Double random SVPWM modulation method based on improved Markov chain

Abstract

The invention provides a double-random SVPWM (Space Vector Pulse Width Modulation) method based on an improved Markov chain. The double-random SVPWM method comprises the following steps: generating a random number by combining double chaotic mapping with the Markov chain; and respectively applying the generated random number to the random switching frequency and the random zero vector action time to realize independent control of the random switching frequency and the random zero vector action time. The random number generation structure comprises the Logistic mapping unit and the Kent mapping unit, and through connection with the Markov chain state switching device, dynamic switching is achieved, the random performance of random numbers is improved, and it is guaranteed that the random numbers are evenly distributed within the value range. The Markov chain state switching device generates complex and diverse pseudo-random number sequences according to a preset probability by controlling the state transition of the mapping unit. A random number generation method is respectively applied to a random zero vector and a random frequency in a double random modulation process, so that the spectrum distribution of the system is expanded, the concentration of high-frequency harmonics is reduced, and the electromagnetic compatibility is improved.

Description

Technical Field

[0001] The present invention belongs to the field of motor control technology, and in particular relates to a method for suppressing high-frequency harmonics of a permanent magnet synchronous motor applicable to various working conditions. The method uses a random number generation technology based on a double chaotic Markov chain and an adaptive simulated annealing algorithm to optimize a double random SVPWM modulation strategy. Background Art

[0002] With the rapid development of power electronics and motor drive technology, random PWM modulation technology has become an important method to reduce harmonic interference and optimize electromagnetic compatibility. In traditional PWM modulation, the switching frequency and zero vector action time are usually fixed. Although this fixed frequency modulation method can provide stable output, it is easy to produce significant harmonic concentration effects. Especially in the case of high-frequency switching, the harmonic energy is concentrated on a specific frequency, resulting in serious electromagnetic interference (EMI) problems, which in turn affects the electromagnetic compatibility (EMC) of the system and the normal operation of other electronic equipment.

[0003] In order to solve this problem, random PWM modulation came into being. By randomizing the switching frequency or zero vector action time, it can effectively expand the spectrum, disperse the harmonic energy, and reduce the concentration of harmonics. However, the effects of the existing random PWM modulation methods under different working conditions are significantly different. Specifically, when the speed is low, random zero vector time modulation can better disperse the harmonic energy, effectively suppress the vibration and noise of the motor, and the torque pulsation is small. When the speed is high, due to the shortened zero vector action time, the effect of random zero vector modulation is significantly reduced, and its high-frequency vibration suppression ability is poor. At this time, random carrier frequency modulation has better high-frequency harmonic suppression ability.

[0004] In addition, there are two main problems in the existing random PWM modulation technology:

[0005] First, the quality of random numbers is low. The existing random number generation methods have poor randomness and uneven distribution, which affects the harmonic dispersion effect. Second, the parameters rely on manual debugging, and random parameters mostly rely on experience adjustment, which makes it difficult to achieve the optimal high-frequency harmonic suppression effect under different working conditions.

[0006] Therefore, it is urgent to propose a dual random SVPWM strategy that can combine random carrier frequency modulation and random zero vector modulation, and optimize the random number generation method and control parameters through an efficient algorithm, so that the system can achieve efficient high-frequency harmonic suppression under different operating conditions and improve the electromagnetic compatibility and operation stability of the permanent magnet synchronous motor. Summary of the invention

[0007] In view of this, the present invention aims to overcome the shortcomings of the above-mentioned problems in the prior art, and proposes a method for suppressing high-frequency harmonics of a permanent magnet synchronous motor based on a double chaotic Markov chain and an adaptive simulated annealing algorithm. By generating high-quality random numbers and optimizing PWM control parameters, double random adjustment of the switching frequency and the zero vector time is achieved, and high-frequency harmonics are effectively suppressed under different working conditions, the high-frequency noise of the motor is reduced, and the electromagnetic compatibility and operation stability of the system are improved.

[0008] The technical solution of the present invention to solve the technical problem is to provide a technical solution to achieve the above purpose, which is implemented as follows:

[0009] A first aspect of the present invention provides a double random SVPWM modulation method based on an improved Markov chain, comprising the following steps:

[0010] Step 1: Generate random numbers by combining double chaotic mapping with Markov chain;

[0011] Step 2: Apply the generated random numbers to the random switching frequency and the random zero vector action time respectively to achieve independent control of the random switching frequency and the random zero vector action time.

