An event-driven online implementation method for digital twins of motor drive systems

Abstract

The application discloses an event-driven motor drive system digital twin online implementation method, specifically comprising the following steps: firstly, constructing an inverter-motor system model and discretizing the same by using a fourth-order Runge-Kutta method; based on an event-driven method, taking the intersection point of a carrier wave and a modulation wave in a modulation algorithm as a discrete switching event, and determining a model discretization calculation step length according to the switching event; then, combining the model calculation step length, programming a discretized digital model and writing the same into a digital microcontroller to realize online implementation, and dividing the digital microcontroller into a control part and an online digital twin part; and using a particle swarm optimization algorithm to iteratively optimize key parameters in the constructed online digital twin model, so as to obtain an optimized digital twin model and realize online estimation of motor parameters. The application effectively shortens the overall solution time of the digital twin model, enables the digital twin model to be realized online without affecting the control process, and simultaneously obtains multiple key parameters of the motor.

Description

Technical Field

[0001] The present invention belongs to the field of power electronics technology, and in particular relates to an event-driven online implementation method of a digital twin of a motor drive system. Background Art

[0002] AC motors have long been the mainstream motor in various industrial applications. Due to their stability, reliability, and fast response, they are widely used in fields such as machinery manufacturing, electric vehicles, and rail transportation. As motors are crucial drive units in machinery, sensing and monitoring their operating status is crucial for their proper operation and service life. For example, key motor parameters are crucial for efficient motor control, condition monitoring, and fault diagnosis. However, motor parameters often vary depending on their operating conditions. Therefore, appropriate methods are needed to obtain these key motor parameters.

[0003] Numerous methods exist for monitoring key motor parameters, both offline and online. Offline methods primarily include stalled-rotor and no-load tests. Offline methods struggle to monitor motor parameters while the motor is running. Online methods primarily include those based on least squares, Kalman filtering, and model reference adaptation. However, online methods struggle to estimate multiple parameters simultaneously, and some require additional sensors. With the rapid development of artificial intelligence (AI) technology, machine learning algorithms have emerged as a new approach for monitoring key motor parameters. However, machine learning algorithms rely on large amounts of training data, which is difficult to implement in many application scenarios.

[0004] Digital twin technology is currently considered one of the top ten strategic technologies for the future and has received widespread attention in the field of power electronics. Some scholars have proposed using digital twin technology to monitor the status of components in AC-DC and DC-DC converters, but these methods are primarily offline. Digital twin technology is rarely used in the field of electric motors. Therefore, some scholars have proposed using digital twin technology to monitor key motor parameters. However, this method is still an offline implementation method, which has the disadvantages of high computational complexity and inability to be implemented and applied online. Summary of the Invention

[0005] In view of the deficiency that the existing digital twin technology can only be implemented offline, the present invention provides an event-driven online implementation method of the digital twin of the motor drive system.

[0006] The present invention provides an event-driven online implementation method for a digital twin of a motor drive system, comprising the following steps:

[0007] Step 1: Taking the asynchronous motor drive system as an example, establish the inverter-motor system state equation in the αβ coordinate system:

[0008]

[0009] Among them, T s =L s / R s ,σ=1-L m 2 / (L s L r ), T r =L r / R r ,ω r is the speed, R s 、R r Represents the stator and rotor resistance, L s 、L r Represents the stator and rotor inductance, L m represents the mutual inductance between the stator and rotor, i sα 、i sβ is the stator current component, ψ rα , ψ rβ is the rotor flux component, ω r is the rotor speed, u sα 、u sβ is the voltage component.

[0010] After obtaining the state equation of the inverter-motor system, the fourth-order Runge-Kutta method is used to discretize the mathematical model of the inverter-motor system to obtain the discretized mathematical model:

[0011]

[0012] Among them, i sα,n+1 、i sβ,n+1 , ψ rα,n+1 , ψ rβ,n+1 and i sα,n 、i sβ,n , ψ rα,n , ψ rβ,n They are the current and rotor flux components at sampling time n+1 and sampling time n, h represents the discretization calculation step, K a1 ~K a4 , K b1 ~K b4 , K c1 ~K c4 and K d1 ~K d4 are coefficients, calculated from the motor differential equation.

[0013] Step 2: Calculate the discretized mathematical model based on a switching event-driven solution method.

[0014] Power electronic systems are often mixed systems of continuous variables and discrete variables. Their state equations can be simply described as:

[0015]

[0016] Among them, x(t) is the state variable, δ(t) is the discrete variable part, u(t) is a constant or other input, and A, B, and C are parameter matrices.

