Motor model prediction control method and system based on magnetic modulation pulse counting disturbance observer

Abstract

The invention discloses a motor model prediction control method and system based on a magnetic modulation pulse counting disturbance observer, and the method employs a dual-vector MPCC to select and synthesize voltage vectors, and the selection of the voltage vectors is more accurate. In the disturbance observer, aiming at the parameter mismatch problem, the influence of internal disturbance and external disturbance is considered, and the obtained disturbance observation value is more accurate; furthermore, the external disturbance and disturbance caused by internal parameter mismatch are combined to form total disturbance, disturbance compensation is carried out through a PI controller, and the control precision of the system can be improved. In addition, the sliding mode gain of the approaching rate of the disturbance observer is improved by using a sectional function, the buffeting of the motor can be reduced, and the stable operation is ensured.

Description

Technical Field

[0001] The invention relates to motor control, and in particular to a motor model predictive control method and system based on a magnetic modulation pulse counting disturbance observer. Background Art

[0002] Model predictive control is applied to flux-adjustable motor control, which can obtain good dynamic performance and good steady-state performance. For model predictive control, the correct establishment of the motor mathematical model is the basis for its implementation. However, affected by factors such as temperature rise, magnetic saturation, and external environment, the parameters of the flux-adjustable motor will change, causing a large deviation in the flux-adjustable motor mathematical model established with the motor parameters, resulting in a deviation in the results predicted by the motor mathematical model, and then causing the optimal voltage vector selected by the evaluation function to be misjudged, resulting in a decrease in the control performance of the system. In order to improve the robustness of model predictive control to parameter mismatch, the mainstream method is to regard the prediction error value caused by parameter mismatch as a disturbance, and then use a disturbance observer to observe and compensate for it. The existing model predictive control only considers internal disturbances, such as inductance or flux linkage, for the parameter mismatch problem, and does not consider the influence of external disturbances. For example, when the flux of the flux-adjustable motor changes, the inductance value and flux linkage value will also change accordingly, thereby generating parameter disturbances and affecting the operation of the system. Summary of the invention

[0003] Purpose of the invention: The purpose of the present invention is to provide a motor model predictive control method and system based on a magnetic modulation pulse count disturbance observer, so as to estimate the disturbance value caused by parameter mismatch when the magnetic flux of a flux-adjustable motor changes and perform real-time compensation, eliminate the prediction error caused by parameter mismatch, and improve the robustness and anti-disturbance performance of motor control.

[0004] Technical solution: The motor model predictive control method based on magnetic modulation pulse counting disturbance observer of the present invention comprises:

[0005] (1) Construct a finite set prediction model for flux-adjustable motors;

[0006] (2) Construct a flux-adjustable motor disturbance observer based on a second-order sliding mode variable structure;

[0007] (3) Using the disturbance observer to obtain the disturbance observation value, constructing the disturbance controller, and calculating the compensation voltage of the d-axis and q-axis;

[0008] (4) constructing d-axis and q-axis control voltages containing compensation voltages in each switching state of the three-phase inverter according to the d-axis and q-axis compensation voltages;

[0009] (5) Construct an objective function and select the switch state on the inverter bridge arm that minimizes the objective function value as the switch control output of the three-phase inverter in the current cycle to suppress the disturbance existing in the flux adjustable motor.

[0010] Furthermore, step (1) includes: a finite set prediction model of the flux-adjustable motor, which is expressed as:

[0011]

[0012] Among them, I d (k+1), I q (k+1) are the estimated d-axis and q-axis current prediction values ​​at time k+1; L d , L q are the d-axis and q-axis inductance values ​​respectively; T is the sampling time; R s is the stator winding resistance; I d (k) I q (k) are the d-axis and q-axis current values ​​at time k; p is the number of motor pole pairs; ω m (k) is the motor speed at time k; U d (k) U q (k) are the d-axis and q-axis voltage values ​​at time k; ψ f is the permanent magnet flux.

[0013] Furthermore, in step (1), a dual-vector model is applied to predict current control, and a second optimal voltage vector is selected based on the optimal voltage vector selected by the traditional MPCC; by allocating the action time of the two voltage vectors, the error between the current prediction value and the reference value of the d-axis and q-axis within one sampling period is minimized; the allocation of the voltage vector action time adopts zero-beat control.

[0014] Furthermore, according to the deadbeat control principle, at the next sampling moment, the predicted current value is equal to the given value, so we can get:

[0015]

[0016] Wherein, s1 is the current slope of the q-axis when the first optimal voltage vector acts; t1 is the action time of the first optimal voltage vector; s2 is the current slope of the q-axis when the second optimal voltage vector acts; T-t1 is the action time of the second optimal voltage vector; is the reference value of the stator current under the q axis;

[0017]

[0018] Where s0 is the current slope of the q-axis when the voltage vector acts at zero; U q1 , U q2is the voltage value of the q axis when two optimal voltage vectors act; R is the stator resistance; ω e is the rotor electrical angular velocity;

[0019] According to the cost function of single-vector MPCC, considering the action time of two voltage vectors at the same time, the predicted voltage vector can be rewritten as:

[0020]

[0021] Among them, U d and U q is the optimal voltage vector value; U d1 , U d2 is the voltage value of the d-axis when two optimal voltage vectors act.

