Common-mode voltage rejection method based on cascaded observers

Abstract

This application relates to a common-mode voltage suppression method based on cascaded observers, specifically for high-precision motor drives. The aim is to address the shortcomings of existing CMV suppression methods, which rely heavily on experience and suffer from low robustness. This method utilizes cascaded observers to observe and compensate for the total disturbance in the motor control system, calculating a reference voltage vector. A discrete virtual vector set is constructed, and the vector plane formed by this set is divided into multiple continuous and non-overlapping sub-regions, ensuring that each sub-region uniquely corresponds to a virtual vector in the discrete virtual vector set. The sub-region to which the reference voltage vector belongs in the vector plane is determined, and the virtual vector corresponding to that sub-region is used as the target vector. A PWM signal is generated based on the target vector to control the motor, achieving common-mode voltage suppression. This application demonstrates significant suppression effects on the first, second, and sixth harmonics, reducing the common-mode voltage to one-third of that achieved by traditional methods, while effectively reducing computation time.

Description

Technical Field

[0001] This application belongs to the field of high-precision motor drive, and in particular relates to a common-mode voltage suppression method based on cascaded observers. Background Technology

[0002] Thanks to their high efficiency, high power density, and high reliability, high-precision permanent magnet synchronous linear motors (PMLSMs) are commonly used in lithography machines, 3C manufacturing, and high-end CNC machine tools. In these industrial scenarios, PMLSMs are often driven by voltage source inverters (VSIs) based on pulse width modulation (PWM) technology to regulate voltage and frequency. However, the PWM modulation process generates common-mode voltage (CMV), which can trigger electromagnetic interference (EMI), induce bearing currents, and cause severe voltage stress on the motor winding insulation. Therefore, suppressing CMV is crucial for improving system reliability.

[0003] For suppressing CMV, the Discrete Control Set-based Model Predictive Control (FCS-MPC) algorithm has inherent advantages due to its ability to solve multi-objective, multi-variable, and nonlinear problems, its simple structure, and its fast dynamic response. FCS-MPC CMV suppression can be divided into two categories: cost function methods and voltage vector selection methods. The former relies heavily on experience, and the design of the weighting coefficients in the cost function is very difficult; any selection error will affect the overall output performance of the MPC. The latter's main idea is to eliminate voltage vectors that will generate high CMV from the candidate vector set, requiring only software modifications and not affecting the controller's fast dynamic response.

[0004] Furthermore, the control accuracy of FCS-MPC is highly dependent on the parameters of the predictive model. DC and AC disturbances in the system can affect the accuracy of the reference voltage output by the control algorithm, thereby affecting the control of the motor current loop. Therefore, improving the robustness of FCS-MPC is equally important. Summary of the Invention

[0005] This application aims to address the problem that existing CMV suppression methods rely too heavily on experience and have low robustness. It provides an improved model predictive control method for reducing common-mode voltage, which can also reduce current harmonics, reduce computational burden, and reduce total harmonic distortion (THD). This method can be applied to three-phase two-level inverters in high-precision linear motor systems.

[0006] This application provides a common-mode voltage suppression method based on cascaded observers, including:

[0007] The total disturbance in the motor control system is observed and compensated using a cascaded observer, and then the reference voltage vector of the motor is calculated. The cascaded observer includes a cascaded extended state observer and a harmonic internal mode observer.

[0008] Construct a discrete virtual vector set, divide the vector plane formed by the discrete virtual vector set into multiple continuous and non-overlapping sub-regions, and ensure that each sub-region uniquely corresponds to a virtual vector in the discrete virtual vector set;

[0009] Determine the sub-region to which the reference voltage vector belongs in the vector plane, and use the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector;

[0010] A PWM signal is generated based on the target vector, and then the motor is controlled to achieve common-mode voltage suppression.

