Dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage rejection

Abstract

This invention discloses a dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression. It proposes incorporating a hard constraint mechanism for common-mode voltage during the search process, effectively suppressing common-mode voltage by pruning high common-mode voltage branch nodes, thus overcoming the problem of excessively high common-mode voltage in traditional spherical decoding predictive control. The method's computational reduction effect becomes more significant with increasing prediction step size. A dynamic phase sequence search strategy is proposed, periodically rotating the search phase sequence to achieve a uniform distribution of inverter switching frequency and reduce the risk of local overheating of the converter. This invention also proposes setting a hard constraint on the upper bound of the number of search nodes for spherical decoding predictive control, which can limit the algorithm's search time as needed and avoid a surge in the number of search nodes under operating condition disturbances. Simulation results based on a three-phase five-level active neutral-point clamped inverter verify that this method can achieve uniform switching frequency distribution, common-mode voltage suppression, capacitor voltage balance, and significantly reduce computational load.

Description

Technical Field

[0001] This invention relates to the field of three-phase multilevel converter control technology, and in particular to a dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression. Background Technology

[0002] Driven by the global energy transition, the demand for high-reliability, high-power-density electric drive systems in industrial equipment has surged. Motor drive systems based on multiphase multilevel converters, with their advantages of high fault tolerance, low torque ripple, and good current quality, have been extensively researched and applied in key areas such as rail traction, ship propulsion, mining, and military and civilian aerospace applications.

[0003] The complex topology of multiphase, multilevel drive systems places high demands on control technology. The complexity and computational load of traditional vector control and finite set model predictive control increase exponentially with the number of phases and levels. Model predictive control based on the spherical decoding algorithm does not exhibit this exponential growth in computational load, effectively reducing the computational burden while maintaining current quality. However, existing spherical decoding predictive control methods use a branch-and-bound approach to search for switching states phase by phase, resulting in a higher node search frequency for the later-searched phase. This leads to locally high switching frequencies, potentially causing localized over-loss and overheating in the inverter. Asymmetrical three-phase voltage and current can also negatively impact the inverter-powered motor control system (e.g., degraded motor torque characteristics and uneven stress on power devices). Existing spherical decoding algorithms do not consider common-mode voltage suppression, which not only affects motor insulation performance but also causes bearing wear and resonance problems. Furthermore, the computational load of existing spherical decoding algorithms increases dramatically during drive system power-on and operational disturbances, easily leading to situations where calculations cannot be completed within the control cycle.

[0004] Therefore, addressing issues such as excessively high common-mode voltage, unstable computational load, and uneven distribution of switching losses is crucial for optimizing multiphase multilevel motor drive control systems. Summary of the Invention

[0005] This invention provides a dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression, which not only solves the problem of excessively high common-mode voltage but also further addresses the issue of uneven distribution of switching losses. This method effectively suppresses common-mode voltage, balances capacitor voltage, reduces computational load, and ensures uniform distribution of switching frequencies across all phases of the converter while optimizing current waveform quality.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution:

[0007] This invention provides a dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression. During the search process, hard constraints on level transitions are inserted between each dimension, and hard constraints on common-mode voltage are inserted between each prediction step. The method includes:

[0008] Determine whether the current search node meets the hard constraint condition of level transition. If it does not meet the constraint condition, then cut off the node and its subordinate branches only in the current control cycle.

[0009] When the dimension of the search node is the last dimension searched in the unit prediction step, it is determined whether the search node and its two parent nodes before it jointly satisfy the common-mode voltage suppression constraint. If they do not meet the constraint, the node and its subordinate branches are cut off only in the current control cycle.

[0010] For nodes that meet the constraints of hard level transition and common-mode voltage suppression, calculate the hypersphere center distance and determine whether it is less than the current hypersphere radius. If it is less than the hypersphere radius, continue searching for the child nodes of the node; if it is greater than the hypersphere radius, only prune the node and its subordinate branches in the current control cycle.

[0011] When the search node is the node of the last dimension of the search tree, satisfies the hard level transition and common-mode voltage suppression constraints, and the distance between the center of the sphere and the hypersphere is less than the hypersphere radius, then the path of the current search node is updated to a local optimum, and the current hypersphere radius is updated to the distance between the center of the sphere. When all nodes on all branches have been searched, the latest local optimum is the global optimum, thus obtaining the optimal switching sequence.

[0012] In one implementation, the dimension of the search tree is determined by the prediction step size N. p0 The number of sub-nodes of each node is determined by the number of phases n of the converter, while the number of switching states of the single-phase bridge arm of the multi-phase multi-level converter is determined by the number of sub-nodes n of the multi-phase multi-level converter. The dimension D of the search tree is expressed as: D = N p *n.

[0013] In one implementation, the hard constraint condition for inserting level transitions between each dimension during the search process is:

[0014] |U xo (k+1)-U xo (k)|≤U dc / 2

[0015] Among them, U xo (k+1) and U xo (k) represents the phase voltage at time k+1 and time k, respectively, U dc This is the DC bus voltage.

[0016] In one implementation, the hard constraint condition for inserting the common-mode voltage between each prediction step is:

[0017] |U ao (k)+U bo (k)+U co (k)| / 3≤U dc / 12

[0018] Among them, U xo (x = a, b, c) represents the three-phase output phase voltage, U dc This is the DC bus voltage.

