Double-random PWM motor noise suppression method based on Mersen rotation algorithm

Abstract

The invention relates to the technical field of motor control, and particularly discloses a double-random PWM motor noise suppression method based on a Mersen rotation algorithm. Comprising the following steps that S01, double random modulation is established, the double random modulation comprises random pulse position modulation and random switching period modulation, the random pulse position modulation is used for conducting controllable random offset on the geometric center position of a carrier signal, and the random switching period modulation is used for conducting random offset on the geometric center position of the carrier signal; the random switching period modulation changes the frequency fs of the triangular carrier wave to realize randomization of the random switching period Tn of the system; and S02, generating a pulse position offset coefficient required by random pulse position modulation and a periodic random number required by random switching period modulation by adopting a Meisson rotation algorithm, and dispersing PWM harmonic energy to other frequency intervals so as to suppress a switching frequency sub-harmonic amplitude and a switching frequency sub-second harmonic amplitude.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of motor control, and particularly discloses a double-random PWM motor noise suppression method based on a Mason rotation algorithm. BACKGROUND

[0002] An alternating current speed regulation system has wide application prospects in the fields of industry, electric vehicles and navigation as an important part of the electrical drive field. A permanent magnet synchronous motor has the advantages of high power density, fast dynamic response, good electromagnetic compatibility, strict speed synchronization and wide speed regulation range, and has become one of the key research directions of alternating current speed regulation in recent years. In modern propulsion systems, the space vector pulse width modulation technology has become one of the most commonly used modulation methods for driving PMSM systems due to its high DC bus voltage utilization rate and low switching loss. However, when the traditional SVPWM technology is used to modulate the switching tubes of the inverter, the high voltage change rate will cause the energy to be highly concentrated near the carrier frequency and its integer multiples, thereby generating high-frequency PWM harmonics. The high-frequency PWM harmonics not only cause sharp noise of the power device, reduce the acoustic comfort of the motor drive system, but also increase the electromagnetic interference of the drive system and affect the normal operation of other common power electronic equipment. Therefore, solving the problem of high-frequency electromagnetic noise caused by PWM harmonics is crucial for the analysis of the vibration and noise mechanism of the speed regulation system and the research of suppression strategies.

[0003] For an integrated motor drive system, there are mainly two ways to reduce high-frequency electromagnetic noise: one is to add a filter outside the system to cut off the propagation path of high-frequency electromagnetic noise, but the dynamic response of the passive filter is slow in the process of rapid change, and the universality is poor. The active filter can obviously improve the shortcomings of the passive filter, but the compensation effect is affected by factors such as compensation circuit, the structure is complex, the space occupation is large, and the volume and weight of the whole system are increased, which is not conducive to miniaturization and light weight. The other is to use random PWM technology to reduce the electromagnetic interference level of the interference source from the noise source by changing the switching mode, the basic idea of which is to randomize the modulation parameters in the switching pulse according to a certain distribution rule, so that the discrete frequency spectrum with concentrated energy output from the fixed frequency is changed into a continuous frequency spectrum with uniform distribution. One advantage of this technology is that it does not need to increase the volume and weight of the system, and the topology structure has high universality, good control performance of switching loss and current ripple, thereby solving the problem of prominent frequency spectrum characteristics caused by PWM modulation.

[0004] In summary, at present, the research on suppressing high-frequency noise by random PWM technology has achieved certain research results at home and abroad, but mainly uses single random PWM technology for suppression, and is mainly applied in the Buck circuit of DC-DC and three-phase induction motor. Therefore, there is still a lot of research space for switching frequency harmonic suppression in the permanent magnet synchronous motor drive system

[0005] The application provides a double-random PWM motor noise suppression method based on a Mason rotation algorithm to solve the above problems. SUMMARY

[0006] The application aims to solve the problem of high-frequency PWM harmonics generated by inverter switches in a permanent magnet synchronous motor drive system under a traditional SVPWM modulation method.

[0007] To achieve the above purpose, the application provides the following basic scheme:

[0008] A double-random PWM motor noise suppression method based on a Mason rotation algorithm, comprising the following steps:

[0009] S01: Establishing double-random modulation, which is random pulse position modulation and random switching period modulation, the random pulse position modulation is used for controllable random offset of the geometric center position of the carrier signal, and the random switching period modulation changes the frequency f of the triangular carrier s Realize the randomization of the system random switching period T n ;

[0010] S02: Using the Mason rotation algorithm to generate the pulse position offset coefficient required by the random pulse position modulation and the period random number required by the random switching period modulation, dispersing the PWM harmonic energy to other frequency intervals, and achieving suppression of the amplitude of the switching frequency sub-harmonic and the amplitude of the switching frequency sub-harmonic.