[0012] Furthermore, the step 1 specifically includes:

[0013] (1) Set the Logistic map to state 1, expressed as:

[0014] X n+1 =λX n (1-X n );

[0015] Among them, λ is set in the range of 3.6 to 4, the system is in a chaotic state, and the initial value is set to λ in the random frequency strategy 1 , the initial value in the random zero vector strategy is set to λ 2 , X n is the sequence iteration value, X 0 is the initial value, where the initial value is set to X in the random frequency strategy 01 , the initial value in the random zero vector strategy is set to X 02 ; The random number R1 generated by the Logistic map is in the interval [-1, 1];

[0016] (2) Set the Kent mapping to state 2, expressed as:

[0017]

[0018] where a∈(0,1), X n * ∈[0, 1], X n* is the sequence iteration value, X 0 * is the initial value, where the initial value is set to X in the random frequency strategy 01 * , parameter is set to a 1 ; The initial value in the random zero vector strategy is set to X 02 * , parameter is set to a 2 , the random number R generated by the Kent map 2 In the interval [-1, 1];

[0019] (3) State switching of Markov chain:

[0020] The Markov chain is used to dynamically switch between state 1 and state 2, and its state transition probability matrix is:

[0021]

[0022] Where P t Represents the transition probability between state 1 and state 2. Through the state control of the Markov chain, the system can dynamically choose between these two different mapping methods and generate a complex random number sequence.

[0023] Furthermore, in step 2, the independent control of the random switching frequency includes:

[0024] By randomizing the PWM switching cycle distribution, the energy of high-frequency harmonics is dispersed. Using the generated random number R 1 Control the switching frequency of the motor. The specific calculation formula is:

[0025] f′=f+K 1 R 1 ;

[0026] Among them, R 1 is a random number that changes within the interval [-1, 1]; f is the center frequency of the carrier; R 1 Generated by double chaotic mapping combined with Markov chain; K 1 is the random switching frequency gain.

[0027] Furthermore, in step 2, the independent control of the random zero vector action time includes:

[0028] Use independently generated random numbers R 2 Control the zero vector action time, the calculation formula is:

[0029] T 0 ′=T 0 +K 2 R 2 T0 ;

[0030] Among them, T 0 ′ is the zero vector time after randomization; T 0 is the initial zero vector time; the random number R 2 Generated by the same random number generation method as the switching frequency control, R 2 ∈[-1, 1]; K 2 is the random zero vector gain.

[0031] Furthermore, it also includes optimizing random parameters using an adaptive simulated annealing algorithm, and the mathematical model is:

[0032] F=w(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 )

[0033] Where w is the permanent magnet synchronous motor model, F is the maximum amplitude percentage of the high-frequency sideband current harmonics of the motor model under this set of independent variables, and the problem is transformed into finding the X when min(F) 01 , X 01 、a 1 、a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 and 2 .

[0034] Furthermore, the method of determining the optimal random parameters by using the adaptive simulated annealing algorithm specifically includes:

[0035] (1) System initial calibration: Calibrate the parameters in the simulation model based on the motor operating status collected from the experimental data to ensure that the simulation model is consistent with the experimental results under actual working conditions;

[0036] (2) Determine the parameter search range and generate initial parameters: a 1 、a 2 , P t1 , Pt2 The range of is set to (0, 1), λ 1 , 2 The range of is set to (3.6, 4), X 01 , X 02 , X 01 * , X 02 * , K 2 The range of K is set to [0, 1], 1 The range is set to [500,2500], and the above parameter range is used as the search space to provide optimization constraints; a set of initial parameter combinations x is randomly generated. 0 =(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 ), the initial solution is used for the first iterative calculation of the algorithm;

[0037] (3) Setting the objective function: The objective function is defined to minimize the percentage of high-frequency harmonic amplitude. The objective function value is calculated through the Simulink simulation model to extract the high-frequency harmonic data under steady-state conditions.