[0019] In one embodiment, the step of calculating the hypersphere center distance includes:

[0020] A quadratic programming model is established for the prediction algorithm of a multi-step finite control set model that only considers reference current tracking and switching frequency constraints.

[0021] The cost function in matrix form is derived from the quadratic programming model and simplified into an unconstrained optimal solution problem. The unconstrained optimal solution of the switch sequence is then obtained.

[0022] The unconstrained optimal solution problem is described as an integer least squares problem;

[0023] Based on the mathematical relationship between the switching state combination of the multilevel converter and the voltage of the three-phase flying capacitor and the DC-side capacitor, the balance constraint of the DC-side capacitor and the flying capacitor voltage is derived.

[0024] Based on the unconstrained optimal solution problem, the voltage balance constraints of the DC-side capacitor and the flying capacitor are added to the search process of the spherical decoding algorithm. The cost function of multi-level converter prediction based on multi-objective optimization and spherical decoding is obtained, which is the distance between the center of the sphere obtained by the switching sequence without level jump.

[0025] In one embodiment, the voltage balance constraint between the DC-side capacitor and the flying capacitor is:

[0026]

[0027] Among them, U dc-upper (k+n) represents the DC-side upper bus capacitor voltage, U dc-lower (k+n) represents the DC-side lower bus capacitor voltage, U fx (k+n)(x=a,b,c) represents the three-phase flying capacitor voltage, λ dc U is the weighting coefficient for the DC-side capacitor voltage constraint term. dc λ is the DC bus voltage. fc J is the weighting coefficient for the three-phase flying capacitor voltage constraint term. xfc(k+n)(x=a,b,c) represents the equivalent sphere center distance converted from the voltage error of the flying capacitor in one of the three phases at time k+n. J dc (k+n) is the equivalent sphere center distance converted from the DC side capacitor voltage error at time k+n.

[0028] In one implementation, the cost function for multi-level converter prediction based on multi-objective optimization and spherical decoding is expressed as:

[0029]

[0030] Among them, J xfc (k+n)(x=a,b,c) represents the equivalent sphere center distance converted from the voltage error of the flying capacitor in one of the three phases at time k+n. J dc (k+n) is the equivalent sphere center distance converted from the DC-side capacitor voltage error at time k+n, J fc (k) represents the equivalent sphere center distance converted from the three-phase flying capacitor voltage error at time k, J sda (k) represents the naked sphere center distance of the switch sequence being searched at time k, J sda-sLANPC (k) represents the total center distance of the sphere corresponding to the switch sequence searched at time k.

[0031] In one implementation, at the beginning of each sampling period, the initial radius of the hypersphere is set, and the square of the initial radius R is... 2 init The optimal switching sequence U at time k-1 opt (k-1) The sum of the squares of the distance from the unconstrained optimal switching sequence shifted by one time step and the weighted distance of the capacitor voltage error at time k, plus the square of the initial hypersphere radius:

[0032]

[0033] Among them, J optdc (k) and J optfc (k) represents the weighted transformation distance of the capacitor voltage error at time k, VU ini (k) is the initial switching sequence for a unit control cycle after processing with a lower triangular matrix. This is the transformed unconstrained optimal solution.

[0034] In one embodiment, the method further includes: employing dynamic phase sequence search to periodically alternate the phase sequence search within adjacent unit sampling periods determined by the number of phases of the converter, so as to ensure uniform distribution of the switching frequency of the model prediction control based on spherical decoding, specifically including:

[0035] At the end of each unit control cycle, the search phase sequence of the search tree is switched for the next unit cycle. When the converter is a three-phase five-level active neutral-point clamping inverter, the specific switching rule is: the search phase sequence is rotated between three phase sequences: A→B→C, B→C→A, and C→A→B. That is, the search phase sequence is switched once in each control cycle, and three adjacent unit control cycles constitute one rotation cycle.

[0036] In one implementation, a hard constraint is set for the upper bound of the number of search nodes, which restricts the search to stop if the number of search nodes reaches a set value. The hard constraint for the upper bound of the number of search nodes is the maximum number of search nodes per unit control cycle under steady-state operating conditions when there is no hard constraint for the upper bound of the number of search nodes. The maximum number of search nodes is adjusted according to the actual calculation time and sampling time setting requirements.

[0037] Beneficial effects of the invention

[0038] (1) Existing spherical decoder predictive control does not consider suppressing common-mode voltage, which affects the insulation performance of the motor and causes wear and resonance problems in the motor bearings. This invention proposes to incorporate a hard constraint on common-mode voltage during the spherical decoder search process, searching only for branch nodes (switching states) with low or zero common-mode voltage. Taking the 3P5LANPC simulation results as an example, this method suppresses common-mode voltage by 75% compared to traditional spherical decoder predictive control, with the common-mode voltage not exceeding U. dc / 12.

[0039] (2) The common-mode voltage suppression method of this invention prunes the branch nodes (switching states) with high common-mode voltage at each step length in the search tree, greatly reducing the search workload. This advantage becomes more and more obvious as the prediction step length increases. 3P5LANPC simulation verification shows that, compared with the traditional spherical decoding search process, this method reduces the number of search nodes by at least 50% for prediction step lengths of three or more, reducing computational load and improving current waveform quality to some extent.

