T s ), reference current vector I ref (( k -2) T s ) is calculated as follows:

[0058] ε ^( kT s )= ε ^(( k -1) T s )+ K c [ I ref (( k -1) T s )- I (( k -1) T s )]- K c ( I 2×2 - T s K p )[ I ref (( k -2) T s )- I (( k -2) T s )]

[0059] Combined with the above ideal error dynamic equation construction link, delay estimation link, and delay estimation error compensation link, the control component u (( k -1) T s ), robot joint motor kT s The estimated value of the time delay of the aggregate uncertainty term at time N 2^( kT s ), kT s The estimated value of the time delay estimation error of the robot joint motor current tracking control at time ε ^( kTs ) are combined to form the control law of improved time delay control, and we get:

[0060] V ( kT s )= A 2 u (( k- 1) T s )+ N 2^( kT s )+ A 2 ε ^( kT s )

[0061] The actual physical meaning of the obtained control law is the d-axis voltage and the q-axis voltage, and the control voltage of the robot joint motor is obtained V ( kT s ) After dq-ab coordinate transformation, the inverter switching function is calculated by SVPWM vector control algorithm S abc By controlling the switch of the switch device, the corresponding abc three-phase drive voltage is obtained and applied to the robot joint motor. The current sensor samples the abc three-phase current and obtains the actual current of the d-axis and the actual current of the q-axis through the abc-dq coordinate transformation, forming kT s The actual current vector of the robot joint motor at this moment I ( kT s ) is fed back to the controller to calculate the control law for the next moment.

[0062] like Figure 2 As shown, the robot joint motor current tracking control delay estimation error compensation method of the present invention is as follows:

[0063] In the specific embodiment of the delay estimation error compensation method of the robot joint motor high-precision current tracking control of the present invention, the control unit in the experiment is composed of a TMS320F28379D control unit and a 5CEFA7F27I7N control unit, the power MOS (Metal-Oxide-Semiconductor) tube adopts a C3M0016120K MOS tube, the control frequency is set to 10kHz, and the switching frequency is set to 10kHz. In order to simulate typical working conditions such as fast-changing reference current and speed changes corresponding to the normal operation of the robot, an ABB motor controlled by an ABB-MFE180 driver is used as a load motor, and is connected to the motor under test by an electromagnetic clutch. The basic parameters of the robot joint motor under test are shown in Table 1 below:

[0064] Table 1

[0065]

[0066] First, the traditional time-delay control algorithm is implemented on the robot joint motor. The control law of the traditional time-delay control algorithm is as follows:

[0067] V ( kT s )= A 2 u (( k -1) T s )+ N 2^( kT s )

[0068] u (( k -1) T s )= I ref ´(( k -1) T s )+ K p I ref (( k -1) T s )- K p I (( k -1) T s )

[0069] N 2^( kT s )= N 2(( k -1) T s )= V (( k -1) T s )- A 2 I ´(( k -1) T s )

[0070] in, V ( ) represents the stator voltage vector to be applied to the robot joint motor,V ( kT s )=[ v d ( kT s ), v q ( kT s )] T , v d ( )and v q ( ) represent the d-axis voltage and q-axis voltage to be applied to the robot joint motors respectively; A 2 represents the gain coefficient, which is a diagonal constant coefficient matrix; u ( ) represents the control component of the traditional time-delay control algorithm control law, which is usually constructed in the form of error dynamics equation; N 2( )and N 2^( ) represent the lumped uncertainties of the robot joint motors and their estimated delays respectively; I ( )and I ´( ) represent the actual current vector and its first-order differential value of the robot joint motor, I (( k -1) T s )=[ i d (( k -1) T s ), i q (( k -1) T s )] T , i d ( )and i q ( ) represent the d-axis current and q-axis current to be applied to the robot joint motors respectively; I ref ( )and I ref ´( ) respectively represent the reference current vector and its first-order differential value of the robot joint motor; K pRepresents a positive definite diagonal constant matrix, which is used to adjust the current tracking error convergence speed.