[0038] (4) Adaptive simulated annealing algorithm optimizes and updates parameters:

[0039] ① Generate new solution:

[0040] The initial temperature T is dynamically determined by the difference in the objective function of the preliminary solution, and a new solution x is generated at the current temperature T. new :

[0041] x new =x current +y·Δx;

[0042] Where y is a random disturbance in the interval [-1, 1], Δx is the search step length adjusted according to the current temperature T. The higher the temperature, the larger the step length; the lower the temperature, the smaller the step length;

[0043] ② Acceptance criteria of solution

[0044] Calculate the objective function f(x new ), compare the current solution f(x current ): If f(x new)<f(x current ), directly accept the new solution; if f(x new )≥f(x current ), according to the Metropolis criterion, it is accepted with probability P:

[0045]

[0046] Where: T is the current temperature. This mechanism allows accepting inferior solutions at higher temperatures to prevent the algorithm from falling into local optimality.

[0047] ③Temperature adaptive update

[0048] Calculate the total number of inferior solutions B generated at the current temperature T 1 and accept inferior solutions B 2 , if B 2 / B 1 >C 1 , indicating that the current temperature is high, and the temperature drop rate will increase in the next step; if B 2 / B 1 <C 2 , indicating that the current temperature is low, and the temperature drop rate will decrease in the next step, C 1 The setting range is 0.8~0.95, C 2 The setting range is 0.05~0.2, and the temperature update formula is:

[0049] T new =T current ·exp(-c·k 1 / D );

[0050] Where: c is the annealing rate constant, k is the current iteration number, and D is the parameter dimension;

[0051] (5) Calculation of objective function and iterative judgment

[0052] In each iteration, the objective function value of the current solution is calculated. If the current solution meets the requirements of the minimum objective function, the optimal parameter combination is output. If the conditions are not met, the ASA algorithm is returned to optimize and update the parameter steps, and the iteration continues until the termination condition is met.

[0053] (6) Output the optimal solution

[0054] When the algorithm meets the termination condition, the global optimal parameter combination obtained by the ASA algorithm is output: best =(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 *, P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 ), apply the optimal parameters to the actual system and verify the high-frequency harmonic suppression effect.

[0055] A second aspect of the present invention provides a dual random SVPWM modulation device based on an improved Markov chain, comprising:

[0056] A random number generating device for generating random numbers by combining a double chaotic map with a Markov chain;

[0057] The independent random adjustment device is used to apply the generated random numbers to the random switching frequency and the random zero vector action time respectively, so as to realize independent control of the random switching frequency and the random zero vector action time.

[0058] A third aspect of the present invention provides an electronic device, comprising a processor and a memory connected to the processor for storing executable instructions of the processor, wherein the processor is used to execute the above-mentioned double random SVPWM modulation method based on an improved Markov chain.

[0059] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned double random SVPWM modulation method based on an improved Markov chain.

[0060] Compared with the prior art, the double random SVPWM modulation method based on the improved Markov chain described in the present invention has the following advantages:

[0061] The present invention adopts a double random SVPWM high-frequency harmonic suppression method based on an improved Markov chain, and significantly improves the high-frequency harmonic suppression effect of a permanent magnet synchronous motor (PMSM) under different working conditions through efficient random number generation and control strategies. The method effectively disperses harmonic energy and reduces the concentration of high-frequency harmonics by adjusting the randomized switching frequency and zero vector time, thereby significantly reducing the high-frequency noise and vibration of the motor and improving the electromagnetic compatibility of the system.

[0062] Combined with the adaptive simulated annealing (ASA) algorithm, the present invention optimizes random number generation and PWM control parameters, so that the system can achieve the best high-frequency harmonic suppression effect under different operating conditions. This adaptability makes the present invention more flexible and applicable when dealing with different operating conditions.

[0063] The present invention utilizes a combination of Logistic and Kent chaotic mapping and dynamically switches states through a Markov chain to generate a pseudo-random number sequence with high randomness and complexity. This high-quality random number generation method enhances the effectiveness of the overall control strategy and ensures the optimization effect of harmonic suppression. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0065] Figure 1 It is a schematic diagram of the double random SVPWM control structure based on double chaotic Markov chains and ASA algorithm of the present invention;

[0066] Figure 2 It is a two-state Markov chain state transition diagram based on double chaos of the present invention;

[0067] Figure 3 It is an overview diagram of the optimization parameters of the present invention;

[0068] Figure 4 This is a flow chart of optimization based on the ASA algorithm of the present invention;

[0069] Figure 5 This is a schematic diagram of simulation results when the rotation speed is 1000 rpm and the torque is 4 N·m in the simulation experiment of the present invention;