[0017] Equation f sw The form of depends on the discrete variable part. When the discrete variable part changes, the equation f sw The form changes accordingly; when the system state changes, the control logic will change the discrete variable part, that is, the occurrence of discrete events determines the form of state change, and the change of system state triggers discrete events. Therefore, the system state equation f can be calculated only when the switch signal changes. sw , to simulate the changes in system states under discrete events, and at the same time, system changes will trigger the control logic to change discrete events.

[0018] In the inverter-motor system, discrete events are mainly changes in the inverter drive signal. To ensure accuracy while improving model solution efficiency, the discretized mathematical model is implemented online. The intersection of the modulation wave and the carrier in the modulation algorithm is used as the discrete switching event. The discrete calculation step size h of the mathematical model is determined based on the switching event:

[0019]

[0020] in, They are the time when the switch signal change event occurs. When the control period is constant and the control strategy is known, the time can be calculated.

[0021] The fourth-order Runge-Kutta method is used to discretize the mathematical model of the inverter-asynchronous motor system, namely:

[0022]

[0023] This switching event-driven method simplifies the calculation process of the inverter-motor system mathematical model and can integrate the inverter-motor system mathematical model into a digital microcontroller to obtain an online twin, which is convenient for embedded implementation.

[0024] Step 3: Combined with the discretization step size of the mathematical model in step 2, the mathematical model of the inverter-motor system is programmed and written into the digital microcontroller, namely the online twin. At this time, the digital microcontroller is mainly divided into a control part and an online digital twin part. The input of the control part is the stator current and speed signal obtained by sampling the existing sensors of the system, and the output is the inverter switching device control signal. The input of the online twin is the stator current and speed signal obtained by the sampling circuit and the modulation signal generated by the modulation link in the control part of the microcontroller. After the online twin is solved, the stator current of the system at the next moment is output to simulate the output characteristics of the actual physical system online in the microcontroller. Among them, the solution of the online twin part does not affect the control process.

[0025] Step 4: Using an intelligent algorithm (e.g., particle swarm optimization) as a bridge between the online twin and the actual system, iteratively optimize the key parameters within the twin (e.g., the stator and rotor resistance, stator and rotor inductance, and mutual inductance between the stator and rotor of the asynchronous motor) to ensure that the output characteristics of the constructed twin are consistent with those of the actual physical system, thereby obtaining an optimized online digital twin model of the motor drive system. At this point, the iterative objective function is:

[0026]

[0027] Among them, N d is the amount of sampled data, i sap,n 、i sbp,n 、i scp,n is the three-phase current sampling value, i sad,n 、i sbd,n 、i scd,n The three-phase current value is output to the digital twin model. At the same time, the process of iteratively optimizing the key parameters of the twin is called online motor parameter estimation. When the difference between the output characteristics of the online twin and the actual physical circuit is less than the threshold, the iterative optimization result is the online motor parameter monitoring result. In addition, the final motor key parameter monitoring result is obtained by averaging multiple monitorings.

[0028] The beneficial technical effects of the present invention include:

[0029] 1) For the mathematical model of the motor drive system discretized by the fourth-order Runge-Kutta method, a switch event-driven digital twin model solution method is used, which effectively shortens the overall online execution time of the digital twin model. This enables the digital twin model to be integrated into the microcontroller without affecting the control process, facilitating the online implementation and embedded development of the digital twin model.

[0030] 2) Combined with a switching event-driven digital twin model, online monitoring of key motor parameters (e.g., stator and rotor resistance, stator and rotor inductance, and mutual inductance between the stator and rotor) is possible. The error in key parameter monitoring is less than 10%, and all parameters are obtained simultaneously, providing an effective means for high-performance motor control and condition monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a three-phase inverter-asynchronous motor system topology.

[0032] Figure 2 This is the vector control block diagram.

[0033] Figure 3 It is the switching signal within the switching cycle of the digital microcontroller.

[0034] Figure 4 Bring processes online for digital twins.

[0035] Figure 5 It is the output characteristic result of the actual physical circuit in the steady state of the system.

[0036] Figure 6 Comparison of the output characteristics of the digital twin model and the actual physical circuit under system steady state.

[0037] Figure 7 This is the output characteristic result of the actual physical circuit when the system reduces the load torque.

[0038] Figure 8 Comparison of the output characteristics of the digital twin model and the actual physical circuit when the system reduces load torque.

[0039] Figure 9 The motor parameter estimation results are under the steady state of the system.

[0040] Figure 10 The average error of motor parameter estimation under different discretization orders.

[0041] Figure 11 Root mean square error of motor parameter estimation under different discretization orders. DETAILED DESCRIPTION

[0042] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] Taking the asynchronous motor drive system as an example, the three-phase two-level inverter-asynchronous motor system is the research object. Its topology is as follows Figure 1 As shown, the inverter-asynchronous motor system control method adopts vector control and space vector pulse width modulation strategy, and its structural block diagram is shown in Figure 2 shown.