[0011] In one possible design, the cascaded extended state observer comprises two cascaded third-order extended state observers, wherein,

[0012] The expression for the first-level extended state observer is:

[0013] ,

[0014] The expression for the second-level extended state observer is:

[0015] ,

[0016] In the formula, , and All are extended state observer gains. To control the gain, For dq axis voltage, For dq axis current, and These represent the predicted current values ​​of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative, and These represent the perturbation estimation errors of the first-level and second-level extended state observers, respectively. and They are respectively and The first derivative, and Let these represent the observation error derivatives of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative.

[0017] In one possible design, the perturbation error transfer function of the cascaded extended state observer The expression is:

[0018] ,

[0019] in, For the Laplace operator, This represents the actual disturbance of the dq axis in the motor control system. To estimate the total disturbance, and These are the estimated perturbations for the first-level and second-level extended state observers, respectively.

[0020] In one possible design, the expression for the harmonic internal mode observer is:

[0021] ,

[0022] In the formula, , and These represent the estimation results for the three harmonic components, respectively.

[0023] , and All are intermediate variables, and each has an expression:

[0024] ;

[0025] For observer gain, ; , and All are resonant bandwidths; The distance between the magnetic poles of the motor; The speed of the motor's rotor;

[0026] Let be the harmonic internal mode term of the motor control system, and have the following expression:

[0027] ,

[0028] To control the gain, For dq axis voltage, This is the predicted current value. This is the sum of the perturbation estimation errors of the two-stage extended state observer.

[0029] In one possible design, the calculation of the reference voltage vector of the motor includes:

[0030] The reference voltage vector of the motor is calculated using the following formula:

[0031] ,

[0032] in, This represents the sampling time in a discrete control system. for The reference voltage vector of the motor at any given time. To control the cycle, This is the current reference value. for The actual total disturbance at any given moment. for Predicted current value at any given time.

[0033] In one possible design, the The actual total disturbance at time t is expressed as:

[0034] ,

[0035] The predicted current value at time t is expressed as:

[0036] ;

[0037] in, for The sum of the perturbation estimation errors of the two-stage extended state observer at time t is for The actual total disturbance at any given moment. and They are respectively Current and voltage along the dq axis at any given time.

[0038] In one possible design, constructing the discrete virtual vector set includes:

[0039] Discrete Virtual Vector The expression is:

[0040] ,

[0041] in, and All are used for synthesis Two different effective voltage vectors, It is a zero vector. The interval for vector partitioning, and They are respectively and The time interval satisfies and .

[0042] In one possible design, dividing the vector plane composed of the discrete virtual vector set into multiple continuous and non-overlapping sub-regions includes:

[0043] The vector plane is divided using three straight lines, each perpendicular to one of the three axes of the vector plane, such that the three straight lines and the three axes of the vector plane can divide the vector plane into multiple continuous and non-overlapping sub-regions.

[0044] In one possible design, determining the sub-region to which the reference voltage vector belongs in the vector plane, and using the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector, includes:

[0045] The reference voltage vector is projected onto the three axes of the vector plane;

[0046] The sub-region to which the projection of the reference voltage vector onto the three axes belongs is determined by the following formula:

[0047] ,

[0048] in, , Indicates a sub-region index. Indicates the reference voltage vector at Projection on axis This refers to the DC bus voltage of the inverter.

[0049] The sub-region to which the reference voltage vector belongs in the vector plane is obtained based on the sub-region index and table lookup method.

[0050] In one possible design, generating the PWM signal based on the target vector includes:

[0051] When the target vector is a single vector, the PWM signal is directly generated using the single vector.

[0052] When the target vector is a synthesized vector, the PWM signal is synthesized according to the synthesized vector order.

