RBFNN-based ship permanent magnet propulsion motor speed pulsation suppression method

Abstract

The invention provides an RBFNN-based ship permanent magnet propulsion motor speed pulsation suppression method, and belongs to the technical field of motor drive control systems. The method comprises the following steps: constructing a permanent magnet propulsion motor vector control system model containing a phase current offset error; designing a radial basis function neural network harmonic suppressor; and connecting the radial basis function neural network harmonic suppressor with a rotating speed ring PI controller of the permanent magnet propulsion motor vector control system in parallel. According to the method, the radial basis function neural network harmonic suppressor is used for compensating rotation speed pulsation caused by current offset errors; even under the condition that the disturbance frequency is unknown, the method has high nonlinear fitting capability and good speed tracking performance and anti-interference capability.

Description

Technical Field

[0001] The present invention relates to the technical field of motor drive control systems, and particularly to a method for suppressing the speed ripple of a ship permanent magnet propulsion motor based on an RBFNN (Radial Basis Function Neural Network). Background Art

[0002] Due to advantages such as high efficiency and high power density, permanent magnet synchronous motors are widely used as ship permanent magnet propulsion motors; during the navigation of a ship, it will face complex working environments such as bad weather, diving areas, narrow channels, and complex working conditions such as propulsion motor failures, steering gear failures, and navigation equipment failures. Non-ideal factors such as equipment failures and complex working environments will cause the ship propulsion motor to be affected by various periodic disturbances. For example, the propeller, as the load of the propulsion motor, will be affected by periodic pulsations when it is in a non-uniform flow field, the non-linear characteristics of the inverter will generate 3rd, 5th, and 7th harmonics, and current measurement errors will cause the generation of 1st and 2nd harmonics. These periodic harmonics will have an adverse impact on the system, resulting in speed fluctuations and a decrease in system stability. In the vector control of a ship permanent magnet propulsion motor, the accuracy of current measurement will directly affect the control accuracy of the motor drive control system. However, in an actual motor control system, due to factors such as the aging, noise, equipment tolerance, and temperature drift of current sensors and components in related circuits, current measurement errors will inevitably be introduced during the sampling process of the drive system. In the vector control system of a ship propulsion motor, the phase current offset error is transformed to the synchronous rotating coordinate system through coordinate transformation, resulting in a pulsating component related to the speed in the q-axis current, thereby causing a 1st pulsating component in the steady-state torque and steady-state speed, which affects the navigation safety of the ship.

[0003] Therefore, a method for suppressing the speed ripple of a ship permanent magnet propulsion motor is needed. Summary of the Invention

[0004] In view of this, the present invention provides a method for suppressing the speed ripple of a ship permanent magnet propulsion motor based on an RBFNN (Radial Basis Function Neural Network). By connecting a radial basis function neural network harmonic suppressor in parallel with the speed loop PI controller, the speed ripple of the ship permanent magnet propulsion motor is suppressed.

[0005] For this purpose, the present invention provides the following technical solutions:

[0006] A method for suppressing the periodic speed ripple of a ship permanent magnet propulsion motor based on an RBFNN includes:

[0007] Constructing a vector control system model of a permanent magnet propulsion motor including a phase current offset error;

[0008] Designing a radial basis function neural network harmonic suppressor;

[0009] Connect the radial basis function neural network harmonic suppressor in parallel with the speed loop PI controller of the permanent magnet propulsion motor vector control system.

[0010] Furthermore, the permanent magnet propulsion motor vector control system model includes: a speed loop PI controller, a current loop PI controller, a coordinate transformation module, a pulse width modulation module, a three-phase inverter, a permanent magnet propulsion motor, and a sensor.

[0011] Furthermore, the output of the radial basis function neural network harmonic suppressor is used as compensation for the speed disturbance of the permanent magnet propulsion motor.

[0012] Furthermore, the activation function of the radial basis function neural network harmonic suppressor:

[0013]

[0014] The output of the radial basis function neural network harmonic suppressor:

[0015]

[0016] where θ is the rotor position; c j is the coordinate vector of the center point of the Gaussian basis function of the j-th neuron in the hidden layer; b j is the width of the Gaussian basis function of the j-th neuron in the hidden layer; is the neural network weight; is the transpose matrix of; neural network output.

[0017] Furthermore, the inputs of the radial basis function neural network harmonic suppressor include: rotor position and speed error.

[0018] Furthermore, connect the radial basis function neural network harmonic suppressor and the speed loop PI controller of the permanent magnet propulsion motor vector control system through the Lyapunov stability criterion.