[0022] Furthermore, step (2) includes: the sliding mode disturbance observer expression is:

[0023]

[0024] in, are the estimated current values ​​of the d-axis and q-axis respectively; are the disturbance estimates of the d-axis and q-axis respectively; H d，smo , H q，smo is the sliding mode control function; b1 and b2 are feedback gains;

[0025] The error between the actual and estimated current values ​​of the d-axis and q-axis is defined as:

[0026]

[0027] The error equation is:

[0028]

[0029] Among them, e I，d and e I，q is the error between the actual value and the estimated value of the current on the d-axis and q-axis;

[0030] In order to ensure that the error can converge quickly, the sliding surface is defined as e I，d =s I，d and e I，q =s I，q ,s I，d and I，q is the sliding surface function of the d-axis and q-axis;

[0031] The traditional constant velocity approach rate is defined as:

[0032]

[0033] Where k is the sliding mode gain, sign(s) is the sign function, and s is the sliding surface function;

[0034]

[0035] On the basis of the constant approach rate, a disturbance observer with improved approach rate is designed. The sliding mode gain of the disturbance observer with improved approach rate is realized by a piecewise function:

[0036]

[0037] Among them, k * is the sliding mode gain of the disturbance observer for improving the approach rate; x1 is the state variable, x2 is the derivative of the state variable; ε>1, z>0, m>0;

[0038] Substituting the traditional constant approach rate, sign function sign(s) and piecewise function into the error equation yields:

[0039]

[0040] From the above formula, the sliding mode control function can be obtained as:

[0041]

[0042] in, and They are respectively represented as the sliding mode gains of the d-axis and q-axis; sign(s I，d ) and sign(s I，q ) is expressed as a sign function of the d-axis and the q-axis; and Represented as e I，d and e I，q The first derivative of .

[0043] Furthermore, the stability of the improved sliding mode approach rate is proved by Lyapunov function, and the Lyapunov function is defined as:

[0044]

[0045] Taking the derivative of V, we get:

[0046]

[0047] Established, according to Lyapunov stability principle, the improved approach rate can ensure the stability of the system.

[0048] Further, step (3) comprises:

[0049] Use the disturbance observer to obtain the d-axis and q-axis disturbance observation values ​​at time k and Construct disturbance control on the d and q axes and calculate the compensation voltages on the d and q axes, which can be expressed as:

[0050]

[0051] Among them, u d_com (k) and u q_com (k) are the compensation voltages of the d-axis and q-axis at time k; k p_d , k i_d are two parameters of the disturbance controller on the d-axis; k p_q , k i_q are two parameters of the disturbance controller on the q axis; f d_err (k), f q_err (k) are the tracking errors of the d-axis and q-axis at time k; u d_com (k-1),u q_com (k-1) are the compensation voltages of the d-axis and q-axis at time k-1, and their initial values ​​are both 0; T is the sampling time;

[0052]

[0053] Further, step (4) comprises:

[0054] The three-phase inverter has k possible switch states at time k. The switch state of the mth group is S m , the control voltage expression is:

[0055]

[0056] Among them, u dm (k) and u qm (k) are the switch states S of the mth group of inverters at time k. m The d and q axis control voltage values ​​under θ(k) are as follows; θ(k) is the rotor position angle at time k; u unm 、u vnm 、u wnm is the switch state S of the mth group of inverters m The three-phase voltage under the condition of m=1, 2, ..., k;

[0057]

[0058] Among them, U dc is the DC side voltage;

[0059] Construct the mth group of switching states S of the inverter after compensation m The d-axis and q-axis control voltages under dm_com (k) and u qm_com (k), its expression is:

[0060]

[0061] Further, step (5) comprises:

[0062] The d-axis and q-axis voltage values ​​u corresponding to the switching vectors of all three-phase inverters after compensation d_com (k) and u q_com (k) is brought into the constructed finite set prediction model of the flux-adjustable motor to obtain the predicted current value i of the inverter in the mth switching state at time k+1 dm (k+1) and i qm (k+1), its expression is:

[0063]

[0064] in, and is the actual value of the d-axis and q-axis inductance when there is a parameter error; R0 represents the actual value of the stator resistance of the flux-adjustable motor when there is a parameter error; It indicates the actual value of the permanent magnet flux linkage of the flux-adjustable motor when there is a parameter error;

[0065] The objective function J of the model predictive control of the flux-adjustable motor is constructed, and its expression is:

[0066]

[0067] in, They are the d-axis and q-axis reference currents respectively;

[0068] Taking the minimum value of the objective function of current tracking as the best control as the optimization goal, a one-step optimization calculation function is established, and the expression is:

[0069] J(S1)=min{J(S m )}

[0070] Where J(S1) represents the minimum value of the objective function when the inverter is in the first switching state S1, S1∈[S1, S2, ..., S k ];

[0071] The first group of switching states of the inverter is used as the control instruction of the flux-adjustable motor control system at time k to control the switching states of each bridge arm of the inverter.

