A method for suppressing circulating current of disc motor based on modular multilevel converter

Abstract

The present invention discloses a method for suppressing circulating current of a disk motor based on a modular multi-level converter, which relates to the technical field of circulating current suppression, including analyzing the circulating current in the modular multi-level converter, determining the mechanism of circulating current generation and suppression mechanism; suppressing the double frequency component in the circulating current based on a circulating current suppression strategy of negative sequence coordinate transformation; and improving the suppression effect of the circulating current through a quasi-PR control strategy of an adaptive filter. The method of the present invention uses an adaptive filter to extract the double frequency component of the circulating current through an improved circulating current suppression strategy based on an adaptive filter and a quasi-PR controller, and uses a quasi-PR controller to accurately suppress the circulating current.

Description

Technical Field

[0001] The invention relates to the technical field of circulating current suppression, and in particular to a method for suppressing circulating current of a disk motor based on a modular multi-level converter. Background Art

[0002] The unique modular structure of MMC will bring about circulating current problems. The circulating current is mainly composed of DC components and harmonic components. The presence of harmonics will increase circuit losses, increase the probability of overcurrent and reduce the life of components. Although passive circulating current suppression can be achieved by increasing the bridge arm inductance, this method has high requirements on cost and volume. Therefore, research on circulating current suppression has always been one of the hot spots of MMC control.

[0003] Some scholars have optimized and improved the shortcomings of existing circulating current suppressors, combining traditional controllers with improved controllers to improve the rapidity, stability and effectiveness of circulating current suppression. Circulating current will bring greater switching losses and device current stress. Most of the existing circulating current suppression targets the double frequency circulating current component, and usually ignores the higher frequency components in the circulating current, such as the quadruple frequency circulating current component. The existence of the quadruple frequency circulating current component will affect the suppression effect of the double frequency circulating current. Therefore, it is necessary to study the circulating current suppression strategy and propose a more effective circulating current suppression method. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above problems or problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that the existing circulation suppression is mostly aimed at the double frequency circulation component, and usually ignores the higher frequency components in the circulation, such as the quadruple frequency circulation component, and the existence of the quadruple frequency circulation component will affect the suppression effect of the double frequency circulation.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a method for suppressing circulating current of a disk motor based on a modular multilevel converter, which includes analyzing the circulating current in the modular multilevel converter to determine the mechanism of circulating current generation and suppression mechanism; suppressing the double frequency component in the circulating current based on a circulating current suppression strategy of negative sequence coordinate transformation; and improving the suppression effect of the circulating current through a quasi-PR control strategy of an adaptive filter.

[0008] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, wherein: the determination of the circulating current generation mechanism and the suppression mechanism includes the analysis of the circulating current generation mechanism and the analysis of the circulating current suppression mechanism;

[0009] The circulation suppression strategy based on negative sequence coordinate transformation includes analyzing the circulation suppression principle;

[0010] The quasi-PR control strategy includes analyzing the principle of the adaptive filter, optimizing the adaptive algorithm, and designing a circulating current suppressor based on the quasi-PR controller.

[0011] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, the circulating current generation mechanism analysis includes:

[0012] Based on the MMC mathematical model, the system circulation is modeled. The equations and are added and simplified. At the same time, the differential equation characterizing the MMC circulation characteristics is substituted into the equation, which is expressed as:

[0013]

[0014] Among them, U dc Represents the DC side voltage; L arm is the bridge arm inductance; u pj and u nj Represent the upper and lower bridge arm voltages respectively; i cirj For circulation;

[0015] The circulating current of MMC is related to the sum of the upper and lower bridge arm voltages, and the common mode voltage is defined as:

[0016] u commj =u nj +u pj

[0017] Re-express the formula as:

[0018]

[0019] Assume that there is no circulating current. When the circulating current is not considered, the sum of the upper and lower bridge arm voltages is exactly the DC side voltage. The upper and lower bridge arm voltages of each phase are supported by the output voltage of the conducting submodule. The on and off states of the submodules are determined by the modulation waves of the upper and lower bridge arms, which come from the closed-loop control output. The single submodule capacitor voltage expression of the upper and lower bridge arms can be obtained by calculation through the formula:

[0020]

[0021] Where:

[0022]

[0023]

[0024] Under the premise of balanced submodule capacitor voltages in a single bridge arm, the common-mode voltage considering capacitor voltage fluctuations is obtained as follows:

[0025]

[0026] According to the formula, the common mode voltage contains not only a DC component, but also a double frequency AC component. The double frequency AC voltage component excites a double frequency circulating current component on the bridge arm inductance:

[0027]

[0028] The doubled frequency circulating current component flows through the submodule capacitor, resulting in a tripled frequency component in the capacitor voltage, which in turn results in a quadruple frequency AC component in the common mode voltage, and finally excites a quadruple frequency circulating current component on the bridge arm inductor. The above derivation process is repeated to obtain the final expression of the circulating current:

[0029]

[0030] Among them, I nm is the peak value of the nth harmonic of the circulating current, θ nm is the initial phase of the nth harmonic of the circulating current.

[0031] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, the circulating current suppression mechanism analysis includes:

[0032] By superimposing the circulating current suppression amount of the same component on the bridge arm voltage modulation wave to generate a reverse inductance voltage drop and thus suppress the circulating current, the phase voltage expression is obtained by subtracting equations 1 and 2 and simplifying them:

[0033]

[0034] From the above formula, we know that the phase voltage has nothing to do with the circulating current, but is only related to the difference between the upper and lower bridge arm voltages. The upper and lower bridge arm voltages are superimposed with the circulating current suppression amount but will not affect the phase voltage. At this time, the reference values ​​of the upper and lower bridge arm voltages are:

[0035]

[0036] Among them, u csj is the circulation suppression amount of phase j.

[0037] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, the analysis of the circulating current suppression principle includes:

[0038] When the third harmonic and above of the circulating current are ignored, the analytical expression of the internal circulating current of the MMC is:

[0039]

[0040] Among them, I cir2f is the amplitude of the double frequency circulation current;

[0041] Based on the principle of vector control, the negative sequence double frequency rotating coordinate system is used to transform the double frequency circulating current into a DC component, and the three-phase stationary coordinate system is used to d -2 q -2 The transformation matrix of the coordinate system is:

[0042]

[0043] The mathematical model of circulation in frequency domain is obtained by negative sequence coordinate transformation and Laplace transformation:

[0044]

[0045] According to the formula, the input-output relationship between the common-mode voltage and circulating current of the bridge arm of the MMC in the negative-sequence double-frequency synchronous rotating coordinate system is obtained, and then the corresponding transfer function block diagram is obtained.

