Sensorless motor control method based on a lyapunov-sliding mode observer

Abstract

The application discloses a Luenberger-sliding mode observer non-inductive control method with a sliding mode saturation function feedback item. A Luenberger-sliding mode observer is established, actual output in the kth control period and input in the kth control period are input into the Luenberger-sliding mode observer for real-time processing to obtain an estimated mechanical angular velocity of a motor rotor and an estimated mechanical angle, which are fed back to a control side, so that non-inductive and non-position control of the motor is realized, and it is ensured that the permanent magnet synchronous motor can stably operate in a medium-high speed section. The application avoids cost and anti-interference problems caused by a position encoder, simultaneously improves parameter sensitivity problems of a traditional Luenberger observer, and enhances robustness of a state observer.

Description

Technical Field

[0001] This invention designs a sensorless control method using a Luneburg-sliding mode observer that incorporates sliding mode feedback, primarily applicable to the sensorless control of high-speed segments in permanent magnet synchronous motors. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) use permanent magnets as the excitation material, offering advantages such as low material consumption, high power factor, and fast dynamic response. With the increasing maturity of control theory and the gradual improvement of hardware power levels, PMSMs are increasingly being applied in various industrial scenarios. In PMSM control systems, obtaining accurate motor position information is crucial; therefore, encoders are often installed to acquire this information in most applications. However, as motor speed increases, encoder accuracy and durability limit the stability of the control system. Therefore, researching sensorless technology has significant practical value and importance.

[0003] To obtain rotor position and speed information, traditional sensorless methods include sliding mode observers, Luneburg observers, and extended Kalman filters. Among these, sliding mode observers and Luneburg observers are widely used due to their advantages of simple structure and low computational cost. Both of these observers operate on the principle of analyzing the back EMF information of the permanent magnet synchronous motor. and An estimate is made, and the speed information is extracted from the back EMF using a phase-locked loop (PLL) structure for closed-loop speed control.

[0004] The aforementioned , stator - Information on back electromotive force in the coordinate system.

[0005] However, the aforementioned observer design methods also have certain drawbacks. Traditional sliding mode observers replace the back EMF in the stator coordinate system current equation with a switching quantity that takes the current estimation error as input. The aim is to represent the back EMF information using a high-frequency switching signal, and then pass it through a low-pass filter to obtain accurate back EMF information. This observer design method has the advantages of low sensitivity to motor parameters and high system robustness, but the use of high-frequency switching signals and low-pass filters also introduces high-frequency noise and phase delay into the control system.

[0006] Traditional Luneburg observers are based on the current equations in the stator coordinate system, using back potential. and Stator current is a state variable. , A state observer is constructed as the output variable. Back EMF information is observed through state estimation, and motor speed and position information are obtained through a phase-locked loop (PLL) structure. This observer design has the advantages of simple structure and no high-frequency chattering in the observation results. However, because the design parameters of the state observer include many motor parameters, when changes in operating conditions cause changes in motor parameters, the single Luneburg observer is prone to divergence in the estimation results.

[0007] Meanwhile, the aforementioned observers can only be applied to medium- and high-speed ranges. In the low-speed range, the motor's back EMF has not yet fully established, resulting in unstable estimations by the observers and significant deviations in the estimated rotor position. Therefore, existing technologies lack a control method to drive the motor to medium- and high-speed ranges before switching to observer operation. Summary of the Invention

[0008] To address the problems existing in the prior art, the purpose of this invention is to provide a Luenberger-sliding mode observer sensorless control method that introduces a sliding mode saturation function feedback term. A Luenberger-sliding mode observer is established, and the actual output and input of the k-th control cycle are input into the Luenberger-sliding mode observer for real-time processing to obtain the estimated mechanical angular velocity and estimated mechanical angle of the motor rotor. These are then fed back to the control side, achieving sensorless and positionless control of the motor and ensuring stable operation of the permanent magnet synchronous motor in the medium- and high-speed range. This invention avoids the cost and interference problems associated with position encoders, while also improving the parameter sensitivity problem of traditional Luenberger observers and enhancing the robustness of the state observer.