[0044] Based on Kirchhoff's law, the mathematical model of the three-phase two-level inverter is:

[0045]

[0046] Among them, u an 、u bn 、u cn They are the three-phase voltage, U dc is the DC side voltage, S1~S6 are the driving signals of power switches T1~T6 (1 is on, 0 is off).

[0047] The three-phase asynchronous motor is powered by a three-phase inverter, and its mathematical model in the αβ coordinate system can be expressed as:

[0048]

[0049] Among them, T s =L s / R s ,σ=1-L m 2 / (L s L r ), T r =L r / R r ,ω r is the speed, R s 、R r Represents the stator and rotor resistance, L s 、L r Represents the stator and rotor inductance, L m represents the mutual inductance between the stator and rotor, i sα 、i sβ is the stator current component, ψ rα , ψ rβ is the rotor flux component, ω r is the rotor speed, u sα 、u sβ is the voltage component. The αβ coordinate axis is transformed from the abc coordinate axis, and the coordinate transformation formula is:

[0050]

[0051] The fourth-order Runge-Kutta method is used to discretize the mathematical model of the inverter-asynchronous motor system, that is,

[0052]

[0053] Among them, i sα,n+1 、i sβ,n+1 , ψ rα,n+1 , ψ rβ,n+1 and i sα,n 、isβ,n , ψ rα,n , ψ rβ,n They are the current and rotor flux at time (n+1) and time (n), h represents the calculation step, K a1 ~K a4 , K b1 ~K b4 , K c1 ~K c4 and K d1 ~K d4 is the coefficient. At this point, the inverter-asynchronous motor system can be described as:

[0054]

[0055] Power electronic systems are often mixed systems of continuous variables and discrete variables. Their state equations can be simply described as:

[0056]

[0057] Among them, x(t) is the state variable, δ(t) is the discrete variable part, u(t) is a constant or other input, and A, B, and C are parameter matrices.

[0058] Equation f sw The form of depends on the discrete variable part. When the discrete variable part changes, the equation f sw The form changes accordingly; when the system state changes, the control logic will change the discrete variable part, that is, the occurrence of discrete events determines the form of state change, and the change of system state triggers discrete events. Therefore, the system state equation f can be calculated only when the switch signal changes. sw , to simulate the changes in system states under discrete events, and at the same time, system changes will trigger the control logic to change discrete events.

[0059] In inverter-motor systems, discrete events primarily involve changes in the inverter drive signal. Considering this, a switching event-driven solution method is proposed for computing the digital twin model. To ensure accuracy while improving model solution efficiency, a discretized mathematical model is implemented online, with the intersection of the modulation wave and the carrier wave as the discrete event. The mathematical model discretization solution step size h is determined based on the switching event:

[0060]

[0061] in, They are the time when the switch signal change event occurs. When the control period is constant and the control strategy is known, the time can be calculated.

[0062] The present invention simplifies the calculation process of the inverter-asynchronous motor system mathematical model, and can integrate the inverter-asynchronous motor system mathematical model into the digital microcontroller to obtain an online twin and realize embedded system. Figure 3 The digital twin online implementation process is as follows. Figure 4 shown.

[0063] An inverter-asynchronous motor experimental platform was built, and the mathematical model of the inverter-asynchronous motor system described above was programmed and written into a digital microcontroller, creating an online twin. The digital microcontroller is primarily divided into a control section and an online digital twin section. The control section's inputs are the stator current and speed signals sampled by the system's existing sensors, and its output is the inverter switching device control signal. The online twin's inputs are the stator current and speed signals obtained by the sampling circuit, as well as the modulation signal generated by the modulation stage in the digital microcontroller's control section. After the online twin solution is applied, the stator current at the next moment is output, simulating the output characteristics of the actual physical system online within the microcontroller.

[0064] In addition, the particle swarm optimization algorithm is used as a bridge between the twin and the actual physical system. The difference in output characteristics between the twin and the actual physical system is used as the optimization objective function. The key internal parameters of the twin (for example, the stator and rotor resistance, stator and rotor inductance, and mutual inductance between the stator and rotor of the asynchronous motor) are iteratively optimized to make the output characteristics of the constructed twin close to the actual physical circuit, and finally obtain the digital twin model of the three-phase inverter-asynchronous motor system. The objective function is:

[0065]

[0066] Among them, N d is the amount of sampled data, i sap,n 、i sbp,n 、i scp,n is the three-phase current sampling value, i sad,n 、i sbd,n 、i scd,n Output three-phase current values ​​for the digital twin model.