[0011] Further, it further comprises an SVPWM module for realizing the random pulse position modulation and the random switching period modulation, a pseudo-random number generator for providing the pulse position offset coefficient and the period random number for the SVPWM module, and an inverter module, wherein the Mason rotation algorithm fills in the pseudo-random number generator to generate the pulse position offset coefficient and the period random number.

[0012] Further, the random pulse position modulation changes the carrier tilt angle θ n , so as to randomize the pulse position offset coefficient σ n of the geometric center position, specifically as follows:

[0013]

[0014] Wherein, d n is the duty ratio, T s is the fixed switching period in the conventional technology, R n is a self-defined random position factor, and Δθ=π / 2-arctan(1 / 2).

[0015] Further, the period random number T nThe relationship between the fixed switching period T s in the conventional technology and the switching period T

[0016] T n =(1+Y n )T s ;

[0017] wherein T s =1 / f s , Y n is a random number between -1 and 1, and f s is a fixed period frequency.

[0018] Further, the Mead rotation algorithm comprises the following steps:

[0019] First, initializing the rotation chain: a non-zero constant is selected as an initial seed of the algorithm input and written into a binary ω-dimensional row vector Z0, the low r bits and the high ω-r bits in the row vector Z r-1 are subjected to an exclusive OR operation, and then the result is multiplied by a constant multiplier f and added by r-1 to obtain a new row vector Z r Based on the formula: a complete basic Mead rotation chain, i.e., {Z0, Z1,..., Z v-1}, can be obtained by repeatedly iterating v-1 times.

[0020] Second, twisting and updating the rotation chain: when the index reaches the iteration upper limit v, the highest bit of the row vector Z of the xth iteration and the lowest ω-1 bits of Z are connected to form a new state vector Z (x) , then the state vector Z offset to the right by m dimensions and Z (x) are subjected to an exclusive OR operation and right shifted by one bit, and finally subjected to an exclusive OR operation with a specific integer l to generate a new state quantity Z

[0021] Finally, generating a pseudo-random number: the output calculation result is subjected to a series of fixed bit operations and multiplied by different coefficients to disturb the bit pattern, so that the random sequence can be better mapped on the interval of R n and Y n , and a pseudo-random number generation formula is obtained, which is as follows: wherein Z is a state quantity, u, g and h are bit numbers of vector shifting of the state quantity, and the values are all less than 32. Further, if the iteration number does not reach the upper limit v, the output calculation result is the basic Mead rotation chain in the first step, and if the iteration number reaches the upper limit v, the output calculation result is the state quantity after the twisting and updating in the second step.

[0022] The principle and effects of the scheme are that:

[0023] 1. Compared with the prior art, the present invention improves the pseudo-random number generation algorithm and uses a double random structure of random pulse position and random switching frequency to reduce the amplitude of the subharmonic of the switching frequency and the harmonic amplitude of the second harmonic of the switching frequency, thereby greatly suppressing the noise caused by PWM harmonics and improving the acoustic comfort of the motor system.

[0024] 2. Compared with existing technologies, the Mason rotation algorithm, based on linear recursion of matrices over a finite binary field, allows for a high loop iteration length of up to 2^n. 19937 -1 can quickly generate high-quality pseudo-random numbers. Taking the 16-bit linear congruential method as an example, the maximum cycle period is 2. 16 The same-digit Mason rotation algorithm has a cycle time as high as 2. 624×16 -1, and the algorithm can achieve an accuracy of 623 dimensions and 32 bits. The generated pseudo-random binary sequence is more uniformly distributed within its cycle period compared to the traditional linear congruent method, and can achieve better random effects. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 The diagram shows a control system block diagram of a dual random PWM motor noise suppression method based on the Mason rotation algorithm proposed in an embodiment of this application;

[0027] Figure 2 This illustration shows a schematic diagram of pulse position offset coefficient modulation in a dual random PWM motor noise suppression method based on Mason rotation algorithm proposed in an embodiment of this application;

[0028] Figure 3 The diagram shows the random switching cycle modulation in a dual random PWM motor noise suppression method based on the Mason rotation algorithm proposed in this application embodiment;

[0029] Figure 4 The flowchart illustrating the initialization of the Mersenne rotation chain in a dual-random PWM motor noise suppression method based on the Mersenne rotation algorithm proposed in this application is shown.

[0030] Figure 5 The flowchart of updating the Mersenne rotation chain in a dual random PWM motor noise suppression method based on the Mersenne rotation algorithm proposed in this application embodiment is shown.

