A t-type three-level model predictive fault-tolerant control method

Abstract

The application discloses a T-type three-level model predictive fault-tolerant control method, which is applied to a fault-tolerant control system of an inverter. Firstly, in order to make the T-type three-level inverter normally operate when open-circuit faults occur in horizontal switch tubes, the application proposes to use a duty cycle thought to synthesize a target vector in a missing vector range, so that the missing vector can be compensated, the total harmonic distortion (THD) can be reduced, and the switching frequency is fixed. Moreover, the calculation amount is reduced through the thought of zero-error control. Meanwhile, a method of controlling the midpoint potential balance according to the current flow direction and a joining factor rho is used. The midpoint potential is controlled while the reference current is accurately tracked, the algorithm is simplified, and the inverter is ensured to normally operate in the case of horizontal switch faults. The method has the technical advantages of simple structure, small calculation amount and easy realization, is applied to an inverter control system, has the advantages of fast system dynamic response, strong robustness, reduced THD and reduced algorithm time consumption.

Description

Technical Field

[0001] The present invention relates to a T-type three-level model prediction fault-tolerant control method, in particular to a model prediction fault-tolerant control method for duty cycle control when a horizontal switch tube fails. Background Art

[0002] With the increasing application of T-type three-level inverters in industry, the control requirements are also becoming increasingly stringent, focusing on two key aspects: first, reference current tracking performance, which is crucial in industry. Good tracking performance ensures better performance of the inverter in industrial applications. Second, midpoint potential balance. Numerous methods have been proposed to improve control in these two areas. Currently, SVPWM modulation is the most commonly used control method for T-type three-level inverters. While it offers good control results, it also suffers from poor dynamic response and complex control processes. Model Predictive Control (MPC), first developed in the 1960s, has become a research hotspot in power electronics in recent years, driven by the rapid development of microcontrollers.

[0003] Model Predictive Control (MPC) is gaining attention in the fields of power electronics and motor drives due to its simple control method, better control effect, ability to realize multivariable control, and great advantages in dealing with complex constrained optimization of nonlinear systems. Its application in industry is becoming more and more extensive.

[0004] However, model predictive control also has many drawbacks, such as over-reliance on model parameters, fixed inverter output voltage, excessively high switching frequency, difficulty selecting weight coefficients, and excessive computational complexity, all of which have impacted industry. Addressing these issues is urgent. Furthermore, the high switching frequency can easily damage switching components in the circuit, causing short circuits or open circuits. Because short circuits are particularly harmful, they typically require immediate circuit closure. Therefore, fault-tolerant control of model-predicted open circuit faults in T-type three-level inverters is crucial. Summary of the Invention

[0005] In response to the improvement needs of the above-mentioned prior art, the present invention provides a model predictive fault-tolerant control method for a T-type three-level inverter. This control method synthesizes a target vector by using the duty cycle control idea for the reference voltage within the range of the inverter's missing voltage vector, thereby ensuring that the T-type three-level inverter can operate normally when the horizontal switch tube fails, greatly reducing the harmonics of the output current. In addition, the use of the zero-beat idea greatly reduces the computational complexity of the control method. At the same time, for the balance of the midpoint potential, a method for selecting the target vector based on the current direction and the addition of ρ is proposed. Compared with traditional control methods, the robustness is improved, while taking into account the steady-state control accuracy and algorithm speed of the controlled system.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A model predictive fault-tolerant control method for a T-type three-level inverter comprises the following steps:

[0008] Step 1: Based on the input and output of the grid-connected inverter, a model of the grid-connected inverter system is established, and then the grid-connected inverter current prediction equation is designed based on the model;

[0009] Step 2: First, according to i a The positive and negative current divides the control process into i a >0 fault part, and i a <0, and combined with the idea of ​​deadbeat control, the output reference voltage U of the grid-connected inverter is obtained from the reference current. ref , perform sector judgment on the reference voltage, narrow the number of candidate voltages, perform model prediction based on the candidate voltage vectors, and obtain the two switching states that minimize the value function;

[0010] Step 3: Combined with the idea of ​​duty cycle control, calculate the action time of the two switch states in the same cycle, and output the two switch states in the same switching cycle so that the current tracks the reference current.