[0053] The beneficial effects of this application are:

[0054] The ICHIO observer (improved cascaded harmonic internal mode observer) proposed in this application has a significant effect on suppressing DC disturbances and first, second, and sixth harmonics, and the common-mode voltage is reduced to one-third of that of traditional methods, while effectively reducing the computation time. Attached Figure Description

[0055] Figure 1 Bode plots for three ESO observers;

[0056] Figure 2 A discrete virtual voltage vector diagram of a three-phase two-level voltage inverter;

[0057] Figure 3 A region partitioning diagram of the discrete virtual voltage vector plane;

[0058] Figure 4 The following are experimental comparisons of the current output performance of different observers: (a) dq-axis current output by a linear ESO observer, (b) dq-axis current output by a third-order CESO observer, (c) dq-axis current output by an ICHIO observer in the embodiment, (d) FFT analysis of the current waveform output by a linear ESO observer, (e) current waveform output by a third-order CESO observer, and (f) FFT analysis of the current waveform output by an ICHIO observer in the embodiment.

[0059] Figure 5 The common-mode voltage and its amplified waveform under traditional SVPWM technology;

[0060] Figure 6 The image shows the common-mode voltage and its amplified waveform under the traditional model predictive control method.

[0061] Figure 7 The diagram shows the common-mode voltage and its amplified waveform under the virtual vector method described in the embodiment.

[0062] Figure 8 The computation time curve for the traditional model predictive control method is shown.

[0063] Figure 9 The computation time curve for the 44-vector method;

[0064] Figure 10 This is a graph showing the computation time of the virtual vector method in the embodiment;

[0065] Figure 11 This is a flowchart of a common-mode voltage suppression method based on cascaded observers. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0067] Specific Implementation Method 1: The common-mode voltage suppression method based on cascaded observers described in this implementation method includes:

[0068] The total disturbance in the motor control system is observed and compensated using a cascaded observer, and then the reference voltage vector of the motor is calculated. The cascaded observer includes a cascaded extended state observer and a harmonic internal mode observer.

[0069] Construct a discrete virtual vector set, divide the vector plane formed by the discrete virtual vector set into multiple continuous and non-overlapping sub-regions, and ensure that each sub-region uniquely corresponds to a virtual vector in the discrete virtual vector set;

[0070] Determine the sub-region to which the reference voltage vector belongs in the vector plane, and use the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector;

[0071] A PWM signal is generated based on the target vector, and then the motor is controlled to achieve common-mode voltage suppression.

[0072] In one embodiment, the cascaded extended state observer includes two cascaded third-order extended state observers, wherein,

[0073] The expression for the first-level extended state observer is:

[0074] ,

[0075] The expression for the second-level extended state observer is:

[0076] ,

[0077] In the formula, , and All are extended state observer gains. To control the gain, For dq axis voltage, For dq axis current, and These represent the predicted current values ​​of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative, and These represent the perturbation estimation errors of the first-level and second-level extended state observers, respectively. and They are respectively and The first derivative, and Let these represent the observation error derivatives of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative.

[0078] In one implementation, the perturbation error transfer function of the cascaded extended state observer The expression is:

[0079] ,

[0080] in, For the Laplace operator, This represents the actual disturbance of the dq axis in the motor control system. To estimate the total disturbance, and These are the estimated perturbations for the first-level and second-level extended state observers, respectively.

[0081] In one implementation, the expression for the harmonic internal mode observer is:

[0082] ,

[0083] In the formula, , and These represent the estimation results for the three harmonic components, respectively.

[0084] , and All are intermediate variables, and each has an expression:

[0085] ;

[0086] For observer gain, ; , and All are resonant bandwidths; The distance between the magnetic poles of the motor; The speed of the motor's rotor;

[0087] Let be the harmonic internal mode term of the motor control system, and have the following expression:

[0088] ,

[0089] To control the gain, For dq axis voltage, This is the predicted current value. This is the sum of the perturbation estimation errors of the two-stage extended state observer.

[0090] In one embodiment, calculating the reference voltage vector of the motor includes:

[0091] The reference voltage vector of the motor is calculated using the following formula:

[0092] ,

[0093] in, This represents the sampling time in a discrete control system. for The reference voltage vector of the motor at any given time. To control the cycle, This is the current reference value. for The actual total disturbance at any given moment. for Predicted current value at any given time.

