阅读说明：本技术 一种磁电编码器角度值噪声主动抑制方法及装置 (Method and device for actively suppressing noise of angle value of magnetoelectric encoder ) 是由 王磊 吴殿昊 肖磊 潘巍 姜金刚 于 2020-06-01 设计创作，主要内容包括：本发明涉及一种磁电编码器角度值噪声主动抑制方法及装置包括：单对极霍尔传感器对角度值信号进行测量,模数转换器对霍尔信号进行模数转换,通过单对极角度计算模块获得单对极角度值；运动学方程神经元自调节模块,用于实现磁电编码器角度值的自适应观测；基于卡尔曼滤波迭代系数主动更新模块,用于提高神经元角度值观测误差的收敛速度,提高磁电编码器角度值噪声抑制效果；本发明采用一种能够主动抑制磁电编码器角度值信号中夹杂高频噪声的方法,用以提高磁电编码器角度值的输出精度。(The invention relates to a method and a device for actively inhibiting noise of an angle value of a magnetoelectric encoder, which comprises the following steps: the single-antipode Hall sensor measures a diagonal value signal, the analog-to-digital converter performs analog-to-digital conversion on the Hall signal, and a single-antipode angle value is obtained through the single-antipode angle calculation module; the kinematic equation neuron self-adjusting module is used for realizing self-adaptive observation of the angle value of the magnetoelectric encoder; the active updating module is used for improving the convergence speed of observation errors of the neuron angle values and improving the noise suppression effect of the magnetoelectric encoder angle values; the method can actively inhibit high-frequency noise mixed in the angle value signal of the magnetoelectric encoder, and is used for improving the output precision of the angle value of the magnetoelectric encoder.)

1. A magnetoelectric encoder angle value noise active suppression method is characterized in that: the method comprises the following concrete implementation processes:

(1) collecting single-antipodal angle value signals A & lt- & gt and A & lt- & gt;

(2) performing analog-to-digital conversion on the single-pair polar angle value signals A & lt + & gt and A & lt- & gt to obtain single-pair polar angle value digital signals HA & lt + & gt and HA & lt- & gt;

(3) solving the single-antipodal angle value according to the single-antipodal angle value digital signals HA + and HA-)θ₁；

(4) Establishing a kinematic state observation equation based on a neuron, and realizing self-adaptive observation of the angle value of the magnetoelectric encoder;

(5) adjusting neuron iteration speed according to a difference value of an angle value obtained through Kalman filtering and a kinematic state observation equation, and improving convergence speed of neuron angle value observation errors;

(6) and judging whether to stop the neuron iterative computation according to the angle value observation error range, and adopting the observation coefficient of the current computation period as an ideal coefficient value.

2. The method for actively suppressing the noise of the angle value of the magnetoelectric encoder according to claim 1 is characterized in that: and (1) obtaining a single-antipodal angle value signal A + and A-through a single-antipodal Hall sensor.

3. The method for actively suppressing the noise of the angle value of the magnetoelectric encoder according to claim 1 is characterized in that: and (2) performing analog-to-digital conversion on the single-antipodal angle value signals A + and A-through an analog-to-digital converter to obtain single-antipodal angle value digital signals HA + and HA-.

4. The method for actively suppressing the noise of the angle value of the magnetoelectric encoder according to claim 1 is characterized in that: and (3) resolving the single-antipodal angle value digital signals HA + and HA-diagonal values obtained in the step (2) to obtain a single-antipodal angle value theta_mThe calculation formula is shown as formula (1):

5. the method for actively suppressing the noise of the angle value of the magnetoelectric encoder according to claim 1 is characterized in that: the step (4) is implemented by adopting the following method:

in a calculation period T_sInner, magnetoelectric encoder rotorThe shaft acceleration α can be considered constant, and the differential equation for the single-pole angle value and the angular velocity is shown in equation (2):

in the formula: theta_mIs a single-antipodal angle value, omega_mSetting motor output torque T for rotor mechanical angular velocity_e(N m), moment of inertia J of motor and load_mLoad torque T_L(N · m), rotor acceleration α is shown as equation (3):

within one calculation period T_sMotor load torque T_LSmall variation, acceleration variation d- Δ α -0, and angular displacement θ_mAngular velocity omega_mAnd the angular acceleration change amount d can be written as equation (4):

in the formula

according to equation (5), the kinematic state observation equation can be obtained as shown in equation (6):

in the formula: l is_iIs the adjustment coefficient of the state observation equation,