[0040] (3) Traditional spherical decoding search methods, due to their layer-by-layer (phase) node (switching state) search characteristics, result in the most frequent searching of the bottom layer (the last phase searched) node (switching state) per unit step size, meaning the last searched phase arm has the highest switching frequency. This leads to uneven switching frequencies in the inverter's three phases, causing uneven distribution of switching losses and local overheating during inverter operation. It also causes excessive asymmetry in the quality of the three-phase output voltage and current waveforms, further negatively impacting the motor control system powered by the inverter (e.g., deterioration of motor torque characteristics and uneven stress on power devices). This invention proposes a dynamic phase sequence spherical decoding search strategy. By periodically rotating the phase sequence search, the switching frequency of the inverter is evenly distributed, reducing the risk of local over-loss and overheating during inverter operation and the asymmetry of the three-phase output voltage and current.

[0041] (4) The traditional spherical decoding algorithm experiences a sharp increase in the number of search nodes during power-on startup and operating condition disturbances. This necessitates setting a longer sampling time to ensure that the calculation can be completed within a unit sampling period. However, under steady-state operation, the number of search nodes for spherical decoding is relatively low, which leads to high time costs and deteriorates the quality of the output current waveform due to the longer sampling period. Therefore, this invention sets a hard constraint on the upper bound of the number of search nodes for the search algorithm, limiting the number of search nodes. This ensures that the calculation time and computational load are controllable while maintaining the quality of the output current waveform, avoiding the problem of incomplete calculation within a unit sampling period. Attached Figure Description

[0042] The accompanying drawings, as part of this invention, are provided to further illustrate the invention. The illustrative embodiments and descriptions of the invention are used to explain the invention, but do not constitute an undue limitation thereof. Clearly, the drawings described below are merely some embodiments, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0043] Figure 1 This is a flowchart of a dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression provided in an embodiment of the present invention;

[0044] Figure 2 This is a topology diagram of a three-phase five-level active neutral-point clamping converter;

[0045] Figure 3 This is a flowchart of the calculation of the hypersphere center distance provided in one embodiment of the present invention;

[0046] Figure 4 A search tree for a dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression in three-phase five-level active neutral-point clamped converters;

[0047] Figure 5This refers to all branch nodes of the spherical decoding search tree for a three-phase five-level active neutral-point clamping converter provided in one embodiment of the present invention.

[0048] Figure 6 This invention provides a three-step prediction of three-phase line voltage, phase voltage, and current for the three-phase five-level active neutral-point clamping converter.

[0049] Figure 7 Three-step prediction of three-phase line voltage, phase voltage, and current for existing three-phase five-level active neutral-point clamping converters;

[0050] Figure 8 This invention provides a three-step prediction of the DC-side capacitor voltage for the three-phase five-level active neutral-point clamping converter.

[0051] Figure 9 Three-step prediction of DC-side capacitor voltage for existing three-phase five-level active neutral-point clamping converters;

[0052] Figure 10 This invention provides a three-step prediction of the three-phase flying capacitor voltage for the three-phase five-level active neutral-point clamping converter.

[0053] Figure 11 Three-step prediction of three-phase flying capacitor voltage for existing three-phase five-level active neutral-point clamping converters;

[0054] Figure 12 This refers to the three-step common-mode voltage of the three-phase five-level active neutral-point clamping converter of this invention;

[0055] Figure 13 The three-step common-mode voltage of the existing three-phase five-level active neutral-point clamping converter;

[0056] Figure 14 The number of search nodes for single-step predictive control of a three-phase five-level active neutral-point clamping converter without upper bound hard constraints on search nodes;

[0057] Figure 15 The number of search nodes for single-step predictive control of a three-phase five-level active neutral-point clamping converter with hard constraints on the upper bound of the search nodes.

[0058] It should be noted that these accompanying drawings and textual descriptions are not intended to limit the scope of the invention in any way, but rather to illustrate the concept of the invention to those skilled in the art by referring to specific embodiments. Detailed Implementation

[0059] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention, providing detailed implementation methods and specific operating procedures, but are not intended to limit its application, and the present invention is not limited to the following embodiments.

[0060] This disclosure provides a dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression. During the search process, hard constraints on level transitions are inserted between each dimension, and hard constraints on common-mode voltage are inserted between each prediction step. (Refer to...) Figure 1 As shown, the method specifically includes the following steps:

[0061] Step S100: Determine whether the current search node meets the hard constraint condition of level transition. If it does not meet the constraint condition, then cut off the node and its subordinate branches only in the current control cycle.

[0062] Furthermore, the dimension of the search tree is determined by the prediction step size N. p The number of sub-nodes of each node is determined by the number of phases n of the converter, and the number of switching states of the single-phase bridge arm of the multi-phase multi-level converter is determined by the number of sub-nodes n of the multi-phase multi-level converter. The expression for the dimension D of the search tree is:

[0063] D = N p *n (1)

[0064] Furthermore, a level transition hard constraint is inserted between each dimension during the search process as follows:

[0065] |U xo (k+1)-U xo (k)|≤U dc / twenty two)

[0066] Among them U xo (k+1) and U xo (k) represents the phase voltage at time k+1 and time k respectively. Nodes and their branches that do not meet the conditions of equation (2) (causing level jumps) are cut off to ensure that the output phase voltage of each phase of the multiphase multilevel converter does not jump across levels.