[0046] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, the principle of analyzing the adaptive filter includes:

[0047] Assume that the input signal vector X(n) = [x(n), x(n-1)..., x(n-1)] T , the output y(n) of the LMS adaptive filter is:

[0048]

[0049] The input signal of the system at time n is x(n), and the output signal y(n) at the corresponding time is obtained through the parameter-adjustable filter. The error signal e(n) is obtained by subtracting y(n) from the expected signal d(n);

[0050] Where L is the coefficient of the filter, and the weight coefficient matrix at time n is W:

[0051] W=[w0,w1…w L ] T

[0052] The error of the LMS algorithm is defined as:

[0053] e(n)=d(n)-y(n)

[0054] Since LMS is a filtering algorithm based on the principle of minimum mean square error, the mean square error is defined as:

[0055] J(n)=E[e 2 (n)] = E[d(n)-y(n)] 2

[0056] J changes with time and represents the convergence speed and accuracy of the filter. When J is minimized, the weight coefficient of the filter can reach the optimal value. The new weight coefficient is obtained by using the stochastic gradient algorithm to minimize the mean square error in a certain sense, satisfying:

[0057]

[0058] Adjust the filter weight coefficient matrix along the negative gradient direction, search for the minimum point of the performance surface, and calculate the weight coefficient update formula:

[0059] W(n+1)=W(n)+2μδ(n)x(n)

[0060] Where μ is the convergence coefficient, μ>0.

[0061] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, wherein: the optimization of the adaptive algorithm includes:

[0062] The convergence rate of the LMS algorithm can be expressed as:

[0063] v=2μλ k

[0064] Among them, λ k is the kth eigenvalue of the autocorrelation matrix of x(n). The above formula shows that the size of μ is proportional to the convergence speed. When μ is larger, the algorithm converges faster, but it will cause the system stability to deteriorate. In order to ensure the robustness of the algorithm, considering the constraint range of μ, the size of μ is adjusted according to the experiment;

[0065] According to the above analysis, the fixed convergence coefficient leads to an irreconcilable contradiction between the fast convergence speed and the small steady-state error. The time-varying convergence coefficient can be adjusted continuously through algorithm iteration to obtain a new weight coefficient update formula:

[0066] W(n+1)=W(n)+2μ(n)e(n)x(n)

[0067] In order to make the function curve more consistent with the convergence coefficient adjustment criteria, the function is transformed based on the Sigmoid function, and a compensation factor is introduced to improve the bottom characteristics of the function, and the convergence coefficient of the algorithm is dynamically adjusted. The update formula is:

[0068]

[0069] Among them, k is the compensation coefficient, and a, b, and c are adjustment coefficients.

[0070] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, wherein: the circulating current suppressor designed based on quasi-PR controller includes:

[0071] A sinusoidal signal of any phase can be composed of a sinusoidal signal and a cosine signal of the same frequency. The harmonics of the MMC circulating current are mainly composed of DC and double frequency signals. Therefore, the expression of the circulating current to be extracted is:

[0072]

[0073] The adaptive filter output, i.e., the circulating current harmonics, is:

[0074] y(n)=w1 cos(2θ)+w2 sin(2θ)=w1x1(n)+w2x2(n)

[0075] The error calculation formula is:

[0076] e(n)=i cirj -y(n)

[0077] The update formula of the LMS algorithm is:

[0078]

[0079] The ideal PR controller has a resonant frequency ω o The transfer function at is:

[0080]

[0081] Among them, k p and k r are proportional gain and resonant gain respectively;

[0082] The PR controller can only achieve infinite gain at the resonant frequency and no gain near the resonant frequency. By applying the optimized quasi-PR controller to expand the bandwidth, the transfer function of the quasi-PR controller is:

[0083]

[0084] Among them, ω c is the cut-off frequency.

[0085] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter of the present invention, wherein: the circulating current suppressor designed based on quasi-PR controller also includes:

[0086] The controller is discretized by bilinear transformation to obtain the difference equation, and its transformation formula is:

[0087]

[0088] After calculation and arrangement, the differential equation of the R controller is:

[0089] y(k)=b1e(k)+b2e(k-2)-a1y(k-1)-a2y(k-2)

[0090] Where:

[0091]

[0092] Where T s is the sampling frequency;

[0093] Let the desired cutoff frequency be ω o , the cutoff frequency changes when the bilinear transformation is used to achieve the conversion from s domain to z domain:

[0094]

[0095] The R controller is accurately discretized by using pre-distortion bilinear transformation, the original design index is pre-distorted to obtain the new design index, and then the R controller is discretized by bilinear transformation according to the new design index; the change formula of pre-distortion bilinear transformation is:

[0096]

[0097] Substituting the equation into the resonant term in the equation, the z-domain expression of the R controller after accurate discretization is obtained:

[0098]

[0099] The differential equation of the R controller after accurate discretization obtained by combining the z-inverse transform is:

[0100]

[0101] Where:

[0102]

[0103] The improved circulating current suppressor based on quasi-PR control of LMS algorithm is designed; cirj and θ as input parameters, the sine and cosine functions are used to extract the specified frequency harmonics, and then the online adjustment of the circulating current harmonic amplitude weight coefficient is realized based on the gradient descent method to obtain the second harmonic in the circulating current;

[0104] Taking the circulating current harmonic as 0 as the target value, the circulating current suppression amount is obtained through the quasi-PR controller to suppress the circulating current and reduce the harmonic components in the circulating current.

[0105] As a preferred solution of the disk motor circulating current suppression method based on modular multi-level converter described in the present invention, wherein: the circulating current suppression strategy based on negative sequence coordinate transformation also includes simulation analysis of circulating current suppression.

[0106] Beneficial effects of the present invention: Since the prior art cannot accurately extract the double frequency circulating current component through coordinate transformation, resulting in poor suppression effect of the traditional method, an improved circulating current suppression strategy based on an adaptive filter and a quasi-PR controller is proposed, and an adaptive filter is used to extract the double frequency component of the circulating current, and a quasi-PR controller is used to accurately suppress the circulating current; finally, the circulating current suppression effect under the circulating current suppression strategy based on negative sequence coordinate transformation and the quasi-PR control circulating current suppression strategy based on the adaptive filter are compared through simulation, which confirms the effectiveness of the proposed improved circulating current suppression strategy in suppressing the double frequency component and its superiority in reducing the harmonic content of the circulating current of the modular multilevel converter. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0108] Figure 1 This is the block diagram of the circulating current transfer function.

[0109] Figure 2 This is a block diagram of the circulation suppression strategy based on negative-sequence coordinate transformation.

[0110] Figure 3 This is the MMC circulation spectrum diagram.

[0111] Figure 4 It is a diagram without circulation suppression, where from top to bottom are (a) three-phase circulating current waveform and (b) circulating current FFT analysis.

[0112] Figure 5 This is a diagram of the circulation suppression module based on negative-sequence coordinate transformation.

[0113] Figure 6 This is the FFT analysis diagram of the d-axis circulation component.

[0114] Figure 7 Figure 2 is a circulating current suppression diagram based on negative-sequence coordinate transformation, where from top to bottom are (a) three-phase circulating current waveform and (b) circulating current FFT analysis.

[0115] Figure 8 This is the structure diagram of the adaptive filter.

[0116] Fig. 9 For improved circulation suppressor diagram.

[0117] Fig.10 This is the block diagram of the circulation suppression system.

[0118] Fig.11 This is the module diagram of the quasi-PR control circulating current suppression based on the adaptive filter.

