An improved model predictive control method suitable for high-speed train traction converter
Patent Information
- Application Number
- CN202211473965.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-11-23
- Publication Date
- 2026-09-22
- Estimated Expiration
- 2042-11-23
AI Technical Summary
但整流输出功率的响应过程是电流环与电压环交互协同动作的过程,即使电流环速度较快,但整个响应过程仍较大的受限于电压环
[0027]1)对模型预测控制方法做出了改进,改进模型预测控制采用负载电流前馈的方式作用于整流侧电压环,从而提高了直流母线电压的调节速度,降低了调节过程中的恢复电压超调。
Smart Images

Figure CN116094380B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of converter control systems in the field of power electronics and electric transmission, and specifically relates to an improved model predictive control method applicable to high-speed train traction converters. Background Technology
[0002] Electrified railways have advantages such as large carrying capacity, high speed, low pollution, and low operating costs. For a long time, their investment and construction have received high attention and strong support from the state. The core power source of electrified railways is the electric traction drive system. The "AC-DC-AC" type electric traction drive system has become the mainstream drive method for electrified railways. Among them, the traction converter, as the upstream power supply device for the traction motor, directly affects the traction and braking performance of the motor through its power output response process and quality.
[0003] Train operation involves frequent switching between multiple operating conditions, such as traction, regenerative braking, and coasting. For the rectifier stage, these operating condition switching often manifests as a continuous change in rectifier output from no-load to full-load. For the inverter and traction motor, the rectifier performance is mainly characterized by a continuous and stable DC bus voltage output, where the DC bus voltage can quickly recover / drop back to the given voltage when load switching occurs. When the rectifier output switches from no-load to full-load, the DC bus voltage will drop. At this time, the voltage loop controller input will increase, thereby increasing the current loop command value. The increase in the current loop command value will cause the current loop to adjust the grid-side input current, increasing the grid-side input power. Since the rectifier input power and output power are conserved, the result will inevitably be an increase in output power, and the DC bus voltage will recover and rise. Similarly, when the rectifier output switches from full-load to no-load, the DC bus voltage will show a positive spike. When the motor's operating conditions switch frequently, it will directly cause continuous positive / negative spikes in the DC bus voltage, such as... Figure 1 As shown. A slow DC bus voltage response can affect the output performance of the traction motor, and a prolonged response time can also trigger LC resonance. Over the years, scholars have proposed control methods for the current loop, significantly improving its response speed. However, the response process of the rectified output power is a process of interactive and coordinated action between the current loop and the voltage loop. Even with a faster current loop, the entire response process is still largely constrained by the voltage loop.
[0004] With the rapid development of rail transit, the development of a new generation of high-speed trains with a speed of 400 km / h has been put on the agenda. Higher speeds place higher performance demands on the converters of the traction motors, and a voltage loop control method for rapid and smooth recovery of the DC bus voltage is essential to ensure the superior performance of the traction motors. Therefore, research on a fast recovery control method for the DC bus voltage in the rectifier stage of the high-speed train traction drive system is of great significance. Summary of the Invention
[0005] To improve the response speed of the voltage loop and smooth out high voltage overshoot during the response process when sudden changes in traction motor speed or load cause DC bus voltage fluctuations, this invention provides an improved model predictive control method suitable for high-speed train traction converters.
[0006] An improved model predictive control method for high-speed train traction converters, as proposed in this invention, includes the following steps:
[0007] Step 1: Sample the secondary voltage of the traction transformer to obtain its amplitude U s And phase information ωt.
[0008] Step 2: Sample the actual speed n of the traction motor k With inverter side current The dq-axis reference voltage vector is obtained through calculation. With current vector
[0009] Step 3: Estimate the bus current based on the inverter-side reference voltage vector, current vector, and bus voltage.
[0010] Step 4: Estimate the value based on the DC bus current. Determining the rectifier-side current command value from the bus voltage error Meanwhile, to eliminate steady-state error, integral compensation is added to the DC bus current term, thus forming an integral compensation term based on DC bus current feedforward.
[0011] Step 5: Use a deadbeat predictive control method to achieve precise control of the rectifier side current.
[0012] Furthermore, step 3 specifically involves:
[0013] The inverter-side reference voltage vector is obtained through the motor control system, and the inverter output current vector is obtained by sampling through an inherent sensor and performing a Park transformation. Therefore, the DC bus current is:
[0014]
[0015] in, and These are the d-axis components of the inverter-side reference voltage and current vectors. and It is the q-axis component of the inverter-side reference voltage and current vector; Let k be the DC bus voltage at time k.
[0016] Furthermore, step 4 specifically involves:
[0017] With control period T sDiscretize the bus voltage in units of , then the DC bus voltage at time k+1 is expressed as:
[0018]
[0019] in, For the current command value, C d This is the value of the DC-side supporting capacitor.
[0020] The rectifier-side voltage loop evaluation function is designed as follows:
[0021]
[0022] Among them, u dcr This is the reference value for the DC bus voltage.