[0071] Then, the gain coefficient of the time delay control algorithm is adjusted, and the robot joint motor is controlled by the traditional time delay control algorithm under the preset initial gain coefficient. The preset initial gain coefficient is a given smaller gain coefficient close to 0 to ensure that the system can be stably controlled; then the gain coefficient is gradually increased to gradually improve the current tracking accuracy of the system; until the current tracking error of the robot joint motor no longer decreases, it is determined that the traditional time delay control algorithm under the gain coefficient at this time is used to finally control the robot joint motor, so that the robot joint motor can operate normally and stably.

[0072] Then, the delay estimation error estimation model for the normal and stable operation of the robot joint motor is established, as follows:

[0073] ε ^( kT s )= ε ^(( k -1) T s )+ K c [ I ref (( k -1) T s )- I (( k -1) T s )]- K c ( I 2×2 - T s K p )[ I ref (( k -2) T s )- I (( k -2) T s )]

[0074] in, ε ^( ) represents the delay estimation error of the current tracking control of the robot joint motor ε ( ) k and T s Respectively represent the moment and control period in discrete time domain;K c and K p denote the diagonal constant coefficient matrix and the positive definite diagonal constant value matrix respectively, K c To correct the step size, K c =diag( K c1 , K c2 ), K c1 and K c2 Denote the diagonal constant coefficient matrices K c The d-axis and q-axis correction step lengths in , K p =diag( K p1 , K p2 ), K p1 and K p2 denote positive definite diagonal constant matrices K p The d-axis and q-axis current tracking error convergence parameters are used to adjust the d-axis and q-axis current tracking error convergence speed; I ( )and I ref ( ) represent the actual current vector and reference current vector of the robot joint motor respectively; I 2×2 represents the 2nd-order identity matrix.

[0075] The present invention introduces a compensation term for the delay estimation error A 2 ε ^( kT s ), design its discrete iterative expression as the delay estimation error estimation model; transform the discrete iterative expression to obtain:

[0076] ε ^´( kT s )= K c e ´(( k -1) T s )+ K c K p e ((k -2) T s )

[0077] in, ε ^´( ) represents the delay estimation error of the current tracking control of the robot joint motor ε ( ) ε ^( )’s first-order differential value; e ( )and e ´( )represent the current tracking error and its first-order differential value of the robot joint motor respectively.

[0078] because A 2 ε ^( kT s ) is the compensation term for the delay estimation error. If this compensation term is added to the control law, the original error dynamics equation will be rewritten as:

[0079] e ´( kT s )+ K p e (( k -1) T s )= ε ( kT s )+△( kT s )- ε ^( kT s ):= ε ~( kT s )

[0080] Among them, △( ) represents the difference in the rate of change of the reference current at adjacent moments; ε ~( ) represents the current tracking control delay estimation error of the compensated robot joint motor; := is a definition symbol.

[0081] At this time, the current tracking error e ( kT s ) will be reduced compared to before.

[0082] By transforming the above discrete iterative expression and comparing it with the new error dynamics equation, we can get:

[0083] ε ^´(kT s )=- K c (0.5× ε ~ T (( k -1) T s ) ε ~(( k -1) T s )) / ( ε ^(( k -1) T s )

[0084] It can be observed that the above formula is consistent with the form of the gradient descent algorithm, and its value function J for:

[0085] J =0.5× ε ~ T (( k -1) T s ) ε ~(( k -1) T s )

[0086] The goal of gradient descent is to make the value function approach 0, even if the delay error compensation value ε ^(( k- 1) T s ) is updated in the direction of the minimum value function. The minimum value of the quadratic value function is 0, which corresponds to the delay estimation error after compensation. ε ~(( k- 1) T s ) is 0. As the given delay estimation error compensation value is continuously iterated, under the action of the gradient descent principle, the compensated delay estimation error ε ~(( k- 1) T s ) will continue to decrease. From the new error dynamics equation, we can see that the current tracking error will also continue to decrease, thus achieving high-precision current tracking of the robot joint motor. This explains the scientificity and rationality of the iterative expression from the perspective of the gradient descent algorithm.