[0072] The motor model predictive control system based on the magnetic modulation pulse counting disturbance observer of the present invention comprises an inverter (2), a coordinate transformation module, a disturbance observer, a PI controller, a prediction module, a dual-vector optimization module, a positive and negative pulse detection module, a dual-arm bridge inverter circuit module and a speed loop adjustment module;

[0073] The coordinate transformation module converts the three-phase current I a (k) I b (k) I c (k)Convert to I d (k) I q (k), and outputs it to the disturbance observer; the positive and negative pulse detection module and the double-arm bridge inverter circuit module detect the positive and negative pulses of the flux-adjustable motor, and output the positive pulse n1 and the negative pulse n2 to the disturbance observer; the disturbance observer adjusts the voltage vector value U according to the optimal voltage vector value U d and U q 、Current I d ,I q And positive pulse n1, negative pulse n2, obtain disturbance observation value and And output it to the PI controller; PI controller according to the disturbance observation value and Get the compensation voltage u of the d-axis and q-axis at the current moment d_com (k) and u q_com (k), and output it to the prediction module; the prediction module and the dual vector optimization module calculate the predicted current value i dm (k+1) and i qm (k+1), and the reference current A switch combination that minimizes the objective function is selected in the inverter; the inverter uses a set of switch combinations that minimize the objective function to control the flux-adjustable motor.

[0074] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0075] The present invention aims at the problem of parameter mismatch in the existing model predictive control, which only considers the internal disturbance but does not consider the influence of the external disturbance. For example, when the magnetic flux of the adjustable flux motor changes, the inductance value and the flux linkage value will also change accordingly, thereby generating parameter disturbance and affecting the operation of the system. The present invention proposes to count the number of positive and negative pulses collected when the magnetic flux of the adjustable flux motor changes, and then use a disturbance observer to observe the disturbance, combine this part of the external disturbance with the disturbance caused by the internal parameter mismatch to form a total disturbance, and then use a PI controller to compensate for the disturbance, thereby improving the control accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 It is a flowchart of a motor model predictive control method based on a magnetic modulation pulse counting disturbance observer provided by an embodiment of the present invention;

[0077] Figure 2is a flowchart of a disturbance observer acquiring d-axis and q-axis disturbances in an embodiment of the present invention;

[0078] Figure 3 It is a flowchart of the internal current estimation of the disturbance observer in the embodiment of the present invention. DETAILED DESCRIPTION

[0079] The present invention will be further described below in conjunction with the accompanying drawings.

[0080] Attached Figures 1 to 3 The reference numerals in the figures are as follows:

[0081] 1. Flux adjustable motor; 2. Inverter; 3. Coordinate transformation module; 4. Disturbance observer; 5. PI controller; 6. Prediction module; 7. Dual vector optimization module; 8. Positive and negative pulse detection module; 9. Dual-arm bridge inverter circuit module; 10. Speed ​​loop adjustment module.

[0082] The embodiment of the present invention provides a motor model predictive control method based on a magnetic modulation pulse count disturbance observer, which specifically includes the following steps:

[0083] (1) Construct a finite set prediction model for flux-adjustable motors;

[0084] See also Figure 1 It should be noted that the model predictive control of the flux-adjustable motor in this embodiment is in the inner loop current control link in the motor double closed-loop control structure, and the system control quantity in this embodiment is voltage.

[0085] (101) The sensor is used to sample the flux-adjustable motor 1 to obtain the motor state of the flux-adjustable motor 1 in real time, including the three-phase current I at time k. a (k) I b (k) I c (k), and the rotor position angle θ(k), the motor speed ω m (k).

[0086] For three-phase current I a (k) I b (k) I c (k) Coordinate transformation is performed to obtain the current value I at time k under the d-axis and q-axis d (k) I q (k):

[0087]

[0088] (102) The mathematical model of the flux-adjustable motor is established and expressed as:

[0089]

[0090] Among them, U d (k) U q (k) are the d-axis and q-axis voltage values ​​at time k; R s is the stator winding resistance; L d , L q are the inductance values ​​of the d-axis and q-axis respectively; p is the number of motor pole pairs; ψ f is the permanent magnet flux.