[0023] To achieve accurate tracking of the DC bus voltage, an error integral compensation term for the bus voltage is added to the DC bus current, thus forming an integral compensation term based on DC bus current feedforward. The rectifier-side current command value is:
[0024]
[0025] Among them, f s The switching frequency, i.e., the switching period T s The reciprocal of ; λ is the weighting coefficient, used to balance the importance of the two indicators on the right side of the equation, and is taken as 0.06 here.
[0026] The beneficial technical effects of this invention compared to the prior art are:
[0027] 1) An improvement was made to the model predictive control method. The improved model predictive control uses load current feedforward to act on the rectifier side voltage loop, thereby improving the regulation speed of DC bus voltage and reducing the recovery voltage overshoot during the regulation process.
[0028] 2) A method for estimating the DC bus current on the rectifier side was designed, thereby achieving accurate extraction of the DC bus current without adding additional sensors.
[0029] 3) The error compensation of the rectifier bus current feedforward value was performed by using an integrator, which improved the open-loop gain of the rectifier side voltage loop and realized accurate tracking of the DC bus voltage. Attached Figure Description
[0030] Figure 1 These are the positive and negative voltage spikes that appear on the DC bus during load switching.
[0031] Figure 2 This is a topology diagram of a traction converter.
[0032] Figure 3 Flowchart for implementing the control method.
[0033] Figure 4 This is a steady-state waveform diagram of the DC bus voltage.
[0034] Figure 5 Comparison of the control effects of switching from full load to no load on the converter output (where a uses the method of this invention, and b uses the traditional PI method).
[0035] Figure 6 Comparison of the no-load to full-load control effect of the converter output (where a uses the method of this invention, and b uses the traditional PI).
[0036] Figure 7 Comparison of the effects of half-load to no-load control on the converter output (where a uses the method of this invention, and b uses the traditional PI method).
[0037] Figure 8 Comparison of the no-load to half-load control effect of the converter output (where a uses the method of this invention, and b uses the traditional PI). Detailed Implementation
[0038] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.
[0039] This invention focuses on a three-phase two-level traction converter, whose topology is as follows: Figure 2 As shown.
[0040] An improved model predictive control method for high-speed train traction converters, as described in this invention, is implemented as follows: Figure 3 As shown. Specifically:
[0041] Step 1: Sample the secondary voltage of the traction transformer to obtain its amplitude U s and phase information ωt;
[0042] Step 2: Sample the actual speed n of the traction motor k With inverter side current The dq-axis reference voltage vector is obtained through calculation. With current vector Permanent magnet synchronous motors use d-axis current i od =0 control mode, therefore the d-axis reference current The q-axis reference current is 0. Specifically, this can be obtained from the speed outer loop PI controller:
[0043]
[0044] The mathematical model of the dq axis of the permanent magnet synchronous motor is as follows:
[0045]
[0046] Among them, u od u oq This is the reference value for the dq axis voltage; L d L q R is the stator inductance; R is the stator resistance; ω e ψ is the rotor's electric angular velocity. f It is a permanent magnet flux linkage.
[0047] With sampling period T s Discretizing the above equation using the given information as a reference, we can obtain the inverter-side dq current at time (k+1) as follows:
[0048]
[0049] Based on the concept of deadbeat control, ideally, the actual current should follow the given current in the next cycle. Therefore, the inverter-side reference voltage vector can be calculated as follows:
[0050]
[0051] At this point, both the reference voltage vector and the output current vector required for DC bus current estimation are known.
[0052] Step 3: Estimate the bus current i based on the inverter-side reference voltage vector, current vector, and bus voltage. k d c In the traction drive system of rail trains, the DC bus current can be estimated using the inverter-side reference voltage vector and the output current vector without adding additional sensors. The inverter-side reference voltage vector can be obtained through the motor control system, and the inverter output current vector can be obtained by sampling and transforming using existing sensors.
[0053] Therefore, the estimated value of the DC bus current is:
[0054]
[0055] Step 4: Estimate the value based on the DC bus current. Determining the rectifier-side current command value from the bus voltage error Meanwhile, to eliminate steady-state errors, integral compensation is added to the DC bus current term, thus forming an integral compensation term based on DC bus current feedforward. For ease of analysis, the traction transformer is approximated as a voltage source connected in series with an inductor, such as... Figure 2 As shown, its internal resistance is negligible. s and i s These represent the secondary side grid voltage and line current of the traction transformer, respectively; L s It is the equivalent inductance of the traction transformer, u ab It is the rectifier-side input voltage; udc It is the DC bus voltage; S1-S4 are insulated gate bipolar transistor (IGBT) modules with anti-parallel diodes on the rectifier side.
[0056] Define rectifier side u s i s and u ab as follows
[0057]
[0058] Among them, u sd u sq i sd i sq u abd u abq These are the dq-axis components of the grid voltage, line current, and rectifier-side input voltage vectors, respectively.
[0059] For the input current in the grid-side boost inductor and the DC bus voltage across the supporting capacitor, applying Kirchhoff's laws yields the following differential equation:
[0060]
[0061] Among them, u ab Let m be the rectifier-side input voltage, m be the rectifier-side modulation function, and i be the input voltage. dc This represents the DC bus current.