[0031] Figure 6 A pseudo-random number generation flow chart in a double random PWM motor noise suppression method based on Mason rotation algorithm is shown;

[0032] Figure 7 An MT-PRNG random number generation situation schematic diagram in a double random PWM motor noise suppression method based on Mason rotation algorithm is shown;

[0033] Figure 8 A permanent magnet synchronous motor three-phase current waveform graph (a) and a traditional SVPWM control A-phase current Fourier analysis spectrum graph (b) in a double random PWM motor noise suppression method based on Mason rotation algorithm are shown;

[0034] Figure 9 A single random modulation method and ordinary double random method PWM waveform graph and A-phase current Fourier analysis spectrum graph in a double random PWM motor noise suppression method based on Mason rotation algorithm are shown;

[0035] Figure 10 A PWM waveform graph and A-phase current Fourier analysis spectrum graph under a Mason rotation double random modulation method in a double random PWM motor noise suppression method based on Mason rotation algorithm are shown. DETAILED DESCRIPTION

[0036] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object of the application, the specific embodiments, structures, features and effects according to the present application are described in detail as follows in combination with the drawings and preferred embodiments.

[0037] The embodiments are shown as follows: Figures 1-10

[0038] A double random PWM motor noise suppression method based on Mason rotation algorithm comprises the following steps:

[0039] S01: Establishing double random modulation, the double random modulation is random pulse position modulation and random switching period modulation, the random pulse position modulation is used for controllable random offset of the geometric center position of the carrier signal, and the random switching period modulation changes the frequency f of the triangular carrier s The randomization of the system random switching period T is realized. n

[0040] On hardware: as Figure 1 ​​As shown, including pseudo-random generator, SVPWM module and inverter module, pseudo-random number generator for SVPWM module provides pulse position offset coefficient and cycle random number, in the mechanical coupling end of permanent magnet motor are installed current sensor, in order to facilitate the motor in the process of running on the current data record; SVPWM module is through changing the switch modulation mode of inverter to obtain random pulse position modulation and random switch cycle modulation, to reduce the electromagnetic interference strength of itself, so as to suppress the noise level of the whole motor system. According to the randomization of switch parameters in inverter output voltage pulse signal;

[0041] About pulse position offset coefficient:

[0042] Random pulse position modulation by changing carrier tilt angle θ n , so that the pulse position offset coefficient σ of the geometric center position is randomized n , as follows:

[0043]

[0044] Where, d n is the duty cycle, T s is the fixed switching period in the conventional technology, R n is the random position factor defined by the user, and Δθ = π / 2-arctan(1 / 2).

[0045] As Figure 2 shown: by introducing random disturbance in the traditional carrier modulation to optimize the system performance. The core of this method is to make controllable random offset to the geometric center position of the carrier signal, break the harmonic concentration phenomenon caused by the symmetry of fixed carrier, so as to realize random pulse position modulation. Compared with the traditional method, this method has more random freedom and better suppression effect.

[0046] About cycle random number:

[0047] The relationship between cycle random number T n and fixed switching period T s in the conventional technology is:

[0048] T n = (1+Y n )T s ;

[0049] In the formula, T s = 1 / f s , Y n is a random number between-1 and 1, and f s is the fixed period frequency.

[0050] As Figure 3 shown:

[0051] Compared with the random pulse position technique, the random period SVPWM module adopts a center-symmetric triangular carrier, and the randomization of the system switching period is realized by changing the frequency of the triangular carrier. By improving the combination of the random pulse position modulation technique and the random switching period modulation technique, the shortcomings of single random modulation are eliminated, and the harmonic amplitude at the switching frequency is greatly weakened by using double randomization, thereby suppressing the noise of the driving inverter.

[0052] S02: Adopting the Mason rotation algorithm: the Mason rotation algorithm fills in the pseudo-random number generator to generate the pulse position offset coefficient and the period random number, so as to generate the pulse position offset coefficient required by the random pulse position modulation and the period random number required by the random switching period modulation, and disperse the PWM harmonic energy to other frequency intervals.

[0053] Before introducing the improved Mason rotation algorithm, the existing pseudo-random number algorithm is introduced.

[0054] In the existing pseudo-random number algorithm, the linear congruential method is the most widely used pseudo-random number generation algorithm. However, when the linear congruential method is used to generate a random number sequence, it is inevitable that a plurality of consecutive random numbers are greater than or less than the expectation, and it is extremely likely that all the random numbers are on one side of the expectation, which cannot maximize the harmonic dispersion characteristics of the random PWM strategy. Therefore, the Mason rotation algorithm is introduced to enhance the randomness of the pseudo-random sequence generated by the computer.