[0091] (103) According to the mathematical model of the flux-adjustable motor, the finite set prediction model expression of the flux-adjustable motor is constructed using the first-order Euler formula:

[0092]

[0093] Among them, I d (k+1), I q (k+1) are the estimated d-axis and q-axis current prediction values ​​at time k+1; T is the sampling time.

[0094] (104) The dual-vector model predictive current control method (MPCC) is applied. Based on the optimal voltage vector selected by the traditional MPCC, the dual-vector MPCC selects a second optimal voltage vector from the remaining voltage vectors. By allocating the action time of the two voltage vectors, the error between the predicted current value and the reference value of the d-axis and q-axis within a sampling period is minimized.

[0095] Calculation of vector action time: The voltage vector action time is allocated using deadbeat control. Before calculating the action time, the slope of the voltage vector needs to be calculated. The slope calculation formula for the voltage vector is:

[0096]

[0097] Where, s0 is the current slope of the q-axis when the voltage vector acts at zero; R is the stator resistance; ω e is the rotor electrical angular velocity; s1 is the current slope of the q-axis when the first optimal voltage vector acts; s2 is the current slope of the q-axis when the second optimal voltage vector acts; U q1 , U q2 is the voltage value of the q-axis when two optimal voltage vectors act.

[0098] (105) According to the deadbeat control principle, at the next sampling moment, the predicted current value is equal to the given value, so we can get:

[0099]

[0100] Among them, t1 is the action time of the first optimal voltage vector; T-t1 is the action time of the second optimal voltage vector; is the reference value of the stator current on the q axis.

[0101] According to the above formula, the action time of the first optimal voltage vector is:

[0102]

[0103] According to the cost function of single-vector MPCC, considering the action time of two voltage vectors at the same time, the predicted voltage vector can be rewritten as:

[0104]

[0105] Among them, U d and U q is the optimal voltage vector value; U d1 , U d2 is the voltage value of the d-axis when two optimal voltage vectors act.

[0106] Since the single-vector MPCC three-phase circuit has a relatively large harmonic content, the d- and q-axis currents fluctuate greatly, which in turn affects the performance of the motor. Therefore, this embodiment applies dual-vector MPCC on the basis of single-vector MPCC. Dual-vector MPCC selects the second optimal voltage vector based on the first optimal voltage vector. Ultimately, the two voltage vectors work together, so that the voltage vector can be selected more accurately, and the given and actual tracking are also more accurate.

[0107] Figure 3 : is a structural block diagram of the internal current estimation of the disturbance observer in an embodiment of the present invention. The present invention simultaneously considers the disturbance caused by the system internal parameter mismatch and the disturbance caused by the change of the flux of the flux-adjustable motor. Therefore, when estimating the current, the actual values ​​of the permanent magnet flux, stator inductance, and stator resistance of the flux-adjustable motor need to consider the influence of the internal parameter mismatch as well as the influence of the flux of the flux-adjustable motor on the flux, inductance, and resistance when the flux changes. d ,I q , U d , U q And the pulse number n1, n2 and other values ​​to estimate the current value The specific implementation methods are as follows:

[0108] (201) The influence of motor parameter error on model prediction current control is analyzed from the perspective of prediction error. When there is a parameter error, the actual values ​​of the permanent magnet flux, stator inductance, and stator resistance of the flux-adjustable motor are expressed as:

[0109]

[0110] Among them, ψ f，0 Indicates the actual value of the permanent magnet flux, Ls，0 represents the actual value of the stator inductance, R0 represents the actual value of the stator resistance; ψ f Indicates the nominal parameter of permanent magnet flux, L s represents the nominal parameter of stator inductance, R represents the nominal parameter of stator resistance; Δψ f , ΔL s , ΔR represents the error between the actual value of the corresponding parameter and the nominal parameter.

[0111] In a flux-adjustable motor, the relationship between flux, inductance L and flux linkage ψ is:

[0112]

[0113] ψ=Nφ

[0114] Among them, φ represents the magnetic flux of the flux adjustable motor; N represents the number of coil turns of the flux adjustable motor; i is the stator current value of the flux adjustable motor.

[0115] It can be seen that when the magnetic flux changes, the corresponding inductance and flux linkage will also change, so ΔL s With Δψ f After adjustment by positive pulse n1 and negative pulse n2, it is:

[0116]

[0117] in, and Indicates the error between the actual value and the nominal parameter of the corresponding parameter after positive and negative pulse adjustment; ΔL s0 is the initial parameter of inductance error; Δψ f0 is the initial parameter of flux error; ΔL is the change of inductance error parameter; Δψ is the change of flux error parameter.

[0118] Therefore, when there is a parameter error, the actual values ​​of the permanent magnet flux, stator inductance, and stator resistance of the flux adjustable motor can be rewritten as:

[0119]

[0120] in, R0 represents the actual values ​​of the permanent magnet flux linkage, stator inductance, and stator resistance of the flux adjustable motor when there is a parameter error.