[0019] Advantages and positive effects of the present invention:

[0020] The present invention uses the rotor position as the input of the radial basis function neural network harmonic suppressor to suppress speed harmonics, reducing the dependence of the system on parameters. And based on the Lyapunov stability theory, the speed loop PI controller is combined with the radial basis function neural network, and the radial basis function neural network harmonic suppressor is used to compensate for the speed pulsation caused by the current offset error, realizing strong non-linear fitting ability, good speed tracking performance and anti-disturbance ability even when the disturbance frequency is unknown. Description of the Drawings

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0022] Figure 1 It is a block diagram of a vector control system for a permanent magnet propulsion motor with a speed loop PI controller in parallel with a radial basis function neural network harmonic suppressor in an embodiment of the present invention;

[0023] Figure 2 It is a block diagram of a vector control system for a permanent magnet propulsion motor including phase current offset error;

[0024] Figure 3 It is a structure diagram of a radial basis function neural network in an embodiment of the present invention;

[0025] Figure 4 It is the speed waveform when using a PI controller in the presence of current offset error;

[0026] Figure 5 It is a Fourier analysis diagram of the harmonic content of the steady-state speed;

[0027] Figure 6 It is the speed waveform when using a speed loop PI controller in parallel with a radial basis function neural network harmonic suppressor in the presence of current offset error in an embodiment of the present invention;

[0028] Figure 7 It is a Fourier analysis diagram of the harmonic content of the steady-state speed after the speed loop PI controller is in parallel with the radial basis function neural network harmonic suppressor in an embodiment of the present invention. Detailed implementation manners

[0029] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0031] The present invention provides a method for suppressing the speed ripple of a ship permanent magnet propulsion motor based on an RBFNN. Based on the advantages of the radial basis function neural network, such as simple structure, fast learning speed, excellent approximation performance and generalization ability, the radial basis function neural network is used as a feedforward compensator to suppress the speed ripple of the propulsion motor. That is, a radial basis function neural network harmonic suppressor is constructed; and the radial basis function neural network harmonic suppressor is connected in parallel with the speed loop PI controller to suppress the influence of periodic harmonic disturbances of the permanent magnet propulsion motor drive system.

[0032] As Figure 1 shown, the method of the present invention is further described as follows:

[0033] Step 1: Establish a vector control system model of a permanent magnet propulsion motor including phase current offset error;

[0034] Figure 2 The vector control system model of the permanent magnet propulsion motor considering the phase current offset error includes: a speed loop PI controller, a current loop PI controller, a coordinate transformation module, a (Pulse width modulation, PWM) pulse width modulation module, a three-phase inverter, a permanent magnet propulsion motor and a sensor. Where ΔI A_ o ffset and ΔI B_offset are the current measurement offset errors of phases A and B respectively.

[0035] Step 2: Determine the first-order pulsation of the motor's steady-state speed caused by the phase current offset error according to the electromagnetic torque equation and mechanical motion equation of the motor:

[0036]

[0037] In the formula, P represents the number of pole pairs of the motor; λ r represents the permanent magnet flux linkage; i d , i q represent the current values on the d-q axes; L d, L q represents the d-q axis inductance value; T e is the electromagnetic torque:

[0038]

[0039] Derive the primary pulsation equation of the motor's steady-state speed caused by the phase current offset errors ΔI A_offset , ΔI B_offset The specific derivation process is as follows:

[0040] For a motor drive system using two current sensors, the three-phase current measurement values are expressed as:

[0041]

[0042] where, i A , i B represent the actual values of the A and B phase currents; ΔI A_o ffset , ΔI B_offset represent the current offset errors.

[0043] The three-phase currents are successively transformed through the Clark transformation and the Park transformation to the d-q axis currents in the synchronous rotating coordinate system; when the three-phase current measurement values contain current offset errors, the d-q axis currents obtained through the coordinate transformation will also contain measurement errors:

[0044]

[0045] In the formula, i d , i q represent the true values; Δi d_offset , Δi q_offset represent the error values; through the measurement errors of the A and B phase currents ΔI A_o ffset , ΔI B_offset calculate to obtain:

[0046]

[0047] In the formula, θ is the rotor position.

[0048] Express Equation (5) in the form of a sine function:

[0049]

[0050] where:

[0051]

[0052] For i d=0 vector control system. At steady state, the d-axis current follows the given value and is always 0. The electromagnetic torque equation of the permanent magnet propulsion motor is calculated by the following formula:

[0053]

[0054] In the formula, P represents the number of pole pairs of the motor; λ r represents the magnetic flux linkage of the permanent magnet; the first term represents the true value of the electromagnetic torque; the second term represents the difference caused by the current offset error.