[0087] The estimated value of the delay estimation error of the current tracking control of the robot joint motor at the previous moment, the actual current vector and the reference current vector at the previous two moments are input into the delay estimation error estimation model. After processing, the delay estimation error estimation model outputs the estimated value of the delay estimation error of the current tracking control of the robot joint motor at the current moment. ε ^(( k -1) T s )and ε ^( kT s ) are respectively the previous ( k -1) T s Moment and Present kT s The estimated value of the time delay estimation error of the current tracking control of the robot joint motor at time.

[0088] An improved delay control algorithm is constructed based on the estimated value of the delay estimation error of the current tracking control of the robot joint motor at the current moment, as follows:

[0089] V ( kT s )= A 2 u (( k- 1) T s )+ N 2^( kT s )+ A 2 ε ^( kT s )

[0090] u (( k- 1) T s )= I ´( kT s )+ ε ( kT s )

[0091] ε ( kT s )= A 2 -1 [ N 2( kT s )- N 2^( kT s )]

[0092] N 2( kT s )=[ N 2d ( kT s ), N 2q ( kT s )] T

[0093] N 2d ( kT s )=( L d - A 21 ) i d ´+ Ri d - w e L d i d

[0094] N 2q ( kT s )=( L q - A 22 ) i q ´+ Ri q - w e ( L d i d + ψ f )

[0095] in, V ( ) represents the stator voltage vector to be applied to the robot joint motor; k and T s Respectively represent the moment and control period in discrete time domain; A 2 represents the gain coefficient, which is a diagonal constant coefficient matrix, A 2=diag( A 21 , A 22 ), A21 and A 22 Denote the diagonal constant coefficient matrices A The gain coefficients of the d-axis and q-axis in 2; u ( ) represents the control component of the control law of the traditional time-delay control algorithm, which is usually constructed in the form of error dynamics equation; N 2^( ) represents the lumped uncertainty of the robot joint motor N 2( )’s estimated delay; ε ^( ) represents the delay estimation error of the current tracking control of the robot joint motor ε ( ) I ´( ) represents the actual current vector of the robot joint motor I ( )’s first-order differential value; N 2d ( )and N 2q ( ) represent the uncertainties of the d-axis and q-axis of the robot joint motors respectively; L d and L q They represent the d-axis inductance and q-axis inductance of the robot joint motor respectively; i d and i d ´represent the d-axis current of the robot joint motor and its first-order differential value, i q and i q ´represent the q-axis current and its first-order differential value of the robot joint motor respectively; R , w e and ψ f They represent the resistance, electrical angular velocity and stator flux of the robot joint motor respectively.

[0096] The control voltage of the robot joint motor finally obtained is the stator voltage vector to be applied to the robot joint motor V ( ). Under the control law of the improved time-delay control algorithm, the right side of the error dynamics equation is made as close to 0 as possible. According to the properties of the first-order differential equation, the current tracking error will also tend to 0.

[0097] An improved time-delay control algorithm is implemented on the robot joint motor, thereby improving the error dynamics equation of the robot joint motor under the action of time-delay control, as follows:

[0098] e ´( kT s )+ K p e (( k -1) T s )= ε ( kT s )+△( kT s )- ε ^( kT s ):= ε ~( kT s )

[0099] △( kT s )= I ref ´( kT s )- I ref ´(( k -1) T s )

[0100] e ( kT s )= I ref ( kT s )- I ( kT s )

[0101] in, e ( )and e ´( )represent the current tracking error and its first-order differential value of the robot joint motor respectively; k and T s Respectively represent the moment and control period in discrete time domain; K p represents a positive definite diagonal constant matrix; ε ( )and ε ^( ) represent the time delay estimation error and its estimated value of the current tracking control of the robot joint motor respectively; △( ) represents the difference in the rate of change of the reference current at adjacent moments; ε ~( ) represents the time delay estimation error of the current tracking control of the robot joint motor after compensation; I ( ) represents the actual current vector of the robot joint motor, I ref ( )and I ref ´( ) represent the reference current vector and its first-order differential value of the robot joint motor respectively.