[0119] Fig.12 FFT analysis of the components extracted for the adaptive filter.

[0120] Fig.13 Figure 2 is a diagram of circulating current suppression for quasi-PR control based on adaptive filter, where from top to bottom are (a) three-phase circulating current waveform and (b) circulating current FFT analysis.

[0121] Fig.14 Figure 2 is a circulating current suppression diagram based on negative-sequence coordinate transformation, where from top to bottom are (a) three-phase circulating current waveform and (b) circulating current FFT analysis.

[0122] Fig.15 Figure 2 is a diagram of circulating current suppression for quasi-PR control based on adaptive filter, where from top to bottom are (a) three-phase circulating current waveform and (b) circulating current FFT analysis. DETAILED DESCRIPTION

[0123] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0124] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0125] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0126] Example 1

[0127] Reference Figure 1 , which is an embodiment of the present invention, provides a disk motor circulating current suppression method based on a modular multi-level converter.

[0128] It should be noted that in the medium and high voltage motor drive occasions, MMC has attracted widespread attention due to its unique modular design. However, the inherent characteristics of MMC will bring about unique circulating current problems. The circulating current will not only increase the loss of the system but also cause the bridge arm current distortion. Therefore, the suppression of the circulating current is crucial. The traditional circulating current suppression strategy usually relies on coordinate transformation to convert the double frequency component in the circulating current into a DC component, and uses a PI controller to track the DC quantity. However, since there is a six-fold AC component in the DC component obtained by coordinate transformation, this circulating current suppression algorithm based on negative sequence coordinate transformation can suppress the double frequency circulating current component, but it will cause the increase of low-frequency circulating current component and six-fold circulating current component. Therefore, an improved circulating current suppression strategy is proposed: an adaptive filter is used to accurately extract the double frequency circulating current component, and a quasi-proportional resonant controller is combined to realize circulating current tracking. Two circulating current suppression modules are built in MATLAB / Simulink, and the effectiveness and superiority of the proposed circulating current suppression strategy are verified through simulation comparison.

[0129] In order to prevent the "bridge arm direct pass" phenomenon from occurring in the inverter, the traditional two-level converter sets a certain dead time on the driving waveform of the upper and lower bridge arms, so as to ensure that no current flows between the upper and lower bridge arms. However, due to the special circuit structure of the MMC, there will be current flowing through the upper and lower bridge arms and the DC side under normal operation, that is, there is a current path between the upper and lower bridge arms in the MMC. The non-fundamental frequency current in this path is usually called circulating current. The existence of circulating current will cause the bridge arm current distortion, increase the current stress of the power device and thus increase the circuit loss.

[0130] In order to further analyze the MMC circulation, the system circulation is modeled based on the MMC mathematical model.

[0131] S1: Analyze the circulating current in the modular multilevel converter and determine the mechanism of circulating current generation and suppression.

[0132] Furthermore, the determination of the circulation generation mechanism and the suppression mechanism includes the analysis of the circulation generation mechanism and the analysis of the circulation suppression mechanism.

[0133] Furthermore, the analysis of the circulation generation mechanism includes:

[0134] Based on the MMC mathematical model, the system circulation is modeled. The equations and are added and simplified. At the same time, the differential equation characterizing the MMC circulation characteristics is substituted into the equation, which is expressed as:

[0135]

[0136] Among them, U dc Represents the DC side voltage; L armis the bridge arm inductance; u pj and u nj Represent the upper and lower bridge arm voltages respectively; i cirj is the circulating current; Idc represents the DC side current; ipj and inj represent the upper and lower bridge arm currents respectively, and isj represents the three-phase current;

[0137] The circulating current of MMC is related to the sum of the upper and lower bridge arm voltages, and the common mode voltage is defined as:

[0138] u commj =u nj +u pj

[0139] Re-express the formula as:

[0140]

[0141] The series capacitors in the MMC bridge arm cannot remain stable at all times due to suspension, which will produce periodic voltage fluctuations. The unbalanced capacitor voltages between the sub-modules in each phase result in not only a DC component but also an AC component that changes with time in the common-mode voltage, which in turn triggers a three-phase circulating current. The capacitor voltage fluctuations and the circulating current have a coupling relationship that triggers and affects each other. The operation of the MMC is to find a balance point in the coupling relationship between the two, so the specific expression of the circulating current can be obtained through the iteration method.

[0142] In order to obtain the specific expression of the circulating current, it is first assumed that there is no circulating current. When the circulating current is not considered, the sum of the upper and lower bridge arm voltages is exactly the DC side voltage. The upper and lower bridge arm voltages of each phase are supported by the output voltage of the conducting submodule. The on and off states of the submodules are determined by the modulation waves of the upper and lower bridge arms, which come from the closed-loop control output. The single submodule capacitor voltage expression of the upper and lower bridge arms can be obtained by calculation through the formula:

[0143]

[0144] Where:

[0145]

[0146] Under the premise of balanced submodule capacitor voltages in a single bridge arm, the common-mode voltage considering capacitor voltage fluctuations is obtained as follows:

[0147]

[0148] According to the formula, the common mode voltage contains not only a DC component, but also a double frequency AC component. The double frequency AC voltage component excites a double frequency circulating current component on the bridge arm inductance:

[0149]

[0150] The doubled frequency circulating current component flows through the submodule capacitor, resulting in a tripled frequency component in the capacitor voltage, which in turn results in a quadruple frequency AC component in the common mode voltage, and finally excites a quadruple frequency circulating current component on the bridge arm inductor. The above derivation process is repeated to obtain the final expression of the circulating current:

[0151]

[0152] Among them, I nm is the peak value of the nth harmonic of the circulating current, θ nm is the initial phase of the nth harmonic of the circulating current.

[0153] It should be noted that the following conclusions can be drawn based on the above circulation analysis:

[0154] (1) The circulating current actually consists of a DC component and an AC component of Idc / 3. The DC component plays the role of AC / DC power transmission, while the existence of the AC component is caused by the potential difference between the upper and lower bridge arm voltages and the DC bus voltage.

[0155] (2) The circulating current AC component is mainly composed of even-order harmonics such as second, fourth, and sixth frequencies. The amplitude of the harmonic component is inversely proportional to the harmonic frequency. The 6n+2 harmonic component has a negative sequence property, the 6n+4 harmonic component has a positive phase sequence, and the 6n+6 harmonic component has a zero sequence property.

[0156] (3) The AC component in the circulating current neither flows into the three-phase circuit nor into the DC line, but flows completely between the three phases of the MMC. It also acts on the bridge arm inductance to generate harmonic voltage. Therefore, it can be considered that the AC component in the circulating current is not the working current and has no effect on the normal operation of the MMC.

[0157] Currently, the methods for suppressing circulating current can be divided into two categories: active circulating current suppression and passive circulating current suppression. Passive circulating current suppression suppresses circulating current by increasing the bridge arm inductance. This type of method will increase the circuit volume and cost. The active circulating current suppression method suppresses the harmonics of the circulating current by superimposing the circulating current suppression amount on the bridge arm reference voltage.