[0070] Figure 6 This is a schematic diagram of simulation results when the rotation speed is 1500 rpm and the torque is 6 N·m in the simulation experiment of the present invention;

[0071] Figure 7 This is a schematic diagram of simulation results when the rotation speed is 2000 rpm and the torque is 8 N·m in the simulation experiment of the present invention;

[0072] Figure 8 This is a schematic diagram of simulation results when the rotation speed is 2500 rpm and the torque is 10 N·m in the simulation experiment of the present invention;

[0073] Fig. 9 A comparison chart of the simulation effects of the present invention. DETAILED DESCRIPTION

[0075] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0076] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and the like are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", and the like may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0077] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood by specific circumstances.

[0078] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0079] Embodiment 1:

[0080] like Figure 1 As shown, the present invention provides a double random SVPWM modulation method based on an improved Markov chain, and the implementation process is as follows:

[0081] The dual random SVPWM module generates randomized switching cycles (switching frequencies) and random zero vectors based on the random number generation method of Logistic and Kent chaos combined with the two-state Markov algorithm. By randomizing the switching frequency and zero vector, high-frequency harmonics and electromagnetic interference are suppressed.

[0082] In the random number generation method, Logistic chaos and Kent chaos are used to generate random numbers with chaotic characteristics, respectively as two states of the Markov chain, as the source of randomness;

[0083] Two-state Markov algorithm: used to optimize the distribution of random number sequences and improve randomness and stability through state transitions.

[0084] ASA: Find the optimal parameters of the random strategy so that it can adapt to different operating conditions and improve overall efficiency.

[0085] The present invention adopts a chaotic random number generation method combining Logistic mapping and Kent mapping, dynamically switches between the two mappings through Markov chain control, generates pseudo-random numbers with high randomness and complexity, and improves the adjustment flexibility of the system.

[0086] (1) Random number generation method based on double chaotic mapping combined with Markov chain

[0087] The specific process is as follows:

[0088] 1) Logistic Mapping (State 1):

[0089] X n+1 =λX n (1-X n )

[0090] Among them, λ is in the range of 3.6 to 4, and the system is in a chaotic state. In the random frequency strategy, the initial value is set to λ 1 , the initial value in the random zero vector strategy is set to λ 2 , X n is the sequence iteration value, X 0 is the initial value. In the random frequency strategy, the initial value is set to X 01 , the initial value in the random zero vector strategy is set to X 02 .

[0091] 2) Kent Mapping (State 2)

[0092]

[0093] where a∈(0,1), X n * ∈[0, 1], X n * is the sequence iteration value, X 0 * is the initial value. In the random frequency strategy, the initial value is set to X 01 * , parameter is set to a 1 ; The initial value in the random zero vector strategy is set to X 02 * , parameter is set to a 2 The random number R generated by the Kent map 2 It is also in the interval [-1, 1].

[0094] 3) State switching of Markov chain:

[0095] The Markov chain is used to dynamically switch between state 1 (Logistic mapping) and state 2 (Kent mapping), and its state transition probability matrix is

[0096]

[0097] Where P t represents the transition probability between state 1 and state 2. Through the state control of the Markov chain, the system can dynamically select between these two different mapping methods based on the transition probability, thereby generating a complex random number sequence. The state transition diagram of the two-state Markov chain based on double chaos is shown in Figure 2 As shown. The two control strategies of random switching frequency and random zero vector select the state transfer matrix P t1 and P t2 .

[0098] (2) High-frequency harmonic suppression control strategy

[0099] The dual random strategy of the present invention will be applied to the random switching frequency (changing the switching period of the PWM wave) and the random zero vector action time (adjusting the zero vector action time in each PWM period) respectively through the above random number generation method. Although they use the same random number generation mechanism, the two random processes are independent of each other and do not interfere with each other. The combination of the dual random strategy makes the harmonic spectrum more dispersed, avoiding vibration and noise caused by harmonic concentration.

[0100] 1) Random switching frequency:

[0101] By randomizing the PWM switching cycle distribution, the energy of high-frequency harmonics is dispersed. Using the generated random number R 1 Control the switching frequency of the motor. The specific calculation formula is:

[0102] f′=f+K 1 R 1

[0103] Among them, R 1 is a random number that changes within the interval [-1, 1]; f is the center frequency of the carrier; R 1 Generated by double chaotic mapping combined with Markov chain; the carrier center frequency of the present invention is set to 8000Hz; K 1 is the random switching frequency gain.