[0011] Furthermore, the step 1 specifically includes:

[0012] According to the input and output of the grid-connected inverter, the relevant model of the inverter system is established

[0013]

[0014] And get the slope of the current under different switching states

[0015]

[0016] Using the forward Euler formula, we get

[0017]

[0018] The prediction equation for the current is deduced as follows:

[0019]

[0020] Furthermore, in step 2, the entire control process is divided into i a >0 fault state and i a <0 normal part;

[0021] For i a >0, the open circuit of the switch tube Sa2 affects the output voltage of the inverter. Phase A cannot output the O vector. The inverter can output 18 switching states.

[0022] At this time, for the midpoint potential, since the small vector is missing, and only i a <0, add ρ to the value function corresponding to the switch state that affects the midpoint potential, similar to the selection of the weight coefficient;

[0023] For i a <0, at this time, the open circuit of the switch tube Sa2 has no effect on the output voltage of the inverter, and the inverter can output 27 switching states.

[0024] At this time, for the midpoint potential, just pay attention to i a For the part <0, the rest can be done according to the traditional method.

[0025] Furthermore, in step 2, for the deadbeat control process:

[0026] a. First, all possible inverter output voltage vectors are divided into 6 large sectors in the left axis of the static two-phase;

[0027] b. Using the idea of ​​deadbeat control, calculate the reference voltage vector U ref , determine the sector where the reference voltage is located, and all switch states in the sector are candidate switch states.

[0028] c. Substitute all candidate switch states into the value function and select the two output vectors V that minimize the value function. opt1 , V opt2 .

[0029] Furthermore, in step b, the reference voltage vector U is calculated ref Method, let the predicted value i at time k+1 x (k+1) is the reference value i refx (k+1), the inverter output reference voltage formula can be obtained from the current prediction equation:

[0030]

[0031] Furthermore, the step three specifically includes:

[0032] According to the two switch states obtained, the current slope is obtained by the current slope formula, and

[0033]

[0034] T opt2 =T s -T opt1

[0035] Calculate the time for each of the two switch states to function.

[0036] Compared with the existing technology, the above technical solution conceived by the present invention has the following advantages over the existing technology:

[0037] (1) The THD of the output current obtained by duty cycle modulation is much smaller than that of other methods

[0038] (2) This method combines fast vector selection and duty cycle prediction control, which improves the steady-state control accuracy of the system while optimizing the algorithm time consumption and improving the control performance of the system.

[0039] (3) This method has a simple structure, small computational complexity, and is easy to implement. When used in a T-type three-level inverter control system, the system has a fast dynamic response speed, small current distortion, low switching frequency, and excellent dynamic and steady-state performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is the main circuit topology diagram of the T-type three-level inverter

[0041] Figure 2 This is the control block diagram of the model prediction fault-tolerant control based on duty cycle control of the present invention

[0042] Figure 3 This is the basic voltage vector distribution diagram of the three-level inverter output

[0043] Figure 4 This is a flow chart of midpoint potential control based on current direction.

[0044] Figure 5 This is a schematic diagram based on duty cycle modulation

[0045] Figure 6 The midpoint potential offset diagram in the matlab simulation of the present invention

[0046] Figure 7 This is the current tracking diagram in the matlab simulation of the present invention DETAILED DESCRIPTION

[0047] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent based on the claims and the following specific embodiments. It should be noted that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0048] Figure 1 It is the topology of the main circuit of the T-type three-level inverter.