[0067] Step S200: When the dimension of the search node is the last dimension searched in the unit prediction step, determine whether the search node and its two previous parent nodes jointly satisfy the common-mode voltage suppression constraint. If they do not meet the constraint, then only the node and its subordinate branches are pruned in the current control cycle.

[0068] Furthermore, common-mode voltage hard constraints are inserted between each prediction step during the search process.

[0069] |U ao (k)+U bo (k)+Uco (k)| / 3≤U dc / 12 (3)

[0070] Among them U xo (x=a,b,c) represents the three-phase output phase voltage. Nodes and their branches that do not meet the conditions of equation (3) (generating high common-mode voltage) are cut off to ensure that the common-mode voltage of the converter is suppressed within a reasonable range.

[0071] When the converter is a three-phase five-level active neutral-point clamping converter, every three adjacent dimensions (A, B, C, and D phases) on the search tree form a unit prediction step, which represents the three phases (bridge arms) of the converter and their switching states.

[0072] Step S300: Calculate the hypersphere center distance for nodes that meet the constraints of hard level transition and common-mode voltage suppression, and determine whether it is less than the current hypersphere radius. If it is less than the hypersphere radius, continue searching for the child nodes of the node; if it is greater than the hypersphere radius, only cut off the node and its subordinate branches in the current control cycle.

[0073] This paper analyzes the working principle of a three-phase five-level active neutral-point clamp converter, constructs a mathematical model and cost function, and analyzes the mathematical relationship between the performance of various circuit parameters (including output phase current, phase voltage, DC-side capacitor voltage, and flying capacitor voltage) and different switching state combinations based on the topological characteristics of the three-phase five-level active neutral-point clamp converter. Taking the three-phase five-level active neutral-point clamp converter as an example... Figure 2 This is a topology diagram of a three-phase five-level active neutral-point clamping converter. (Refer to...) Figure 3 As shown, the steps for calculating the distance between the centers of the hypersphere and the center of the hypersphere include:

[0074] Step S310: Establish a quadratic programming model for the prediction algorithm of a multi-step finite control set model that only considers reference current tracking and switching frequency constraints.

[0075] In this embodiment, a quadratic programming model is established for the prediction algorithm of a multi-step finite control set model that only considers reference current tracking and switching frequency constraints, including:

[0076] Step S311: Based on the state variable coefficient matrix and input variable coefficient matrix of the current-voltage space state system predicted by the single-step model, establish the traditional single-step model for predicting the current-space state system.

[0077] The traditional single-step model for predicting the current-space state system can be represented as:

[0078] i s (k+1)=Ai s (k)+Bu(k) (4)

[0079]

[0080] In the formula, i α (k+1) and i β (k+1) represents the current value in a two-dimensional rotating coordinate system, u(k) represents the three-phase output phase voltage level, Δu(k) represents the sum of the number of three-phase level transitions per unit control cycle, R represents the load resistance, L represents the load inductance, and T represents the current value. s For the sampling period, U dc This is the DC bus voltage.

[0081] Step S312: Extend the mathematical model of the current space state system predicted by the traditional single-step model to obtain the prediction model of the multi-step model.

[0082] Traditional single-step models for predicting current-space state systems can be extended to multi-step models for prediction:

[0083] I s (k+1)=DI s (k)+γU(k) (8)

[0084]

[0085] I s (k+1)=[i s (k+1) i s (k+2)…i s (k+N p (10)

[0086] U(k)=[u(k) u(k+1)…u(k+N p -1)] (11)

[0087] Step S130: Based on the mathematical model predicted by the multi-step model, establish a quadratic programming model for the multi-step finite control set model prediction algorithm that only considers reference current tracking and switching frequency constraints.

[0088] The quadratic programming model considering only reference current tracking and switching frequency constraints can be expressed as:

[0089]

[0090] In the formula, N p To predict the step size, i sx (i+n)(x=a,b,c) represents the three-phase current at time k+n, and λ s These are the weighting coefficients for the switching frequency constraint term. Let Δu(k+n) be the three-phase output reference phase current at time k+n, and let Δu(k+n) be the total number of three-phase phase voltage level jumps per unit control cycle at time k+n.

[0091] Step S313: Based on the matrix form of the cost function derived from the quadratic programming model, simplify it into an unconstrained optimal solution problem, and find the unconstrained optimal solution of the switch sequence.

[0092] J SDA =U T (k)HU(k)+2Θ T (k)U(k)+θ(k) (13)

[0093]

[0094] Right now,

[0095]

[0096] Where, N p To predict the step size, i sx (x = a, b, c) represents the three-phase current, λ s These are the weighting coefficients for the switching frequency constraint term.

[0097] This can be further simplified to an unconstrained optimal solution problem:

[0098]

[0099] The unconstrained optimal solution for the switching sequence is:

[0100] U unc (k)=-H -1 Θ(k) (18)

[0101] The Cholesky decomposition of a symmetric positive definite matrix H is:

[0102] H = VV T (19)

[0103] Step S314: Describe the unconstrained optimal solution problem as an integer least squares problem.