[0102] The dynamics of the robot joint motors can be modeled as:

[0103] V ( kT s )= A 2 I ´( kT s )+ N 2( kT s )

[0104] N 2( kT s )=[ N 2d ( kT s ), N 2q ( kT s )] T

[0105] N 2d ( kT s )=( L d - A 21 ) i d ´+ Ri d - w e L d i d

[0106] N 2q ( kT s )=( Lq - A 22 ) i q ´+ Ri q - w e ( L d i d + ψ f )

[0107] in, I ´( ) represents the actual current vector of the robot joint motor I ( ) is the first-order differential value of .

[0108] Considering the lumped uncertainty N 2( kT s ) cannot be predicted and obtained in advance, so it needs to be obtained through delay estimation. The error caused by delay estimation will affect the current tracking accuracy, especially under the typical working conditions of fast-changing reference current and fast-changing speed corresponding to the operation of the robot. The current tracking control delay estimation error is defined as ε ( kT s )as follows:

[0109] ε ( kT s ) := A 2 -1 [ N 2( kT s )- N 2^( kT s )]

[0110] Combining the robot dynamics modeling equation with the traditional time-delay control law, we get the following relationship:

[0111] u (( k- 1) T s )- I ´( kT s )= ε ( kT s )

[0112] The control component that can be set manually in the time delay control law u (( k- 1)T s ) is substituted into the specific expression, and the error dynamics equation is further obtained as follows:

[0113] e ´( kT s )+ K p e (( k -1) T s )= ε ( kT s )+△( kT s )

[0114] △( kT s )= I ref ´( kT s )- I ref ´(( k -1) T s

[0115] e ( kT s )= I ref ( kT s )- I ( kT s )

[0116] From the error dynamics equation, we know that the current tracking error e ( kT s ) Mainly affected ε ( kT s ) and △( kT s ) together, and the difference in the reference current change rate at adjacent moments △( kT s ) cannot be determined manually when encountering unexpected working conditions, such as when encountering sudden external forces and moments, △( kT s ) will suddenly increase or decrease. Therefore, the delay estimation error is compensated to reduce ε ( kT s ), to reduce the current tracking error e ( kTs ) is of great significance.

[0117] Finally, the three-phase driving voltage of the robot joint motor is obtained according to the control voltage of the robot joint motor. The control voltage of the robot joint motor includes the d-axis voltage and the q-axis voltage to be applied to the robot joint motor. The d-axis voltage and the q-axis voltage to be applied to the robot joint motor are converted into the α-axis voltage and the β-axis voltage after dq-αβ coordinate transformation, and then the switching devices of the inverter of the robot joint motor are controlled to be turned on and off after space vector pulse width modulation SVPWM, so as to obtain the three-phase driving voltage of the robot joint motor, and finally the operation of the robot joint motor is controlled to perform current tracking control delay estimation error compensation of the robot joint motor, so as to realize high-precision current tracking of the robot joint motor.

[0118] In the specific implementation of the present invention, in order to compare the current tracking accuracy before and after adding the delay estimation error compensation link, all the constant coefficients in the expressions of all delay control laws involved are set to A 2=diag(0.0048, 0.0048), K p =diag(50, 50), K c =diag(5000, 5000). In order to observe and compare the current tracking effect more intuitively, the d-axis current tracking error is compared. i d-error = i d-ref -i d and q-axis current tracking error i q-error = i q-ref -i q , i d-ref and i q-ref They represent the d-axis reference current and the q-axis reference current respectively.