[0067] When the output data difference between the mathematical model and the actual physical circuit is less than the set threshold or reaches the number of iterations, the digital twin model of the inverter-motor can be obtained. This twin model can simulate the steady-state and dynamic processes of the actual physical circuit, such as Figure 5 、 Figure 6 The following table shows the output characteristics of the actual physical circuit in the steady state of the system, as well as the comparison between the output characteristics of the digital twin model and the actual physical circuit. Figure 7 、 Figure 8The figure shows the output characteristics of the actual physical circuit when the system reduces the load torque, as well as the comparison between the output characteristics of the digital twin model and the actual physical circuit.

[0068] In order to further verify the effectiveness of the proposed digital twin online implementation method, the key parameters of the motor are estimated online under the steady state of the system, and 6 estimated values ​​are obtained. The final estimated values ​​are compared with the nominal values. Figure 9 As shown in Table 1, the optimization results of key parameters are statistically analyzed and the error is analyzed. The execution time of the proposed digital twin model online method and offline implementation method is compared and analyzed, as shown in Table 2.

[0069] Table 1 Monitoring results of key motor parameters

[0070]

[0071]

[0072] Table 2 Comparison of online and offline implementation methods of digital twin models

[0073]

[0074] In addition, the execution time of the mathematical model of the inverter-asynchronous motor system under different discretization orders is compared, as shown in Table 3.

[0075] Table 3 Online execution time of digital twin model under different discretization orders

[0076]

[0077] Parameter estimation is performed using the mathematical model of the system under different discretization orders, and the errors are counted and analyzed, such as Figure 10 、 11 shown.

Claims

1. An event-driven online implementation method for digital twins of a motor drive system, characterized in that: The following steps are involved: Step 1: Taking the asynchronous motor drive system as an example, establish the inverter-motor system state equation in the αβ coordinate system: Among them, T s =L s / R s ,σ=1-L m 2 / (L s L r ), T r =L r / R r ,ω r is the speed, R s 、R r Represents the stator and rotor resistance, L s 、L r Represents the stator and rotor inductance, L m represents the mutual inductance between the stator and rotor, i sα 、i sβ is the stator current component, ψ rα , ψ rβ is the rotor flux component, ω r is the rotor speed, u sα 、u sβ is the voltage component; After obtaining the state equation of the inverter-motor system, the fourth-order Runge-Kutta method is used to discretize the mathematical model of the inverter-motor system to obtain the discretized mathematical model: Among them, i sα,n+1 、i sβ,n+1 , ψ rα,n+1 , ψ rβ,n+1 and i sα,n 、i sβ,n , ψ rα,n , ψ rβ,n They are the current and rotor flux components at sampling time n+1 and sampling time n, h represents the discretization calculation step, K a1 ~K a4 , K b1 ~K b4 , K c1 ~K c4 and K d1 ~K d4 is the coefficient, calculated according to the motor differential equation; Step 2: Based on the event-driven method, in order to ensure accuracy while improving the efficiency of solving the motor system mathematical model, the mathematical model discretized by the fourth-order Runge-Kutta method is implemented online. The intersection of the carrier and modulation wave in the modulation algorithm is used as the discrete switching event, and the solution step size h of the model discretization is determined according to the switching event: in, are the time when the switch signal change event occurs, which can be calculated when the control period is constant and the control strategy is known; Step 3: Combined with the discretization step size of the mathematical model in Step 2, the mathematical model of the inverter-motor system is programmed and written into the digital microcontroller, i.e., the online twin. At this point, the digital microcontroller is divided into a control part and an online digital twin part. The input of the control part is the stator current and speed signals obtained by sampling the system's existing sensors, and the output is the inverter switching device control signal. The input of the online twin is the stator current and speed signal obtained by the sampling circuit and the modulation signal generated by the modulation link in the microcontroller control part. After the online twin is solved, the stator current of the system at the next moment is output to simulate the output characteristics of the actual physical system online in the microcontroller. The solution of the online twin part does not affect the control process. Step 4: Using the intelligent algorithm as a bridge between the online twin and the actual system, iteratively optimize the key parameters within the twin. These parameters are the stator and rotor resistance, stator and rotor inductance, and mutual inductance of the asynchronous motor. This ensures that the output characteristics of the constructed twin are consistent with those of the actual physical system, thereby obtaining an optimized online digital twin model of the motor drive system. At this point, the iterative objective function is: Among them, N d is the amount of sampled data, i sap,n 、i sbp,n 、i scp,n is the three-phase current sampling value, i sad,n 、i sbd,n 、i scd,n The three-phase current value is output to the digital twin model. At the same time, the process of iteratively optimizing the key parameters of the twin is called online motor parameter estimation. When the difference between the output characteristics of the online twin and the actual physical circuit is less than the threshold, the iterative optimization result is the online motor parameter monitoring result. In addition, the final motor key parameter monitoring result is obtained by averaging multiple monitorings.

Images