Magnetic Levitation Control System and Control Method

Abstract

A magnetic levitation control system and a control method are disclosed. The magnetic levitation control system includes a gap sensor, a chopper, a magnetic levitation controller, and a levitation electromagnet. The gap sensor is connected to the magnetic levitation controller through the chopper. The levitation electromagnet is connected to the magnetic levitation controller through peripheral hardware. The gap sensor is installed on the levitation electromagnet. The magnetic levitation controller includes an anti-saturation levitation controller.

Description

Technical Field

[0001] The present invention relates to a maglev train, and more particularly to a novel suspension control system and control method for an EMS maglev train. Background Art

[0002] A maglev train is a modern transportation mode with a non-contact electromagnetic suspension, guidance and drive system. It relies on electromagnetic attraction or repulsion to suspend the train in the air, so as to achieve no mechanical contact between the train and the track, and is driven by a linear motor. Maglev trains have become an ideal means of transportation due to their high speed, low energy consumption, comfortable ride and low noise. Currently, according to the different suspension principles and methods adopted by maglev vehicles, maglev trains are generally divided into two categories. One is Electrodynamic Suspension, abbreviated as EDS type; the other is Electromagnetic Suspension, abbreviated as EMS type. The EDS maglev system uses electromagnetic repulsion to suspend the vehicle above the track, while the EMS maglev system uses the attraction generated by an electromagnet located below the track to lift the vehicle to ensure non-contact with the track. The EDS maglev system can stably suspend without applying control, while the EMS maglev system needs to apply active control to ensure stable suspension. Currently, all commercially operated maglev trains are of the EMS type.

[0003] For an EMS maglev train, the suspension system is the key and core. However, the suspension system has strong nonlinearity and open-loop instability. In addition, the system parameters are uncertain, and the system will be subject to external interference during operation. Therefore, it poses a high challenge to the design of a high-performance suspension controller.

[0004] The most urgent problems faced by the suspension system of an EMS maglev train are model uncertainties (such as wind load, track irregularity, number of passengers, etc.) and actuator output saturation. Currently, most maglev vehicle suspension controllers are linear PID controllers. It cannot ensure that the controller output is always within the saturation range. When the controller output exceeds the saturation range, the actuator cannot provide the required control amount, the system control performance will decrease, and the system stability will decline or even be lost. Summary of the Invention

[0005] The present invention discloses a maglev control system and control method for a maglev train. Compared with the prior art, the controller in the present invention can ensure that the control output does not exceed the saturation range, has a small static error, strong anti-interference ability, and has broad application value and commercial promotion value.

[0006] According to one aspect of the present invention, a magnetic levitation control system is provided. The magnetic levitation control system includes a gap sensor, a chopper, a magnetic levitation controller, and a levitation electromagnet. The gap sensor is connected to the magnetic levitation controller through the chopper. The levitation electromagnet is connected to the magnetic levitation controller through peripheral hardware. The gap sensor is installed on the levitation electromagnet. It is characterized in that the magnetic levitation controller includes an anti-saturation levitation controller.

[0007] According to another aspect of the present invention, a magnetic levitation control method for a maglev train is provided. The method includes: providing a magnetic levitation control system for the maglev train, the magnetic levitation control system including a gap sensor, a chopper, a magnetic levitation controller, and a levitation electromagnet, the gap sensor being connected to the magnetic levitation controller through the chopper, the magnetic levitation controller being connected to a computer device, the levitation electromagnet being connected to the magnetic levitation controller through peripheral hardware, the gap sensor being installed on the levitation electromagnet; using the gap sensor to measure and collect suspension distance data in real time; performing analog-to-digital conversion and filter modulation on the suspension distance data; transmitting the suspension distance data to the computer device connected to the magnetic levitation controller through a communication line; enabling the computer device carrying the magnetic levitation controller to output the control quantity obtained by the magnetic levitation controller to the peripheral hardware, thereby driving the levitation electromagnet to work and levitating the train. The magnetic levitation controller includes an anti-saturation levitation controller. Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments recorded in the present application, and are not limitations for those skilled in the art or ordinary technicians.