[0121] like Figure 3 As shown in Figure 2, when there is a parameter disturbance, the current prediction model is modified as follows:

[0122]

[0123] In the formula, I′ d (k+1) and I′q (k+1) is the predicted current value of the direct axis and quadrature axis at the next moment when there is a parameter error; and is the actual value of the d-axis and q-axis inductance when there is a parameter error; and is the back electromotive force value of the direct axis and quadrature axis at the current moment when there is a parameter error.

[0124] (202) Considering the disturbance effect caused by motor parameter mismatch, the stator current equation is written as:

[0125]

[0126] In the formula, f d 、f q is the total disturbance of the direct axis and the quadrature axis; It indicates the error between the actual value of the d-axis and q-axis inductance and the nominal parameters after positive and negative pulse adjustment.

[0127] Since the sampling period is very short, the conversion rate of the parameter disturbance error can be considered to be 0, so the above equation is rewritten as:

[0128]

[0129] Figure 2 The present invention considers the disturbance caused by the mismatch of the internal parameters of the system and the disturbance caused by the change of the flux of the adjustable flux motor. With the measured value I d ,I q The calculated error value e I，d 、e I，q and the sliding mode control function H d，smo , H q，smo The total disturbance value of the disturbance caused by the internal parameter mismatch and the disturbance caused by the flux change of the flux-adjustable motor is calculated. The specific implementation methods are as follows:

[0130] The sliding mode disturbance observer is designed as:

[0131]

[0132]

[0133] in, are the estimated current values ​​of the d-axis and q-axis respectively; t represents time; are the disturbance estimates of the d-axis and q-axis respectively; H d，smo , H q，smoare the sliding mode control functions of the d-axis and q-axis respectively; b1 and b2 are the feedback gains;

[0134] The error between the actual and estimated current values ​​of the d-axis and q-axis is defined as:

[0135]

[0136] The error equation is:

[0137]

[0138] Among them, e I，d and e I，q is the error between the actual value and the estimated value of the current on the d-axis and q-axis.

[0139] In order to ensure that the error can converge quickly, the sliding surface is defined as e I，d =s I，d and e I，q =s I，q ,s I，d and I，q is the sliding surface function of the d-axis and q-axis;

[0140] The traditional constant velocity approach rate is defined as:

[0141]

[0142] Where k is the sliding mode gain, sign(s) is the sign function, and s is the sliding surface function;

[0143]

[0144] The discontinuity of the sign function in the constant approach rate used by the traditional disturbance observer will cause "jittering" in the sliding mode. The approach speed is mainly determined by the sliding mode gain k. The larger the k, the faster the approach speed and the larger the "jittering". Conversely, the smaller the k, the slower the approach speed, but the smaller the "jittering". To this end, based on the constant approach rate, a disturbance observer with improved approach rate is designed. The sliding mode gain of the disturbance observer with improved approach rate is realized by a piecewise function:

[0145]

[0146] Where, k* is the sliding mode gain of the disturbance observer with improved approach rate; x1 is the state variable, x2 is the derivative of the state variable; ε>1, z>0, m>0;

[0147] The parameter selection method of the sliding mode gain of the disturbance observer with improved approaching rate is as follows: using the disturbance observer with constant approaching rate, setting the sliding mode gain k * When the sliding mode gain k *When the “jittering” is small, it takes a long time for the d and q axis currents to reach a steady state. * When the "jitter" is large, the time required for the d and q axis currents to reach steady state is shorter. Sliding mode gain k * The value of should ensure that the "jitter" is small, and ε>1 should be taken to ensure that it can quickly approach the sliding surface. The value of m is determined by the size of the sliding surface.

[0148] The stability of the improved sliding mode approach rate is proved by Lyapunov function, and the Lyapunov function is defined as:

[0149]

[0150] Taking the derivative of V, we get:

[0151]

[0152] From the above formula we can see that Established, according to Lyapunov stability principle, the improved approach rate can ensure the stability of the system.

[0153] Substituting the traditional constant approach rate, sign function sign(s) and piecewise function into the error equation yields:

[0154]

[0155] From the above formula, the sliding mode control function can be obtained as:

[0156]

[0157] in, and They are respectively represented as the sliding mode gains of the d-axis and q-axis; sign(s I，d ) and sign(s I，q ) is expressed as a sign function of the d-axis and the q-axis; and Represented as e I，d and e I，q The first derivative of .

[0158] (3) Using the disturbance observer to obtain the disturbance observation value, constructing the disturbance controller, and calculating the compensation voltage of the d-axis and q-axis;

[0159] (301) The interference value f d and f q Controlled to 0, the current estimated by the measured motor parameters will be equal to the accurate result. In order to achieve this goal, the proposed disturbance observer is used to obtain the d-axis and q-axis disturbance observation values ​​at time k. and After that, the tracking error f with a disturbance reference value of zero is calculated.d_err (k) and f q_err (k) and then a PI controller is used to regulate the disturbance in the motor to track the required compensation voltage.