[0104] 2) Random zero vector:

[0105] By randomly adjusting the distribution of zero vector time, the concentration effect of high-frequency harmonics can be further reduced. 2 Control the zero vector action time, the calculation formula is T 0 ′=T0 +K 2 R 2 T 0

[0106] Among them, T 0 ′ is the zero vector time after randomization; T 0 is the initial zero vector time; the random number R 2 Generated by the same random number generation method as the switching frequency control, R 2 ∈[-1, 1]; K 2 is the random zero vector gain.

[0107] (3) Random parameter optimization based on ASA algorithm

[0108] Through analysis, we can know that the initial parameters (X 01 , X 02 、a 1 、a 2 , X 01 * , X 02 * , P t1 , P t2 , 1 , 2 ), random switching frequency gain (K 1 ) and random zero vector gain (K 2 ), all correspond to different permanent magnet synchronous motor output current waveform spectra, thus affecting the high-frequency sideband harmonic suppression effect. At the same time, these parameters need to limit the value range in the system parameter design to avoid occupying a large computing memory of the main control chip. Therefore, in the present invention, an adaptive simulated annealing (ASA) algorithm is used to find the best parameter combination by fast optimization within a certain parameter range. This problem can be abstracted into the following mathematical expression

[0109] F=w(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 );

[0110] Where w is the permanent magnet synchronous motor model selected in this paper, and F is the maximum amplitude percentage of the high-frequency sideband current harmonics of the motor model under this set of independent variables. The problem is then transformed into finding the X when min(F) is 01 , X 01 、a 1 、a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 and 2 The optimization parameters overview is shown in the figure below: Figure 3 shown.

[0111] The ASA algorithm is a global optimization algorithm that searches for the optimal solution by simulating the energy changes in the physical annealing process. Unlike traditional simulated annealing (SA), the ASA algorithm dynamically adjusts the annealing parameters according to the characteristics of the problem, making the search process more flexible and efficient. The core idea of ​​ASA is to dynamically change the temperature and search step size so that the algorithm can explore more solution spaces in the early stage and more accurately approach the global optimal solution in the later stage. The optimization flow chart based on the ASA algorithm is as follows: Figure 4 shown.

[0112] The screening and solving process of the global optimal solution is as follows:

[0113] 1) Initial system calibration: According to the motor operating status (such as phase current data) collected from the experimental data, calibrate the parameters in the simulation model (such as d-axis inductance, winding resistance, etc.) to ensure that the simulation model is consistent with the experimental results under actual working conditions, providing an accurate objective function evaluation basis for the ASA algorithm.

[0114] 2) Determine the parameter search range and generate initial parameters: a 1 、a 2 , P t1 , P t2 The range of is set to (0, 1), λ 1 , 2 The range of is set to (3.6, 4), X 01 , X 02 , X 01 * , X 02 * , K 2 The range of K is set to [0, 1], 1 The range is set to [500,2500], and the above parameter range is used as the search space to provide optimization constraints. A set of initial parameter combinations X is randomly generated.0 =(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 ), the initial solution is used for the first iterative calculation of the ASA algorithm.

[0115] 3) Setting the objective function: The objective function is defined to minimize the percentage of high-frequency harmonic amplitude. The objective function value is calculated through the Simulink simulation model to extract the high-frequency harmonic data under steady-state conditions.

[0116] 4) ASA algorithm optimizes and updates parameters:

[0117] ① Generate new solution:

[0118] The initial temperature T is dynamically determined by the difference in the objective function of the preliminary solution. At the current temperature T, a new solution x is generated new :

[0119] x new =x current +y·Δx

[0120] Where y is a random disturbance in the interval [-1, 1], △x is the search step adjusted according to the current temperature T. The higher the temperature, the larger the step; the lower the temperature, the smaller the step.

[0121] ② Acceptance criteria of solution

[0122] Calculate the objective function f(x new ), compare the current solution f(x current ): If f(x new )<f(x current ), directly accept the new solution; if f(x new )≥f(x current ), according to the Metropolis criterion, it is accepted with probability P:

[0123]

[0124] Where: T is the current temperature. This mechanism allows accepting inferior solutions at higher temperatures to prevent the algorithm from falling into a local optimum.