[0158] Furthermore, the analysis of the circulation suppression mechanism includes:

[0159] Since the specific manifestation of the circulating current is the voltage drop on the bridge arm inductance, the circulating current can be suppressed by superimposing the circulating current suppression amount of the same component on the bridge arm voltage modulation wave to generate a reverse inductance voltage drop. Subtracting equation 1 from equation 2 and simplifying them, the phase voltage expression is obtained:

[0160]

[0161] From the above formula, we know that the phase voltage has nothing to do with the circulating current, but is only related to the difference between the upper and lower bridge arm voltages. The upper and lower bridge arm voltages are superimposed with the circulating current suppression amount but will not affect the phase voltage. At this time, the reference values ​​of the upper and lower bridge arm voltages are:

[0162]

[0163] Among them, u csj is the circulation suppression amount of phase j.

[0164] It should be noted that, according to the above analysis, suppressing the circulating current harmonic component will not affect the normal operation of the MMC, but can also reduce the bridge arm current of the MMC to achieve the purpose of reducing switching losses.

[0165] S2: A circulation suppression strategy based on negative sequence coordinate transformation to suppress the double frequency component in the circulation.

[0166] Furthermore, the circulation suppression strategy based on negative-sequence coordinate transformation includes analysis of the circulation suppression principle and simulation analysis of the circulation suppression.

[0167] Furthermore, the analysis of the circulation suppression principle includes:

[0168] When the third harmonic and above of the circulating current are ignored, the analytical expression of the internal circulating current of the MMC is:

[0169]

[0170] Among them, I cir2f is the amplitude of the double frequency circulation current;

[0171] Based on the principle of vector control, the negative sequence double frequency rotating coordinate system is used to transform the double frequency circulating current into a DC component, and the three-phase stationary coordinate system is used to d -2 q -2 The transformation matrix of the coordinate system is:

[0172]

[0173] The mathematical model of circulation in frequency domain is obtained by negative sequence coordinate transformation and Laplace transformation:

[0174]

[0175] According to the formula, the input-output relationship between the common-mode voltage and circulating current of the bridge arm of the MMC in the negative-sequence double frequency synchronous rotating coordinate system is obtained, and then the corresponding transfer function block diagram is obtained, as shown in Figure 1 shown.

[0176] It should be noted that since the PI controller has good tracking performance for DC flow, PI control is widely used in engineering. The final MMC vector control loop decoupling control block diagram can be obtained as follows: Figure 2 This strategy first calculates the MMC by current sampling to obtain the three-phase circulating current, and then obtains the easily controllable DC quantity I by coordinate transformation. cird and I cirq , set the circulating current reference value to 0, and then set I cird ,I cirq The difference from 0 is processed by a PI controller and the decoupling component 2ωL is introduced arm In this way, through the negative sequence coordinate inverse transformation, the control quantity u for three-phase circulating current suppression can be obtained. csj .

[0177] The circulating current double harmonic component can be transformed into a DC component by synchronously rotating the coordinate system. However, due to the existence of other harmonics in the circulating current, the component obtained after coordinate transformation is not a complete DC quantity. From the above analysis, it can be seen that the circulating current double harmonic rotates negatively and the quadruple harmonic rotates positively. In order to better analyze their relationship, Figure 3 The spectrum of the MMC circulating current is drawn in the figure. The horizontal axis represents the direction, the vertical axis represents the amplitude, the second harmonic of the circulating current is on the left half axis of the horizontal axis, and the fourth harmonic is on the right half axis. It can be seen that the distance between the fourth harmonic and the second harmonic is six times the frequency, that is to say, the four-fold circulating current harmonic will generate the sixth harmonic in the negative-sequence synchronous rotating coordinate system. The existence of the sixth harmonic leads to the poor tracking effect of the PI controller on the circulating current.

[0178] It should be noted that, from the above analysis, the key problem of suppressing circulating current is to extract harmonics. Notch filter is one of the common methods to solve this problem, which can achieve signal filtering within the specified frequency band. The existing technology uses notch filter to extract the harmonic component of the circulating current at the second harmonic frequency, and adopts the parallel connection of PI and PR controller to suppress the circulating current. Notch filter is a kind of band-stop filter, which can achieve signal filtering within the specified frequency band, but it is still an open-loop extraction method in essence. Notch filter is designed based on the principle of resonant circuit, which is sensitive to the change of temperature and component parameters, and cannot adjust the filter coefficient according to the random change of external signal. The filtering effect is poor, the phase delay is serious, the detection accuracy is low, and the phase-to-phase decoupling link also increases the amount of MMC control calculation, and the system stability and control accuracy are limited. In order to further improve the effect of circulating current suppression, this paper proposes a high-precision closed-loop harmonic extraction method-adaptive filter harmonic extraction method based on improved least mean square (LMS) algorithm, and adopts quasi-PR control to achieve zero static error tracking of circulating current to avoid phase-to-phase decoupling.

[0179] S3: Improve the suppression effect of circulating current through the quasi-PR control strategy of the adaptive filter.

[0180] It should be noted that, unlike the notch filter, the adaptive filter is a time-varying nonlinear system. In practical applications, the filter coefficients can be adaptively updated using the filter parameter results of the previous moment. Its greatest feature is that it can adapt to the statistical characteristics of signals and noise that change over time through iterative learning in an unknown environment to achieve optimal signal processing. It has the advantages of simple structure, small amount of calculation and good noise resistance. Since the adaptive filtering method has good dynamic response and anti-interference capabilities, it has been widely used in signal processing. However, in MMC systems, there is little research on the use of adaptive filters to suppress circulating harmonics. The specific composition of the adaptive filter is as follows: Figure 8 Shown

[0181] Furthermore, the quasi-PR control strategy includes analyzing the principle of an adaptive filter, optimizing an adaptive algorithm, and designing a circulating current suppressor based on a quasi-PR controller.

[0182] It should be noted that by introducing the adaptive filter into the field of circulating current harmonic suppression of the present invention, the specific harmonics of the circulating current can be regarded as a kind of interference, and the filter coefficients are adjusted by the adaptive algorithm using known information, so that its characteristics change with the changes of the circulating current harmonics. Since the adaptive filter is based on the filter and the adaptive algorithm is optimized, the filtering effect and anti-interference ability are highly correlated with the algorithm performance. The LMS algorithm was first proposed by Widrow and Hoff in the 1960s. The algorithm is an optimized extension of the steepest descent method in the Wiener filtering theory. Thanks to its simple principle, few parameters, easy implementation, small amount of calculation and strong robustness, it is widely used in the field of digital signal processing, and can also obtain good adaptability in circulating current suppression.

[0183] Furthermore, the principle of analyzing the adaptive filter includes:

[0184] Assume that the input signal vector X(n) = [x(n), x(n-1)..., x(n-1)] T , the output y(n) of the LMS adaptive filter is:

[0185]

[0186] The input signal of the system at time n is x(n), and the output signal y(n) at the corresponding time is obtained through the parameter-adjustable filter. The error signal e(n) is obtained by subtracting y(n) from the expected signal d(n). This signal is used as a variable of the adaptive algorithm, and the algorithm continuously updates the weight coefficient to obtain the best filtering effect.