[0009] Figure 1 Schematic diagram of the suspension system structure of a suspension controller for an EMS maglev train according to the present invention;

[0010] Figure 2 Schematic diagram of the control flow of a suspension controller for an EMS maglev train according to the present invention;

[0011] Figure 3 Schematic diagram of the suspension control dynamic model of an EMS maglev train according to the present invention;

[0012] Figure 4 Schematic diagram of the control system of a suspension controller for an EMS maglev train according to the present invention;

[0013] Figure 5 Schematic diagram of the RBF neural network structure of a suspension controller for an EMS maglev train according to the present invention;

[0014] Figure 6 This is a schematic diagram of the RBF neuron model structure of a suspension controller for an EMS maglev train according to the present invention. Detailed implementation manner

[0015] Magnetic levitation control system

[0016] Figure 1 A maglev control system 10 according to an embodiment of the present invention is shown. As shown in the figure, the maglev control system 10 includes a gap sensor 12, a chopper 14, a maglev controller 16, and a suspension electromagnet 18. The gap sensor 12 is connected to the maglev controller 16 through the chopper 14. The suspension electromagnet 18 is connected to the maglev controller 16 through peripheral hardware. The gap sensor 12 is installed on the suspension electromagnet 18. The gap sensor 12 is connected to the chopper 14 through a gap processing board 242, a control board 244, and an interface conversion board 246. The maglev controller 16 includes computer hardware and algorithm software. The algorithm software of the maglev controller 16 is compiled and stored in the computer hardware. The suspension electromagnet 18 is installed on the vehicle body 20 such that the vehicle body 20 floats on the track 22.

[0017] The maglev controller 16 can include any suitable controller, such as a differential controller, an integral controller, and so on. In one embodiment, the maglev controller 16 includes an anti-saturation suspension controller. The anti-saturation suspension controller is executed based on the error system e and the maglev minimum model according to the following controller model:

[0018]

[0019] Wherein, the error system e is the difference between the actual suspension gap and the ideal suspension distance, u(t) is the control quantity, which is a variable of time t and refers to the output value of the controller. The maglev train adjusts the actual suspension gap closer to the ideal gap based on this output value; m is the mass of the train, g is the acceleration due to gravity, and mg is the nominal suspension weight. The control quantity u(t) can be any suitable control quantity. According to an embodiment of the present invention, the control quantity u(t) is the electromagnetic suction force, that is, u(t) = F m .

[0020] At the same time, the current required for the electromagnet coil is:

[0021]

[0022] Wherein, z is the suspension distance, k p , k d The value of needs to satisfy:

[0023]

[0024]

[0025] Among them, F max is the maximum electromagnetic suction force. The derivation of the above controller model will be described in detail later.

[0026] When the maglev train is running, it will inevitably be disturbed by various other factors, such as wind force, track deformation, signal transmission feedback delay, etc. In order to further improve the performance of the above anti-saturation suspension controller, a control compensation term can also be introduced to the above control quantity. That is,

[0027] u all (t) = u(t) + U(t)

[0028] Among them, u all (t) is the total control quantity, u(t) is the control quantity of the anti-saturation suspension controller, and U(t) is the compensation term.

[0029] According to an embodiment of the present invention, the control compensation term can be an RBF enhanced control term based on the RBF neural network. In this embodiment, when the magnetic suspension controller 16 works, it is responsible for executing the operation of the neural network approximation algorithm, obtaining the set physical parameters of the suspension system input, and obtaining the gap data between the track and the car body in real time and calculating and controlling the output control signal.

[0030] For the RBF enhanced control term, the compensation term U(t) is:

[0031]

[0032] Among them, the network weight matrix of the output layer is w = [w1, w2,..., w M .

[0033]

[0034] In the formula, R j (x) is the output of the j-th node in the hidden layer, X = [x1, x2,..., x n T is the input vector of the network, is the coordinate matrix of the center points of the Gaussian basis functions of the hidden layer neurons, σ = [σ1, σ2,..., σ M is the width vector of the Gaussian basis functions, σ j (j = 1, 2,..., M) is the width of the j-th hidden layer neuron, and N is the number of hidden layer nodes.