[0049] Figure 2 A schematic diagram of a T-type three-level model predictive fault-tolerant control method disclosed in an embodiment of the present invention adopts i q =0 control mode, the outer speed loop uses the PI regulator, and the inner current loop adopts the proposed model predictive control method based on duty cycle control. The specific strategy is as follows:

[0050] First, based on the input and output of the grid-connected inverter, the inverter system model is established. The relationship between the inverter output voltage and the circuit parameters is expressed as follows:

[0051]

[0052] Where: x = a, b, c, e x is the grid voltage, U x is the output voltage of the inverter, i x is the output current of the inverter, R is the resistance parameter of the inverter side, and L is the inductance parameter of the inverter side.

[0053] By 3 / 2 transformation, the expression of (1) converted to α, β coordinate axis is:

[0054]

[0055] From (2), we can get the current slope in a certain switching state is

[0056]

[0057] Sα, Sβ are the current slopes in this switching state.

[0058] Using the forward Euler formula, we can get

[0059]

[0060] From (1) and (4), the current prediction equation is:

[0061]

[0062] Figure 3 In order to divide the 18 switch states (excluding the 9 switch states that cannot be used due to faults) into 6 large sectors, according to the deadbeat idea, let the predicted value i at time k+1 be x (k+1) is the reference value i refx (k+1), the inverter output reference voltage formula can be obtained from (5):

[0063]

[0064] According to the sector judgment, the sector of the inverter's output reference voltage at time k can be obtained, and the candidate vector can be narrowed down to achieve the purpose of reducing the amount of calculation.

[0065] Then, the two switch states that minimize J are obtained through the cost function (7).

[0066] J=(i refα (k+1)-i α (k+1)) 2 +(i refβ (k+1)-i β (k+1)) 2 (7)

[0067] Figure 4 For the flow chart of the midpoint potential balance method, this paper proposes a method of using current flow direction and adding ρ.

[0068] Since for the neutral point current

[0069] i np =(S a2 -S a1 )i a +(S b2 -S b1 )i b +(S c2 -S c1 )i c (8)

[0070] First, determine i a The positive and negative directions of the current flow, because when the Sa2 switch tube is open, when i a <0, there is no effect on the circuit, and the output of the inverter can be controlled in normal state. However, since i a <0, for the switch states OPN, ONP, OPP, NOO, POO, ONN the scope of action is limited, for example, for POO, i np =-i a

[0071] Because I a<0, so i np >0. At this time, Uc1 decreases and Uc2 increases. When working in a circuit where Uc1 < Uc2, the offset will become more serious. At this time, ρ is added to reduce the probability of selecting this switch state. That is:

[0072]

[0073] ρ increases as the deviation between Uc1 and Uc2 increases. For other small vectors, only the magnitudes of Uc1 and Uc2 need to be considered. For example:

[0074] When Uc1 > Uc2, output PPO. When Uc1 < Uc2, output OON

[0075] For medium vectors, such as PON, at this time, i np = i b

[0076]

[0077] When i a >0, phase A cannot output the O state. At this time, the midpoint potential balance is related to the selection of the medium vector as described above

[0078] Figure 5 This is a schematic diagram of duty cycle debugging. Two switching vectors act for a period of time respectively, so that the current tracks the reference current. According to the duty cycle control principle, the acting time of the two vectors is obtained

[0079] The predicted current value at the k + 1 moment in duty cycle prediction:

[0080]

[0081] Where T opt1 , T opt2 are the acting times of the current slopes corresponding to switch states 1 and 2. And T opt1 + T opt2 = T s

[0082] Substituting (11) into (7), a quadratic equation of one variable about T opt1 can be obtained. By finding the T opt1 that can make J reach the minimum value, it can be done

[0083] Taking the derivative of J gives

[0084] J′ = 2(S 2α - S 1α )(i refα (k + 1)-i α (k)-T s S2α )+2(S 2α -S 1α )(S 2α -S 1α )T opt1 +2(S 2β -S 1β )i refβ (k+1)-i β (k)-T s S 2β )+2(S 2β -S 1β )(S 2β -S 1β )T opt1 (12