from equations (5) and (6), a state observation error function expression can be obtained, as shown in equation (7):

in the formula: the state observation error is

in the formula ξ₁，ξ₂Is the damping coefficient of the state observer; omega_nIs the state observer response frequency;

formula (9) is obtainable from formula (8):

changing the error adjustment coefficient l₁、l₂、l₃And adjusting the convergence speed of the observation error of the kinematic state, wherein based on the formula (4), the formula (5) and the formula (6), the kinematic single-pair polar angle value state observer can write the formula (10):

the euler discretization state equation can be obtained from the formula (10), and is shown in the formula (11):

from equation (11), the observation of the single-pole angle in the k-th calculation cycleAnd observing angular velocityThe observation value of the single-antipodal angle of the k +1 th calculation cycle can be calculatedAnd observing angular velocityCarrying out prediction;

the observation precision of the state observer is improved by adopting a neuron adaptive kinematics state equation angle value observation method, and a single-pair-pole angle value observation correction target function is set

setting an approximation function of a single-antipodal angle value of a magnetoelectric encoder as

in the formula: b is_i(γ) represents a reference function; omega_iRepresenting error correction weight coefficients;

function V for measuring error of angle value_(x,γ,ω)As shown in formula (14):

single antipodal angle value theta_mAs a reference function B_iThe input value of (γ) is as shown in equation (15):

B_i(γ)＝B_i(θ_m) (15)

in order to improve the compensation calculation efficiency of the observation precision of the angle value, a neuron compensation linear correction function is established, and the optimal weight coefficient omega is obtained through neuron iterative calculation₁、ω₂、ω₃Adjustment coefficient of error for state observation₁、l₂、l₃Correcting to obtain a corrected observation error adjustment coefficient l_1f、l_2f、l_3f；

Constructing an error measurement function V (X) as shown in formula (16):

in the iterative calculation process, the correction weight coefficient omega is adaptively adjusted₁、ω₂、ω₃Making the observation value of the single-antipodal angleAngle theta towards single antipole of magnetoelectric encoder_mContinuously approximating, in order to minimize the iterative computation steps of the neuron and ensure the fastest decrease of the error measurement function, firstly finding the error measurement function V (X) along the weight correction coefficient ξ₁，ξ₂，ω_nAs shown in formulas (17) to (19):

iterative formula of neuron adaptive correction weight coefficient is as shown in formula (20) to formula (22):

obtaining an error measure function edge ξ according to the observation equation of state₁，ξ₂，ω_nPartial differentiation of direction and update of the next calculation cycle ξ_1(k+1)，ξ_2(k+1)，ω_n(k+1)Further obtain the updated error adjustment system_1f、l_2f、l_3fAnd calculating to obtain an observation angle value of the magnetoelectric encoder in the current calculation period, judging whether the observation angle value is within an ideal error range according to an error measurement function value, and stopping iterative updating calculation if the observation angle value meets the condition.

6. The method for actively suppressing the noise of the angle value of the magnetoelectric encoder according to claim 1 is characterized in that: the step (5) is realized by the following method:

establishing a magnetoelectric encoder angle value feedback prediction model as shown in formula (23):

in the formula: f is a state transition matrix, and F is a state transition matrix,T_sis a calculation cycle;is a predictive model state variable;the angular position is predicted for the kalman filter,

establishing a single-antipole angle value feedback updating model of the magnetoelectric encoder by an equation (23), as shown in an equation (24):

in the formula:to update model state variables;in order for the kalman filter to observe the angular position,

in order to improve the error convergence speed of the angle value, the observation iterative adjustment speed of the neuron needs to be improved, and if the deviation between the feedback angle value and the observation angle value of the motion state equation is directly used as an error monitoring scalar, the error monitoring scalar can be obtainedHigh-frequency signal noise mixed in the feedback angle value is introduced, the angle value error monitoring is not facilitated, and the single-antipodal angle observed value

further, the weight correction coefficient ξ₁、ξ₂、ω_nIs expressed by the equations (26), (27), (28):

obtaining a single antipodal angle observation value according to the methodSingle-pole angle value theta_mAnd synchronously outputting the error measurement function, determining a calculation termination position by judging the range of the error measurement function V (X), and actively improving the error noise suppression capability of the angle value of the magnetoelectric encoder by taking the error adjustment coefficient of the current calculation period as an ideal coefficient value.

