CN117691909A - Linear induction motor three-level dual-vector model predictive thrust control method and system - Google Patents

Abstract

The invention discloses a method and a system for controlling a three-level double-vector model predictive thrust of a linear induction motor, which belong to the technical field of linear motor control, and comprise the following steps: collecting system state parameters, and estimating thrust and primary flux linkage by combining a motor model; inputting the thrust reference value, the estimated thrust, the primary flux linkage and the like into a reference primary flux linkage vector generator, and calculating and outputting a reference primary flux linkage vector; solving a reference voltage vector according to the reference primary flux linkage vector, determining candidate voltage vector combinations according to the reference voltage vector position, and calculating respective corresponding duty ratios; substituting each candidate voltage vector combination and the duty ratio thereof into the defined cost function to determine the optimal voltage vector combination; and generating switching pulse according to the optimal voltage vector combination and the duty ratio thereof, and realizing the control of the linear induction motor. The invention can effectively reduce the implementation complexity of the double-vector model predictive thrust control in the linear induction motor system driven by three-level frequency conversion, and remarkably improves the steady-state performance of the system.

Description

Linear induction motor three-level dual-vector model predictive thrust control method and system

Technical field

The invention belongs to the technical field of linear motor control, and more specifically, relates to a linear induction motor three-level dual vector model predictive thrust control method and system.

Background technique

Compared with rotary induction motor-driven rail transit systems, linear induction motors can directly produce linear motion without the need for transmission mechanisms such as gearboxes. They have the advantages of stronger climbing ability, smaller turning radius and smaller cross-sectional area, and are highly suitable for intricate tracks. transportation network development needs. At the same time, as the passenger flow of rail transit continues to increase, newly built lines mostly adopt the DC 1500V power supply method to meet the needs of higher power level traction motors. Compared with the two-level topology, the three-level inverter topology (NPC three-level inverter topology is the most widely used) has the characteristics of high output voltage sinusoidality, small voltage stress of a single power device, and small output current distortion rate, etc. advantage. Therefore, the three-level variable frequency drive linear induction motor traction system has broad application prospects in rail transit.

However, due to the disconnection of the primary iron core, the linear induction motor faces severe end effects, which causes the mutual inductance to change drastically during motor operation and causes severe thrust attenuation during high-speed operation. Model predictive thrust control combines direct thrust control with model predictive control, takes thrust and primary flux as control targets, and uses online optimization to determine the pseudo-acting voltage vector. It is easy to handle multi-objective constraints and can effectively alleviate end effects. The resulting thrust attenuation, while maintaining the fast dynamic response capability of direct thrust control, has great potential for development in linear induction motor drive systems.

Since only one voltage vector acts during the entire control cycle, the traditional single vector model predicts the existence of large thrust fluctuations and current harmonics in thrust control. In order to further improve the steady-state performance, a method of applying two voltage vectors in each control cycle, that is, dual-vector model predictive thrust control, is usually used to significantly reduce thrust fluctuations with a small increase in switching frequency. Related methods have been used in two It is applied in linear induction motor system driven by level inverter. However, there are still certain difficulties when the above method is directly applied to three-level inverters. First, the significantly increased discrete voltage vectors of the three-level inverter dramatically increase the number of voltage vector combinations. Without considering simplification, there are only 49 dual-vector combinations for two-level inverters, while there are 729 dual-vector combinations for three-level inverters. It is difficult to determine the optimal voltage vector combination using existing methods. Secondly, it is difficult for existing methods to consider the inherent control requirements brought by the three-level inverter topology, such as neutral point voltage balance and smooth transition of switches. Therefore, although the core idea of dual-vector model predictive thrust control is universal, existing methods are difficult to effectively extend to three-level variable frequency driven linear induction motor systems to achieve steady-state performance improvement.

Contents of the invention

In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a linear induction motor three-level dual-vector model predictive thrust control method and system that considers voltage vector optimization, aiming to solve the problem of existing dual-vector model predictive thrust control applied to three-level systems. The linear induction motor system driven by level variable frequency has problems such as complex algorithm and difficulty in effectively considering the inherent control requirements of the three-level inverter.