[0085] J″=(S 2α -S 1α ) 2 +(S 2β -S 1β ) 2 (13)

[0086] From J″>0, we can get J′ as a monotonically increasing function. opt1 When J′=0, that is, at t=T opt1 When t<0, J takes the minimum value at 0, t>T s When J is in T s (minimum value at

[0087] Let J′=0, we can get

[0088]

[0089] That is, the switch state is 1, and the action time is T opt1 , the action time of switch state 2 is T opt2 =T s -T opt1

[0090] At the same time, output T opt1 , and two switching states, the conduction of the switching device can be controlled by PWM regulation.

[0091] Figure 6 This is a midpoint potential offset diagram after MATLAB simulation. It can be seen from the figure that the midpoint potential offset of the invention is within an acceptable range.

[0092] Figure 7 The current tracking diagram after MATLAB simulation shows that the dynamic response of the invention is very good. After FFT analysis, the THD value of the invention is only 4.56%, which meets the control requirements.

[0093] It will be easily understood by those skilled in the art that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A model predictive fault-tolerant control method for a T-type three-level inverter, characterized by: The following steps are involved: Step 1: Based on the input and output of the grid-connected inverter, a model of the grid-connected inverter system is established. Then, based on the model, the grid-connected inverter current prediction equation is designed. The current prediction equation is: ； Step 2: First, according to The positive and negative currents divide the control process into >0 of the faulty portion, and <0 normal part, and combined with the idea of ​​deadbeat control, the output reference voltage of the grid-connected inverter is obtained from the reference current , perform sector judgment on the reference voltage, narrow the number of candidate voltages, perform model prediction based on the candidate voltage vectors, and obtain the two switching states that minimize the value function; The reference voltage vector U ref Calculation method: Let the predicted value at time k+1 Reference value , the inverter output reference voltage formula can be obtained from the current prediction equation: ， The value function is: ; Step 3: Combined with the idea of ​​duty cycle control, calculate the action time of the two switch states in the same cycle, and output the two switch states in the same switching cycle so that the current tracks the reference current.

2. The T-type three-level inverter model predictive fault-tolerant control method according to claim 1, characterized in that: The step 1 specifically includes: According to the input and output of the grid-connected inverter, the relevant model of the inverter system is established ， And get the slope of the current under different switching states ， Using the forward Euler formula, we get ， The prediction equation for the current is deduced as follows: 。 3. The model predictive fault-tolerant control method for a T-type three-level inverter according to claim 1, characterized in that: In the step 2, The entire control process is divided into >0 fault status and <0 normal part; for >0, the open circuit of the switch tube Sa2 affects the output voltage of the inverter. Phase A cannot output the O vector. The inverter can output 18 switching states. At this time, for the midpoint potential, since the small vector is missing, there is only <0, add the value function corresponding to the switch state affecting the midpoint potential , similar to the selection of weight coefficients; for <0, at this time, the open circuit of the switch tube Sa2 has no effect on the inverter output voltage, and the inverter can output 27 switching states; At this time, for the midpoint potential, just pay attention to For the part <0, the rest can be done according to the traditional method.

4. The T-type three-level inverter model predictive fault-tolerant control method according to claim 1, characterized in that: In the step 2, for the deadbeat control process: a. First, all possible inverter output voltage vectors are divided into 6 large sectors in the left axis of the static two-phase; b. Using the idea of ​​deadbeat control, calculate the reference voltage vector U ref , determine the sector where the reference voltage is located, and all switch states in the sector are candidate switch states; c. Substitute all candidate switch states into the value function and select the two output vectors V that minimize the value function. opt1， V opt2 .

5. The model predictive fault-tolerant control method for a T-type three-level inverter according to claim 1, characterized in that: The step three specifically includes: According to the two switch states obtained, the current slope is obtained by the current slope formula, and ， ， Calculate the time for each of the two switch states to function.

Images