7. The device for actively suppressing the noise of the angle value of the magnetoelectric encoder is characterized in that: the device for actively suppressing the noise of the angle value of the magnetoelectric encoder comprises:

the single-antipode Hall sensor is used for acquiring a magnetic field signal generated by the single-antipode magnetic steel and converting the magnetic field signal into a voltage signal to obtain single-antipode angle value signals A & lt + & gt and A & lt- & gt;

the analog-to-digital converter is used for converting the single antipodal angle value signals A & lt + & gt and A & lt- & gt into single antipodal angle value digital signals HA & lt + & gt and HA & lt- & gt;

a single-dipole angle calculation module for converting the obtained digital value into a single-dipole angle value theta₁；

The kinematic equation neuron self-adjusting module is used for realizing self-adaptive observation of the angle value of the magnetoelectric encoder;

and the active updating module is based on a Kalman filtering iteration coefficient and is used for improving the convergence speed of the observation error of the neuron angle value and improving the noise suppression effect of the magnetoelectric encoder angle value.

The technical field is as follows:

the invention relates to an active suppression method and device for angle value noise of a magnetoelectric encoder, belonging to the technical field of magnetoelectric encoder manufacturing.

Background art:

the encoder is used for measuring angles, is a core element for realizing motor control, is widely applied to the high-tech fields of mechanical engineering, robots, aviation, precise optical instruments and the like, and plays a vital role in modern industry. With the accelerated development of the industrialization process, the requirements on indexes such as resolution, precision and the like of the angular displacement sensor are higher. The angular displacement sensor widely adopted by the high-precision servo platform in the field of industrial control at present comprises a rotary transformer, a photoelectric encoder and a magnetoelectric encoder.

The optical devices of the photoelectric encoder are sensitive to environmental conditions such as pollution, vibration and temperature. The resolver structure includes an excitation winding and an output winding, which causes the volume and weight of the resolver to be key factors that restrict miniaturization thereof. In addition, the angle value of the rotary transformer has limited measuring range, relatively complex manufacturing process, high cost and poor real-time performance. Compared with the prior art, the magnetoelectric encoder has the advantages of simple structure, high temperature resistance, oil stain resistance, impact resistance, small volume, low cost and the like, and has unique advantages in the application places of miniaturization and severe environmental conditions.

For example, FIG. 1 is an exemplary Hall element distribution diagram for a multi-pair pole magnetoelectric encoder. In the figure, 2 Hall elements A +, A-are present, of which the single-pole Hall elements A +, A-are distributed at 90 DEG intervals around the circumference of the stator. Under the action of the single-antipode magnetic steel, voltage signals with the phase difference of 90 degrees are generated on the 2 Hall elements, and then the current single-antipode angle value is obtained through the angle value resolving processing.

However, in the technical field of application of magnetoelectric encoders, the angle value of the magnetoelectric encoder calculated due to vibration, temperature change, magnetic field change and other reasons can still bounce back and forth within a certain range, so that high-frequency noise is mixed in an output angle value signal, and in order to suppress the high-frequency noise mixed in the angle value signal, the invention provides an active suppression method and device for the angle value noise of the magnetoelectric encoder by taking the thought.

The invention content is as follows:

aiming at the problems, the invention provides a scheme, aiming at adopting a method capable of actively inhibiting high-frequency noise mixed in an angle value signal of a magnetoelectric encoder so as to improve the output precision of the angle value of the magnetoelectric encoder.

The invention discloses an active suppression method for angle value noise of a magnetoelectric encoder, which comprises the following steps:

(1) collecting single-antipodal angle value signals A & lt- & gt and A & lt- & gt;

(2) performing analog-to-digital conversion on the single-pair polar angle value signals A & lt + & gt and A & lt- & gt to obtain single-pair polar angle value digital signals HA & lt + & gt and HA & lt- & gt;

(3) solving the single-epipolar angle value theta according to the single-epipolar angle value digital signals HA + and HA-)₁；

(4) Establishing a kinematic state observation equation based on a neuron, and realizing self-adaptive observation of the angle value of the magnetoelectric encoder;

(5) adjusting neuron iteration speed according to a difference value of an angle value obtained through Kalman filtering and a kinematic state observation equation, and improving convergence speed of neuron angle value observation errors;

(6) and judging whether to stop the neuron iterative computation according to the angle value observation error range, and adopting the observation coefficient of the current computation period as an ideal coefficient value.

Preferably, in the step (1), the single-dipole angle value signal a +, a-is obtained by a single-dipole hall sensor.