[0094] In one embodiment, the The actual total disturbance at time t is expressed as:

[0095] ,

[0096] The predicted current value at time t is expressed as:

[0097] ;

[0098] in, for The sum of the perturbation estimation errors of the two-stage extended state observer at time t is for The actual total disturbance at any given moment. and They are respectively Current and voltage along the dq axis at any given time.

[0099] In one implementation, constructing the discrete virtual vector set includes:

[0100] Discrete Virtual Vector The expression is:

[0101] ,

[0102] in, and All are used for synthesis Two different effective voltage vectors, It is a zero vector. The interval for vector partitioning, and They are respectively and The time interval satisfies and .

[0103] In one embodiment, dividing the vector plane formed by the discrete virtual vector set into multiple continuous and non-overlapping sub-regions includes:

[0104] The vector plane is divided using three straight lines, each perpendicular to one of the three axes of the vector plane, such that the three straight lines and the three axes of the vector plane can divide the vector plane into multiple continuous and non-overlapping sub-regions.

[0105] In one embodiment, determining the sub-region to which the reference voltage vector belongs in the vector plane, and using the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector, includes:

[0106] The reference voltage vector is projected onto the three axes of the vector plane;

[0107] The sub-region to which the projection of the reference voltage vector onto the three axes belongs is determined by the following formula:

[0108] ,

[0109] in, , Indicates a sub-region index. Indicates the reference voltage vector at Projection on axis This refers to the DC bus voltage of the inverter.

[0110] The sub-region to which the reference voltage vector belongs in the vector plane is obtained based on the sub-region index and table lookup method.

[0111] In one implementation, generating the PWM signal based on the target vector includes:

[0112] When the target vector is a single vector, the PWM signal is directly generated using the single vector.

[0113] When the target vector is a synthesized vector, a combination of base voltage vectors that generate low common-mode voltage is selected, and their preset action sequence and time allocation are optimized to synthesize a PWM signal.

[0114] To further illustrate the implementation scheme of this application, Figure 11 A common-mode voltage suppression method based on cascaded observers is provided, wherein the numbering of the steps does not necessarily restrict their execution order. Each step is described in detail below:

[0115] This embodiment relates to an improved discrete control set model predictive control method that takes into account total harmonic distortion, computational burden, and reduction of AC side common-mode voltage.

[0116] I. Current Disturbance Suppression Scheme

[0117] The model expression for the three-phase two-level inverter system of PMLSM is as follows:

[0118] (1),

[0119] in, , ,

[0120] , , ,

[0121] and These represent the d-axis and q-axis currents, respectively. and These represent the voltages along the d-axis and q-axis, respectively. and These represent the d-axis and q-axis inductance, respectively. and These represent the stator resistance and magnetic flux, respectively. The distance between the magnetic poles of the motor. This represents the speed of the motor's rotor.

[0122] The above formula can be simplified to Equation (2) using a hyperlocal model. In this way, the complex linear motor model is simply divided into motor parameters and total disturbance, thereby achieving control that does not depend on model parameters.

[0123] (2),

[0124] In the formula, To control the gain, This is the nominal stator inductance. This is the perturbation matrix, which includes perturbations along the d-axis and q-axis. and These disturbances can be further classified into DC disturbances and AC disturbances. DC disturbances are caused by load transients, parameter changes, cross-coupling terms, etc., while AC harmonics mainly include the first and second harmonics caused by proportional and bias errors in current sampling, and the sixth harmonic caused by dead-zone effects, inverter nonlinearity, etc. These DC disturbances and low-frequency AC harmonics cannot be filtered out by the LCL filter in the system, so they need to be suppressed by control algorithms.

[0125] This embodiment proposes a novel cascaded observer (ICHIO), which cascades a cascaded extended state observer (CESO) and a harmonic internal-model observer (HIO). The CESO is responsible for observing and suppressing DC and low-frequency disturbances, while the HIO can observe and suppress high-frequency disturbances by dynamically embedding harmonics into the observation error. This reduces the system order, improves the system accuracy, and enables ultra-precision control.