[0119] like Figure 3 As shown in the figure, it is the current tracking control experimental result diagram of the traditional time delay control algorithm and the algorithm of the present invention under the condition of constant speed and variable reference current. The specific working condition set in the experiment is: the load motor provides a constant speed condition of 1200r / min, the q-axis current setting of the motor under test is set to a sinusoidal variation reference current with a frequency of 100Hz, an amplitude of 2A, and a bias of 4A, and the d-axis current setting is 0A, that is, I ref =[ i d-ref , iq-ref ] T =[0, 4+2sin(200pt)] T A. It should be pointed out that due to the use of i d-ref =0A control, so i d-error = -i d For convenience, the actual current is used i d Reflects the d-axis current tracking error i d-error , and the q-axis current tracking error i q-error The experimental results show that if Figure 3 As shown in (a), before compensating the delay estimation error, the d-axis current tracking error i d-error The maximum and minimum values ​​of - i q-ref The maximum and minimum values ​​of the corresponding moments are the same, and its variation range is 0.38A; the q-axis current tracking error i q-error The maximum and minimum values ​​correspond to the phase advance i q-ref The maximum and minimum values ​​of correspond to a phase of 90°, and its variation range is 0.77A. After compensating for the delay estimation error, Figure 3 As shown in (b), the d-axis current tracking error i d-error The variation range is reduced to 0.25A, and the q-axis current tracking error i q-error The variation range is reduced to 0.23 A, and the periodic variation is no longer obvious. This verifies the effectiveness of the proposed algorithm for high-precision current tracking under constant speed and variable reference current conditions.

[0120] like Figure 4 As shown in FIG. 1 , the current tracking control experimental results of the conventional time delay control algorithm and the algorithm of the present invention under the condition of variable speed and constant reference current are shown. The specific working condition set in the experiment is: the load motor provides a variable speed condition with a sinusoidal change of 2 Hz, an amplitude of 300 r / min, and a bias of 900 r / min, that is, n =900+300sin(4pt) r / min, n Indicates the speed. The d-axis and q-axis currents of the motor under test are set to 0A and 6A respectively, that is, I ref =[ i d-ref , iq-ref ] T =[0, 6] T A. Experimental results show that Figure 4 As shown in (a), before compensating the delay estimation error, the d-axis current tracking error i d-error Maximum and minimum hysteresis speed w e The maximum and minimum values ​​of the phase correspond to 90°, and its variation range is 0.38A; the q-axis current tracking error i q-error The maximum and minimum values ​​correspond to the phase leading speed w e The maximum and minimum values ​​of correspond to a phase of 90°, and its variation range is 0.60A. After compensating for the delay estimation error, Figure 4 As shown in (b), the d-axis current tracking error i d-error The variation range is reduced to 0.33A, and the q-axis current tracking error i q-error The variation range is reduced to 0.21A, at which point the maximum and minimum values ​​of the current tracking error are related to the speed. w e The maximum and minimum values ​​coincide with each other. This verifies the effectiveness of the proposed algorithm in high-precision current tracking under constant reference current and variable speed conditions.

[0121] like Figure 5 As shown in the figure, the current tracking control experimental results of the traditional time delay control algorithm and the algorithm of the present invention under the condition that the speed and reference current change simultaneously, zoom represents local amplification, and the specific working condition set in the experiment is: the load motor provides n =900+300sin(4pt) r / min variable speed condition, the current tracking reference of the motor under test is set to I ref =[ i d-ref , i q-ref ] T =[0, 4+2sin(200pt)] T A. The experimental results show that due to the large difference between the speed change cycle and the current change cycle, the current tracking error is affected by different factors at different time scales and exhibits different tracking characteristics. Figure 5 As shown in (a), when the delay estimation error is not compensated, the d-axis current tracking error i d-error Maximum and minimum hysteresis speed w eThe maximum and minimum values ​​of the phase correspond to 90°, and its variation range is 0.42A; the q-axis current tracking error i q-error The maximum and minimum values ​​correspond to the phase leading speed w e The maximum and minimum values ​​of correspond to a phase of 90°, and its variation range is 1.20A. After compensating for the delay estimation error, Figure 5 As shown in (b), the d-axis current tracking error i d-error The variation range is reduced to 0.32A, and the q-axis current tracking error i q-error The variation range is reduced to 0.32A, at which point the maximum and minimum values ​​of the current tracking error are related to the speed. w e The maximum and minimum values ​​coincide. However, at a smaller time scale, such as Figure 5 As shown in zoom1~zoom4 in (a), taking the q-axis current as an example, before compensating the delay estimation error, its tracking error is i q-error The maximum and minimum values ​​correspond to the phase advance i q-ref The maximum and minimum values ​​correspond to a phase of 79°. In addition, the sinusoidal speed w e This causes the current tracking to advance periodically ( Figure 5 (a) zoom1) and hysteresis ( Figure 5 (a) zoom3), where the peak-to-peak current fluctuation errors are 0.75A and 0.78A respectively. After compensating for the delay estimation error, as Figure 5 As shown in zoom5 and zoom7 in (b), the periodic advance and lag of the q-axis current tracking error at the corresponding time is no longer obvious, and the peak-to-peak value of the current fluctuation error is reduced to 0.16 A and 0.21 A respectively. This verifies the effectiveness of the proposed algorithm in high-precision current tracking under the condition of both speed and reference current changing.