In order to achieve the above objectives, the present invention provides a linear induction motor three-level dual-vector model predictive thrust control method, which includes the following steps:

S1: Real-time collection of the status parameters of the linear induction motor control system driven by the Neutral-Point-Clamped (NPC) NPC three-level inverter, including motor phase current, speed and inverter midpoint voltage; Use the thrust and flux observer to calculate the motor thrust and primary flux at the current moment based on the sampling results;

S2: Input the thrust reference value generated by the speed controller, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux vector generator into the reference primary flux vector generator, and then calculate it through the reference primary flux vector generator Output reference primary flux vector;

S3: Solve the reference voltage vector based on the reference primary flux vector, determine the candidate voltage vector combinations based on the reference voltage vector position, and calculate the duty cycle corresponding to each candidate voltage vector combination; substitute each candidate voltage vector combination and its duty cycle into the defined value function to determine the optimal voltage vector combination;

S4: Generate switching pulses for each bridge arm based on the optimal voltage vector combination and its duty cycle, which act on the three-level inverter to control the linear induction motor.

Further preferably, the method for determining candidate voltage vector combinations based on reference voltage vector positions includes the following steps:

All discrete voltage vectors that can be generated by the NPC three-level inverter are divided into large vectors, medium vectors, small vectors and zero vectors according to their amplitudes; among them, the large vector amplitude is 2/3 times u _dc ; medium The vector magnitude is times u _dc ; the small vector amplitude is 1/3 times u _dc ; u _dc is the DC bus voltage;

Using the six median vectors as dividing lines, the inverter output voltage plane is divided into six large sectors I to VI; among them, the angular bisector direction of sector I is defined as the α-axis direction, sector II and sector III The dividing line direction of is defined as the β axis direction;

Taking sector I as an example, using u _1α = u _dc /3 and u _1β = 0 as the dividing line, sector I is further divided into four small areas: R1 ~ R4; combined according to the specific size of the reference voltage vector Table 1 determines the candidate voltage vector combinations, and calculates the duty cycle corresponding to each candidate voltage vector combination; for the case where the reference voltage vector is in other sectors, first rotate it to the I-th sector through coordinate transformation, and then use the I-th sector The conditions in the area are used to determine the area, and then the corresponding candidate voltage vector combinations are determined;

Table 1

Substitute each candidate voltage vector combination and its duty cycle into the defined value function to determine the optimal voltage vector combination;

Further preferably, for the candidate voltage vector combination (u _i , u _j ), the value function designed is in the form of:

Among them, k _u is the weight coefficient, In order to consider the midpoint voltage prediction value after delay compensation, d _opt is the duty cycle corresponding to the candidate voltage vector combination (u _i , u _j ), and the expression is:

H _m is the constraint to avoid direct switching of P and N states according to the switch state. The expression is:

Among them, S _x (x will be a, b and c respectively) represents the switching state of the 1 represents time k and k+1 respectively.

On the other hand, the present invention provides a linear induction motor three-level dual-vector model predictive thrust control system, including:

The thrust and primary flux observation module is used to calculate the current thrust and primary flux of the motor using the thrust and flux observer based on the motor phase current, speed and the primary flux calculated at the previous moment;

The reference conversion module is used to input the thrust reference value generated by the speed controller, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux into the reference primary flux vector generator, and then output the reference value after calculation Primary flux linkage vector, and further solve the reference voltage vector based on the optimal reference flux linkage vector;

The optimal voltage vector combination selection module is used to determine candidate voltage vector combinations based on the position of the reference voltage vector, calculate the duty cycle corresponding to each combination, and substitute the candidate voltage vector combination and its duty cycle into the defined value function, Select the candidate voltage vector combination that minimizes the value function as the optimal voltage vector combination;

The pulse sequence control module is used to generate three-phase bridge arm pulses based on the optimal voltage vector combination and its duty cycle and act on the corresponding bridge arm of the NPC three-level inverter to control the linear induction motor.

Generally speaking, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:

(1) The linear induction motor three-level dual vector model predictive thrust control method and system provided by the present invention proposes a new voltage vector plane sector division method and corresponding simplified search method for the three-level inverter, which can greatly Reduces the number of voltage vector combinations that need to be evaluated, and effectively reduces the implementation complexity of dual-vector model predictive thrust control in a three-level variable frequency drive linear induction motor system;

(2) The linear induction motor three-level dual-vector model predictive thrust control method and system provided by the present invention analyzes the switching times under different voltage vector combinations and uses the characteristics of redundant small vectors to optimize the voltage vector combination to be evaluated, effectively reducing switching frequency; and through analysis and mathematical derivation, the inherent control requirements of the three-level inverter such as mid-point voltage balance and voltage vector smooth switching are included in the value function to ensure the control performance of the system.