Preferably, in the step (2), the single-antipodal angle value signals a +, a-are subjected to analog-to-digital conversion through an analog-to-digital converter to obtain single-antipodal angle value digital signals HA +, HA-.

Preferably, the step (3) is performed by resolving the single pair polar angle value digital signal HA +, HA-diagonal value obtained in the step (2) to obtain a single pair polar angle value theta_mThe calculation formula is shown as formula (1):

preferably, the step (4) is performed by the following method:

in a calculation period T_sIn the above, the acceleration α of the rotating shaft of the magnetoelectric encoder can be regarded as constant, and then the differential equation of the single-pole angle value and the angular velocity is shown in equation (2):

in the formula: theta_mIs a single-antipodal angle value, omega_mSetting motor output torque T for rotor mechanical angular velocity_e(N m), moment of inertia J of motor and load_mLoad torque T_L(N · m), rotor acceleration α is shown as equation (3):

t within one calculation period_sMotor load torque T_LSmall variation, acceleration variation d- Δ α -0, and angular displacement θ_mAngular velocity omega_mAnd the angular acceleration change amount d can be written as equation (4):

in the formulaC＝[100]From this, formula (5) can be obtained:

according to equation (5), the kinematic state observation equation can be obtained as shown in equation (6):

in the formula: l is_iIs the adjustment coefficient of the state observation equation,

from equations (5) and (6), a state observation error function expression can be obtained, as shown in equation (7):

in the formula: the state observation error isThe characteristic equation of the state observation error function is shown in formula (8):

in the formula ξ₁，ξ₂Is the damping coefficient of the state observer; omega_nIs the state observer response frequency;

formula (9) is obtainable from formula (8):

changing the error adjustment coefficient l₁、l₂、l₃And adjusting the convergence speed of the observation error of the kinematic state. The kinematic single-pole angle value state observer based on equations (4), (5) and (6) can be written as equation (10):

the euler discretization state equation can be obtained from the formula (10), and is shown in the formula (11):

from equation (11), the observation of the single-pole angle in the k-th calculation cycle

And observing angular velocityThe observation value of the single-antipodal angle of the k +1 th calculation cycle can be calculatedAnd observing angular velocityCarrying out prediction;

preferably, the step (5) is realized by the following method:

establishing a magnetoelectric encoder angle value feedback prediction model as shown in formula (23):

in the formula: f is a state transition matrix, and F is a state transition matrix,T_sis a calculation cycle;is a predictive model state variable;

the angular position is predicted for the kalman filter,predicting angular velocity for a kalman filter; b is a control matrix;

a covariance matrix that is a state variable of the prediction model; q is a noise matrix introduced by the prediction model;

establishing a single-antipole angle value feedback updating model of the magnetoelectric encoder by an equation (23), as shown in an equation (24):

in the formula:

to update model state variables;

in order for the kalman filter to observe the angular position,

observing angular velocity for a Kalman filter; k_kIs a Kalman filter coefficient; p_kA covariance matrix that is an observation model state variable; r is an observation noise covariance; h is a state variable extraction matrix; z_kThe actual observed value of the system state is obtained;

in order to improve the convergence speed of the angle value error, the iterative adjustment speed of neuron observation needs to be improved, if the deviation between the feedback angle value and the observation angle value of the motion state equation is directly used as an error monitoring scalar, high-frequency signal noise mixed in the feedback angle value can be introduced, the error monitoring of the angle value is not facilitated, and the single-antipodal angle observation value is used for monitoring the error of the angle valueAnd Kalman filter observation angle positionAnd (3) outputting the difference value, and taking the difference value as a monitoring basis of the change trend of the angle observation error to obtain an adaptive adjustment gain coefficient lambda, which is expressed by the formula (25):

further, the weight correction coefficient ξ₁、ξ₂、ω_nIs represented by equation (26):

obtaining a single antipodal angle observation value according to the methodSingle-pole angle value theta_mAnd synchronously outputting the error measurement function, determining a calculation termination position by judging the range of the error measurement function V (X), and actively improving the error noise suppression capability of the angle value of the magnetoelectric encoder by taking the error adjustment coefficient of the current calculation period as an ideal coefficient value.