[0009] The technical solution of the present invention is as follows: This invention establishes a Luenberger-sliding mode observer, which measures the actual output of the k-th control cycle. and the input quantity in the kth control cycle The estimated state variables are updated in real time by inputting them into the Luneburg-sliding mode observer. The estimated mechanical angular velocity of the motor rotor is obtained through processing. and estimating mechanical angles This allows the permanent magnet synchronous motor to operate stably at medium and high speeds by feeding back the rotor position information.

[0010] The expression for the Lumberjack-sliding mode observer is: in, This represents the input quantity in the k-th control cycle; This represents the estimated state variable for the k-th control period. Represents the matrix of a discrete system. Represents the discrete control matrix. This represents the actual output of the motor model in the k-th control cycle. This represents the estimated output for the k-th control cycle; Represents the Luneburg gain coefficient matrix. Represents the sliding mode gain coefficient matrix. Represents a saturation function. This represents the discrete output matrix.

[0011] The input quantity Including the motor Rotating coordinate system of axis shaft voltage , motor Rotating coordinate system of axis shaft voltage and motor load torque ; The estimated state variables Including the motor Estimation of axis-rotating coordinate system shaft current , motor Estimation of axis-rotating coordinate system shaft current And the estimated mechanical angular velocity of the motor ; The actual output quantity and estimated output These refer to the motor in Rotating coordinate system of axis shaft current and shaft current sum and its estimated value .

[0012] In addition to the aforementioned Luneburg-sliding mode observer: the input quantity of the current k-th control cycle. The actual output of the motor model in the k-th control cycle is obtained by inputting the motor inverter system control processing. ; In the aforementioned Luneburg-sliding mode observer: the estimated state variable for the current k-th control cycle. After discrete output matrix Multiplication yields the estimated output of the current k-th control cycle generated internally by the observer. From the actual output and the estimated output generated internally by the observer The subtraction is fed into the error feedback module, and then the input value of the current k-th control cycle is added. Discrete control matrix The result of the multiplication, the estimated state variables for the current k-th control period Matrix of discrete system The result of the multiplication is added together with the output of the error feedback module to obtain the estimated state variables for the next control cycle, i.e., the (k+1)th control cycle. Then after the delay function The delayed operation processing feedback obtains the estimated state variables for the current k-th control cycle. .

[0013] The error feedback module described above converts the actual output quantity... and estimated output The difference between the subtraction and the sliding mode gain are calculated separately and then superimposed to obtain the result.

[0014] From the estimated state variables The estimated mechanical angular velocity of the motor is obtained by extracting it. And will estimate the mechanical angular velocity The estimated machine angle is obtained through integration. .

[0015] In the Romberg sliding mode observer, the actual output quantity With feedback output Obtain the error signal by subtraction. - And obtain the error signal - Multiply by the Lumberjack gain coefficient matrix The state variables are fed back to the observer's input side for state estimation, pending the acquisition of the estimated state variables again. Then, the estimated mechanical angular velocity Integrating is performed to obtain the estimated machine angle. .

[0016] Compared to the switching function in traditional sliding mode observers This paper introduces a saturation function in the sliding mode gain calculation of the error feedback module. Using saturation function Compensation for mechanical angular velocity estimation error in output quantity error .

[0017] In obtaining estimated state variables Estimated mechanical angular velocity Then, integration can be performed to obtain the estimated location information. However, pure integration can easily lead to cumulative integration errors, and the Romberg observer itself is sensitive to motor parameters. Therefore, a sliding mode saturation function is introduced into the feedback term. It can compensate for estimation errors and solve the problem of inaccurate estimation.

[0018] The saturation function Specifically: .

[0019] The Lumberjack-sliding mode observer ultimately outputs the estimated mechanical angular velocity of the motor. Estimated mechanical angles of the motor Ultimately, the estimated mechanical angular velocity of the motor will be determined. Feedback is sent to the input of the velocity loop, which has both a setpoint and a feedback input to estimate the mechanical angular velocity. It is the input terminal corresponding to the feedback quantity input to the speed loop; The estimated mechanical angle of the motor Feedback to the motor In the coordinate transformation matrix between the axis rotating coordinate system and the three-phase coordinate system, sensorless control is achieved.

[0020] The speed loop uses a PI controller.

[0021] When constructing the state observer in this invention, the selected observation equation is as follows: The current equations and motion equations of the axis rotating coordinate system differ from traditional observer design schemes. Instead of estimating back EMF information, they directly estimate the motor speed, enabling closed-loop start-up of the observer for wind turbine-type loads.