[0187] Where L is the coefficient of the filter, and the weight coefficient matrix at time n is W:

[0188] W=[w0,w1…w L ] T

[0189] The error of the LMS algorithm is defined as:

[0190] e(n)=d(n)-y(n)

[0191] Since LMS is a filtering algorithm based on the principle of minimum mean square error, the mean square error is defined as:

[0192] J(n)=E[e 2 (n)] = E[d(n)-y(n)] 2

[0193] J changes with time and represents the convergence speed and accuracy of the filter. When J is minimized, the weight coefficient of the filter can reach the optimal value. The new weight coefficient is obtained by using the stochastic gradient algorithm to minimize the mean square error in a certain sense, satisfying:

[0194]

[0195] Adjust the filter weight coefficient matrix along the negative gradient direction, search for the minimum point of the performance surface, and calculate the weight coefficient update formula:

[0196] W(n+1)=W(n)+2μδ(n)x(n)

[0197] Where μ is the convergence coefficient, μ>0.

[0198] In summary, the LMS adaptive filtering algorithm process can be obtained. The core of applying adaptive filter technology to the extraction of circulating harmonics is to utilize the basic principles of the filter and combine it with the flexibility of the adaptive algorithm to achieve effective separation and extraction of harmonics. The performance of the adaptive algorithm directly determines the ability of the adaptive filter to resist external interference and improve the quality of signal processing. In view of this, subsequent research will focus on the optimization of the adaptive algorithm to enhance the application effect of the adaptive filter in the extraction of circulating harmonics.

[0199] Furthermore, the optimization of the adaptive algorithm includes:

[0200] The two important indicators for evaluating adaptive algorithms are steady-state error and convergence speed. An excellent adaptive filter should have a faster convergence speed and smaller steady-state error when subjected to external interference. The convergence speed of the LMS algorithm can be expressed as:

[0201] v=2μλ k

[0202] Among them, λ k is the kth eigenvalue of the autocorrelation matrix of x(n). The above formula shows that the size of μ is proportional to the convergence speed. When μ is larger, the algorithm converges faster, but it will cause the system stability to deteriorate. In order to ensure the robustness of the algorithm, considering the constraint range of μ, the size of μ is adjusted according to the experiment;

[0203] According to the above analysis, the fixed convergence coefficient leads to an irreconcilable contradiction between the fast convergence speed and the small steady-state error. The time-varying convergence coefficient can be adjusted continuously through algorithm iteration to obtain a new weight coefficient update formula:

[0204] W(n+1)=W(n)+2μ(n)e(n)x(n)

[0205] The principle of the variable convergence coefficient LMS algorithm is: the convergence coefficient is larger in the early stage of system iteration to obtain a fast convergence speed, and the convergence coefficient is smaller in the steady-state stage of the system in the later stage of iteration to obtain better steady-state performance, solve the contradiction between dynamic performance and steady-state performance, and improve the overall performance of the system.

[0206] The symmetric Sigmoid function is a variant of the hyperbolic tangent function. It has the characteristics of even symmetry, monotonicity, and continuous boundedness. The dependent variable increases monotonically with the independent variable. When the independent variable approaches 0, the dependent variable also approaches 0. Therefore, it can be used as a function model for the variable convergence coefficient calculation formula.

[0207] In order to make the function curve more consistent with the convergence coefficient adjustment criteria, the function is transformed based on the Sigmoid function, and a compensation factor is introduced to improve the bottom characteristics of the function, and the convergence coefficient of the algorithm is dynamically adjusted. The update formula is:

[0208]

[0209] Among them, k is the compensation coefficient, which is 0.01, and a, b, and c are adjustment coefficients.

[0210] The adjustment factor works as follows:

[0211] (1) Parameter a can set an upper bound for the convergence coefficient. Although the convergence coefficient is time-varying, in order to ensure the stability of the algorithm, a range of the convergence coefficient must be given so that the value range of μ is (0, a].

[0212] (2) Parameter b mainly affects the bottom characteristics of the convergence coefficient change curve. The smaller b is, the flatter the bottom of the curve is and the better the steady-state performance is. Therefore, by adjusting b, the steady-state performance and dynamic performance of the algorithm can be compromised.

[0213] (3) The parameter c mainly affects the variation of the convergence coefficient when e(n) approaches 0. When the filter is close to a stable state, the error is close to 0. In order to avoid excessive variation of μ in the late convergence period, which may lead to poor system stability, the value of c should not be too large.

[0214] Furthermore, the design of the circulating current suppressor based on the quasi-PR controller includes:

[0215] The principle of using adaptive filters to extract circulating current harmonics is to use adaptive interference cancellation to process the signal so that the input signal and the desired signal interfere with each other and cancel each other out. The convergence coefficient and weight coefficient are adjusted through the error feedback of the adaptive algorithm to better track the harmonic current.

[0216] A sinusoidal signal of any phase can be composed of a sinusoidal signal and a cosine signal of the same frequency. The harmonics of the MMC circulating current are mainly composed of DC and double frequency signals. Therefore, the expression of the circulating current to be extracted is:

[0217]

[0218] The adaptive filter output, i.e., the circulating current harmonics, is:

[0219] y(n)=w1 cos(2θ)+w2 sin(2θ)=w1x1(n)+w2x2(n)

[0220] The error calculation formula is:

[0221] e(n)=i cirj -y(n)

[0222] The update formula of the LMS algorithm is:

[0223]

[0224] Traditional PI control can only achieve zero-error tracking of DC quantities. If the control object becomes an AC signal, the PI control effect is poor. Therefore, proportional resonance (PR) control is introduced to achieve circulating current suppression. PR control developed from the inner membrane principle can achieve zero-error tracking of AC signals of specific frequencies. The PR controller is applied to the circulating current control of MMC, and the circulating current harmonics of specific frequencies can be eliminated without the decoupling process. This method is not only applicable to MMC three-phase systems, but also to single-phase systems.

[0225] The ideal PR controller has a resonant frequency ω o The transfer function at is:

[0226]

[0227] Among them, kp and k r are proportional gain and resonant gain respectively;

[0228] The PR controller can only achieve infinite gain at the resonant frequency and no gain near the resonant frequency. In order to improve the anti-frequency disturbance performance and avoid the frequency offset affecting the system stability, the optimized quasi-PR controller is applied to expand the bandwidth. The transfer function of the quasi-PR controller is:

[0229]

[0230] Among them, ω c is the cut-off frequency. In order to ensure the stability of the controller, the value is generally 3-15rad / s.

[0231] According to k p =15, k r =2000 and ω o =50Hz and the Bode plots of the quasi-PR controller with a cutoff frequency of 0.5Hz. It can be seen that the bandwidth of the quasi-PR controller near the resonant frequency is increased, which can speed up the response speed.