[0126] (a) Design of a novel cascaded extended state observer

[0127] Using a cascaded, third-order structure in an ESO (extended state observer) to observe the current, disturbance, and the derivative of the disturbance separately can improve the observer bandwidth and expand the range of observed currents.

[0128] The expression for the first-level observer is as follows:

[0129] (3),

[0130] The second-level observer is as follows:

[0131] (4),

[0132] In the formula, , and All are observer gains for CESO, typically set to 0. , , , This refers to the bandwidth of CESO. and These represent the predicted current values ​​generated by the first-stage and second-stage observers, respectively. and These represent the perturbation estimation errors of the first-stage and second-stage observers, respectively. and These represent the observation error derivatives of the first-stage and second-stage observers, respectively.

[0133] By performing a Laplace transform on equations (3) and (4), the perturbation error transfer function of the designed third-order CESO observer is obtained. :

[0134] (5).

[0135] in, For the Laplace operator, This represents the actual disturbance of the dq axis in the motor control system. To estimate the total disturbance, and These are the estimated perturbations for the first-level and second-level extended state observers, respectively.

[0136] To better illustrate the effectiveness of different observers in perturbation estimation, perturbation error transfer functions were plotted for 2nd-order ESO, 3rd-order ESO, and 3rd-order CESO. Bode plot, such as Figure 1 As shown in the diagram, the solid line represents a 2nd-order ESO, the dashed line represents a 3rd-order ESO, and the dashed line represents a 3rd-order CESO. It can be observed that compared to the non-cascaded structure, CESO has a larger estimation bandwidth and a faster dynamic response; higher-order ESO observers can provide stronger disturbance tracking performance and estimation bandwidth. Therefore, the 3rd-order CESO achieves the best disturbance estimation accuracy and bandwidth.

[0137] (b) Design of harmonic internal mode observer

[0138] Will Substituting into formula (5), we obtain the following expressions for amplitude frequency and phase frequency:

[0139] (6),

[0140] in, and These represent the amplitude frequency and phase frequency, respectively. The imaginary unit, The harmonic angular frequency of the AC disturbance to be observed.

[0141] This indicates that the third-order CESO suffers from amplitude attenuation and phase shift when observing AC disturbances, making it difficult to accurately observe and suppress AC disturbances. Therefore, this embodiment designs a harmonic internal mode observer, which is cascaded with the third-order CESO described above to form a novel observer.

[0142] According to equation (2), the expression for the harmonic internal mode term of the system is:

[0143] (7),

[0144] in, This is the predicted current value. The sum of the perturbation estimation errors of the two-stage extended state observer is: .

[0145] By using harmonic internal mode terms Introduced into the error correction section of the HIO observer, the HIO structure can be described as follows:

[0146] (8),

[0147] In the formula, , and The estimated results represent the three harmonic components; , and It is an intermediate variable; , and All are resonant bandwidths. , , ; It is the observer gain. .

[0148] (c) Design and analysis of cascaded ICHIO observers

[0149] The aforementioned third-order CESO and HIO are cascaded into ICHIO, resulting in a total actual perturbation. The estimation result is expressed as:

[0150] .

[0151] Based on the hyperlocal model, i.e., equation (2), considering the one-step delay introduced by the digital control system, the improved MPCC (Model Predictive Current Control) reference voltage vector with disturbance and harmonic compensation can be derived, and its expression is:

[0152] (9),

[0153] in, This represents the sampling time in a discrete control system. for The reference voltage vector of the motor at any given time. To control the cycle, This is the current reference value. for The sum of the perturbation estimation errors of the two-stage extended state observer at time t is for Predicted current value at any given time.

[0154] Therefore, based on the proposed ICHIO observer, and taking into account both DC disturbances and AC harmonics in the motor current loop, a more accurate MPCC reference voltage vector can be derived.