The invention also discloses a device for actively suppressing the noise of the angle value of the magnetoelectric encoder, which comprises the following components:

the single-antipode Hall sensor is used for acquiring a magnetic field signal generated by the single-antipode magnetic steel and converting the magnetic field signal into a voltage signal to obtain single-antipode angle value signals A & lt + & gt and A & lt- & gt;

the analog-to-digital converter is used for converting the single antipodal angle value signals A & lt + & gt and A & lt- & gt into single antipodal angle value digital signals HA & lt + & gt and HA & lt- & gt;

a single-dipole angle calculation module for converting the obtained digital value into a single-dipole angle value theta₁；

The kinematic equation neuron self-adjusting module is used for realizing self-adaptive observation of the angle value of the magnetoelectric encoder;

and the active updating module is based on a Kalman filtering iteration coefficient and is used for improving the convergence speed of the observation error of the neuron angle value and improving the noise suppression effect of the magnetoelectric encoder angle value.

The invention has the beneficial effects that:

1. based on a kinematic state equation, the self-adaptive observation of the angle value of the magnetoelectric encoder is realized, and high-frequency noise signals mixed in the angle value of the magnetoelectric encoder are reduced.

2. The neuron iteration coefficient is adaptively adjusted based on the difference between the angle value after Kalman filter filtering and the angle value before the observation of the kinematic state, so that the observation speed of the neuron is increased, and the noise suppression effect of the angle value is improved.

3. A jumping phenomenon of an angle value exists when the angle value of the magnetoelectric encoder crosses a zero point, a neuron observation coefficient with strong tracking characteristic is used under the condition, the magnetoelectric encoder is positioned at a non-zero-crossing working position, and the neuron observation coefficient with better filtering effect is adopted, so that high-frequency noise filtering mixed in the angle value is realized.

Drawings

For ease of illustration, the invention is described in detail by the following detailed description and the accompanying drawings.

FIG. 1 is an exemplary block diagram of an encoder according to the present invention;

FIG. 2 illustrates the operation of the encoder of the present invention;

FIG. 3 is a comparison waveform of position feedback before and after position tracking using a kinematic angular position state observer;

FIG. 4-a is a waveform diagram illustrating the difference between the single antipodal angle value and the single antipodal angle observation value;

FIG. 4-b shows the output of forced elimination of differential value trip points for single antipodal angle values;

FIG. 5 is a block diagram of neuron adaptive error adjustment coefficient correction;

FIG. 6 is a flow chart of neuron adaptive error adjustment coefficient correction;

FIG. 7 is a comparison graph of angle value observations using a neuron adaptation method and without the method;

FIG. 8 illustrates damping coefficients ξ for a state observer₁Adaptive tuning of the Process map, State observer damping coefficient ξ₁A self-adaptive adjustment process;

FIG. 9 shows damping coefficients ξ for a state observer₂Is an adaptive adjustment process diagram;

FIG. 10-a is the output waveform of the angular error measurement function V (X)' without neuron adaptation;

FIG. 10-b is the output waveform of the neuron adaptive angle error measurement function V (X)'s;

FIG. 11 is a graph of the results after Kalman filtering;

FIG. 12 is a graph of the Kalman filtering back angle value and angle feedback value difference value calculation results;

FIG. 13-a is a diagram of a neuron adaptive method and Kalman filtering angle value deviation calculation;

FIG. 13-b is an enlarged view of a neuron adaptive method and a Kalman filtering angle value deviation calculation;

FIG. 14 is a graph of the monitoring coefficient λ adjustment variation process;

FIG. 15 is a graph illustrating equation of state damping coefficients ξ of motion through a Kalman filtering supervised tuning process₁The adaptive adjustment process diagram of (1);

FIG. 16 is a state of motion equation damping coefficient ξ for a Kalman filtering supervised tuning process₂The adaptive adjustment process diagram of (1);

FIG. 17-a is a graph of the output result after the noise active suppression method is adopted;

FIG. 17-b is a partial enlarged view of an output result after the noise active suppression method is adopted;

FIG. 17-c is a graph of the angle value tracking process using the present method;

FIG. 18-a is a graph of the output noise after the method has been applied;

FIG. 18-b is a partial magnified view of the output noise after the method is employed;

detailed description of the preferred embodiments

The following detailed description of embodiments of the invention refers to the accompanying drawings.

The embodiments/examples described herein are specific embodiments of the present invention, are intended to be illustrative of the concepts of the present invention, are intended to be illustrative and exemplary, and should not be construed as limiting the embodiments and scope of the invention. In addition to the embodiments described herein, those skilled in the art will be able to employ other technical solutions which are obvious based on the disclosure of the claims and the specification of the present application, and these technical solutions include those which make any obvious replacement or modification of the embodiments described herein, and all of which are within the scope of the present invention.