[0125] ③Temperature adaptive update

[0126] Calculate the total number of inferior solutions B1 generated at the current temperature T and the number of inferior solutions B accepted 2 If B 2 / B 1 >C 1 , indicating that the current temperature is high, and the temperature drop rate will increase in the next step; if B 2 / B 1 <C 2 , indicating that the current temperature is low and the temperature drop rate will decrease in the next step. 1 The setting range is 0.8~0.95, C 2 The setting range is 0.05~0.2. The temperature update formula is:

[0127] T new =T current ·exp(-c·k 1 / D )

[0128] Where: c is the annealing rate constant, k is the current number of iterations, and D is the parameter dimension.

[0129] 5) Calculation of objective function and iterative judgment

[0130] In each iteration, the objective function value of the current solution is calculated. If the current solution meets the requirements of the minimum objective function, the optimal parameter combination is output. If the conditions are not met, the ASA algorithm is returned to the optimization and parameter update step, and the iteration continues until the termination condition is met.

[0131] 6) Output the optimal solution

[0132] When the algorithm meets the termination condition, the global optimal parameter combination obtained by the ASA algorithm is output: best =(X 01 , X 02 , a 1 , a 2 , X 01 * , X 02 * , P t1 , P t2 , K 1 , K 2 ,λ 1 ,λ 2 ), apply the optimal parameters to the actual system and verify the high-frequency harmonic suppression effect.

[0133] In the Simulink simulation environment, simulation models of the traditional SVPWM control method, the dual random SVPWM control method and the new dual random SVPWM control method proposed in this invention under different working conditions are built. The parameters are set as follows: bus voltage is 311V and switching frequency is 8kHz. Working conditions are set as follows: ① speed is 1000rpm and torque is 4N·m; ② speed is 1500rpm and torque is 6N·m; ③ speed is 2000rpm and torque is 8N·m; ④ speed is 2500rpm and torque is 10 N·m. The simulation results are as follows: Figure 5-9 shown.

[0134] The optimal parameters obtained by optimization are shown in Table 1 (retain four decimal places):

[0135] Table 1

[0136]

[0137]

[0138] Through simulation experiments, it can be seen that the high-frequency current harmonic suppression effect is significant. The maximum amplitude percentage of high-frequency current harmonics before suppression fluctuates between 0.80% and 1.33% at different speeds and loads, indicating that high-frequency harmonics have a greater impact on the system, especially at medium speed and medium load (1500rpm, 6N.m), reaching a peak of 1.33%. After suppression, the optimization strategy reduces the maximum amplitude percentage of high-frequency harmonics to 0.19% to 0.29%, greatly reducing the high-frequency harmonic components. The reduction is significant. Under all working conditions, the reduction of high-frequency harmonics exceeds 75%, the maximum reduction reaches 80% (2000rpm, 8N.m), and the minimum reduction is 76.25% (2500rpm, 10N.m), which fully verifies the effectiveness of the optimization strategy. Therefore, this optimization method can significantly reduce the amplitude of high-frequency harmonics at different speeds and loads, providing strong support for reducing high-frequency noise and vibration of motors. After suppression, the power spectrum density amplitude dropped significantly, ranging from 68.95% to 88.10%. The significant drop in the power spectrum density amplitude shows that the optimization strategy has shown good results in electromagnetic vibration and noise suppression, effectively reducing high-frequency vibration and noise problems during motor operation.

[0139] In summary, the present invention generates random number sequences independently through Logistic mapping and Kent mapping, and uses the Markov chain control module to dynamically switch the mapping state to generate a pseudo-random number sequence with high randomness and complexity. This method effectively solves the problems of insufficient randomness and uneven distribution of existing random number generation methods, and provides high-quality randomness support for random switching frequency and random zero vector time modulation.

[0140] The present invention applies high-quality random number generation methods to dual random SVPWM modulation, including random switching frequency modulation and random zero vector time modulation. The random switching frequency disperses high-frequency harmonic energy and avoids harmonic concentration by changing the switching cycle distribution of PWM; the random zero vector time further reduces high-frequency noise and vibration by randomly adjusting the zero vector action time. Although the two modulation methods are independent of each other, they achieve optimized coupling through the same random number generation mechanism, significantly improving the harmonic dispersion effect and electromagnetic compatibility.