[0035] ​The working principle of this embodiment is as follows: The gap sensor 12 measures and collects suspension gap data in real time, at high speed, and continuously. After being converted by analog-to-digital conversion and filtered and modulated, the suspension gap data is transmitted to the computer device carrying the maglev controller 16 through a communication line. The computer device of the maglev controller 16 outputs the control quantity obtained by the maglev controller 16 to the peripheral hardware to drive the suspension electromagnet 18 to work and suspend the train car body 20. The error between the suspension gap and the target position is calculated by the maglev controller 16 to obtain the control output quantity. With the real-time and continuous operation of the gap sensor 12, the computer device of the maglev controller 16, and the peripheral hardware of the suspension electromagnet 18, the train car body 20 will move to the target position within a limited time and remain within the range of the position error limit, achieving a stable and reliable suspension control effect.

[0036] Magnetic levitation control method

[0037] Based on the above maglev control system, the present invention also proposes a corresponding maglev control method. According to one embodiment, the maglev control method for a maglev train may include the following steps:

[0038] 1. Provide a maglev control system for the maglev train, where the maglev control system includes a gap sensor, a chopper, a maglev controller, and a suspension electromagnet. The gap sensor is connected to the maglev controller through the chopper. The maglev controller is connected to a computer device. The suspension electromagnet is connected to the maglev controller through peripheral hardware. The gap sensor is installed on the suspension electromagnet.

[0039] 2. Use the gap sensor to measure and collect suspension gap data in real time.

[0040] 3. Perform analog-to-digital conversion and filtering modulation on the suspension gap data.

[0041] 4. Transmit the suspension gap data to the computer device connected to the maglev controller through a communication line.

[0042] 5. Cause the computer device carrying the maglev controller to output the control quantity obtained by the maglev controller to the peripheral hardware, thereby driving the suspension electromagnet to work and suspending the train.

[0043] The maglev controller in the above steps can be configured according to the aforementioned maglev control system.

[0044] Construct the anti-saturation magnetic levitation controller model of the present invention and the RBF enhanced control term of the neural network

[0045] To deeply understand the aforementioned maglev control system and control method, the following is combined with Figures 2 - 6Briefly describe the process and method of constructing the anti-saturation magnetic suspension controller model of the present invention and the RBF enhanced control term of the neural network.

[0046] Figure 2 The process and method of constructing the above anti-saturation suspension controller model of the present invention and the RBF enhanced control term of the neural network are shown. The method includes steps S1 - S4. In step S1, establish the dynamic model of the maglev train suspension control, design the controller and analyze the stability of the system, and introduce the RBF supervisory control. In step S2, input the set physical parameters of the suspension system into the magnetic suspension controller. In step S3, enable the magnetic suspension controller to obtain the gap data between the track and the car body in real time and then output the control signal. In step S4, enable the peripheral hardware of the suspension system to receive the control signal, drive the suspension electromagnet to move to the target position within a limited time, and maintain it within the range of the position error limit.

[0047] Combined with Figure 3 , in the figure, z(t) is the suspension gap. There is a designed ideal value for the suspension gap. During the operation of the vehicle, the actual suspension gap fluctuates around the ideal value to ensure the smooth operation of the train. For most maglev vehicles in China, the ideal suspension gap between the vehicle and the track, that is, the target suspension distance, is usually 8 mm. The final controller model in step S1 is expressed as: for a fixed target suspension distance, based on the error system e (that is, the difference between the actual suspension gap and the ideal suspension distance) and the maglev minimum model, an anti-saturation suspension controller is proposed. Here, saturation refers to integral saturation. Integral saturation means that there is a deviation in one direction in the system, and the output of the controller expands due to the continuous accumulation of the integral action, resulting in the output of the controller continuously increasing beyond the normal range and entering the saturation region. When there is a reverse deviation in the system, it is necessary to first exit the saturation region and cannot respond quickly to the reverse deviation. Anti-saturation means avoiding the above situation. According to an embodiment, the anti-saturation suspension controller is as follows:

[0048]

[0049] where, u(t) represents the control quantity, which is a variable of time t and refers to the output value of the controller. The train adjusts the actual suspension gap closer to the ideal gap based on this output value; g is the acceleration due to gravity, and mg is the nominal suspension weight. At the same time, the current required for the electromagnet coil is:

[0050]

[0051] where, z is the suspension distance. To ensure that the controller always provides a force greater than 0 and operates within the amplitude saturation, k p , k d should satisfy:

[0052]

[0053]

[0054] The above method for constructing the final controller model includes: S11. Establish the mathematical model of the maglev train suspension controller to obtain the control current of the controller; S12. Introduce the RBF neural network.