[0232] Furthermore, the design of the circulating current suppressor based on the quasi-PR controller also includes:

[0233] The controller is discretized by bilinear transformation to obtain the difference equation, and its transformation formula is:

[0234]

[0235] After calculation and arrangement, the differential equation of the R controller is:

[0236] y(k)=b1e(k)+b2e(k-2)-a1y(k-1)-a2y(k-2)

[0237] Where:

[0238]

[0239] Where T s is the sampling frequency; the formula of the R controller obtained by bilinear transformation is relatively simple, but it also has very obvious disadvantages. When the s-domain controller is discretized, there is a high-frequency pole mismatch problem: the poles after discretization are different from the s-domain poles, which greatly reduces the controller performance. When the system control frequency is lower and the set resonant frequency is higher, the resonant frequency offset caused by discretization is more obvious and the harmonic suppression ability is weaker;

[0240] Let the desired cutoff frequency be ω o, the cutoff frequency changes when the bilinear transformation is used to achieve the conversion from s domain to z domain:

[0241]

[0242] The R controller is accurately discretized by using pre-distortion bilinear transformation, the original design index is pre-distorted to obtain the new design index, and then the R controller is discretized by bilinear transformation according to the new design index; the change formula of pre-distortion bilinear transformation is:

[0243]

[0244] Substituting the equation into the resonant term in the equation, the z-domain expression of the R controller after accurate discretization is obtained:

[0245]

[0246] The differential equation of the R controller after accurate discretization obtained by combining the z-inverse transform is:

[0247]

[0248] Where:

[0249]

[0250] The designed improved circulating current suppressor based on quasi-PR control of LMS algorithm is obtained, as Fig. 9 As shown; the suppressor is based on the circulating current i cirj and θ as input parameters, the sine and cosine functions are used to extract the specified frequency harmonics, and then the online adjustment of the circulating current harmonic amplitude weight coefficient is realized based on the gradient descent method to obtain the second harmonic in the circulating current;

[0251] Taking the circulating current harmonic as 0 as the target value, the circulating current suppression amount is obtained through the quasi-PR controller to suppress the circulating current and reduce the harmonic components in the circulating current.

[0252] It should be noted that the control strategies of the disc motor drive system based on modular multilevel converters are different at high and low frequencies. In the low-frequency stage, since high-frequency circulating current injection is required to suppress capacitor voltage fluctuations, circulating current suppression is usually not added. Therefore, the circulating current suppression strategy proposed in this paper is applied when the motor is in a medium- and high-frequency steady state. Finally, the control block diagram of the disc motor drive system based on modular multilevel converters during the medium- and high-frequency operation stage of the motor is as follows: Fig.10 As shown, at this time, only the MMC output current, internal circulating current and capacitor voltage balance need to be controlled.

[0253] In summary, the present invention analyzes the causes of circulating current and establishes a mathematical model of circulating current, and determines that the circulating current harmonics are composed of even-order harmonics such as double frequency and quadruple frequency, and the main component is the double frequency component; the traditional circulating current suppression method is analyzed: the circulating current suppression strategy based on negative-sequence coordinate transformation, because the coordinate transformation cannot accurately extract the double frequency circulating current component, resulting in poor suppression effect of the traditional method, therefore an improved circulating current suppression strategy based on an adaptive filter and a quasi-PR controller is proposed, an adaptive filter is used to extract the double frequency component of the circulating current, and a quasi-PR controller is used to accurately suppress the circulating current; finally, the circulating current suppression effects of the circulating current suppression strategy based on negative-sequence coordinate transformation and the quasi-PR control circulating current suppression strategy based on the adaptive filter are compared through simulation, which confirms the effectiveness of the proposed improved circulating current suppression strategy in suppressing the double frequency component and its superiority in reducing the harmonic content of the circulating current of the modular multilevel converter.

[0254] Example 2

[0255] Reference Figures 1 to 15 , as an embodiment of the present invention, a method for suppressing circulating current of a disc motor based on a modular multi-level converter is provided, which is verified through simulation experiments.

[0256] The simulation analysis of circulation suppression includes:

[0257] In order to verify the effectiveness of the circulation suppression strategy based on negative sequence coordinate transformation introduced above, a simulation study was carried out in MATLAB / Simulink. The basic parameters of the simulation model are shown in Table 1:

[0258] Table 1MMC-AFPM simulation parameters

[0259]

[0260] In order to further analyze the harmonic situation of the circulating current, the harmonic content rate H of the circulating current is defined as:

[0261]

[0262] Where Vx is the amplitude of the harmonic component with frequency x, and Vdc is the size of the DC component. The effect of circulating current suppression is measured by comparing the amplitude of the harmonic component in the circulating current with the size of the DC component. The smaller H means the better the circulating current suppression effect. At the same time, the percentage of the amplitude of the circulating current harmonic component Mag is defined as:

[0263]

[0264] Mag is the percentage of the amplitude of the harmonic component with frequency x to the magnitude of the DC component.

[0265] First, a simulation analysis is performed without adopting the circulation suppression strategy. Figure 4 (a) is the three-phase circulating current waveform under the working conditions of speed of 900 rpm and load torque of 3 N·m. It can be seen from the figure that the MMC circulating current mainly fluctuates in the form of a sine wave. Figure 4 (b) is the result of the circulating current FFT analysis with the base frequency set to 60 Hz. It can be seen from the figure that the DC component of the A-phase circulating current harmonic is 0.484 A, and the frequency of the main harmonic component is 120 Hz. This verifies the correctness of the conclusion that the MMC circulating current harmonics obtained from the circulating current analysis are mainly double-frequency harmonics.

[0266] Table 2 gives in detail the Mag of each circulating current component when there is no circulating current suppression. At this time, the Mag of the double frequency circulating current component is 74.05%, and the Mag of the quadruple frequency circulating current component is 2.12%.

[0267] Table 2 Percentage of the amplitude of each circulating current component when there is no circulating current suppression

[0268]

[0269] The simulation module of the circulation suppression strategy based on negative sequence coordinate transformation is as follows Figure 5 Before the circulation suppression algorithm is enabled, the FFT analysis of the icird waveform obtained by the coordinate change is performed with the base frequency set to 60Hz, and the results are as follows Figure 6 As shown, it is verified that the fourth harmonic component changes into the sixth harmonic in the synchronous rotating coordinate system.

[0270] Figure 7 (a) is the three-phase circulating current waveform after the circulating current suppression algorithm is put into use. From the waveform diagram, it can be seen that the three-phase circulating current has been suppressed: the circulating current fluctuation amplitude has become smaller. The FFT analysis of the circulating current at this time is as follows: Figure 7 As shown in (b), the DC component of the circulating current is 0.485A and H is 13.08%.

[0271] Table 3 gives the detailed results of the circulation FFT at this time. The Mag of the double frequency circulation component is 8.87%, and the Mag of the quadruple frequency circulation component is 3.21%, which are 65.18% and 1.09% lower than before the circulation suppression strategy was put into use, respectively, but the Mag of the 36Hz and 48Hz circulation components increased by 0.45% and 2.96%, respectively. It can be concluded that the circulation suppression strategy based on negative sequence coordinate transformation can significantly reduce the H and double frequency circulation component amplitude of the circulation, among which the double frequency circulation component is significantly suppressed, but the quadruple frequency circulation component and low-frequency component of the circulation will be slightly increased.