[0155] II. Common-mode voltage suppression scheme

[0156] Since the number of available voltage vectors in a traditional three-phase two-level voltage source inverter (VSI) is limited, this embodiment proposes an extended control set strategy based on discrete virtual vectors. This strategy can effectively increase the number of candidate voltage vectors, thereby reducing the voltage error between the desired reference voltage vector and the pre-selected candidate vectors. To alleviate the computational burden in the digital control system, an optimization algorithm for selecting discrete virtual vectors is further proposed, which can achieve fast and accurate identification of the optimal switching vector. In addition, considering the synthesis characteristics of the selected virtual vectors, current ripple, switching frequency of power devices, and the impact of dead time on common-mode voltage, the switching sequence of the selected voltage vectors is optimized. Specifically:

[0157] Discrete Virtual Vector Defined as:

[0158] (10)

[0159] In the formula, and All are used for synthesis Two different effective voltage vectors, It is a zero vector. The interval for vector partitioning, The larger the value, the more virtual vectors are divided within the region. and They are and The time interval satisfies and This embodiment takes A schematic diagram of discrete virtual voltage vectors is shown below. Figure 2 As shown, there are a total of 37 virtual vectors to choose from, which is more than the number of vectors available in the traditional virtual vector method, improving the accuracy of current control and reducing THD.

[0160] Traditional virtual vector methods select virtual vectors by substituting each virtual vector into a cost function, which significantly increases the computational burden. Therefore, this embodiment introduces three lines perpendicular to the A, B, and C coordinate axes, dividing the entire region into 37 sub-regions, each containing one virtual vector, such as... Figure 3 As shown. The reference voltage vector... After inverse Clarke and inverse Park transformations, the projections become the projections of the A, B, and C axes. The transformation process is shown in equation (11). By directly selecting the corresponding virtual vectors based on the projections on each coordinate axis, the calculation time can be accelerated.

[0161] (11),

[0162] In the formula, , , These are the projected values ​​of the voltage on axes A, B, and C, respectively. This represents the displacement of the PMLSM.

[0163] Since the A, B, and C axes are divided into multiple sub-regions by a series of vertical boundaries, the reference voltage can be mapped as discrete amplitude intervals along these axes. These intervals are indexed by the sub-regions. The representation is defined as follows:

[0164] (12)

[0165] in, This refers to the DC bus voltage of the inverter. .

[0166] pass , and The sub-regions can be determined, and the corresponding relationships are summarized in Table 1. The blank areas in the table represent areas that would not appear in practice. , and Combinations. Table 1 shows how to quickly, accurately, and efficiently select discrete virtual vectors.

[0167] Table 1. Correspondence Table of Discrete Virtual Vector Sets

[0168]

[0169] The traditional FCS-MPC scheme introduces zero vector , The generated common-mode voltage is ,and The generated common-mode voltage is only Therefore, this embodiment uses only six non-zero vectors. Synthesize all the virtual vectors. The 37 virtual vectors can be divided into three groups: single vectors, double vectors, and quad vectors. Let's take the first sector as an example for analysis:

[0170] Case 1 (Single Vector Composition): For The synthesized voltage remains constant throughout the entire control cycle, requiring no vector switching, and is classified as single-vector operation. Because single-vector synthesis does not introduce a zero vector, the common-mode voltage can be maintained at... Therefore, a single vector is used to generate a PWM signal, which is then applied to the three-phase inverter to control the motor.

[0171] Case 2 (Two-vector composition): Vector and It can be done by simply and Synthesis. To reduce output current ripple, The vector sequence is used to synthesize a PWM signal, which is applied to the three-phase inverter to control the motor and ultimately reduce the common-mode voltage.

[0172] vector and need And zero vector, however, in order to suppress the high common-mode voltage caused by the zero vector, use and The zero vector is synthesized, and the order of vector synthesis is as follows: .

[0173] Case 3 (Four-Vector Composition): Vector need , And zero vector, considering the simultaneous reduction of CMV and current ripple, the switching sequence is selected as The vector sequence is used to synthesize the PWM signal, which is then applied to the three-phase inverter.