[0055] In the above step S11, the maglev train suspension control mathematical model shows that within a certain range, it can be considered that each maglev train suspension controller is independent of each other, and the single-point suspension system can be regarded as the minimum dynamic model for controller design. As Figure 3 shown, the single-point suspension system consists of a single suspension electromagnet and coil, a rigid track, and a control input. In the figure, N m is the number of turns of the electromagnet coil, z(t) is the suspension gap, A m is the pole area of the electromagnet, F m (i m (t), x m (t)) represents the electromagnetic suction force, u m (t) and i m (t) represent the voltage and current of the electromagnet coil respectively, and mg is the nominal suspension weight.

[0056]

[0057] where z is the suspension gap, m is the mass, g is the acceleration due to gravity, and sat imax (F m ) is defined as:

[0058]

[0059] F max is the maximum electromagnetic suction force that the system can provide, which is mainly related to the maximum current provided by the chopper. F m represents the suction force generated by the electromagnet and can be expressed as:

[0060]

[0061] where, i(t) represents the current of the electromagnet coil; N is the number of turns of the electromagnet coil; A is the pole area of the electromagnet, and μ0 is the air permeability.

[0062] Define the electromagnetic suction force as the control quantity, then u(t) = F m . Since the suspension gap z(t) can be measured in real time by the sensor, as long as F m (t) is determined, the control current i(t) required by the system can be easily calculated by the following formula:

[0063]

[0064] Therefore, we can either use F m or use i(t) as the control input to be designed. The control input means: giving this value a limited range and keeping it within this range all the time to avoid saturation. However, when using F m as the control input, the one-way constraint must be ensured, that is, u(t) = F m .

[0065] If the amplitude of u(t) can be ensured to be always between [0, Fmax] through an appropriate controller, then u(t) (that is, F m ) will never saturate, and sat imax {μ(t)} = μ(t). Therefore, the saturation non-linearity can be removed by a bilateral constraint, and the dynamic equation is converted as follows:

[0066]

[0067] At this time, the control current required by the electromagnet can be conveniently calculated by the following formula:

[0068]

[0069] The anti-saturation suspension controller is proposed as follows:

[0070]

[0071] Meanwhile, the current required by the electromagnet coil is:

[0072]

[0073] where z is the suspension gap. To ensure that the proposed controller always provides a force greater than 0 and works within the amplitude saturation, the values of kp and kd need to satisfy:

[0074]

[0075]

[0076] The control objective of the maglev train is to keep the suspension gap z(t) at the target suspension gap. To quantitatively describe the control objective, the error signal is defined as:

[0077]

[0078] where z d is the target suspension gap, and is the first derivative and the second derivative of the target suspension distance. If z d is a constant value, then and are zero. Combining (2) and (4), the error system e can be derived as follows:

[0079]

[0080] To prove the stability, according to the designed controller (3) and the error system (5), construct a Lyapunov equation:

[0081]

[0082] Define an auxiliary function as follows:

[0083]

[0084] It is easy to obtain that E aux (0) = 0. Differentiating E aux with respect to e gives:

[0085]

[0086] Obviously, when e = 0, It can be clearly seen from (7) that when e > 0, When e < 0, Therefore, combining E aux (0) = 0, it can be concluded that:

[0087] Eaux(e) ≥ 0, that is: 2

[0088] Then formula (6) can be transformed into:

[0089]

[0090]

[0091] When and only when e = 0 and , V = 0; therefore V is a positive definite function.