Description of the drawings

Figure 1 is a flow chart of the linear induction motor three-level dual vector model predictive thrust control method provided by the present invention;

Figure 2 is a T-shaped equivalent circuit diagram of the linear induction motor provided by the present invention;

Figure 3 is a structural diagram of the reference primary flux vector generator provided by the present invention;

Figure 4 is an NPC type three-level inverter topology and its voltage vector distribution diagram provided by the present invention, (a) NPC type three-level inverter topology, (b) NPC type three-level inverter discrete voltage vector Distribution;

Figure 5 is a schematic diagram of the voltage vector sector division provided by the present invention, (a) the specific location of the R1 area, (b) the specific location of the R2 area;

Figure 6 is a switching pulse diagram of different voltage vector combinations provided by the present invention, (a) (u ₁₃ , u ₂₅ ) voltage vector combination switching pulse diagram, (b) (u ₁₄ , u ₂₅ ) voltage vector combination switching pulse diagram;

Figure 7 is an architectural diagram of the linear induction motor three-level dual vector model predictive thrust control system provided by the present invention;

Figure 8 is a comparison diagram of the effect of direct switching whether to consider P and N states according to an embodiment of the present invention. (a) The effect of direct switching is not considered to be P and N states. (b) The effect of direct switching is considered to be P and N states;

Figure 9 is a voltage, current waveform and harmonic analysis diagram of each model predictive thrust control method under the same switching frequency provided by the embodiment of the present invention, (a) traditional single vector model predictive thrust control, (b) dual vector without voltage vector optimization Model predictive thrust control, (c) method of the invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain 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 can be combined with each other as long as they do not conflict with each other.

The flow chart of the linear induction motor three-level dual-vector model predictive thrust control method provided by the present invention is shown in Figure 1, and specifically includes the following steps:

S1: Use the current sensor and speed sensor to sample the motor phase current and speed signals respectively; use the thrust and flux observers to calculate the motor thrust and primary flux at the current moment based on the sampling results;

Specifically, the T-shaped equivalent circuit of the linear induction motor is shown in Figure 2. Compared with the rotary induction motor, due to the primary core breaking structure, the edge effect occurs, resulting in changes in the excitation inductance during motor operation. In order to quantitatively describe this Mutual inductance changes, define function f(Q):

Among them: Q=lR ₂ /v ₂ (L _m0 +L _l2 ); l is the initial length of the motor; v ₂ is the linear speed of the motor; R ₂ is the secondary resistance of the motor; L _l2 is the secondary leakage inductance of the motor; L _m0 is Excitation inductance when the motor is stationary;

According to the equivalent circuit shown in Figure 2, the mathematical model of the linear induction motor can be expressed as:

Among them, u ₁ =u _1α +ju _1β and u ₂ =u _2α +ju _2β are the primary and secondary voltage vectors respectively; i ₁ =i _1α +ji _1β and i ₂ =i _2α +ji _2β are the primary and secondary voltage vectors respectively. Current vector; ψ ₁ ＝ψ _1α +jψ _1β and ψ ₂ ＝ψ _2α +jψ _2β are the primary and secondary flux linkage vectors; L ₁ ＝L _meq +L _l1 and L ₂ ＝L _meq +L _l2 are the primary and secondary flux vectors. Primary inductance; R ₁ and R ₂ are the primary and secondary resistances of the motor; p is the differential operator, ω ₂ is the secondary angular velocity, ω ₂ =v ₂ π/τ, τ is the motor pole pitch; L _meq is the edge considered The equivalent magnetizing inductance after the effect can be expressed as:

L _meq =L _m0 (1-f(Q)) (3)

Further specifically, based on the mathematical model in formula (2), the thrust and flux observer can be established as follows:

Among them, k and k-1 represent the motor state variables at time k and k-1 respectively, and the superscript "^" represents the observation quantity. is the magnetic leakage coefficient, F _e is the electromagnetic thrust, and T _s is the control period.