[0104] The cost function of the multi-step predictive control method for multi-level converters based on multi-objective optimization and spherical decoding can be described as an integer least squares problem:

[0105]

[0106] Right now,

[0107]

[0108] Among them, J sdaJ(k) represents the distance between the naked sphere centers of the currently searched switching sequence, and U(k) represents the multi-step three-phase switching sequence (or multi-step three-phase voltage vector) being searched at time k. sda (k) represents the naked sphere center distance of the current searched switch sequence, U unc (k) is the unconstrained optimal switching sequence derived at time k. This is the transformed unconstrained optimal solution.

[0109] Step S340: Based on the mathematical relationship between the switching state combination of the multilevel converter and the voltage of the three-phase flying capacitor and the DC-side capacitor, the voltage balance constraint of the DC-side capacitor and the flying capacitor is obtained.

[0110] The mathematical relationship between the switching state combination and the DC-side capacitor voltage of a three-phase five-level active neutral-point clamping converter can be expressed as:

[0111] U dc-upper (k+n)=U dc-upper (k+n-1)+(|S a1 (k+n-1)-S a3 (k+n-1)|*i sa (k+n-1)+|S b1 (k+n-1)-S b3 (k+n-1)|*i sb (k+n-1)+|S c1 (k+n-1)-S c3 (k+n-1)|*i sc (k+n-1))*T s / C dc / 2 (23)

[0112] U dc-lower (k+n)=U dc -U dc-upper (k+n) (24)

[0113] Among them U dc-upper U dc-lower U is the DC-side capacitor voltage; fx (x = a, b, c) represents the three-phase flying capacitor voltage; S xy (x = a, b, c; y = 1, 3) represents the switching state of the three-phase bridge arm (see Table 1 for details), i sx (x = a, b, c) represents the three-phase current, U xn (x = a, b, c) represents the sum of the output phase voltage and the common-mode voltage, T s For the sampling period, C dc C is the DC-side capacitance value. fx (x = a, b, c) represents the three-phase flying capacitor value, Udc This is the DC bus voltage.

[0114] Table 1

[0115]

[0116] The mathematical relationship between the switching state combination of a three-phase five-level active neutral-point clamping converter and the voltage of the three-phase flying capacitor can be expressed as:

[0117]

[0118] Among them U fx (x = a, b, c) represents the three-phase flying capacitor voltage; S xy (x = a, b, c; y = 3, 4) represents the switching state of the three-phase bridge arm, i sx (x = a, b, c) represents the three-phase current, T s For the sampling period, C fx (x=a,b,c) represents the three-phase flying capacitor values.

[0119] The cost functions for the DC-side capacitor voltage and the flying capacitor voltage can be expressed as:

[0120]

[0121] Among them, U dc-upper (k+n) represents the DC-side upper bus capacitor voltage, U dc-lower (k+n) represents the DC-side lower bus capacitor voltage, U fx (k+n)(x=a,b,c) represents the three-phase flying capacitor voltage, λ dc λ is the weighting coefficient for the DC-side capacitor voltage constraint term. fc J is the weighting coefficient for the three-phase flying capacitor voltage constraint term. xfc (i+n)(x=a,b,c) is the equivalent sphere center distance converted from the flying capacitor voltage error of one phase in the three phases at time k+n, J dc (k+n) is the equivalent sphere center distance converted from the DC side capacitor voltage error at time k+n.

[0122] Step S350: Based on the unconstrained optimal solution problem, the voltage balance constraints of the DC side capacitor and the flying capacitor are added to the search process of the spherical decoding algorithm to obtain the cost function of the multi-level converter prediction based on multi-objective optimization and spherical decoding, that is, the distance between the center of the sphere obtained by the switching sequence without level jump.

[0123] Furthermore, the cost function of the multi-level converter predictive control method based on multi-objective optimization and spherical decoding, i.e., the formula for calculating the distance between the center of the sphere, can be derived from the switching sequence without level transitions:

[0124]

[0125] Among them, J xfc (k+n)(x=a,b,c) represents the equivalent sphere center distance converted from the voltage error of the flying capacitor in one of the three phases at time k+n. J dc (k+n) is the equivalent sphere center distance converted from the DC-side capacitor voltage error at time k+n, J fc (k) represents the equivalent sphere center distance converted from the three-phase flying capacitor voltage error at time k, K sda (k) represents the naked sphere center distance of the switch sequence being searched at time k, J sda-5LANPC (k) represents the total center distance of the sphere corresponding to the switch sequence searched at time k.

[0126] At the beginning of each sampling period, the initial radius of the hypersphere needs to be set, and the square of the initial radius R is... 2 init The optimal switching sequence U at time k-1 opt (k-1) Shift by one time step and the unconstrained optimal switching sequence (center of the ball) The distance is the sum of the squares of the distance converted by the weighted capacitor voltage error at time k.

[0127] Therefore, the square of the initial hypersphere radius can be calculated:

[0128]

[0129] Among them, J optdc (k) and J optfc (k) represents the weighted transformation distance of the capacitor voltage error at time k, VU ini (k) is the initial switching sequence for a unit control cycle after processing with a lower triangular matrix. This is the transformed unconstrained optimal solution.

[0130]

[0131] Without crossing level transitions, the distance to the center of the sphere can be calculated using equation (25) to determine whether the sphere is inside the hypersphere.