[0092] Differentiating both sides of the equation of V with respect to time, we get:

[0093]

[0094] Because So Therefore, the maglev closed-loop system is Lyapunov stable, and

[0095]

[0096] To further prove the convergence of the error, define the set as follows

[0097]

[0098] Then define M as the largest invariant set contained in S. The following conclusions can be drawn in the M set:

[0099]

[0100] Combined with the foregoing, we have:

[0101]

[0102] Therefore, it can be concluded that M only contains equilibrium points Therefore, according to LaSalle’s invariance theorem, we can obtain

[0103]

[0104] Therefore, the asymptotic stability of the closed-loop system is proved.

[0105] Furthermore, in the step S12, the design and stability analysis of the proposed anti-saturation suspension controller are carried out under the assumption of no interference. However, the maglev train will inevitably be affected by various other factors during operation, such as wind force, track deformation, signal transmission feedback delay, etc. To further improve the performance of the proposed anti-saturation suspension controller, a control compensation term is proposed based on the RBF neural network.

[0106] There are many structures of artificial neural networks. Among them, the RBF neural network has received attention in control systems because of its good generalization ability, simple network structure, and avoidance of lengthy calculations. In this paper, the RBF neural network is selected to design the supervisory control structure. The RBF neural network can effectively improve the performance of the controller when the system has large uncertainties. As Figure 5 shown, the RBF neural network is a feedforward network with three layers of neurons. The input layer signal is X = [x1, x2, …, x n T 、the hidden layer signal is R = [R1, R2, …, R M , and the activation function of Rj (j = 1, 2, …, M) usually adopts the Gaussian basis function, which defines the random input vector X ∈ R (X is an input sample set) as:

[0107]

[0108] In the formula, R j (x) is the output of the jth node in the hidden layer, X = [x1, x2, …, x​n T is the input vector of the network, is the coordinate matrix of the centers of the Gaussian basis functions of the hidden layer neurons, σ = [σ1, σ2, …, σ M is the width vector of the Gaussian basis functions, σ j (j = 1, 2, …, M) is the width of the j-th hidden layer neuron, and N is the number of hidden layer nodes. The network weight matrix of the output layer is w = [w1, w2, …, w M , and the actual output signal obtained is:

[0109]

[0110] The hidden layer neuron model containing the Gaussian kernel function is as Figure 4 shown.

[0111] The RBF neural network compensation control is expected to gradually replace the anti-saturation suspension controller u(t) with the neural network output control quantity U(t) after online learning during the control process.

[0112] The overall controller u all (t) is designed as follows:

[0113] u all (t) = u(t) + U(t)

[0114] The schematic diagram of the controller structure is as Figure 4 shown.

[0115] According to the Least Mean Square (LMS) algorithm,

[0116]

[0117] According to the stochastic gradient descent method, the weights are adjusted in the following way

[0118]

[0119] w(N) = w(N - 1) + Δw(N) + ((N - 1) - (N - 2))

[0120] where, Δw j (N) is the iterative correction amount of w j , η ∈ [0, 1] is the learning rate, and α ∈ [0, 1] is the momentum factor. This embodiment is implemented through the cooperation of the software and hardware of the suspension system.

[0121] ​Although the present invention has been described above with reference to exemplary embodiments, the above embodiments are only for explaining the technical concept and features of the present invention, and cannot be used to limit the protection scope of the present invention. Any equivalent variation or modification made according to the spirit and essence of the present invention shall be covered by the protection scope of the present invention.

[0122] It should be understood that those of ordinary skill in the art can make many modifications and changes to the parameters without creative labor according to the concept of the present invention. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of the present invention through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art shall be within the protection scope determined by the claims.

Claims

1. A magnetic levitation control system, The magnetic levitation control system includes a gap sensor, a chopper, a magnetic levitation controller, and a levitation electromagnet. The gap sensor is connected to the magnetic levitation controller through the chopper. The levitation electromagnet is connected to the magnetic levitation controller through peripheral hardware. The gap sensor is installed on the levitation electromagnet. It is characterized in that The magnetic levitation controller includes an anti-saturation levitation controller. Wherein, the anti-saturation levitation controller controls the control quantity u(t) based on the error system e and the magnetic levitation mathematical model according to the following controller model: Wherein, the error system e is the difference between the actual levitation gap and the ideal levitation distance. m is the mass of the train, g is the acceleration due to gravity, and mg is the nominal levitation weight. At the same time, the current required for the electromagnet coil is: where z is the suspension distance, k p , k d shall satisfy: Among them, F max is the maximum electromagnetic suction force.