[0160]

[0161] Among them, u d_com (k) and u q_com (k) are the compensation voltages of the d-axis and q-axis at time k; k p_d , k i_d are two parameters of the disturbance controller (usually a PI controller) on the d-axis; k p_q , k i_q are two parameters of the disturbance controller (usually a PI controller) on the q axis; f d_err (k), f q_err (k) are the tracking errors of the d-axis and q-axis at time k; u d_com (k-1),u q_com (k-1) are the compensation voltages of the d-axis and q-axis at time k-1, and their initial values ​​are both 0.

[0162] (302) According to Routh’s stability criterion, in order for the system to be stable, the following conditions must be met:

[0163]

[0164] (4) constructing d-axis and q-axis control voltages containing compensation voltages in each switching state of the three-phase inverter according to the d-axis and q-axis compensation voltages;

[0165] (401) There are 27 switching states of the neutral point clamped three-level inverter. There are k possible switching states for the control operation at time k. The mth group of switching states is denoted as S m , whose expression is:

[0166] S m =[S U S V S W ]

[0167] Among them, S U Indicates the switch state of the U bridge arm, S V Indicates the switch state of the U bridge arm, S W Indicates the switch status of the U bridge arm.

[0168] A model is established between the output phase voltage and the inverter bridge arm switch state, and its expression is:

[0169]

[0170] Among them, uunm 、u vnm 、u wnm is the switch state S of the mth group of inverters m The three-phase voltage under dc is the DC side voltage.

[0171] (402) The d-axis and q-axis control voltages corresponding to the m-th group of switch states of the inverter are obtained by coordinate transformation, and the expressions thereof are:

[0172]

[0173] Among them, u dm (k) and u qm (k) are the switch states S of the mth group of inverters at time k. m The d and q axis control voltage values ​​under θ(k) are as follows; θ(k) is the rotor position angle at time k; m=1, 2,…, k.

[0174] (403) The d-axis and q-axis voltage values ​​of the inverter in the mth switching state, i.e., u dm (k) and u qm (k), plus the d-axis and q-axis compensation voltages at time k when the disturbance exists, the switch state S of the mth group of the inverter after compensation is obtained. m The dq axis control voltage value under dm_com (k) and u qm_com (k), its expression is:

[0175]

[0176] (5) Construct an objective function and select the switch state on the inverter bridge arm that minimizes the objective function value as the switch control output of the three-phase inverter in the current cycle to suppress the disturbance existing in the flux adjustable motor.

[0177] (501) The d-axis and q-axis voltage values ​​u corresponding to the switching vectors of all three-phase inverters after compensation are d_com (k) and u q_com (k) is brought into the constructed finite set prediction model of the flux-adjustable motor to obtain the predicted current value i of the inverter in the mth switching state at time k+1 dm (k+1) and i qm (k+1), its expression is:

[0178]

[0179] (502) The objective function J of the flux-adjustable motor model predictive control is constructed, and its expression is:

[0180]

[0181] in, are the d-axis and q-axis reference currents, respectively. In this embodiment, the d-axis and q-axis reference currents of the flux adjustable motor control system, namely and It is given after calculation by the system control outer loop.

[0182] (503) Taking the current tracking as the optimal control objective function value as the minimum as the optimization goal, a one-step optimization calculation function is established, and the expression is:

[0183] J(S1)=min{J(S m )}

[0184] Where J(S1) represents the minimum value of the objective function when the inverter switches in the first group are combined in S1, S1∈[S1,S2,…,S k ];

[0185] (504) The first switch combination of the inverter is used as a control instruction of the flux-adjustable motor control system at time k to control the switch state of each bridge arm of the inverter, thereby improving the robustness of the dual-vector model predictive current control system of the flux-adjustable motor.

[0186] See also Figure 1 The embodiment of the present invention also provides a motor model predictive control system based on a magnetic modulation pulse counting disturbance observer, including an inverter 2, a coordinate transformation module 3, a disturbance observer 4, a PI controller 5, a prediction module 6, a dual-vector optimization module 7, a positive and negative pulse detection module 8, a dual-arm bridge inverter circuit module 9 and a speed loop adjustment module 10.