[0272] Table 3 Percentage of the amplitude of each circulating current component when the circulating current is suppressed based on negative sequence coordinate transformation

[0273]

[0274] Furthermore, the quasi-PR control based on adaptive filter is used to improve the circulation suppression module. Fig.11 The improved circulating current suppressor is mainly composed of an adaptive filter and a quasi-PR controller. The adaptive filter is used to extract the second harmonic component of the circulating current, and the quasi-PR controller is used to achieve circulating current tracking.

[0275] Before enabling the circulating current suppression algorithm, the FFT analysis of the circulating current double harmonic component extracted by the adaptive filter with the base frequency set to 120Hz is performed, and the results are as follows: Fig.12 ,It can be seen from the figure that the AC component of the circulating current waveform extracted by the adaptive filter is basically composed of double frequency components, ,which realizes the accurate extraction of double frequency circulating current components.

[0276] Fig.13 (a) is the three-phase circulating current waveform when the circulating current suppression algorithm is used under the conditions of a speed of 900 rpm and a load torque of 3 N·m. The FFT analysis of the circulating current is shown in Fig.13 As shown in (b), the DC component of the circulating current is 0.4878A and H is 3.20%.

[0277] Table 4 gives the FFT data of the quasi-PR control circulation suppression algorithm based on the adaptive filter in detail. Comparing Tables 2, 3 and 4, it can be seen that compared with the case without circulation suppression strategy, the Mag of the 36Hz circulation component under the proposed circulation suppression strategy is reduced by 4.56%, the Mag of the 48Hz circulation is reduced by 2.99%, the Mag of the double frequency circulation component is reduced by 72.33%, the Mag of the quadruple frequency circulation component is reduced by 1.78%, and H is reduced by 70.94%. Compared with the circulation suppression strategy based on negative sequence coordinate transformation, the Mag of the 36Hz circulation component under the proposed circulation suppression strategy is reduced by 5.01%, the Mag of the 48Hz circulation component is reduced by 5.95%, the Mag of the double frequency circulation component is reduced by 7.15%, the Mag of the quadruple frequency circulation component is reduced by 2.87%, and H is reduced by 9.88%.

[0278] The smaller H is, the smaller the proportion of AC component of circulating current is, and the less system loss is. Through the comparative analysis of the above data, it can be seen that under the same working conditions, the circulating current suppression strategy proposed in this paper has lower content of double frequency circulating current component, quadruple frequency circulating current component and circulating current harmonic. Therefore, it is concluded that the proposed improved circulating current suppression strategy can better reduce system loss and improve the overall efficiency of MMC control system.

[0279] Table 4 Percentage of the amplitude of each circulation component when the circulation suppression strategy proposed in this paper is used

[0280]

[0281]

[0282] In order to further verify the effectiveness of the proposed quasi-PR control circulating current suppression strategy based on adaptive filter when the motor operating frequency is 50Hz, the circulating current harmonic content rate and the Mag of each harmonic component of the circulating current under the two circulating current suppression strategies are compared under the working conditions of motor speed of 750rpm and load torque of 3N·m. The circulating current suppression results based on negative sequence coordinate transformation are shown in Figure 2. Fig.14 As shown, the circulating current DC is 0.4041A and H is 10.27%. Table 5 shows the FFT analysis results of the circulating current at this time.

[0283] The results of circulating current suppression based on quasi-PR control of adaptive filter are as follows Fig.15 As shown, the circulating current DC under this circulation suppression strategy is 0.4054A and H is 1.43%. Table 6 shows the FFT analysis results under this circulation suppression method. Comparing Tables 5 and 6, it can be seen that compared with the circulation suppression strategy based on negative sequence coordinate transformation, the H of the quasi-PR control circulation suppression strategy based on the adaptive filter is reduced by 8.84%, the Mag of the 30Hz circulation component is reduced by 1.94%, the Mag of the 40Hz circulation component is reduced by 4.18%, the Mag of the double frequency circulation component is reduced by 5.94%, and the Mag of the quadruple frequency circulation component is reduced by 4.4%.

[0284] Table 5 Percentage of the amplitude of each circulating current component when the circulating current is suppressed based on negative sequence coordinate transformation (750rpm)

[0285]

[0286] Table 6 Percentage of the amplitude of each circulation component when the circulation suppression strategy proposed in this paper is used (750rpm)

[0287]

[0288] In summary, compared with the circulating current suppression strategy based on negative-sequence coordinate transformation, the quasi-PR control circulating current suppression strategy based on adaptive filter proposed in this paper can achieve the best suppression effect on the circulating current double frequency component without increasing the proportion of the circulating current low-frequency component and the fourth-frequency component, and can better reduce the circulating current harmonic content and improve the overall efficiency of the MMC control system.

[0289] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for suppressing circulating current of a disk motor based on a modular multi-level converter, characterized in that: include, Analyze the circulating current in modular multilevel converters and determine the generation and suppression mechanisms of circulating currents; A circulation suppression strategy based on negative sequence coordinate transformation is used to suppress the double frequency component in the circulation; Improve the suppression effect of circulating current through the quasi-PR control strategy of adaptive filter; The determination of the generation mechanism and suppression mechanism of the circulation includes the analysis of the generation mechanism of the circulation and the analysis of the suppression mechanism of the circulation; The circulation suppression strategy based on negative sequence coordinate transformation includes analyzing the circulation suppression principle; The quasi-PR control strategy includes analyzing the principle of the adaptive filter, optimizing the adaptive algorithm, and designing a circulating current suppressor based on the quasi-PR controller; The principle of the analytical adaptive filter includes: Assume that the input signal vector X(n) = [x(n), x(n-1)..., x(n-1)] T , the output y(n) of the LMS adaptive filter is: The input signal of the system at time n is x(n), and the output signal y(n) at the corresponding time is obtained through the parameter-adjustable filter. The error signal e(n) is obtained by subtracting y(n) from the expected signal d(n); Where L is the coefficient of the filter, and the weight coefficient matrix at time n is W: W=[w0,w1…w L ] T The error of the LMS algorithm is defined as: e(n)=d(n)-y(n) Since LMS is a filtering algorithm based on the principle of minimum mean square error, the mean square error is defined as: J(n)=E[e 2 (n)]=E[d(n)-y(n)] 2 J changes with time and represents the convergence speed and accuracy of the filter. When J is minimized, the weight coefficient of the filter can reach the optimal value. The new weight coefficient is obtained by using the stochastic gradient algorithm to minimize the mean square error in a certain sense, satisfying: Adjust the filter weight coefficient matrix along the negative gradient direction, search for the minimum point of the performance surface, and calculate the weight coefficient update formula as follows: W(n+1)=W(n)+2μδ(n)x(n) Where, μ is the convergence coefficient, μ>0; The design of the circulating current suppressor based on the quasi-PR controller includes: A sinusoidal signal of any phase can be composed of a sinusoidal signal and a cosine signal of the same frequency. The harmonics of the MMC circulating current are mainly composed of DC and double frequency signals. Therefore, the expression of the circulating current to be extracted is: Among them, I dc Represents the DC side current; The adaptive filter output, i.e., the circulating current harmonics, is: y(n)=w1cos(2θ)+w2sin(2θ)=w1x1(n)+w2x2(n) The error calculation formula is: e(n)=i cirj -y(n) The update formula of the LMS algorithm is: The ideal PR controller has a resonant frequency ω o The transfer function at is: Among them, k p and k r are proportional gain and resonant gain respectively; The PR controller can only achieve infinite gain at the resonant frequency and no gain near the resonant frequency. Applying the optimized quasi-PR controller to expand the bandwidth, the transfer function of the quasi-PR controller is: Among them, ω c is the cut-off frequency; The design of the circulating current suppressor based on the quasi-PR controller also includes: The controller is discretized by bilinear transformation to obtain the difference equation, and its transformation formula is: After calculation and arrangement, the differential equation of the R controller is: y(k)=b1e(k)+b2e(k-2)-a1y(k-1)-a2y(k-2) Where: Where T s is the sampling frequency; Let the desired cutoff frequency be ω o , the cutoff frequency changes when the bilinear transformation is used to achieve the conversion from s domain to z domain: The R controller is accurately discretized by using pre-distortion bilinear transformation, the original design index is pre-distorted to obtain the new design index, and then the R controller is discretized by bilinear transformation according to the new design index; the change formula of pre-distortion bilinear transformation is: Substituting the equation into the resonant term in the equation, the z-domain expression of the R controller after accurate discretization is obtained as follows: The differential equation of the R controller after accurate discretization obtained by combining the z-inverse transform is: Where: The improved circulating current suppressor based on quasi-PR control of LMS algorithm is designed; cirj and θ as input parameters, the sine and cosine functions are used to extract the specified frequency harmonics, and then the online adjustment of the circulating current harmonic amplitude weight coefficient is realized based on the gradient descent method to obtain the second harmonic in the circulating current; Taking the circulating current harmonic as 0 as the target value, the circulating current suppression amount is obtained through the quasi-PR controller to suppress the circulating current and reduce the harmonic components in the circulating current.