[0174] The above summarizes the switching sequence of the seven vectors in the first sector proposed in this embodiment, as shown in Table 2. The analysis strategy for the other five sectors follows the same process, and therefore will not be repeated here.

[0175] Table 2

[0176]

[0177] Figure 4 This is a comparative experimental graph showing the current performance of different observers. Figure 4 In the figure, (a) represents the dq-axis current output by the linear ESO, with dq-axis current ripples of 0.3262 A and 0.4070 A, respectively. Figure 4 In (b), the dq-axis current output by the third-order CESO is shown, with dq-axis current ripples of 0.3184 A and 0.3285 A, respectively. Figure 4 In the figure, (c) represents the dq-axis current output by ICHIO, with dq-axis current ripple of 0.3176 A and 0.2857 A, respectively. It can be seen that the dq-axis current ripple gradually decreases. The third-order CESO has a smaller ripple than the linear ESO due to the cascading of the observers. After adding the internal mode observer, the ripple is further reduced. Figure 4 In the figure, (d), (e), and (f) are FFT analysis plots of the dq axis current. It can be seen that the ICHIO observer proposed in this embodiment has a significant effect on suppressing the first, second, and sixth harmonics.

[0178] Figures 5-7 This is a diagram of the common-mode voltage and its amplified waveform, where... Figure 5This is a waveform diagram of the common-mode voltage on the AC side of an inverter under a traditional SVPWM modulation strategy. It can be seen that the common-mode voltage jumps frequently between -100V and 100V. . Figure 7 The graph shows the common-mode voltage and amplified waveform under traditional model predictive control methods. The common-mode voltage is also... Fluctuations. Figure 8 The graph shows the common-mode voltage and its amplified waveform under the improved model predictive control virtual vector method. It can be seen that the improved method can reduce the common-mode voltage amplitude to [value missing]. The common-mode voltage is significantly reduced, to one-third of that achieved with traditional methods.

[0179] Figures 8-10 The computation time used for different modulation strategies is shown, among which, Figure 8 The traditional FCS-MPC method substitutes eight vectors into the cost function in each cycle, selects the vector with the smallest cost function value as the optimal control quantity, and applies it to the next control cycle. The calculation time is 19.2 ns. Figure 9 As a variant of the FCS-MPC method, 44 virtual vectors were used, which were substituted into the cost function for calculation, and the calculation time was 29.55ns. Figure 10 The virtual vector method proposed in the embodiment divides the space into 37 regions by adding three sets of parallel lines, which greatly reduces the computation time to 19.1 ns, making it the fastest method in all experiments and effectively reducing computation time.

[0180] Specific Implementation Method Two: The common-mode voltage suppression device based on cascaded observers described in this embodiment includes a processor and a memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the common-mode voltage suppression method based on cascaded observers as described in Specific Implementation Method One.

[0181] Specific Implementation Method 3: A computer storage medium as described in this embodiment stores at least one instruction, which is loaded and executed by a processor to implement the common-mode voltage suppression method based on cascaded observers as described in Specific Implementation Method 1.

[0182] While specific embodiments of this application have been described herein with reference to them, it should be understood that these embodiments are merely examples of the principles and applications of this application. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of this application as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A common-mode voltage rejection method based on a cascade observer, characterized in that, include: The total disturbance in the motor control system is observed and compensated using a cascaded observer, and then the reference voltage vector of the motor is calculated. The cascaded observer includes a cascaded extended state observer and a harmonic internal mode observer. Construct a discrete virtual vector set, divide the vector plane formed by the discrete virtual vector set into multiple continuous and non-overlapping sub-regions, and ensure that each sub-region uniquely corresponds to a virtual vector in the discrete virtual vector set; Determine the sub-region to which the reference voltage vector belongs in the vector plane, and use the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector; A PWM signal is generated based on the target vector, and then the motor is controlled to achieve common-mode voltage suppression.