[0187] The coordinate transformation module 3 transforms the three-phase current I of the flux adjustable motor 1 a (k) I b (k) I c (k)Convert to I d (k) I q (k), and outputs it to the disturbance observer 4; the positive and negative pulse detection module 8 and the double-arm bridge inverter circuit module 9 perform positive and negative pulse detection on the flux adjustable motor 1, and output positive pulse n1 and negative pulse n2 to the disturbance observer 4; the disturbance observer 4 adjusts the voltage vector value U according to the optimal voltage vector value U d and U q 、Current I d ,I q And positive pulse n1, negative pulse n2, obtain disturbance observation value and And output it to the PI controller 5; PI controller 5 according to the disturbance observation value and Get the compensation voltage u of the d-axis and q-axis at the current moment d_com(k) and u q_com (k), and output it to the prediction module 6; the prediction module 6 and the dual vector optimization module 7 calculate the predicted current value i dm (k+1) and i qm (k+1), and the reference current A switch combination that minimizes the objective function is selected in the inverter 2; the inverter 2 uses the switch combination that minimizes the objective function to control the flux-adjustable motor 1.

Claims

1. A motor model predictive control method based on a magnetic modulation pulse counting disturbance observer, characterized in that: include: (1) Construct a finite set prediction model for flux-adjustable motors; (2) Construct a flux-adjustable motor disturbance observer based on a second-order sliding mode variable structure; (3) Using the disturbance observer to obtain the disturbance observation value, constructing the disturbance controller, and calculating the compensation voltage of the d-axis and q-axis; (4) constructing d-axis and q-axis control voltages containing compensation voltages in each switching state of the three-phase inverter according to the d-axis and q-axis compensation voltages; (5) Construct an objective function and select the switch state on the inverter bridge arm that minimizes the objective function value as the switch control output of the three-phase inverter in the current cycle to suppress the disturbance existing in the flux adjustable motor.

2. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 1 is characterized in that: Step (1) includes: a finite set prediction model of a flux-adjustable motor, which is expressed as: Among them, Id(k+1) and Iq(k+1) are the estimated d-axis and q-axis current prediction values ​​at time k+1, respectively; Ld and Lq are the d-axis and q-axis inductance values, respectively; T is the sampling time; Rs is the stator winding resistance; Id(k) and Iq(k) are the d-axis and q-axis current values ​​at time k, respectively; p is the number of motor pole pairs; ωm(k) is the motor speed at time k; Ud(k) and Uq(k) are the d-axis and q-axis voltage values ​​at time k, respectively; ψf is the permanent magnet flux.

3. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 2 is characterized in that: In step (1), the dual-vector model is used to predict current control. Based on the optimal voltage vector selected by the traditional MPCC, a second optimal voltage vector is selected. By allocating the action time of the two voltage vectors, the error between the predicted current value and the reference value of the d-axis and q-axis within one sampling period is minimized. The allocation of the voltage vector action time adopts zero-beat control.

4. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 3 is characterized in that: According to the deadbeat control principle, at the next sampling moment, the predicted current value is equal to the given value, so we can get: Wherein, s1 is the current slope of the q-axis when the first optimal voltage vector acts; t1 is the action time of the first optimal voltage vector; s2 is the current slope of the q-axis when the second optimal voltage vector acts; T-t1 is the action time of the second optimal voltage vector; is the reference value of the stator current under the q axis; Among them, s0 is the current slope of the q-axis when the 0 voltage vector acts; Uq1 and Uq2 are the voltage values ​​of the q-axis when the two optimal voltage vectors act; R is the stator resistance; ωe is the rotor electrical angular velocity; According to the cost function of single-vector MPCC, considering the action time of two voltage vectors at the same time, the predicted voltage vector can be rewritten as: Among them, Ud and Uq are the synthesized optimal voltage vector values; Ud1 and Ud2 are the voltage values ​​of the d-axis when the two optimal voltage vectors act.

5. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to any one of claims 2 to 4, characterized in that: Step (2) includes: The sliding mode disturbance observer expression is: in, are the estimated current values ​​of the d-axis and q-axis respectively; are the disturbance estimates of the d-axis and q-axis respectively; Hd,smo and Hq,smo are sliding mode control functions; b1 and b2 are feedback gains; The error between the actual and estimated current values ​​of the d-axis and q-axis is defined as: The error equation is: Where, eI,d and eI,q are the errors between the actual and estimated current values ​​of the d-axis and q-axis; In order to ensure that the error can converge quickly, the sliding surface is defined as eI,d = sI,d and eI,q = sI,q, where sI,d and sI,q are the sliding surface functions of the d-axis and q-axis; The traditional constant velocity approach rate is defined as: Where k is the sliding mode gain, sign(s) is the sign function, and s is the sliding surface function; On the basis of the constant approach rate, a disturbance observer with improved approach rate is designed. The sliding mode gain of the disturbance observer with improved approach rate is realized by a piecewise function: Where, k* is the sliding mode gain of the disturbance observer with improved approach rate; x1 is the state variable, x2 is the derivative of the state variable; ε>1, z>0, m>0; Substituting the traditional constant approach rate, sign function sign(s) and piecewise function into the error equation yields: From the above formula, the sliding mode control function can be obtained as: in, and They are respectively represented as the sliding mode gains of the d-axis and q-axis; sign(sI,d) and sign(sI,q) are represented as the sign functions of the d-axis and q-axis; and Expressed as the first-order derivatives of eI,d and eI,q.

6. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 5 is characterized in that: The stability of the improved sliding mode approach rate is proved by Lyapunov function, and the Lyapunov function is defined as: Taking the derivative of V, we get: Established, according to Lyapunov stability principle, the improved approach rate can ensure the stability of the system.

7. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 5 is characterized in that: Step (3) includes: Use the disturbance observer to obtain the d-axis and q-axis disturbance observation values ​​at time k and Construct disturbance control on the d and q axes and calculate the compensation voltages on the d and q axes, which can be expressed as: Among them, ud_com(k) and uq_com(k) are the compensation voltages of the d-axis and q-axis at time k respectively; kp_d and ki_d are the two parameters of the disturbance controller on the d-axis; kp_q and ki_q are the two parameters of the disturbance controller on the q-axis; fd_err(k) and fq_err(k) are the tracking errors of the d-axis and q-axis at time k respectively; ud_com(k-1) and uq_com(k-1) are the compensation voltages of the d-axis and q-axis at time k-1 respectively, and their initial values ​​are both 0; T is the sampling time; 8. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 7 is characterized in that: Step (4) comprises: The three-phase inverter has k groups of possible switch states for control operation at time k. The switch state of the mth group is denoted as Sm, and its control voltage expression is: Among them, udm(k) and uqm(k) are the d-axis and q-axis control voltage values ​​of the inverter in the mth switching state Sm at time k, respectively; θ(k) is the rotor position angle at time k; uunm, uvnm, and uwnm are the three-phase voltages of the inverter in the mth switching state Sm; m=1,2,…,k; Wherein, Udc is the DC side voltage; Construct the d-axis and q-axis control voltages of the compensated inverter under the mth switching state Sm, namely udm_com(k) and uqm_com(k), whose expressions are:

9. The motor model predictive control method based on magnetic modulation pulse count disturbance observer according to claim 8 is characterized in that: Step (5) comprises: The d-axis and q-axis voltage values ​​ud_com(k) and uq_com(k) corresponding to the switching vectors of all three-phase inverters after compensation are brought into the constructed finite set prediction model of the flux-adjustable motor, and the predicted current values ​​idm(k+1) and iqm(k+1) of the inverter in the mth switching state at time k+1 are obtained, and the expressions are: in, and is the actual value of the d-axis and q-axis inductance when there is a parameter error; R0 represents the actual value of the stator resistance of the flux-adjustable motor when there is a parameter error; It indicates the actual value of the permanent magnet flux linkage of the flux-adjustable motor when there is a parameter error; The objective function J of the model predictive control of the flux-adjustable motor is constructed, and its expression is: in, They are the d-axis and q-axis reference currents respectively; Taking the minimum value of the objective function of current tracking as the best control as the optimization goal, a one-step optimization calculation function is established, and the expression is: J(S1)=min{J(Sm)} Where J(S1) represents the minimum value of the objective function when the inverter is in the first switching state S1, S1∈[S1,S2,…,S k ]; The first group of switching states of the inverter is used as the control instruction of the flux-adjustable motor control system at time k to control the switching states of each bridge arm of the inverter.

10. A motor model predictive control system based on a magnetic modulation pulse counting disturbance observer, characterized in that: It includes an inverter (2), a coordinate transformation module (3), a disturbance observer (4), a PI controller (5), a prediction module (6), a dual-vector optimization module (7), a positive and negative pulse detection module (8), a dual-arm bridge inverter circuit module (9) and a speed loop adjustment module (10); The coordinate transformation module (3) converts the three-phase currents Ia(k), Ib(k), and Ic(k) of the flux adjustable motor (1) into Id(k) and Iq(k), and outputs them to the disturbance observer (4); the positive and negative pulse detection module (8) and the double-arm bridge inverter circuit module (9) perform positive and negative pulse detection on the flux adjustable motor (1), and output positive pulses n1 and negative pulses n2 to the disturbance observer (4); the disturbance observer (4) obtains disturbance observation values ​​according to the optimal voltage vector values ​​Ud and Uq, the currents Id and Iq, and the positive pulses n1 and negative pulses n2. and And output it to the PI controller (5); PI controller (5) according to the disturbance observation value and The compensation voltages ud_com(k) and uq_com(k) of the d-axis and the q-axis at the current moment are obtained and output to the prediction module (6); the prediction module (6) and the dual-vector optimization module (7) calculate the predicted current values ​​idm(k+1) and iqm(k+1) and the reference current Selecting a switch combination that minimizes the objective function in the inverter (2); The inverter (2) controls the flux-adjustable motor (1) by using a set of switch combinations that minimize the objective function.

Images