[0141] In order to further optimize the random number generation method and PWM control parameters, the present invention introduces the adaptive simulated annealing (ASA) algorithm to achieve global optimization of random parameters. The ASA algorithm dynamically adjusts the annealing rate and search step size to minimize the maximum amplitude percentage of high-frequency harmonics, and dynamically optimizes key parameters such as random switching frequency gain, random zero vector gain, and initial parameters of chaos mapping. Through this optimization process, the optimal harmonic suppression effect is ensured in the full range of operating conditions.

[0142] The applicability and compatibility of the present invention are excellent. The optimization strategy shows good stability and consistency under different speeds (1000rpm~2500rpm) and loads (4N·m~10N·m), especially under high-speed and high-load conditions. By optimizing the switching frequency and zero vector time, the present invention effectively disperses the harmonic energy, reduces the vibration and noise of the motor, and maintains high efficiency and stable operating performance.

[0143] Embodiment 2:

[0144] A double random SVPWM modulation device based on an improved Markov chain, comprising:

[0145] A random number generating device for generating random numbers by combining a double chaotic map with a Markov chain;

[0146] The independent random adjustment device is used to apply the generated random numbers to the random switching frequency and the random zero vector action time respectively, so as to realize the control of the random switching frequency and the random zero vector action time.

[0147] Embodiment three:

[0148] An electronic device comprises a processor and a memory which is in communication with the processor and is used to store executable instructions of the processor. The processor is used to execute the above-mentioned double random SVPWM modulation method based on an improved Markov chain.

[0149] Embodiment 4:

[0150] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the double random SVPWM modulation method based on an improved Markov chain is implemented.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

[0152] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A double random SVPWM modulation method based on an improved Markov chain, characterized in that: The steps include: Step 1: Generate random numbers by combining double chaotic mapping with Markov chain; Step 2: Apply the generated random numbers to the random switching frequency and the random zero vector action time respectively to achieve independent control of the random switching frequency and the random zero vector action time.

2. The random PWM modulation method based on a double random strategy according to claim 1, characterized in that: The step 1 specifically includes: (1) Set the Logistic map to state 1, expressed as: X n+1 =λX n (1-X n ); Among them, λ is set in the range of 3.6 to 4, the system is in a chaotic state, where the initial value is set to λ1 in the random frequency strategy, and the initial value is set to λ2 in the random zero vector strategy, X n is the sequence iteration value, X0 is the initial value, and the initial value is set to X in the random frequency strategy 01 , the initial value in the random zero vector strategy is set to X 02 ; The random number R1 generated by the Logistic map is in the interval [-1,1]; (2) Set the Kent mapping to state 2, expressed as: where a∈(0,1), X n * ∈[0,1],X n * is the sequence iteration value, X0 * is the initial value, where the initial value is set to X in the random frequency strategy 01 * , the parameter is set to a1; the initial value in the random zero vector strategy is set to Xx2 * , the parameter is a2, the random number R2 generated by the Kent mapping is in the interval [-1,1]; (3) State switching of Markov chain: The Markov chain is used to dynamically switch between state 1 and state 2, and its state transition probability matrix is: Where P t Represents the transition probability between state 1 and state 2. Through the state control of the Markov chain, the system can dynamically choose between these two different mapping methods and generate a complex random number sequence.

3. The random PWM modulation method based on a double random strategy according to claim 2, characterized in that: In step 2, the independent control of the random switching frequency includes: By randomizing the PWM switching cycle distribution, the energy of high-frequency harmonics is dispersed, and the switching frequency of the motor is controlled by using the generated random number R1. The specific calculation formula is: f′=f+K1R1; Among them, R1 is a random number that changes in the interval [-1,1]; f is the carrier center frequency; R1 is generated by double chaotic mapping combined with Markov chain; K1 is the random switching frequency gain.

4. The random PWM modulation method based on a double random strategy according to claim 2, characterized in that: In step 2, the independent control of the random zero vector action time includes: Use the independently generated random number R2 to control the zero vector action time. The calculation formula is: T0′=T0+K2R2T0; Among them, T0′ is the zero vector time after randomization; T0 is the initial zero vector time; the random number R2 is generated by the same random number generation method as the switching frequency control, R2∈[-1,1]; K2 is the random zero vector gain.