[0055] Substituting the q-axis current error in Equation (6) gives:

[0056]

[0057] Ignoring the influence of the damping coefficient, the motion equation of the permanent magnet propulsion motor is:

[0058]

[0059] Among them, T L represents the load torque of the motor, which is a constant; J represents the moment of inertia; ω m represents the angular velocity;

[0060] Substituting Equation (9) into (10) gives the speed error:

[0061]

[0062] Observing Equations (6), (10), and (11), it can be found that when the current offset error is included in the vector control system of the permanent magnet propulsion motor, a periodic harmonic with a fundamental frequency of 1 will be generated in the q-axis current, resulting in a periodic harmonic with a fundamental frequency of 1 in the electromagnetic torque and the motor speed.

[0063] Step 3: Construct a harmonic suppressor based on a three-layer radial basis function neural network;

[0064] S31. The activation function of the neural network is:

[0065]

[0066] The output of the neural network is:

[0067]

[0068] In the formula, θ is the rotor position; cj is the coordinate vector of the center point of the Gaussian function of the jth neuron in the hidden layer; b j is the width of the Gaussian function of the jth neuron in the hidden layer; is the weight of the neural network; is the transpose matrix of; Neural network output.

[0069] The input signals of the neural network include the rotor position signal and the speed error, and its structure is as Figure 3 shown, where θ is the rotor position, and the range is within [0, 2π]; therefore, the centers of the Gaussian functions can be directly selected equidistantly within [0, 2π]. To accelerate the convergence speed of the neural network, the width b is fixed at 0.6, and the number of neural networks is 40.

[0070] Step 4: Connect the radial basis function neural network harmonic suppressor in parallel to the speed loop PI controller:

[0071] Connect the radial basis function neural network harmonic suppressor with the input of speed error and rotor position in parallel to the speed loop PI controller, and the output of the harmonic suppressor is used as the compensation for the speed disturbance of the permanent magnet propulsion motor.

[0072] Step 5: Prove the stability of the system using the Lyapunov stability criterion;

[0073] Since the mapping from the input layer to the hidden layer of the radial basis function neural network is linear and the mapping from the hidden layer to the output layer is non-linear, the stability of the system cannot be proved by the transfer function method, and the Lyapunov stability criterion is used to prove the stability of the system.

[0074] The mechanical motion equation of the permanent magnet propulsion motor is:

[0075]

[0076] In the formula, ω m is the angular velocity; T e and T e * are the electromagnetic torque and its reference value; J is the moment of inertia; B is the viscous friction torque coefficient; T L is the load torque; is the control gain; T d = T e * - T e + Bω m + T L T d is the total disturbance torque;

[0077]

[0078] Among them, d ap = -[(b n - b)T e * + bT d is the non-periodic disturbance; and Jn are the nominal control gain and the nominal moment of inertia, respectively; d rip is a periodic disturbance.

[0079] The derivative of the speed tracking error can be obtained according to as follows:

[0080]

[0081] The convergence law of the tracking error is expressed as:

[0082]

[0083] where K p > 0 is the proportional gain.

[0084] Substituting (17) into (16) gives:

[0085]

[0086] The actual values of the aperiodic and periodic disturbances of the system are unknown. The aperiodic and periodic disturbance observation components and are used to replace the aperiodic and periodic disturbances, obtaining:

[0087]

[0088] The actual periodic disturbance is written in the format of a neural network:

[0089] d rip = W T h - ε(20)

[0090] |ε| ≤ ε N , ε N is the upper limit of the approximation error, which is a known constant, meaning that the approximation error is bounded and reasonable for the control system.

[0091]

[0092] where is the weight vector updated online, is the aperiodic disturbance observation error.

[0093] The adaptive law is designed as:

[0094]

[0095] where K I > 0 and η > 0 are two gains.

[0096] Select the Lyapunov function:

[0097]

[0098] Differentiate the selected function:

[0099]

[0100] When the neural network completes the training process, the optimal weight vector should be a constant, d ap is also a constant or slowly varying quantity. Therefore, their derivatives can be regarded as zero, that is: and Substituting (21), (22), and (23) into (25) gives:

[0101]

[0102] According to Young's inequality, we have:

[0103]

[0104] Substituting (27) into (26) gives:

[0105]

[0106] To

[0107] Assumption 1: Consider the dynamic variation (15) of a PMSM (Permanent Magnet Synchronous Motor) with bounded initial conditions. By applying the speed controller (19) and the disturbance observation laws (22), (23), the signals ε, The closed-loop system is semi-globally uniformly ultimately bounded (SGUUB).