[0055] The Mason rotation algorithm is based on the matrix linear recursion on the finite binary field, and the cycle iteration length of the pseudo-random number sequence is as high as 2 19937 -1, which can quickly generate high-quality pseudo-random numbers. The process of generating pseudo-random numbers by the Mason rotation algorithm can be roughly divided into three stages: initializing the rotation chain, twisting the update rotation chain, and generating pseudo-random numbers.

[0056] Specifically as follows:

[0057] Initializing the rotation chain:

[0058] As shown in the accompanying Figure 4 : To construct the Mason rotation pseudo-random number generator, a non-zero constant needs to be selected as the initial seed of the algorithm input and written into the binary ω-dimensional row vector Z0. The low r bits and the high ω-r bits in the row vector Z r-1 are subjected to an exclusive OR operation, and then the result is multiplied by the constant multiplier f and added to r-1 to obtain a new row vector Z r . As shown in formula (1), a complete basic Mason rotation chain {Z0, Z1,…, Z v-1} can be obtained by repeatedly iterating v-1 times in this way.

[0059] Twisting the update rotation chain:

[0060] When the index reaches the maximum iteration count v, the entire state array must be regenerated to continue generating different random numbers. Therefore, the initial rotation chain needs to be updated; this process is called twisting. (See attached image) Figure 5 As shown, the row vector of the x-th iteration highest bit and The lowest ω-1 bits are concatenated to form a new state vector Z. (x) Then, shift the state vector to the right by m dimensions. and Z (x) Perform an XOR operation and shift right by one bit, then finally perform an XOR operation with a specific integer l to generate a new state variable.

[0061] Pseudo-random number generation:

[0062] As attached Figure 6 As shown: The output calculation result is subjected to a series of fixed bit operations and multiplied by different coefficients to shuffle its bit pattern, so that the random sequence can be better mapped to R. n and Y n To improve the statistical characteristics of the output over the specified interval, the calculation formula is shown in equation (2) below:

[0063] Where Z is the state variable, and u, g, and h are the number of bits used to vector-shift the state variable, all of which are less than 32.

[0064] Regarding the output calculation results: If the number of iterations has not reached the upper limit v, the output result is the basic Mason swivel chain in the first step. If the upper limit v is reached, the output result is the state quantity after the torsion update in the second step.

[0065] The overall logic is as follows: First, the random periodic formula can be referenced. Figure 3 Tn is defined here as a variable determined by Yn, which is a random number between -1 and 1 obtained through the Mason twitch algorithm. Therefore, Tn is calculated using the random period formula, thus achieving the purpose of random periodicity for the triangular carrier wave. Then there's the random pulse position, which can be found in [reference needed]. Figure 2 It can be observed that triangular carrier waves with different tilt angles generate pulse widths. Therefore, I changed the tilt angle θ by outputting a random position factor Rn. n In the construction of the control model, Rn directly determines θ. n The size is not given, but no specific formula is provided. This formula clearly defines the relationship between the offset coefficient and Rn.

[0066] The algorithm is used to generate 10000 pseudo-random number sequences, and 0-1 is divided into 10 equal intervals. The distribution of random numbers after reduction is shown in the following table: Figure 7 Figure 7 (a) As can be seen, the theoretical frequency of each interval is 1000, and the actual frequency is very close to the theoretical frequency. The chi-square test statistic is 1.97, which indicates that the pseudo-random number generator has good uniform distribution characteristics, Figure 7 (b) It can be seen that the random numbers generated by the MT algorithm avoid long-term ideal expectations or below, which is conducive to improving the performance of random PWM strategy.

[0067] The comparison is as follows:

[0068] The influence of different pseudo-random number generation methods on the distribution of PWM harmonic energy is compared and verified. The PMSM double closed-loop speed regulation simulation system is used for observation and verification. The motor model parameters used in the simulation are: P n = 4, R s = 2.875 Ω, L d = L q = 8.5 mH, J = 0.003 kg·m 2 , ψ f = 0.175 Wb, B = 0.008 N·m·s. The verification objects include single random (random switching frequency or random pulse position), ordinary double random and double random based on Mason rotation algorithm. On this basis, the pseudo-random number generator is introduced to build a random SVPWM technology simulation model. The Fourier analysis spectrum of the output current under different control methods is compared and analyzed to obtain the suppression of switching frequency sub-harmonic and multiple frequency harmonic, Figure 8 (a) is the three-phase current waveform of the permanent magnet synchronous motor, Figure 8 (b) is the A-phase current Fourier analysis spectrum of the traditional SVPWM technology, Figure 9 and Figure 10 are the A-phase current Fourier analysis spectrum and PWM waveform under single random, ordinary double random and double random modulation based on Mason rotation, respectively. Figure 9 From the comparative analysis, Figure 9 (a) and (b) are the single random method, which has the highest suppression of switching frequency sub-harmonic and 2 times frequency harmonic, only 68% and 61%, Figure 9 (c) and (d) are the ordinary double random method, which has a suppression of switching frequency sub-harmonic and 2 times frequency harmonic of 76% and 62%, and from Figure 10 Comparative analysis shows that, Figure 10 (e) and (f) are the modulation method based on Mason rotation double random, which has a suppression of switching frequency sub-harmonic of 90%, Figure 10 ​(g) and (h) are based on Mason rotation double random modulation method 2 times the frequency of harmonic suppression degree can be as high as 94%, therefore, based on Mason rotation algorithm double random PWM modulation strategy has better harmonic suppression effect.