[0132] J sda-5LANPC (k)≤r 2 (k) (31)

[0133] Where, r 2 (k) represents the center-to-center distance of the currently searched optimal switch sequence, which is also the radius of the currently updated hypersphere. J sda-5LANPC (k) represents the total center-to-center distance of the current search sequence of switches.

[0134] By leveraging the properties of the lower triangular matrix V, multidimensional problems can be transformed into one-dimensional problems requiring successive computations, which also facilitates the software design of the algorithm.

[0135]

[0136] Step S400: Set the initial radius of the hypersphere. Taking the single-phase switch state of a single prediction step as a node, search for nodes in each dimension in turn. When a switch sequence is found within the hypersphere, update the hypersphere radius. After multiple searches and updates, until there is one and only one switch sequence within the hypersphere, which is the optimal switch sequence. The updated hypersphere radius is the sum of the squares of the distance from the node of the switch sequence to the unconstrained optimal switch sequence and the capacitor voltage error conversion distance.

[0137] Furthermore, taking the 3P5LANPC as the controlled object, and using three adjacent unit control cycles as one dynamic phase sequence cycle, a dynamic phase sequence search strategy is adopted to periodically rotate the phase sequence search to avoid local over-loss and overheating of the converter caused by uneven switching frequencies of each phase arm.

[0138] The dynamic phase sequence search strategy for the 3P5LANPC as the control object is as follows: Under a unit dynamic phase sequence cycle, the search phase sequence is rotated among three phase sequences: A→B→C, B→C→A, and C→A→B. That is, the search phase sequence is switched once in each control cycle, and three adjacent unit control cycles constitute one rotation cycle.

[0139] Furthermore, a hard constraint is set for the upper bound of the number of search nodes, which restricts the search to stop if the number of search nodes reaches the set value. The hard constraint for the upper bound of the number of search nodes is the maximum number of search nodes per unit control cycle under steady-state working conditions when there is no hard constraint for the upper bound of the number of search nodes. The maximum number of search nodes is adjusted according to the actual calculation time and sampling time setting requirements.

[0140] Taking 3P5LANPC as an example, an upper bound on the number of search nodes is set for the search algorithm. This means that the search terminates immediately when the number of search nodes reaches the set value per unit control cycle, and the current optimal solution is output as the global optimal solution. This ensures that the number of search nodes and the computational load are controllable, avoiding adverse consequences caused by incomplete computation within a unit control cycle. This upper bound hard constraint on the number of search nodes can typically be selected as the maximum number of search nodes per unit control cycle under steady-state operating conditions when there is no upper bound hard constraint on the number of search nodes. The upper bound constraint on the number of search nodes can be adjusted according to the actual computation time and sampling time requirements.

[0141] The specific implementation process of the dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression based on a three-phase five-level active neutral-point clamping converter is presented, aiming to reduce the difficulty and complexity of software design.

[0142] The search rules and process of the dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression based on a three-phase five-level active neutral-point clamping converter are described in the case study. Figure 4 (A search rule for a dynamic phase sequence spherical decoding multi-step predictive control method for common-mode voltage suppression in three-phase five-level active neutral-point clamped converters) can also be referenced. Figure 5 (All branch nodes of the spherical decoding search tree for a three-phase five-level active neutral-point clamping converter), in Figure 4 Nodes or branches not shown in the table are considered to have been removed because they do not meet the jump level constraint or the common-mode voltage suppression constraint.

[0143] It should be noted that after pruning the current search node in steps S100, S200, and S300 above, the search should continue to the next node in the dimension of the branch where the pruned node is located. If all nodes in the dimension of the branch where the pruned node is located have been searched, the search should proceed to the next node in the dimension of the branch where the parent node of the pruned node is located.

[0144] In an optional embodiment, the method further includes: periodically alternating the phase sequence search within a certain number of adjacent unit sampling periods using dynamic phase sequence search to ensure uniform distribution of the switching frequency of the model prediction control based on spherical decoding, specifically including:

[0145] At the end of each unit control cycle, the search phase order of the search tree is switched for the next unit cycle. The specific switching rule is as follows: the search phase order is rotated among three phase orders: A→B→C, B→C→A, and C→A→B. That is, the search phase order is switched once every control cycle, and three adjacent unit control cycles constitute one rotation cycle.

[0146] Taking a three-phase five-level active neutral-point clamping converter as an example, it has three-phase bridge arms, so the phase sequence is rotated in a dynamic phase sequence cycle with three adjacent unit control cycles.

[0147] In one optional embodiment, a hard constraint is set for the upper bound of the number of search nodes for the search algorithm, restricting the search to stop if the number of search nodes reaches the set value, and outputting the current optimal solution as the global optimal solution, so as to ensure that the computation time and computation amount are controllable.

[0148] Taking a three-phase five-level active neutral-point clamping converter as an example, an upper bound on the number of search nodes is set for the search algorithm. That is, the search is terminated immediately when the number of search nodes reaches the set value in a unit control cycle, and the current optimal solution is output as the global optimal solution. This ensures that the number of search nodes and the amount of computation are controllable and avoids the adverse consequences caused by incomplete calculation within a unit control cycle.