2. The magnetic levitation control system according to claim 1, wherein, The magnetic levitation controller further includes an RBF enhancement control term.

3. The magnetic levitation control system according to claim 2, wherein, The magnetic levitation controller is executed according to the following controller model: u all (t) = u(t) + U(t) where, u all (t) is the total control quantity, u(t) is the control quantity of the saturated suspension controller, and U(t) is the control quantity of the RBF enhancement control term. Among them, the network weight matrix of the output layer is w = [w1, w2, …, w M , Among them, R j (x) is the output of the j-th node in the hidden layer, X = [x1, x2, …, x n T is the input vector of the network, is the coordinate matrix of the centers of the Gaussian basis functions of the hidden layer neurons, σ = [σ1, σ2, …, σ M j is the width vector of the Gaussian basis functions, σ j (j = 1, 2, …, M) is the width of the j-th hidden layer neuron, and N is the number of hidden layer nodes.

4. The magnetic levitation control system according to claim 1, wherein The control quantity u(t) is the electromagnetic suction force, i.e., u(t) = F m .

5. The magnetic levitation control system according to claim 1, wherein, The gap sensor is connected to the chopper through a gap processing board, a control board, and an interface conversion board.

6. The magnetic levitation control system according to claim 1, wherein, The magnetic levitation controller includes computer hardware and algorithm software, which are used to obtain the set physical parameters of the levitation system input, and to obtain the gap data between the track and the car body in real time and output a control signal.

7. A magnetic levitation control method for a maglev train, the method comprising: Providing a magnetic levitation control system for the maglev train. The magnetic levitation control system includes a gap sensor, a chopper, a magnetic levitation controller, and a levitation electromagnet. The gap sensor is connected to the magnetic levitation controller through the chopper. The magnetic levitation controller is connected to a computer device. The levitation electromagnet is connected to the magnetic levitation controller through peripheral hardware. The gap sensor is installed on the levitation electromagnet; Using the gap sensor to measure and collect the levitation distance data in real time; Performing analog-to-digital conversion and filter modulation on the levitation distance data; Transmitting the levitation distance data to the computer device connected to the magnetic levitation controller through a communication line; Enabling the computer device carrying the magnetic levitation controller to output the control quantity obtained by the magnetic levitation controller to the peripheral hardware, thereby driving the levitation electromagnet to work and levitating the train; Characterized in that the magnetic levitation controller includes an anti-saturation levitation controller, Wherein, the anti-saturation levitation controller controls the control quantity u(t) based on the error system e and the magnetic levitation mathematical model according to the following controller model: Wherein, the error system e is the difference between the actual levitation gap and the ideal levitation distance. m is the mass of the train, g is the acceleration due to gravity, and mg is the nominal levitation weight. At the same time, the current required for the electromagnet coil is: where z is the suspension distance, k p , k d shall satisfy the following: Among them, F max is the maximum electromagnetic suction force.

8. The magnetic levitation control method according to claim 7, wherein, The magnetic levitation controller further includes an RBF enhancement control term.

9. The magnetic levitation control method according to claim 8, wherein, The magnetic levitation controller is executed according to the following controller model: u all (t) = u(t) + U(t) where, u all (t) is the total control quantity, u(t) is the control quantity of the saturated suspension controller, and U(t) is the control quantity of the RBF enhancement control term. Among them, the network weight matrix of the output layer is w = [w1, w2, …, w M , where R j (x) is the output of the j-th node in the hidden layer, X = [x1, x2, …, x n T is the input vector of the network, is the coordinate matrix of the centers of the Gaussian basis functions of the hidden layer neurons, σ = [σ1, σ2, …, σ M j is the width vector of the Gaussian basis functions, σ j (j = 1, 2, …, M) is the width of the j-th hidden layer neuron, and N is the number of hidden layer nodes.

10. The magnetic levitation control method according to claim 7, wherein, The control quantity u(t) is the electromagnetic suction force, i.e., u(t) = F m .

Images