In order that the objects, aspects and advantages of the invention will become more apparent, the invention will be described by way of example only, and in connection with the accompanying drawings. It is to be understood that such description is merely illustrative and not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.

As shown in fig. 1, fig. 2, fig. 3, fig. 4, fig. 5, fig. 6, fig. 7, fig. 8, fig. 9, fig. 10, fig. 11, fig. 12, fig. 13, fig. 14, fig. 15, fig. 16, fig. 17, and fig. 18, the present embodiment adopts the following technical solutions:

fig. 2 is a schematic structural diagram of an angle value fine-dividing process according to an embodiment of the present invention, which includes:

the single-dipole Hall sensor 101 is used for collecting a magnetic field signal generated by the single-dipole magnetic steel and converting the magnetic field signal into a voltage signal to obtain a single-dipole angle value signal A + and A-.

An analog-to-digital converter 102 for converting the single-antipodal voltage signal a +, a-into a single-antipodal angle value digital signal HA +, HA-.

A single-dipole angle calculation module 103 for converting the obtained digital quantity into a single-dipole angle value theta₁The phase angle deviation of the digital signal HA + and HA-of the single-pair polar angle value is 90 degrees, and the single-pair polar angle value theta is solved through an arc tangent formula (1)_m：

The kinematics equation neuron self-tuning module 104 performs a computation cycle T_sIn the above, the acceleration α of the rotating shaft of the magnetoelectric encoder can be regarded as constant, and then the differential equation of the single-pole angle value and the angular velocity is shown in equation (2):

in the formula: theta_mIs a single-antipodal angle value, omega_mSetting motor output torque T for rotor mechanical angular velocity_e(N·m)，Moment of inertia J of motor and load_mLoad torque T_L(N · m), rotor acceleration α is shown as equation (3):

within one calculation period T_sMotor load torque T_LSmall variation, acceleration variation d- Δ α -0, and angular displacement θ_mAngular velocity omega_mAnd the angular acceleration change amount d can be written as equation (4):

in the formulaC＝[100]From this, formula (5) can be obtained:

according to equation (5), the kinematic state observation equation can be obtained as shown in equation (6):

in the formula: l is_iIs the adjustment coefficient of the state observation equation,

from equations (5) and (6), a state observation error function expression can be obtained, as shown in equation (7):

in the formula: the state observation error isThe characteristic equation of the state observation error function is shown in formula (8):

in the formula ξ₁，ξ₂Is the damping coefficient of the state observer; omega_nIs the state observer response frequency;

formula (9) is obtainable from formula (8):

changing the error adjustment coefficient l₁、l₂、l₃And adjusting the convergence speed of the observation error of the kinematic state, wherein based on the formula (4), the formula (5) and the formula (6), the kinematic single-pair polar angle value state observer can write the formula (10):

the euler discretization state equation can be obtained from the formula (10), and is shown in the formula (11):

from equation (11), the observation of the single-pole angle in the k-th calculation cycle

And observing angular velocityThe observation value of the single-antipodal angle of the k +1 th calculation cycle can be calculated

And observing angular velocityCarrying out prediction;

FIG. 3 is a graph of position feedback versus waveforms before and after position tracking using a kinematic angular position state observer, at ξ₁＝0.5；ξ₂＝0.5；ω_nIn the case of 1000, the angle value output of the angular position state observer is obtained, as shown in fig. 3, it can be seen that the angular position state observer more accurately tracks the current angle value;

fig. 4-a shows a waveform diagram obtained by respectively performing differential calculation processing on a single-dipole angle value and a single-dipole angle observed value, and it can be seen that a zero-crossing point (0LSB transits to 65535LSB) exists in the original angle value, a jump point exists in the single-dipole angle value after differential processing, and a jump phenomenon does not exist in differential output of the state observed angle value because the zero-crossing point of the angle value cannot be accurately tracked by the current error adjustment coefficient of the single-dipole angle observed value. In order to facilitate the comparison of the differential value output, as shown in fig. 4- (b), the output is obtained after the jump point of the differential value of the single antipodal angle value is forcibly eliminated, and it can be seen that the noise of the observation value of the single antipodal angle is effectively suppressed, but the observation precision is low;

the observation precision of the state observer is improved by adopting a neuron adaptive kinematics state equation angle value observation method, and a single-pair-pole angle value observation correction target function is setAs shown in equation (12):

setting an approximation function of a single-antipodal angle value of a magnetoelectric encoder as