[0122] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for compensating delay estimation error in current tracking control of robot joint motors, characterized in that: include: Step 1) Implementing a time delay control algorithm on the robot joint motor and adjusting the gain coefficient of the time delay control algorithm so that the robot joint motor operates normally and stably; Step 2) Establish a delay estimation error estimation model when the robot joint motor is running normally and stably, input the estimated value of the delay estimation error of the current tracking control of the robot joint motor at the previous moment, the actual current vector and the reference current vector at the previous two moments into the delay estimation error estimation model, and output the estimated value of the delay estimation error of the current tracking control of the robot joint motor at the current moment after processing by the delay estimation error estimation model; Step 3) constructing an improved time delay control algorithm according to the estimated value of the current tracking control time delay estimation error of the robot joint motor at the current moment, and implementing the improved time delay control algorithm on the robot joint motor, thereby improving the error dynamics equation of the robot joint motor under the action of time delay control, and finally obtaining the control voltage of the robot joint motor; Step 4) The three-phase drive voltage of the robot joint motor is obtained according to the control voltage of the robot joint motor, so as to control the robot joint motor to perform current tracking control delay estimation error compensation of the robot joint motor and realize current tracking of the robot joint motor.

2. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 1, characterized in that: In the step 1), the robot joint motor is first controlled by a time delay control algorithm under a preset initial gain coefficient, and then the gain coefficient is gradually increased until the current tracking error of the robot joint motor no longer decreases. The time delay control algorithm under the gain coefficient at this time is determined to perform final control on the robot joint motor, so that the robot joint motor can operate normally and stably.

3. The method for compensating time delay estimation error of current tracking control of robot joint motor according to claim 1, characterized in that: In the step 2), the delay estimation error estimation model when the robot joint motor is running normally and stably is as follows: ε ^( kT s )= ε ^(( k -1) T s )+ K c [ I ref (( k -1) T s )- I (( k -1) T s )]- K c ( I 2×2 - T s K p )[ I ref (( k -2) T s )- I (( k -2) T s )] in, ε ^() represents the delay estimation error of the current tracking control of the robot joint motor ε () estimated value; k and T s Respectively represent the moment and control period in discrete time domain; K c and K p denote the diagonal constant coefficient matrix and the positive definite diagonal constant value matrix respectively, K c =diag( K c1 , K c2 ), K c1 and K c2 Denote the diagonal constant coefficient matrices K c The d-axis and q-axis correction step lengths in , K p =diag( K p1 , K p2 ), K p1 and K p2 denote positive definite diagonal constant matrices K p The d-axis and q-axis current tracking error convergence parameters in; I ()and I ref () represent the actual current vector and reference current vector of the robot joint motor respectively; I 2×2 represents the 2nd-order identity matrix; ε ^(( k -1) T s )and ε ^( kT s ) are respectively the previous ( k -1) T s Moment and Present kT s The estimated value of the time delay estimation error of the current tracking control of the robot joint motor at time.

4. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 1, characterized in that: In the step 3), the improved delay control algorithm is as follows: V ( kT s )= A 2 u (( k- 1) T s )+ N 2^( kT s )+ A 2 ε ^( kT s ) in, V () represents the stator voltage vector to be applied to the robot joint motor; k and T s Respectively represent the moment and control period in discrete time domain; A 2 represents the gain factor, A 2=diag( A 21 , A 22 ), A 21 and A 22 Denote the diagonal constant coefficient matrices A The gain coefficients of the d-axis and q-axis in 2; u () represents the control component of the delay control algorithm; N 2^() represents the lumped uncertainty of the robot joint motor N 2() delay estimate; ε ^() represents the delay estimation error of the current tracking control of the robot joint motor ε () estimated value; The control voltage of the robot joint motor is the stator voltage vector to be applied to the robot joint motor V ().

5. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 4, characterized in that: The control component of the delay control algorithm u ()as follows: u (( k- 1) T s )= I ´( kT s )+ ε ( kT s ) ε ( kT s )= A 2 -1 [ N 2( kT s )- N 2^( kT s )] in, I ´() represents the actual current vector of the robot joint motor I The first-order differential value of ().

6. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 4, characterized in that: The lumped uncertainty of the robot joint motor is N 2() is as follows: N 2( kT s )=[ N 2d ( kT s )， N 2q ( kT s )] T N 2d ( kT s )=( L d - A 21 ) i d ´+ Ri d - w e L d i d N 2q ( kT s )=( L q - A 22 ) i q ´+ Ri q - w e ( L d i d + ψ f ) in, N 2d ()and N 2q () represent the uncertainties of the d-axis and q-axis of the robot joint motors respectively; L d and L q They represent the d-axis inductance and q-axis inductance of the robot joint motor respectively; i d and i d ´represent the d-axis current of the robot joint motor and its first-order differential value, i q and i q ´represent the q-axis current and its first-order differential value of the robot joint motor respectively; R , w e and ψ f They represent the resistance, electrical angular velocity and stator flux of the robot joint motor respectively.

7. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 1, characterized in that: In the step 3), the error dynamics equation of the improved robot joint motor under the time delay control is as follows: e ´( kT s )+ K p e (( k -1) T s )= ε ( kT s )+△( kT s )- ε ^( kT s ) △( kT s )= I ref ´( kT s )- I ref ´(( k -1) T s ) e ( kT s )= I ref ( kT s )- I ( kT s ) in, e ( )and e ´( )represent the current tracking error and its first-order differential value of the robot joint motor respectively; k and T s Respectively represent the moment and control period in discrete time domain; K p represents a positive definite diagonal constant matrix; ε ()and ε ^() represents the time delay estimation error and its estimated value of the current tracking control of the robot joint motor respectively; △() represents the difference of the reference current change rate at adjacent moments; ε ~() represents the time delay estimation error of the current tracking control of the compensated robot joint motor; I () represents the actual current vector of the robot joint motor, I ref ()and I ref ´() represent the reference current vector and its first-order differential value of the robot joint motor respectively.

8. The method for compensating delay estimation error of current tracking control of robot joint motor according to claim 1, characterized in that: In the step 4), the control voltage of the robot joint motor includes the d-axis voltage and the q-axis voltage to be applied to the robot joint motor. The d-axis voltage and the q-axis voltage to be applied to the robot joint motor are converted into the α-axis voltage and the β-axis voltage after dq-αβ coordinate transformation, and then the switching devices of the inverter of the robot joint motor are controlled to be turned on and off after space vector pulse width modulation SVPWM, so as to obtain the three-phase drive voltage of the robot joint motor, and finally control the operation of the robot joint motor.

9. An electronic device, characterized in that: include: A memory and a processor coupled to each other, wherein the memory stores program data, and the processor calls the program data to execute the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having program data stored thereon, characterized in that: When the program data is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images