[0108] Proof: According to equations (24) and (28), it can be further expanded as:

[0109]

[0110] where to ensure that α > 0.

[0111] Multiply both sides of equation (29) by e αt , and integrate over the interval [0, t] to obtain:

[0112]

[0113] According to equation (30), it can be obtained that V is bounded, and ε, It is SGUUB, that is, the speed tracking error and the disturbance observation error asymptotically converge to a small neighborhood of zero. Therefore, the proposed speed control system remains stable.

[0114] The effectiveness of this method is verified by the following simulation:

[0115] The simulation parameters are: the rated power P of the permanent magnet propulsion motor N = 200W, the rated torque T N = 0.14 N·m, the number of pole pairs P = 4, the stator resistance R = 0.36 Ω, the inductance L = 0.201 mH, the rotor magnetic flux ψ r = 0.00655 Wb, the given speed is 300 rpm, and the offset errors of the added phase A and phase B currents are 0.1 A and 0.15 A respectively.

[0116] If the learning rate is too small, the system will not converge or the convergence speed will be slow. If the learning rate is too large, the system will overfit. In this simulation, the learning rate η of the system is set to 0.8.

[0117] At 10 s, the current error is added, and the steady-state speed waveform of the motor is as Figure 4 shown. Fourier analysis is performed on it, and the harmonic content is as Figure 5 shown; combining Figure 4 and Figure 5 it can be seen that when there is an offset error in the current link, there is an obvious first-order fluctuation in the speed. The first-order pulsation content is 0.76%, and the amplitude is 2.25 rpm.

[0118] When the RBFNN harmonic suppressor is connected in parallel with the PI controller, the steady-state speed waveform of the motor is as Figure 6 shown. Fourier analysis is performed on it, and the harmonic content is as Figure 7 shown; combining Figure 6 and Figure 7 it can be seen that after adding the RBFNN harmonic suppressor, there is almost no first-order fluctuation in the steady-state speed, and the first-order pulsation content is 0.12%. Thus, the method of the present invention is effective and accurate in suppressing the first-order pulsation of torque speed.

[0119] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for suppressing the periodic speed pulsation of a ship's permanent magnet propulsion motor based on RBFNN, characterized in that, Including: Construct a vector control system model of a permanent magnet propulsion motor including phase current offset error; Design a radial basis function neural network harmonic suppressor; Connect the radial basis function neural network harmonic suppressor in parallel with the speed loop PI controller of the permanent magnet propulsion motor vector control system.

2. The periodic rotational speed pulsation suppression method for a ship permanent magnet propulsion motor based on RBFNN according to claim 1, wherein The permanent magnet propulsion motor vector control system model includes: a speed loop PI controller, a current loop PI controller, a coordinate transformation module, a pulse width modulation module, a three-phase inverter, a permanent magnet propulsion motor, and a sensor.

3. A method for suppressing the periodic speed pulsation of a ship's permanent magnet propulsion motor based on RBFNN according to claim 1, characterized in that, The output of the radial basis function neural network harmonic suppressor is used as compensation for the speed disturbance of the permanent magnet propulsion motor.

4. A method for suppressing the periodic speed pulsation of a ship's permanent magnet propulsion motor based on RBFNN according to claim 3, characterized in that, The activation function of the radial basis function neural network harmonic suppressor: The output of the radial basis function neural network harmonic suppressor: where θ is the rotor position; c j is the coordinate vector of the center point of the Gaussian basis function of the j-th neuron in the hidden layer; b j is the width of the Gaussian basis function of the j-th neuron in the hidden layer; is the neural network weight; is the transpose matrix of; the neural network output.

5. A method for suppressing the periodic speed pulsation of a ship's permanent magnet propulsion motor based on RBFNN according to claim 3, characterized in that, The inputs of the radial basis function neural network harmonic suppressor include: rotor position and speed error.

6. The method for suppressing the periodic speed pulsation of a ship's permanent magnet propulsion motor based on RBFNN according to claim 1, characterized in that, Connect the radial basis function neural network harmonic suppressor and the speed loop PI controller of the permanent magnet propulsion motor vector control system in parallel through the Lyapunov stability criterion.

Images