5. The random PWM modulation method based on double random strategy according to claim 2, characterized in that: It also includes using an adaptive simulated annealing algorithm to determine the optimal random parameters. The mathematical model is: F=w(X 01 ,X 02 ,a1,a2,X 01 *,X 02 *,P t1 ,P t2 ,K1,K2,λ1,λ2 ) ; Where w is the permanent magnet synchronous motor model, F is the maximum amplitude percentage of the high-frequency sideband current harmonics of the motor model under this set of independent variables, and the problem is transformed into finding the X when min(F) 01 , X 01 , a1, a2, X 01 *、X 02 *、P t1 , P t2 , K1, K2, λ1 and λ2.

6. The random PWM modulation method based on double random strategy according to claim 5, characterized in that: The method of using the adaptive simulated annealing algorithm to determine the optimization random parameters specifically includes: (1) System initial calibration: Calibrate the parameters in the simulation model based on the motor operating status collected from the experimental data to ensure that the simulation model is consistent with the experimental results under actual working conditions; (2) Determine the parameter search range and generate initial parameters: a1, a2, P t1 , P t2 The range of is set to (0,1), the range of λ1,λ2 is set to (3.6,4), and X 01 , X 02 , X 01 * , X 02 * , K2 is set to [0,1], K1 is set to [500,2500], and the above parameter ranges are used as search space to provide optimization constraints; a set of initial parameter combinations x0=(x0 1， x 02 , a1, a2, X 01 * , X 02 * , P t1 , P t2 , K1, K2, λ1, λ2), the initial solution is used for the first iterative calculation of the algorithm; (3) Setting the objective function: The objective function is defined to minimize the percentage of high-frequency harmonic amplitude. The objective function value is calculated through the Simulink simulation model to extract the high-frequency harmonic data under steady-state conditions. (4) Adaptive simulated annealing algorithm optimizes and updates parameters: ① Generate new solution: The initial temperature T is dynamically determined by the difference in the objective function of the preliminary solution, and a new solution x is generated at the current temperature T. new :x new =x current +y·Δx; Where y is a random disturbance in the interval [-1,1], Δx is the search step length adjusted according to the current temperature T. The higher the temperature, the larger the step length; the lower the temperature, the smaller the step length; ② Acceptance criteria of solution Calculate the objective function f(x new ), compare the current solution f(x current ): If f(x new ) <f(x current ), directly accept the new solution; if f(x new )≥f(x current ), according to the Metropolis criterion, it is accepted with probability P: Where: T is the current temperature. This mechanism allows accepting inferior solutions at higher temperatures to prevent the algorithm from falling into local optimality. ③Temperature adaptive update Calculate the total number of inferior solutions B1 and the number of accepted inferior solutions B2 generated at the current temperature T. If B2 / B1>C1, it means that the current temperature is high and the temperature drop rate in the next step will increase; if B2 / B1<C2, it means that the current temperature is low and the temperature drop rate in the next step will decrease. The setting range of C1 is 0.8~0.95, and the setting range of C2 is 0.05~0.

2. The temperature update formula is: T new =T current ·exp(-c·k 1 / D ); Where: c is the annealing rate constant, k is the current iteration number, and D is the parameter dimension; (5) Calculation of objective function and iterative judgment In each iteration, the objective function value of the current solution is calculated. If the current solution meets the requirements of the minimum objective function, the optimal parameter combination is output. If the conditions are not met, the ASA algorithm is returned to optimize and update the parameter steps, and the iteration continues until the termination condition is met. (6) Output the optimal solution When the algorithm meets the termination condition, the global optimal parameter combination obtained by the ASA algorithm is output: best =(X 01 , X 02 , a1, a2, X 01 * , X 02 * , P t1 , P t2 , K1, K2, λ1, λ2), apply the optimal parameters to the actual system and verify the high-frequency harmonic suppression effect.

7. A double random SVPWM modulation device based on an improved Markov chain, characterized in that: include: A random number generating device for generating random numbers by combining a double chaotic map with a Markov chain; The independent random adjustment device is used to apply the generated random numbers to the random switching frequency and the random zero vector action time respectively, so as to realize independent control of the random switching frequency and the random zero vector action time.

8. An electronic device, comprising a processor and a memory connected to the processor for storing instructions executable by the processor, characterized in that: The processor is used to execute the double random SVPWM modulation method based on improved Markov chain as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the double random SVPWM modulation method based on an improved Markov chain as described in any one of claims 1 to 6 is implemented.

Images