2. The method of claim 1, wherein the plurality of cascaded observers are configured to generate the plurality of error signals based on a plurality of differential mode signals and a common mode signal. The cascaded extended state observer includes two cascaded third-order extended state observers, wherein, The expression for the first-level extended state observer is: ， The expression for the second-level extended state observer is: ， In the formula, , and All are extended state observer gains. To control the gain, This is the dq-axis voltage. For dq axis current, and These represent the predicted current values ​​of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative, and These represent the perturbation estimation errors of the first-level and second-level extended state observers, respectively. and They are respectively and The first derivative, and Let these represent the observation error derivatives of the first-stage and second-stage extended state observers, respectively. and They are respectively and The first derivative.

3. The common-mode voltage suppression method based on cascaded observers according to claim 2, characterized in that, The perturbation error transfer function of the cascaded extended state observer The expression is: ， in, For the Laplace operator, This represents the actual disturbance of the dq axis in the motor control system. To estimate the total disturbance, and These are the estimated perturbations for the first-level and second-level extended state observers, respectively.

4. The common-mode voltage suppression method based on cascaded observers according to claim 2 or 3, characterized in that, The expression for the harmonic internal mode observer is: ， In the formula, , and These represent the estimation results for the three harmonic components; , and All are intermediate variables, and they all contain expressions: ； For observer gain, ; , and All are resonant bandwidths; The distance between the magnetic poles of the motor; The speed of the motor's rotor; Let be the harmonic internal mode term of the motor control system, and have the following expression: ， To control the gain, This is the dq-axis voltage. This is the predicted current value. This is the sum of the perturbation estimation errors of the two-stage extended state observer.

5. The common-mode voltage suppression method based on cascaded observers according to claim 4, characterized in that, The calculation of the reference voltage vector of the motor includes: The reference voltage vector of the motor is calculated using the following formula: ， in, This represents the sampling time in a discrete control system. for The reference voltage vector of the motor at any given time. To control the cycle, This is the current reference value. for The actual total disturbance at any given moment. for Predicted current value at any given time.

6. The common-mode voltage suppression method based on cascaded observers according to claim 5, characterized in that, The The actual total disturbance at time t is expressed as: ， The predicted current value at time t is expressed as: ； in, for The sum of the perturbation estimation errors of the two-stage extended state observer at time t is for The actual total disturbance at any given moment. and They are respectively Current and voltage along the dq axis at any given time.

7. The common-mode voltage suppression method based on cascaded observers according to claim 1, characterized in that, The construction of the discrete virtual vector set includes: Discrete Virtual Vector The expression is: ， in, and All are used for synthesis Two different effective voltage vectors, It is a zero vector. The interval for vector partitioning, and They are respectively and The time interval satisfies and .

8. The common-mode voltage suppression method based on cascaded observers according to claim 1, characterized in that, The step of dividing the vector plane composed of the discrete virtual vector set into multiple continuous and non-overlapping sub-regions includes: The vector plane is divided using three straight lines, each perpendicular to one of the three axes of the vector plane, such that the three straight lines and the three axes of the vector plane can divide the vector plane into multiple continuous and non-overlapping sub-regions.

9. The common-mode voltage suppression method based on cascaded observers according to claim 1, characterized in that, The step of determining the sub-region to which the reference voltage vector belongs in the vector plane, and using the virtual vector corresponding to the sub-region to which the reference voltage vector belongs as the target vector, includes: The reference voltage vector is projected onto the three axes of the vector plane; The sub-region to which the projection of the reference voltage vector onto the three axes belongs is determined by the following formula: ， in, , Indicates a sub-region index. Indicates the reference voltage vector at Projection on the axis This refers to the DC bus voltage of the inverter. The sub-region to which the reference voltage vector belongs in the vector plane is obtained based on the sub-region index and table lookup method.

10. The common-mode voltage suppression method based on cascaded observers according to claim 1, characterized in that, The step of generating a PWM signal based on the target vector includes: When the target vector is a single vector, the PWM signal is directly generated using the single vector. When the target vector is a synthesized vector, the PWM signal is synthesized according to the synthesized vector order.

Images