S2: Input the thrust reference value generated by the speed controller, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux vector generator into the reference primary flux vector generator, and then calculate it through the reference primary flux vector generator Output reference primary flux vector;

The structure of the reference primary flux generator is shown in Figure 3. In order to clearly explain the working principle of the reference primary flux generator, first the thrust expression of the linear induction motor is simplified as:

Among them, ω ₁ is the rotation angular speed of the primary magnetic link, that is, the synchronous angular speed, τ _c =σL ₂ /R ₂ . Apply Laplace transform to equation (6) to get:

Among them, ω _s is the slip angular velocity.

It can be seen from equation (7) that the thrust transfer function with slip angular velocity as input is equivalent to a first-order system. Therefore, a PI controller can be used to achieve good thrust tracking performance, and thereby convert the thrust error into equivalent slip angular velocity ω _s . The position change angle Δθ ₁ of the primary flux linkage is calculated from the equivalent slip angular velocity and the secondary angular velocity. The phase angle of the reference primary flux vector can be further calculated based on the primary flux phase angle θ ₁ and the position change angle Δθ ₁ at the current moment. After estimating the primary flux linkage vector at the current moment in equation (5), its phase angle θ ₁ is calculated from its α-β component:

Finally, the reference primary flux vector is obtained from the primary flux amplitude reference value and phase angle reference value, which is used as a new control target. Among them, the thrust control target is reflected in its phase angle, and the flux control target is reflected in its amplitude. By introducing a reference primary flux generator, it is possible to avoid using two reference values of different dimensions (thrust and flux) to calculate the reference voltage vector.

S3: Solve the reference voltage vector based on the reference primary flux vector, determine the candidate voltage vector combinations based on the reference voltage vector position, and calculate the duty cycle corresponding to each candidate voltage vector combination; substitute each candidate voltage vector combination and its duty cycle into the defined value function to determine the optimal voltage vector combination;

Specifically, for the delay caused by the calculation time of the actual control system, it is first necessary to further combine the motor mathematical model prediction to compensate for the impact of the delay, thereby improving the control accuracy of the controller. Based on the sampling and observation values at the current k time, the k+1 time is predicted. The prediction expression is:

Among them, the superscript k+1 represents the motor state variable at time k+1, u _opt is the equivalent primary voltage acting on the inverter at the last moment.

Based on the deadbeat principle, the reference voltage vector can be calculated by making the predicted value of the primary flux linkage at time k+2 equal to its reference value:

(a) in Figure 4 is the topology diagram of the NPC three-level inverter. In different switching states, each phase of the motor can be connected to the positive pole (P), negative pole (N) or neutral of the DC bus. point (O); therefore, the three-phase combination can produce 27 voltage vectors to be evaluated, which are divided into four categories according to their amplitudes: large vectors {u ₁ , u ₃ ,...u ₁₁ }, medium vectors {u ₂ ,u ₄ ,...u ₁₂ }, small vector {u ₁₃ ,u ₁₄ ,...u ₂₄ }, zero vector {u ₂₅ ,u ₂₆ ,u ₂₇ }, the specific distribution of each voltage vector is shown in Figure 4 As shown in (b). In (b) of Figure 4, the points where the two vectors can actually be synthesized are represented by solid lines in the figure, and the points closest to the reference voltage vector will minimize the value of the value function, thereby producing the optimal control effect; Therefore, based on the reference voltage vector to guide the selection of the optimal voltage vector combination, it can effectively avoid substituting each voltage vector combination into the value function for repeated calculation and comparison, and effectively reduce the complexity of the algorithm.

Furthermore, in order to facilitate the elimination of as many voltage vector combinations as possible based on the reference voltage vector, the inverter output voltage range is divided into six large sectors I to VI using the six mid-vectors as dividing lines. Taking sector I as an example, each boundary line corresponds to 1 to 2 voltage vector combinations: (u ₁₃ ,u ₂₅ ), (u ₁₄ ,u ₂₅ ), (u ₁₄ ,u ₁₅ ), (u ₁₃ ,u ₁₆ ), (u ₁ ,u ₁₃ ), (u ₁ ,u ₁₄ ), (u ₂ ,u ₁₃ ), (u ₂ ,u ₁₄ ), (u ₁ ,u ₂ ), (u ₁₃ ,u ₂₄ ) , (u ₁₄ ,u ₂₃ ), (u ₁ ,u ₁₂ ), (u ₁₂ ,u ₁₃ ), (u ₁₂ ,u ₁₄ ). In order to further reduce the number of voltage vector combinations that need to be evaluated, using u _1α = u _dc /3 and u _1β = 0 as the dividing line, sector I is further divided into four small areas: R1 ~ R4, as shown in Figure 5 , and further consider the number of switching actions to optimize the voltage vector combination. For example, (u ₁₃ , u ₂₅ ) and (u ₁₄ , u ₂₅ ) have exactly the same effect on thrust and flux control. However, the former produces two more switching actions than the latter, as shown in Figure 6, so only (u ₁₄ , u ₂₅ ) as candidate voltage vector combinations.