[0022] This invention replaces the stator coordinate system current equation in the Luneburger observer with the rotating coordinate system current equation, and introduces the motor motion equation to construct a system based on... , and For a full-order Luneburger observer with state variables, in order to address the instability caused by the parameter sensitivity of the Luneburger observer, a sliding mode saturation function is introduced into the observer feedback term to compensate for the speed estimation error.

[0023] This invention estimates the motor speed by constructing a Luneburg state observer. The specific steps are as follows: First, construct the full-order Romberg state equations as follows: In the formula, , For the motor in Rotating coordinate system of axis shaft current and shaft current, Let be the mechanical angular velocity of the motor rotor. , These are the phase resistance and phase inductance of the motor stator, respectively. For rotor flux linkage, This represents the number of pole pairs of the motor. , These are the motor load viscosity coefficient and moment of inertia, respectively. Since the above state equations are not linear state equations, they cannot be directly applied using mature linear system theory. Therefore, they are piecewise linearized to obtain linear full-order Romberg state equations. In the formula, , These are the d-axis currents. q-axis current The measured value; By discretizing the above linear full-order state equations, we can obtain the discrete linear full-order Luenberger state equations. Furthermore, by introducing Luenberger and sliding mode feedback terms, we can obtain the linear full-order Luenberger-sliding mode observer. In the above formula , and These refer to the d-axis current respectively. , shaft current and the mechanical angular velocity of the motor rotor The estimated value, and These refer to the (k+1)th control cycle and the kth control cycle, respectively. shaft current estimate Refers to the system sampling period. , , Refers to the Luneburg gain coefficient. , , Refers to the sliding mode gain coefficient.

[0024] The above formula can be simplified as: The beneficial effects of this invention are: This invention is essentially a sensorless control algorithm, which avoids the cost and reliability problems associated with position encoders; Traditional Luneburg observers are simple in structure, provide good speed estimation, and are free from high-frequency jitter, but they are sensitive to motor parameters and have low system robustness. This invention integrates the Luneburg observer and the sliding mode observer, and enhances the robustness of the observer to motor parameters by introducing a sliding mode saturation function into the observer structure. Traditional sliding mode observers are robust, but their estimated rotational speed exhibits high-frequency jitter, which affects control performance to some extent. This invention introduces a sliding mode saturation function, rather than a switching function, which reduces system jitter input to a certain degree. Attached Figure Description

[0025] Figure 1 This is a block diagram of the overall structure controlled by the present invention; Figure 2 This is a block diagram of the Luneburg-sliding mode observer of the present invention in discrete state. Detailed Implementation

[0026] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0027] The embodiments of the present invention and their implementation process are as follows: The method of this invention is used for a surface-mounted high-speed permanent magnet synchronous motor with a fan load driven by a three-phase inverter. The overall control block diagram is as follows: Figure 1 As shown.

[0028] The current equation for a permanent magnet synchronous motor in a two-phase rotating coordinate system is: (1) By introducing the motion equations of the permanent magnet synchronous motor into the above equation, we can obtain the state equations: (2) The second term on the right side of the equation, i.e. and Because there is cross-coupling between current and rotational speed, the above equation is a nonlinear state equation and cannot be expressed as a standard linear state equation.

[0029] Therefore, mature linear system theory cannot be used directly. However, by using piecewise linearization, equation (2) can be made into a linear system within a certain region, and linear system theory can then be applied.

[0030] In practical permanent magnet synchronous motor control systems, the AC and DC axis currents , It can be measured using a current sensor, so , Equation (2) can be obtained through measurement and is transformed into: (3) In the formula, , This represents the measured direct and quadrature axis currents. Discretizing the above linear full-order state equations yields the discrete linear full-order Luenberger state equations. Introducing Luenberger and sliding mode feedback terms, we obtain the linear full-order Luenberger-sliding mode observer as follows: (4) The above formula can be expressed as: (5) In the formula , , , , , .

[0031] Based on equation (6), the rotational speed estimation equation for the Luneburg observer can be obtained as follows: (6) In the formula , They represent Shaft current estimation error , .

[0032] The motor speed can be estimated from equation (6), and the motor rotor position information can be obtained by superimposing the estimated speed information: (7) Once the motor rotor position information is obtained, it can be applied to coordinate transformation to ensure stable motor operation.