As shown in formula (13):

in the formula: b is_i(γ) represents a reference function; omega_iRepresenting error correction weight coefficients;

function V for measuring error of angle value_(x,γ,ω)As shown in formula (14):

single antipodal angle value theta_mAs a reference function B_iThe input value of (γ) is as shown in equation (15):

B_i(γ)＝B_i(θ_m) (15)

FIG. 5 is a block diagram of correction of adaptive error adjustment coefficients of neurons, in which to improve the efficiency of compensation calculation for observation accuracy of angle values, linear correction functions of neuron compensation are established, and optimal weight coefficients ω are obtained by iterative calculation of neurons₁、ω₂、ω₃Adjustment coefficient of error for state observation₁、l₂、l₃Correcting to obtain a corrected observation error adjustment coefficient l_1f、l_2f、l_3fAs shown in fig. 5;

constructing an error measurement function V (X) as shown in formula (16):

in the iterative calculation process, adaptively adjusting and correcting the weight coefficient omega₁、ω₂、ω₃Making the observation value of the single-antipodal angle

Angle theta towards single antipole of magnetoelectric encoder_mContinuously approximating, in order to minimize the iterative computation steps of the neuron and ensure the fastest decrease of the error measurement function, firstly finding the error measurement function V (X) along the weight correction coefficient ξ₁，ξ₂，ω_nAs shown in formulas (17) to (19):

iterative formula of neuron adaptive correction weight coefficient is as shown in formula (20) to formula (22):

FIG. 6 is a flow chart of neuron adaptive error adjustment coefficient correction, which is performed according to the observation equation of state to obtain an error measure function edge ξ₁、ξ₂、ω_nPartial differentiation of direction and update of the next calculation cycle ξ_1(k+1)、ξ_2(k+1)、ω_n(k+1)And then obtaining the updated error adjustment coefficient l_1f、l_2f、l_3fCalculating to obtain an observation angle value of the magnetoelectric encoder in the current calculation period, judging whether the observation angle value is within an ideal error range according to an error measurement function value, and stopping iterative updating calculation if the observation angle value meets the conditions, as shown in fig. 6;

FIG. 7 is a comparison graph of angle observation values using a neuron adaptive method and without using the method, and it can be seen that the angle observation values using an adaptive neuron adaptive algorithm are gradually stable and tend to a true feedback value, as shown in FIG. 7;

FIG. 8 illustrates damping coefficients ξ for a state observer₁Adaptive tuning of the Process map, State observer damping coefficient ξ₁Adaptive adjustment processes, e.g.FIG. 8 is a schematic view;

FIG. 9 shows damping coefficients ξ for a state observer₂For adaptive tuning of the process map, the state observer damping coefficient ξ is shown₂The adaptive adjustment process, as shown in fig. 9;

FIG. 10-a shows the output waveform of the angular error measurement function V (X)' without neuron adaptation; FIG. 10- (b) shows the output waveform of the angle error measurement function V (X)' after adaptive adjustment by neuronsWherein sgn (f) represents the positive and negative signs of the function f;

as can be seen from fig. 10-a and 10-b, the angle value observation error is gradually converged by using the neuron adaptive method, but the observation accuracy still has a large error with the single-antipodal angle value, and the error convergence rate is low, and no good observation accuracy is obtained after 220000 calculation cycles of iterative calculation.

Based on the kalman filtering iteration coefficient active updating module 105, establishing a magnetoelectric encoder angle value feedback prediction model, as shown in formula (23):

in the formula: f is a state transition matrix, and F is a state transition matrix,T_sis a calculation cycle;

is a predictive model state variable;the angular position is predicted for the kalman filter,predicting angular velocity for a kalman filter; b is a control matrix;

a covariance matrix that is a state variable of the prediction model; q is a noise matrix introduced by the prediction model;

establishing a single-antipole angle value feedback updating model of the magnetoelectric encoder by an equation (23), as shown in an equation (24):

in the formula:to update model state variables;in order for the kalman filter to observe the angular position,observing angular velocity for a Kalman filter; k_kIs a Kalman filter coefficient; p_kA covariance matrix that is an observation model state variable; r is an observation noise covariance; h is a state variable extraction matrix; z_kThe actual observed value of the system state is obtained;

FIG. 11 is a graph of the results of Kalman filtering, with updated model state variables as