When the reference voltage vector is in the I-th sector, according to the specific area where the reference voltage vector is located and considering the optimization of the number of switches, the candidate voltage vector combinations under different situations are shown in Table 1, where and/> are the α-axis and β-axis components of the reference voltage vector respectively.

Table 1

When the reference voltage is in other sectors n, it can first be rotated to the I-th sector through coordinate transformation, and then refer to Table 1 to determine its specific area, and then determine the corresponding candidate voltage vector combinations. The coordinate transformation formula is:

Furthermore, after determining the candidate voltage vector combination, it is necessary to further determine the respective action time of the two voltage vectors, that is, the duty cycle. When the reference voltage vector is When , for the voltage vector combination (u _i , u _j ), the duty cycle d _opt of u _i can be expressed as:

Among them, · represents the dot product of two vectors, ||u _i -u _j || represents the amplitude of the vector (u _i -u _j ). Correspondingly, the duty cycle of u _j is (1-d _opt ). The duty cycle calculated by Equation (12) needs to be further constrained to the range [0,1].

On the other hand, compared with the two-level inverter, the NPC three-level inverter has a neutral point lead. As the current flowing through the neutral point changes during the operation of the motor, the mid-point voltage of the inverter changes. Changes will occur. A large offset in the inverter midpoint voltage will cause fluctuations in current and thrust. Therefore, in order to ensure the control performance of the motor drive system, the midpoint voltage needs to be controlled. The inverter midpoint voltage is defined as:

ΔU _c12 =U _dc1 -U _dc2 (13)

Among them, U _dc1 and U _dc2 are the voltages on the capacitors C ₁ and C ₂ respectively. Based on Kirchhoff's current law and the circuit topology shown in Figure 4, the midpoint voltage at k+1 moment can be predicted based on the motor phase current, capacitor voltage and the switching state that will act at the current moment:

in, is the switching state matrix of each phase bridge arm of the voltage vector u _i at time k. The numbers "1", "0" and "-1" are used to represent the switching states P, O and N respectively,/> is the three-phase current matrix, C=C ₁ =C ₂ .

On the basis of equation (14), when there is a combination of voltage vectors (u _i , u _j ) acting simultaneously on a control period, the prediction expression of the midpoint voltage at time k+1 is:

At the same time, when the switching state of a certain phase is directly switched between P and N, the phase voltage will undergo a high voltage jump, causing motor current distortion and even causing IGBT shoot-through during the state switching process. Therefore, measures need to be taken to avoid P and N Direct switching between states. It should be noted that in the switching pulse mode shown in Figure 6, there are only two situations (regardless of the order) in which direct switching of P and N can occur, as shown in Table 2.

Table 2

According to the switch state relationship given in Table 2, the constraint condition H _m is defined to avoid direct switching between P and N states, which is specifically expressed as:

where x will be a, b and c respectively. Combining the control objectives of thrust, primary flux linkage, midpoint voltage and switch state switching constraints, the designed value function is:

in, In order to further consider the midpoint voltage of delay compensation according to equation (15), k _u is the weight coefficient. Substituting each candidate voltage vector combination and its duty cycle into equation (17) in turn to calculate the value function, select the candidate voltage vector combination with the smallest value function as the optimal voltage vector combination.

S4: Generate switching pulses for each bridge arm based on the optimal voltage vector combination and its duty cycle, which act on the three-level inverter to control the linear induction motor.

Specifically, after determining the optimal voltage vector combination, the switching pulses of each bridge arm can be synthesized according to the switching states and their duty cycles of the two voltage vectors in the combination, and directly applied to the 12 NPC three-level inverters. On the IGBT, the control of the linear induction motor is completed.