[0033] This invention is a sensorless control algorithm that can save the cost of adding an encoder in practical applications, while also having higher reliability and no physical losses. At the same time, it samples the Luneburg observer frame, avoiding the problems of high-frequency jitter and phase delay of sliding mode observers. The introduction of a sliding mode saturation function into the Luneburg observer feedback term improves the robustness of the system to changes in motor parameters.

Claims

1. A sensorless motor control method based on a Luenberger-sliding mode observer, characterized in that: Establish a Romberg-sliding mode observer to measure the actual output in the k-th control cycle. and the input quantity in the kth control cycle The input is processed in real time in the Luneburg-sliding mode observer to obtain the estimated mechanical angular velocity of the motor rotor. and estimating mechanical angles This, in turn, provides feedback on the motor rotor position information; The expression for the Lumberjack-sliding mode observer is: In the above formula , and These refer to the d-axis current respectively. , shaft current and the mechanical angular velocity of the motor rotor The estimated value, and These refer to the (k+1)th control cycle and the kth control cycle, respectively. shaft current estimate Refers to the system sampling period. , , Refers to the Luneburg gain coefficient. , , Refers to the sliding mode gain coefficient; The above formula can be simplified as follows: in, This represents the input quantity in the k-th control cycle; This represents the estimated state variable for the k-th control period. Represents the matrix of a discrete system. Represents the discrete control matrix. This represents the actual output of the motor model in the k-th control cycle. This represents the estimated output for the k-th control cycle; Represents the Luneburg gain coefficient matrix. Represents the sliding mode gain coefficient matrix. Represents a saturation function. Represents a discrete output matrix; The saturation function Specifically: ； The rotational speed estimation equation for the Luneburger observer is: In the formula , They represent Shaft current estimation error , ; For rotor flux linkage, This represents the number of pole pairs of the motor. , These are the motor load viscosity coefficient and moment of inertia, respectively. The motor speed is estimated from this, and the motor rotor position information can be obtained by superimposing the estimated speed information: 。 2. The sensorless motor control method based on a Luenberger-sliding mode observer according to claim 1, characterized in that: The input quantity Including the motor Rotating coordinate system of axis shaft voltage , motor Rotating coordinate system of axis shaft voltage and motor load torque ; The estimated state variables Including the motor Estimation of axis-rotating coordinate system shaft current , motor Estimation of axis-rotating coordinate system shaft current And the estimated mechanical angular velocity of the motor ; The actual output quantity and estimated output These refer to the motor in Rotating coordinate system of axis shaft current and shaft current sum and its estimated value .

3. The sensorless motor control method based on a Luenberger-sliding mode observer according to claim 1, characterized in that: In addition to the aforementioned Luneburg-sliding mode observer: the input quantity of the current k-th control cycle. The actual output of the motor model in the k-th control cycle is obtained by inputting the motor inverter system control processing. ; In the aforementioned Luneburg-sliding mode observer: the estimated state variable for the current k-th control cycle. After discrete output matrix Multiply to obtain the estimated output for the current k-th control cycle. From the actual output and estimated output The subtraction is fed into the error feedback module, and then the input value of the current k-th control cycle is added. Discrete control matrix The result of the multiplication, the estimated state variables for the current k-th control period Matrix of discrete system The result of the multiplication is added together with the output of the error feedback module to obtain the estimated state variables for the next control cycle, i.e., the (k+1)th control cycle. Then after the delay function The delayed operation process obtains the estimated state variables for the current k-th control cycle. .

4. The sensorless motor control method based on a Luenberger-sliding mode observer according to claim 3, characterized in that: From the estimated state variables The estimated mechanical angular velocity of the motor is obtained by extracting it. And will estimate the mechanical angular velocity The estimated machine angle is obtained through integration. .

5. A sensorless motor control method based on a Luenberger-sliding mode observer according to claim 1 or 4, characterized in that: The Lumberjack-sliding mode observer ultimately outputs the estimated mechanical angular velocity of the motor. Estimated mechanical angles of the motor Ultimately, the estimated mechanical angular velocity of the motor will be determined. Feedback is sent to the input of the speed loop; the estimated mechanical angle of the motor is then fed back. Feedback to the motor In the coordinate transformation matrix between the axis rotating coordinate system and the three-phase coordinate system, sensorless control is achieved.

Images