[0149] The embodiments of this invention use MATLAB Simulink as the simulation platform. The simulation parameters of the dynamic phase sequence spherical decoding multi-step predictive control method based on common-mode voltage suppression of a three-phase five-level active neutral-point clamping converter are shown in Table 1:

[0150]

[0151] Simulation results verified the feasibility of the method of the present invention and its four main contributions.

[0152] (1) Reduced computational load

[0153] Table 2 shows the number of spherical decoding search nodes obtained from simulation of a three-phase five-level active neutral-point clamping converter.

[0154] Table 2

[0155]

[0156] Table 2 concludes that the proposed dynamic phase sequence spherical decoding multi-step predictive control method with common-mode voltage suppression can further reduce computational complexity compared to existing spherical decoding predictive control methods. Its average number of search nodes is reduced by 2.8% for single-step prediction, 40.7% for two-step prediction, and 49.2% for three-step prediction compared to traditional spherical decoding predictive control methods. Therefore, the proposed method's advantage in reducing computational complexity becomes more significant as the prediction step size increases.

[0157] (2) Common-mode voltage suppression

[0158] contrast Figure 12 and Figure 13 Existing spherical decoding predictive control methods lack the ability to suppress common-mode voltage, which can reach U. dc / 3. The method proposed in this invention can suppress the common-mode voltage to within U. dc Within 12 seconds. Compared to existing methods, this method suppresses common-mode voltage by 75%.

[0159] (3) Switching frequency uniformity

[0160] contrast Figure 6 and Figure 7 As can be seen, the waveform asymmetry of the three-phase phase voltage and line voltage in the existing spherical decoding predictive control method is quite serious, with the voltage switching of phase C being more frequent and the phase current quality of phase C being better than that of phase A. The method proposed in this invention, however, can make the three-phase voltage waveforms more symmetrical and the waveform quality of the three-phase phase currents similar.

[0161] The three-phase five-level active neutral-point clamping converter can be divided into high-voltage and high-frequency stages due to its topological characteristics. Table 3 shows the single-step, two-step, and three-step switching frequencies of existing spherical decoder predictive control and the control method proposed in this invention, derived from simulations of this topology.

[0162] Table 3

[0163]

[0164] It can be seen that the fixed phase sequence search strategy (default phase sequence: A→B→C) used in existing traditional spherical decoder predictive control results in unequal switching frequencies for each phase bridge arm, which easily leads to localized overheating and affects the converter's service life. The method proposed in this invention uses a dynamic phase sequence search strategy, which can effectively homogenize the distribution of the three-phase switching frequencies.

[0165] (4) Current quality improvement and capacitor voltage balance

[0166] The simulation results in Table 4 verify that the method proposed in this invention can further improve the current waveform quality based on the existing spherical decoding predictive control method.

[0167] Table 4

[0168]

[0169] contrast Figure 8 and Figure 9 It can be seen that the method proposed in this invention has a better effect on DC-side capacitor voltage balance control under three-step predictive control than the existing three-step spherical decoding predictive control method.

[0170] contrast Figure 10 and Figure 11 It can be seen that the method proposed in this invention has a better effect on the voltage balance control of the three-phase flying capacitor under three-step predictive control than the existing three-step spherical decoding predictive control method.

[0171] Depend on Figure 8 and Figure 10 Thus, the present invention stably controls the DC-side capacitor voltage error and the flying capacitor voltage error to within 1.95% and 2.78%, respectively.

[0172] (5) Hard constraint on the upper bound of the search node

[0173] Figure 14 and Figure 15 This is a single-step prediction control method proposed in this invention. Figure 14 The number of search nodes increases dramatically at the moment of power-on and during changes in operating conditions on the DC bus of the inverter. At the moment of power-on, the number of search nodes reaches as high as 120, and at 0.05s, the number of search nodes reaches as high as 80 during the operating condition disturbance (reference current jumps from 2A to 3A). Figure 15 To compare the simulation results of the search node number with the addition of a hard constraint on the upper bound of the search node number (set to 60), Figure 14 and Figure 15 Therefore, the method proposed in this invention can effectively avoid the problem of a dramatic increase in the number of search nodes and computational load, while ensuring the current waveform quality under steady-state operation, and can also flexibly control the computational load and sampling time according to system requirements.

[0174] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0175] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features found in other embodiments but not others, combinations of features from different embodiments are also within the scope of protection of this invention and form different embodiments. For example, in the embodiments described above, those skilled in the art can use them in combination based on known technical solutions and the technical problems to be solved by this application.

Claims

1. A dynamic phase sequence spherical decoding multi-step length predictive control method of common-mode voltage rejection, characterized in that, The method includes inserting hard constraints on level transitions between each dimension and common-mode voltage suppression constraints between each prediction step during the search process. Determine whether the current search node meets the hard constraint condition of level transition. If it does not meet the constraint condition, then cut off the node and its subordinate branches only in the current control cycle. When the dimension of the search node is the last dimension searched in the unit prediction step, it is determined whether the search node and its two parent nodes before it jointly satisfy the common-mode voltage suppression constraint. If they do not meet the constraint, the node and its subordinate branches are cut off only in the current control cycle. For nodes that satisfy the hard constraint of level transition and the common-mode voltage suppression constraint, calculate the hypersphere center distance and determine whether it is less than the current hypersphere radius. If it is less than the hypersphere radius, continue searching for the child nodes of the node; if it is greater than the hypersphere radius, only prune the node and its subordinate branches in the current control cycle. When the search node is the node of the last dimension of the search tree, satisfies the hard constraint of level transition and the common-mode voltage suppression constraint, and the distance between the center of the sphere and the hypersphere is less than the hypersphere radius, then the path where the current search node is located is updated to a local optimum, and the current hypersphere radius is updated to the distance between the center of the sphere. When all nodes on all branches have been searched, the latest local optimum is the global optimum, thus obtaining the optimal switching sequence.

2. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 1, characterized in that, The dimension of the search tree is determined by the prediction step and the number of phases of the converter The dimension of each node is determined by the number of switching states of the single-phase bridge arm of the multi-phase multi-level converter; the dimension of the search tree is determined by the prediction step The expression is: .

3. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 1, characterized in that, The hard constraint condition for inserting level transitions between each dimension during the search process is as follows: ， in, and Let be the phase voltages at time (k+1) and time (k), respectively. This is the DC bus voltage.

4. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 1, characterized in that, The hard constraint condition for inserting common-mode voltage between each prediction step is: ， in, This refers to the three-phase output phase voltage. This is the DC bus voltage.

5. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 1, characterized in that, The steps for calculating the distance between the centers of the hypersphere include: A quadratic programming model is established for the prediction algorithm of a multi-step finite control set model that only considers reference current tracking and switching frequency constraints. The cost function in matrix form is derived from the quadratic programming model and simplified into an unconstrained optimal solution problem. The unconstrained optimal solution of the switch sequence is then obtained. The unconstrained optimal solution problem is described as an integer least squares problem; Based on the mathematical relationship between the switching state combination of the multilevel converter and the voltage of the three-phase flying capacitor and the DC-side capacitor, the balance constraint of the DC-side capacitor and the flying capacitor voltage is derived. Based on the unconstrained optimal solution problem, the voltage balance constraints of the DC-side capacitor and the flying capacitor are added to the search process of the spherical decoding algorithm. The cost function of multi-level converter prediction based on multi-objective optimization and spherical decoding is obtained, which is the distance between the center of the sphere obtained by the switching sequence without level jump.

6. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 5, characterized in that, The voltage balance constraint for the DC-side capacitor and the flying capacitor is: ， ， in, Let be the DC-side upper bus capacitor voltage at time k+n. Let be the DC-side lower bus capacitor voltage at time k+n. (x=a,b,c) represents the three-phase flying capacitor voltage at time k+n. The weighting coefficient for the DC-side capacitor voltage constraint term. This is the DC bus voltage. The weighting coefficients for the three-phase flying capacitor voltage constraint term. (x=a,b,c) represents the equivalent sphere center distance converted from the voltage error of the flying capacitor in one of the three phases at time k+n. The equivalent center distance of the sphere is the result of the DC-side capacitor voltage error at time k+n.

7. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 6, characterized in that, The cost function for multi-level converter prediction based on multi-objective optimization and spherical decoding is expressed as: ， in, This refers to a multi-step three-phase switching sequence or a multi-step three-phase voltage vector. This is the transformed unconstrained optimal solution. (x=a,b,c) represents the equivalent sphere center distance converted from the voltage error of the flying capacitor in one of the three phases at time k+n. Let be the equivalent sphere center distance converted from the DC-side capacitor voltage error at time k+n. Let be the equivalent sphere center distance converted from the three-phase flying capacitor voltage error at time k. The distance between the naked sphere centers and the switch sequence being searched at time k. This represents the total sphere center distance corresponding to the switch sequence searched at time k. To predict the step size.

8. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 7, characterized in that, At the beginning of each sampling period, the initial radius of the hypersphere is set, and the square of the initial radius of the hypersphere is... yes Optimal switching sequence at time 1 The distance between shifting one time step and the unconstrained optimal switch sequence and the first... The sum of squares of the weighted distances of the capacitor voltage error at each moment, and the square of the initial hypersphere radius: ， in, and For the first The distance of the capacitor voltage error weight conversion at any given moment. The initial switching sequence for each unit control cycle is processed by the lower triangular matrix.

9. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 1 or 8, characterized in that, The method further includes: employing dynamic phase sequence search to periodically alternate the phase sequence search within adjacent unit sampling periods determined by the number of phases of the converter, in order to ensure uniform distribution of the switching frequency of the model prediction control based on spherical decoding, specifically including: At the end of each unit control cycle, the search phase sequence of the search tree is switched for the next unit cycle. When the converter is a three-phase five-level active neutral-point clamping inverter, the specific switching rule is: the search phase sequence is rotated between three phase sequences: A→B→C, B→C→A, and C→A→B. That is, the search phase sequence is switched once in each control cycle, and three adjacent unit control cycles constitute one rotation cycle.

10. The dynamic phase sequence spherical decoding multi-step prediction control method for common-mode voltage suppression according to claim 9, characterized in that, A hard constraint is set for the upper bound of the number of search nodes, which restricts the search to stop if the number of search nodes reaches the set value. The upper bound of the number of search nodes is the maximum number of search nodes per unit control cycle under steady-state operation when there is no hard constraint for the upper bound of the number of search nodes. The maximum number of search nodes is adjusted according to the actual calculation time and sampling time setting requirements.

Images