2. The method for suppressing circulating current of a disk motor based on a modular multi-level converter according to claim 1, characterized in that: The analysis of the circulation generation mechanism includes: Based on the MMC mathematical model, the system circulation is modeled. The equations and are added and simplified. At the same time, the differential equation characterizing the MMC circulation characteristics is substituted into the equation, which is expressed as: Among them, U dc Represents the DC side voltage; L arm is the bridge arm inductance; u pj and u nj Represent the upper and lower bridge arm voltages respectively; i cirj For circulation; The circulating current of MMC is related to the sum of the upper and lower bridge arm voltages, and the common mode voltage is defined as: in commj =in nj +in pj Re-express the formula as: Assume that there is no circulating current. When the circulating current is not considered, the sum of the upper and lower bridge arm voltages is exactly the DC side voltage. The upper and lower bridge arm voltages of each phase are supported by the output voltage of the conducting submodule. The on and off states of the submodules are determined by the modulation waves of the upper and lower bridge arms, which come from the closed-loop control output. The single submodule capacitor voltage expression of the upper and lower bridge arms can be obtained by calculation through the formula: Where: Under the premise of balanced submodule capacitor voltages in a single bridge arm, the common-mode voltage considering capacitor voltage fluctuations is obtained as follows: According to the formula, the common mode voltage contains not only a DC component, but also a double frequency AC component. The double frequency AC voltage component excites a double frequency circulating current component on the bridge arm inductance: The doubled frequency circulating current component flows through the submodule capacitor, resulting in a tripled frequency component in the capacitor voltage, which in turn results in a quadruple frequency AC component in the common mode voltage, and finally excites a quadruple frequency circulating current component on the bridge arm inductor. The above derivation process is repeated to obtain the final expression of the circulating current: Among them, I nm is the peak value of the nth harmonic of the circulating current, θ nm is the initial phase of the nth harmonic of the circulating current.

3. The method for suppressing circulating current of a disk motor based on a modular multi-level converter according to claim 2, characterized in that: The analysis of the circulation suppression mechanism includes: By superimposing the circulating current suppression amount of the same component on the bridge arm voltage modulation wave to generate a reverse inductance voltage drop and thus suppress the circulating current, the phase voltage expression is obtained by subtracting equations 1 and 2 and simplifying them: From the above formula, we know that the phase voltage has nothing to do with the circulating current, but is only related to the difference between the upper and lower bridge arm voltages. The upper and lower bridge arm voltages are superimposed with the circulating current suppression amount but will not affect the phase voltage. At this time, the reference values ​​of the upper and lower bridge arm voltages are: Among them, u csj is the circulation suppression amount of phase j, i sj Represents three-phase current.

4. The method for suppressing circulating current of a disk-type motor based on a modular multi-level converter according to claim 3, characterized in that: The analysis of the circulation suppression principle includes: When the third harmonic and above of the circulating current are ignored, the analytical expression of the internal circulating current of the MMC is: Among them, I cir2f is the amplitude of the double frequency circulation current; Based on the principle of vector control, the negative sequence double frequency rotating coordinate system is used to transform the double frequency circulating current into a DC component, and the three-phase stationary coordinate system is used to d -2 q -2 The transformation matrix of the coordinate system is: The mathematical model of circulation in frequency domain is obtained by negative sequence coordinate transformation and Laplace transformation: Among them, i cirq (s) and i cird (s) are the components of the circulation on the q-axis and d-axis in the s-domain, respectively; According to the formula, the input-output relationship between the common-mode voltage and circulating current of the bridge arm of the MMC in the negative-sequence double-frequency synchronous rotating coordinate system is obtained, and then the corresponding transfer function block diagram is obtained.

5. The method for suppressing circulating current of a disk motor based on a modular multi-level converter according to claim 4, characterized in that: The optimization of the adaptive algorithm comprises: The convergence rate of the LMS algorithm can be expressed as: v=2μl k Among them, λ k is the kth eigenvalue of the autocorrelation matrix of x(n). The above formula shows that the size of μ is proportional to the convergence speed. When μ is larger, the algorithm converges faster, but it will cause the system stability to deteriorate. In order to ensure the robustness of the algorithm, considering the constraint range of μ, the size of μ is adjusted according to the experiment; According to the above analysis, the fixed convergence coefficient leads to an irreconcilable contradiction between the fast convergence speed and the small steady-state error. The time-varying convergence coefficient can be adjusted continuously through algorithm iteration to obtain a new weight coefficient update formula: W(n+1)=W(n)+2μ(n)e(n)x(n) In order to make the function curve more consistent with the convergence coefficient adjustment criteria, the function is transformed based on the Sigmoid function, and a compensation factor is introduced to improve the bottom characteristics of the function, and the convergence coefficient of the algorithm is dynamically adjusted. The update formula is: Among them, k is the compensation coefficient, and a, b, and c are adjustment coefficients.

6. The method for suppressing circulating current of a disk-type motor based on a modular multi-level converter according to any one of claims 1 to 5, characterized in that: The circulation suppression strategy based on negative-sequence coordinate transformation also includes simulation analysis of circulation suppression.

Images