[0069] The main purpose of the present application is to weaken the amplitude of the switching frequency sub-harmonic and the amplitude of the switching frequency sub-double frequency harmonic by improving the pseudo-random number generation algorithm and using the double random structure of random pulse position modulation and random switching period modulation, greatly suppressing the noise caused by PWM harmonic, and improving the acoustic comfort of the motor system.

[0070] The above is only the preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with the preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any indirect modification, equivalent change and modification of the above embodiments made according to the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm, characterized in that, Includes the following steps: S01: Establish dual random modulation, which consists of random pulse position modulation and random switching period modulation. Random pulse position modulation is used to controllably and randomly offset the geometric center position of the carrier signal, and random switching period modulation changes the frequency f of the triangular carrier. s Achieve random switching period T of the system n Randomization; S02: The Mason rotation algorithm is used to generate the pulse position offset coefficient required for random pulse position modulation and the period random number required for random switch period modulation, thereby dispersing the PWM harmonic energy to other frequency ranges and suppressing the amplitude of the second harmonic of the switching frequency and the amplitude of the second harmonic of the switching frequency.

2. The noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm according to claim 1, characterized in that, It also includes an SVPWM module for implementing random pulse position modulation and random switching cycle modulation, a pseudo-random number generator and an inverter module that provide pulse position offset coefficients and periodic random numbers for the SVPWM module, wherein the Mason rotation algorithm is filled in the pseudo-random number generator to generate pulse position offset coefficients and periodic random numbers.

3. The noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm according to claim 2, characterized in that, Random pulse position modulation modulates the carrier tilt angle θ. n Thus, the pulse position offset coefficient σ at the geometric center position is randomized. n The details are as follows: Where, d n T is the duty cycle. s In traditional technology with a fixed switching cycle, R n For a user-defined random position factor, Δθ = π / 2 - arctan(1 / 2).

4. The noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm according to claim 3, characterized in that, Periodic random number T n Compared with the fixed switching period T in traditional technology s The relationship is: T n =(1+Y n )T s ; In the formula, T s =1 / f s Y n f is a random number between -1 and 1. s It has a fixed period frequency.

5. A noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm according to claim 4, characterized in that, Mason's rotation algorithm includes the following steps: First, initialize the rotation chain: select a non-zero constant as the initial seed for the algorithm input and write it into the binary ω-dimensional row vector Z0, then take the row vector Z... r-1 Perform an XOR operation on the lower r bits and the higher ω-r bits, then multiply the result by the constant multiplier f, and add r-1 to obtain a new row vector Z. r Based on the formula: By iterating v-1 times, a complete set of basic Mason rotations can be obtained, namely {Z0, Z1, ..., Z...}. v-1 }; Secondly, rotate the chain by twisting: when the index reaches the upper limit v of the number of iterations, the row vector of the x-th iteration is rotated. highest bit and The lowest ω-1 bits are concatenated to form a new state vector Z. (x) Then, shift the state vector to the right by m dimensions. and Z (x) Perform an XOR operation and shift right by one bit, then finally perform an XOR operation with a specific integer l to generate a new state variable. Finally, pseudo-random number generation: the output calculation result is subjected to a series of fixed bit operations and multiplied by different coefficients to shuffle its bit pattern, so that the random sequence can be better mapped to R. n and Y n Within the given interval, the formula for generating pseudo-random numbers is obtained, as follows: Where Z is the state variable, and u, g, and h are the number of bits used to vector-shift the state variable, all of which are less than 32.

6. A noise suppression method for a dual-random PWM motor based on the Mason rotation algorithm according to claim 5, characterized in that, If the number of iterations does not reach the upper limit v, the output result is the basic Mason swirl chain in the first step. If the upper limit v is reached, the output result is the state variable after the torsion update in the second step.

Images