The covariance matrix of the state variables of the prediction model is

The noise matrix introduced by the prediction model is Q ═ 0.0001, 0; 0,0.0001]The angular displacement state variable extraction matrix is H₁＝[1,0]And the observation noise covariance matrix is R1. Then, the single antipodal angle value is used as the actual observed value of the system state, and the single antipodal angle value theta_mVariation range ∈ [0,65535 ]]LSB, derived Kalman Filter observed angular positionAs shown in fig. 11;

fig. 12 is a view showing a calculation result of a difference value of the angle value and a difference value of the angle feedback value after kalman filtering, and it can be seen that the output noise of the angle value after kalman filtering is effectively suppressed by performing the difference calculation on the single-pole angle value and the observation angle position of the kalman filter, as shown in fig. 12;

FIG. 13-a is a diagram of calculating the deviation between the neuron adaptive method and the Kalman filtering angle value, FIG. 13-b is a partial enlarged view of calculating the deviation between the neuron adaptive method and the Kalman filtering angle value, in order to increase the error convergence rate of the angle value, the iterative adjustment rate of neuron observation needs to be increased, if the deviation between the feedback angle value and the observation angle value of the motion state equation is directly used as an error monitoring scalar, high-frequency signal noise mixed in the feedback angle value is introduced, which is not beneficial to the error monitoring of the angle value, and the single-antipodal angle observation value is used

And Kalman filter observation angle positionThe difference value output is used as a monitoring basis for the change trend of the angle observation error as shown in the figure 13-a and the figure 13-b, and active adjustment is carried out according to the error;

obtaining an adaptive adjustment gain factor λ, as represented by equation (25):

FIG. 14 is a diagram of the process of adjusting the change of the monitoring coefficient λ to obtain the adaptive adjustment gain coefficient λ, as shown in FIG. 14;

further, the weight correction coefficient ξ₁、ξ₂、ω_nIs represented by equation (26):

FIG. 15 is a graph illustrating equation of state damping coefficients ξ of motion through a Kalman filtering supervised tuning process₁The damping coefficient ξ of the equation of motion state through the Kalman filtering supervision and adjustment process₁As shown in fig. 15;

FIG. 16 is a state of motion equation damping coefficient ξ for a Kalman filtering supervised tuning process₂The damping coefficient ξ of the equation of motion state through the Kalman filtering supervision and adjustment process₂The adaptive adjustment process of (2), as shown in fig. 16;

FIG. 17-a is a graph of an output result after a noise active suppression method is adopted, and FIG. 17-b is a partial enlarged view of an output result after the noise active suppression method is adopted, according to which a single antipodal angle observation value is obtainedSingle-pole angle value theta_mAnd synchronously outputting the error measurement function, wherein the result is shown in fig. 17-a and fig. 17-b, and it can be seen that the angle error is converged to the best value in 176000 iterative calculation cycles, the calculation termination position is determined by judging the range of the error measurement function, and the error adjustment coefficient of the current calculation cycle is taken as an ideal coefficient value, so that the active improvement of the error noise suppression capability of the angle value of the magnetoelectric encoder is realized.

It can be seen from FIG. 17-a that the best error adjustment coefficient, equation of state adjustment coefficient ξ, was obtained at 176000 th iteration calculation cycle₁＝65.9959，ξ₂＝0.8375，ω_n1000, at this timeThe state observer has the best tracking performance, when the single-pole angle value crosses zero, the error adjusting coefficient with the best tracking characteristic is adopted to obtain better tracking capability, and the state observation coefficient ξ with better filtering effect is adopted at the position where the single-pole angle value continuously changes₁＝11.2，ξ₂＝0.95，ω_n1000 to reduce the effect of angle value noise. As shown in fig. 17-b, at a partial magnification. Fig. 17-c shows the tracking process of the angle value by using the method, which can be seen that the method not only ensures the fast tracking capability of the zero crossing point, but also reduces the feedback noise of the angle value and improves the noise suppression capability of the single-pole angle value.

Fig. 18-a is a graph of the angle value output noise after the method is adopted, fig. 18-b is a partial enlarged graph of the output noise after the method is adopted, a good zero crossing point tracking characteristic is obtained by adopting an angle value noise active suppression method of a magnetoelectric encoder, in addition, in a stage of continuous smooth change of a single-pole angle value, the angle value noise is effectively suppressed, and the angle value noise is improved to +/-5 LSB from the initial +/-25 LSB, as shown in fig. 18-b.

The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.