Figure 7 is an architecture diagram of the linear induction motor three-level dual vector model predictive thrust control system provided by the present invention, including a thrust and primary flux observation module 100, a reference conversion module 200, an optimal voltage vector combination selection module 300 and a pulse sequence. control module 400, wherein,

The thrust and primary flux observation module 100 is used to calculate the thrust and primary flux of the motor at the current moment by using the thrust and flux observer based on the motor phase current, speed and the primary flux calculated at the previous moment;

The reference conversion module 200 is used to input the thrust reference value generated by the speed controller, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux into the reference primary flux vector generator, and output it after calculation Refer to the primary flux vector, and further solve the reference voltage vector based on the optimal reference flux vector;

The optimal voltage vector combination selection module 300 is used to determine candidate voltage vector combinations based on the position of the reference voltage vector, calculate the duty cycle corresponding to each combination, and substitute the candidate voltage vector combination and its duty cycle into the defined value function. , select the candidate voltage vector combination that minimizes the value function as the optimal voltage vector combination;

The pulse sequence control module 400 is used to generate three-phase bridge arm pulses based on the optimal voltage vector combination and its duty cycle and act on the corresponding bridge arm of the NPC three-level inverter to control the linear induction motor.

The experimental results of the dual-vector model predictive thrust control method proposed by the present invention and the traditional dual-vector model predictive thrust control without voltage vector optimization are shown in Figure 8, where the motor speed is 11m/s, the load is 50N, and the state indicates whether there is P , direct switching between N, State=1 means there is direct switching in a certain phase, State=0 means there is no direct switching. Judging from the implementation effect, the traditional method sometimes causes direct switching between P and N states, which causes large distortions in line voltage and phase current, thus affecting the control performance of the motor. The proposed method is analyzed in Table 2 and introduces constraints in the value function, which effectively prevents direct switching between P and N states.

In order to further verify the improvement effect of the proposed control method on the steady-state performance of the linear induction motor drive system, the steady-state performance of two model-predicted thrust control methods in the existing literature were compared at the same switching frequency (1.23kHz), as shown in Figure (a) in 9 is the traditional single vector model predictive thrust control, (b) is the dual vector model predictive thrust control without voltage vector optimization, and (c) is the method proposed by the present invention. Figure 9 shows the line voltage, phase current and harmonic analysis of the three methods at a motor speed of 11m/s and a load of 180N. It can be seen that the proposed method exhibits better steady-state performance at the same switching frequency by introducing dual vector control and considering voltage vector optimization.

It is easy for those skilled in the art to understand that the above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements, etc., made within the spirit and principles of the present invention, All should be included in the protection scope of the present invention.

Claims (8)