[0062] By combining the above equations, we can obtain the dynamic model of the rectifier side dq axis as follows:
[0063]
[0064] In pulse rectifiers, due to the power frequency characteristics of the grid-side voltage and input current, the output voltage inevitably contains ripple at twice the power frequency (i.e., 100Hz). Figure 4 As shown. After being regulated by the voltage loop and entering the current loop, the ripple multiplies with the sinusoidal component of the grid-side voltage, generating a large number of 150Hz harmonics, which severely interfere with the current loop. Therefore, a simple first-order low-pass filter is designed to reduce the interference of the 100Hz ripple on the current loop. That is, the DC bus voltage at time k is...
[0065]
[0066] in, Let ω be the sampled value of the DC bus voltage at time k. c The cutoff frequency is ω, which is usually chosen to achieve better filtering results. c =25.
[0067] Similarly, the evaluation function for the rectifier-side voltage loop is designed as follows:
[0068]
[0069] To achieve accurate tracking of the rectifier-side voltage loop, an integral compensation stage can be added to the DC bus current, i.e., the rectifier-side current loop command value is...
[0070]
[0071] Step 5: Implement precise control of the rectifier-side current using deadbeat predictive control. To achieve sinusoidal grid-side input current and fast dynamic performance, deadbeat predictive control can be applied to the rectifier-side current loop. The control period T... s Discretizing the input current in units of time, its discrete-time dynamic model is as follows:
[0072]
[0073] To achieve accurate tracking of the rectifier-side input current and the command value, the error between the input current and the command value should be minimized after one control cycle; that is, the evaluation function is:
[0074]
[0075] Furthermore, the optimal input voltage vector on the rectifier side can be expressed in the dq coordinate system as follows:
[0076]
[0077] The optimal modulation function on the rectifier side is then:
[0078]
[0079] in, ωt is the angular velocity of the grid-side voltage, which can be obtained through a phase-locked loop. After the optimal modulation function is calculated, the duty cycle of each switch can be allocated according to the principle of space vector modulation.
[0080] Figure 5-8 This is a comparison chart showing the control effect when the converter experiences sudden load changes. Figure 5 (b) Using traditional PI control, when the converter load switches from full load to no load, the DC bus voltage overshoot is approximately 13V, and the settling time is approximately 290ms. Figure 5 (a) Using the improved model predictive control method, the DC bus voltage overshoot is approximately 8V when switching from full load to no load, and its dynamic response time is approximately 40ms. Figure 6 In (b), when switching from no-load to full-load, the entire response process of the DC bus voltage using traditional PI control takes approximately 190ms. However, the improved model predictive control of the DC bus voltage, such as... Figure 6 (a) has a relatively fast response time of approximately 80ms. Similarly, Figure 7 (a), (b) and Figure 8 (a) and (b) show that when the load switches between no load and half load, the improved model predictive control method has a better control effect than the traditional PI control method.
Claims
1. An improved model predictive control method applicable to high-speed train traction converters, characterized in that, Includes the following steps: Step 1: Sample the secondary voltage of the traction transformer to obtain its amplitude U s and phase information ωt; Step 2: Sample the actual speed n of the traction motor k With inverter side current The dq-axis reference voltage vector is obtained through calculation. With current vector ; Step 3: Estimate the bus current based on the inverter-side reference voltage vector, current vector, and bus voltage. ; Step 4: Estimate the value based on the DC bus current. Determining the rectifier-side current command value from the bus voltage error Meanwhile, in order to eliminate steady-state error, integral compensation is added to the DC bus current term, thus forming an integral compensation term based on DC bus current feedforward. With control period T s Discretize the bus voltage in units of , then the DC bus voltage at time k+1 is expressed as: ; in, For the current command value, C d The value of the DC-side supporting capacitor; The evaluation function for the rectifier-side voltage loop is designed as follows: ; Among them, u dcr This is the reference value for the DC bus voltage; To achieve accurate tracking of the DC bus voltage, an error integral compensation term for the bus voltage is added to the DC bus current, thus forming an integral compensation term based on DC bus current feedforward. The rectifier-side current command value is: ; Among them, f s The switching frequency, i.e., the switching period T s The reciprocal of ; λ is the weighting coefficient, used to balance the importance of the two indicators on the right side of the equation, and is taken as 0.06 here; Step 5: Use a deadbeat predictive control method to achieve precise control of the rectifier side current.
2. The improved model predictive control method for high-speed train traction converters according to claim 1, characterized in that, Step 3 specifically involves: The inverter-side reference voltage vector is obtained through the motor control system, and the inverter output current vector is obtained by sampling through an inherent sensor and performing a Park transformation. Therefore, the DC bus current is: ; in, and These are the d-axis components of the inverter-side reference voltage and current vectors. and It is the q-axis component of the inverter-side reference voltage and current vector; Let k be the DC bus voltage at time k.
Citation Information
Patent Citations
DC bus current estimation method based on motor controller
CN112701986A