1. Three-level dual-vector model predictive thrust control method for linear induction motor, which is characterized by including the following steps: Collect motor phase current, motor speed and inverter midpoint voltage in real time; calculate the motor thrust and primary flux linkage at the current moment based on the sampling results; Taking the thrust reference value, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux as input, the reference primary flux vector is output through vector calculation; Solve the reference voltage vector based on the deadbeat principle based on the reference primary flux vector, determine the candidate voltage vector combinations based on the position of the reference voltage vector, and calculate the duty cycle corresponding to each candidate voltage vector combination; combine each candidate voltage vector combination and its duty Substitute into the defined value function, and select the candidate voltage vector combination that minimizes the value function as the optimal voltage vector combination; According to the optimal voltage vector combination and its duty cycle, each bridge arm switching pulse is generated, which acts on the three-level inverter to realize the control of the linear induction motor. 2. The control method according to claim 1, characterized in that the method of determining candidate voltage vector combinations according to reference voltage vector positions includes the following steps: All discrete voltage vectors that can be generated by the three-level inverter are divided into 6 large vectors {u ₁ , u ₃ ,...u ₁₁ } and 6 medium vectors {u ₂ , u ₄ , according to their amplitudes. ..u ₁₂ }, 12 small vectors {u ₁₃ , u ₁₄ ,...u ₂₄ } and 3 zero vectors {u ₂₅ , u ₂₆ , u ₂₇ }; among them, the amplitude of the large vector is 2/3 times u _dc ; the mid-vector amplitude is times u _dc ; the small vector amplitude is 1/3 times u _dc ; u _dc is the DC bus voltage; Using the six median vectors as dividing lines, the inverter output voltage plane is divided into six large sectors I to VI; among them, the angular bisector direction of sector I is defined as the α-axis direction, sector II and sector III The dividing line direction of is defined as the β axis direction; For the I-th sector, taking u _1α = u _dc /3 and u _1β = 0 as the dividing line, the I-th sector is further divided into 4 small areas: R1 ~ R4; determine the candidate voltage according to the specific position of the reference voltage vector vector combination, and calculate the duty cycle corresponding to each candidate voltage vector combination; for the case where the reference voltage vector is in other sectors, it can be rotated to the I-th sector through coordinate transformation, and then the conditions in the I-th sector are used Determine the area you are in, and then determine the corresponding candidate voltage vector combinations. 3. The control method according to claim 2, wherein determining the candidate voltage vector combination according to the reference voltage vector position includes: for the candidate voltage vector combination ( _ui , u _j ), comprehensive thrust, primary flux linkage, The control objective of the midpoint voltage and switch state switching constraints, the designed value function form is: Among them, k _u is the weight coefficient, In order to consider the midpoint voltage prediction value after delay compensation, d _opt is the duty cycle corresponding to the candidate voltage vector combination (u _i , u _j ), and the expression is: in, is the reference voltage vector, 1≤i,j≤27. 4. The control method according to claim 3, characterized in that _Hm is a constraint designed according to the switch state to avoid direct switching of the P and N states, and the expression is: Among them, S _x represents the switch state of the k and k+1 moments. 5. Linear induction motor three-level dual-vector model predictive thrust control system, which is characterized by: The thrust and primary flux observation module is used to calculate the motor thrust and primary flux at the current moment based on the real-time collected motor phase current, motor speed and inverter midpoint voltage; The reference conversion module is used to take the thrust reference value, the primary flux amplitude reference value, the motor thrust at the current moment, the motor speed and the primary flux as input, and output the reference primary flux through vector calculation, and further calculate the reference primary flux according to the reference value. The chain vector solves the reference voltage vector based on the deadbeat principle; The optimal voltage vector combination selection module is used to determine candidate voltage vector combinations based on the position of the reference voltage vector, calculate the duty cycle corresponding to each candidate voltage vector combination, and substitute each candidate voltage vector combination and its duty cycle into the defined value function, select the candidate voltage vector combination that minimizes the value function as the optimal voltage vector combination; The pulse sequence control module is used to generate switching pulses for each bridge arm based on the optimal voltage vector combination and its duty cycle, which acts on the three-level inverter to control the linear induction motor. 6. The control system according to claim 5, wherein the method of determining the candidate voltage vector combination according to the reference voltage vector position includes: All discrete voltage vectors that can be generated by the three-level inverter are divided into 6 large vectors, 6 medium vectors, 12 small vectors and 3 zero vectors according to their amplitude; among them, the amplitude of the large vector is 2/3 times u _dc ; the mid-vector amplitude is times u _dc ; the small vector amplitude is 1/3 times u _dc ; u _dc is the DC bus voltage; Using the six median vectors as dividing lines, the inverter output voltage plane is divided into six large sectors I to VI; among them, the angular bisector direction of sector I is defined as the α-axis direction, sector II and sector III The dividing line direction of is defined as the β axis direction; For the I-th sector, taking u _1α = u _dc /3 and u _1β = 0 as the dividing line, the I-th sector is further divided into 4 small areas: R1 ~ R4; determine the candidate voltage according to the specific position of the reference voltage vector vector combination, and calculate the duty cycle corresponding to each candidate voltage vector combination; for the case where the reference voltage vector is in other sectors, it can be rotated to the I-th sector through coordinate transformation, and then the conditions in the I-th sector are used Determine the area you are in, and then determine the corresponding candidate voltage vector combinations. 7. The control system according to claim 6, wherein determining the candidate voltage vector combination according to the reference voltage vector position includes: for the candidate voltage vector combination ( _ui , u _j ), comprehensive thrust, primary flux linkage, The control objective of the midpoint voltage and switch state switching constraints, the designed value function form is: Among them, k _u is the weight coefficient, In order to consider the midpoint voltage prediction value after delay compensation, d _opt is the duty cycle corresponding to the candidate voltage vector combination (u _i , u _j ), and the expression is: in, is the reference voltage vector, 1≤i,j≤27. 8. The control system according to claim 7, characterized in that _Hm is a constraint designed according to the switch state to avoid direct switching of the P and N states, and the expression is: Among them, S _x represents the switch state of the k and k+1 moments.