Method for improving anti-interference capability of Hall current sensor

Abstract

The invention provides a method for improving the anti-interference capability of a Hall current sensor, and the method employs three-stage filtering, and comprises the steps: S1, Kalman filtering: processing linear Gaussian noise; s2, second-order low-pass filtering: the cut-off frequency is 20Hz, and high-frequency noise is inhibited in a targeted manner; and S3, moving average filtering: smoothing transient fluctuation. According to the invention, a plurality of filtering modes are fused together, so that the anti-interference capability of the product in a complex electromagnetic environment is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of current sensor, in particular to a method for improving the anti-interference ability of Hall current sensor. BACKGROUND

[0002] The core technology of Hall current sensor is the Hall effect: when an electric current passes through a conductor or semiconductor placed in a magnetic field, a voltage (Hall voltage) proportional to the size of the original current is generated in the direction perpendicular to the current and the magnetic field. By measuring the Hall voltage, the size of the original current can be known. With the advancement of semiconductor technology, the consistency of Hall elements has been improved, and Hall current sensors are widely used in current detection in industrial automation and automotive fields.

[0003] Modern Hall current sensors mainly adopt two technical architectures: open-loop and closed-loop designs. The open-loop structure is relatively simple, consisting of a Hall element, a magnetic core, and a signal processing circuit. When the measured current passes through the magnetic core, a magnetic field is generated, which is sensed by the Hall element and generates a corresponding Hall voltage. The signal processing circuit amplifies and converts the standard output signal. This design has the advantages of simple structure, low cost, and fast response speed, but also has the limitations of relatively low measurement accuracy and susceptibility to external magnetic field interference.

[0004] In contrast, the closed-loop Hall current sensor adds a feedback coil to the open-loop structure, forming a more precise design. When the measured current passes through the magnetic core to generate a magnetic field, the voltage signal generated by the Hall element is amplified to drive the feedback coil, generating a magnetic field opposite to the direction of the measured current. Through the negative feedback mechanism, the original magnetic field is offset, so that the Hall element is always in a state of near-zero magnetic field. At this time, the current in the feedback coil is proportional to the measured current, and the measured current value can be accurately determined by measuring the feedback current. The advantage of closed-loop design is high measurement accuracy and good stability. SUMMARY

[0005] Although the closed-loop Hall current sensor increases the feedback coil and uses the feedback coil to close the loop for signal processing, it has improved the anti-interference ability of the product to some extent, but in the industrial environment with complex electromagnetic environment, its anti-interference ability is still insufficient to output stable measurement data. To solve the above problems, the present application discloses a method for improving the anti-interference ability of Hall current sensor.

[0006] The specific scheme is as follows:

[0007] A method for improving the anti-interference ability of Hall current sensor adopts three-stage filtering, including:

[0008] S1, Kalman filtering: processing linear Gaussian noise;

[0009] S2, second-order low-pass filter: cut-off frequency 20 Hz, to suppress high-frequency noise;

[0010] S3, moving average filter: smooth transient fluctuations.

[0011] Further, in step S1, under laboratory conditions, the noise of the complex environment is simulated by using a signal generator and a power amplifier, and the direct current is measured with different parameters to confirm the optimal filtering parameters; Kalman filter is a Bayesian recursive estimator, which finds the optimal estimate in uncertainty through iteration of prior prediction (system model) and posterior correction (observation data), and can effectively process linear and Gaussian noise. Four parameters are designed, including estimated value X, estimated covariance P, process noise covariance Q and measurement noise covariance R; the lower the value of process noise covariance Q and the higher the value of measurement noise covariance R, the more obvious the suppression effect of noise and the better the filtering effect, but at the same time, the longer the filtering time, which is generally confirmed by laboratory test.

[0012] Further, step S1 includes the following steps:

[0013] S11, make a prediction: P=P+Q;

[0014] S12, update: calculate Kalman gain K=P / (P+R); data update: X=X+K·(measured value-X); estimated covariance update: P=(1-K)·P.

[0015] Further, in step S2, to improve the frequency response characteristics of the product and suppress nonlinear or non-Gaussian noise, a second-order low-pass filter designed by using the second-order bilinear transformation method is superimposed on the Kalman filter. Bilinear transformation is a method of converting continuous-time system (analog filter) into discrete-time system (digital filter), and its design process is as follows:

[0016] The analog transfer function of the second-order Butterworth low-pass filter is:

[0017] H(s)=ω02 / (s 2 +(ω0 / Q)s+ω02)

[0018] Where ω0=2πf0, f0 is the cut-off frequency; Q is the quality factor, Q=1 / √2≈0.7071, to achieve maximum flat response;

[0019] Calculate the pre-distortion factor to compensate for the frequency nonlinear distortion caused by bilinear transformation:

[0020] K=tan(π·f_c / f_s);

[0021] wherein f c is the cut-off frequency; f s is the sampling frequency;

[0022] The normalization factor is calculated as:

[0023] norm = 1 / (1+√2.K+K2);

[0024] The coefficients are calculated as: b0=K 2 norm; b1=2.b0; b2=b0; a1=2.(K 2 -1).norm; a2=(1-√2.K+K 2 )norm;

[0025] After the bilinear transformation, the difference equation of the digital filter is obtained as:

[0026] y[n]=b0x[n]+b1x[n-1]+b2x[n-2]-a1y[n-1]-a2y[n-2]

[0027] Further, in step S3, the N data points continuously collected are regarded as a window, the oldest data is removed and new data is inserted each time the window is shifted, and the arithmetic average of all data in the window is calculated, so as to eliminate random noise and short-term fluctuations in the signal by calculating the arithmetic average of the data points in the sliding window.

[0028] The application has the beneficial effect that the application combines multiple filtering methods to effectively improve the anti-interference ability of the product in a complex electromagnetic environment. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 is a filter flowchart;

[0030] Figure 2 is a data comparison chart before and after filtering of 5A source side DC current superimposed with 50Hz 8A AC interference;

[0031] Figure 3 is a data comparison chart before and after filtering of 5A source side DC current superimposed with 100Hz 8A AC interference;

[0032] Figure 4 is a data comparison chart before and after filtering of 5A source side DC current superimposed with 300Hz 8A AC interference. DETAILED DESCRIPTION

[0033] The application will be further illustrated below in conjunction with the drawings and specific embodiments, and it should be understood that the following specific embodiments are only used to illustrate the application and not to limit the scope of the application.

[0034] The traditional filtering scheme generally adopts Kalman filter, and estimates by using the mean and covariance matrix of the estimated quantity. When linear and quasi-linear transformation is performed, the mean and covariance estimation can be effectively maintained. For non-linear or non-Gaussian system, the estimation result is greatly deviated. In order to ensure that the filtering effect is optimal, as shown in the description, the application provides a method for improving the anti-interference ability of Hall current sensor, adopting three-stage filtering architecture, including the following steps: Figure 1

[0035] I. Kalman filtering

[0036] The filtering parameters (initial value X, estimation error covariance P, process noise variance Q and measurement noise variance R) of Kalman filtering are defined and initialized, wherein the process noise variance and the measurement noise variance are obtained by experimentally measuring multiple times, and the optimal values of the product are 0.025 and 1.0 respectively. The process of one Kalman filtering is as follows:

[0037] A) Prediction: P=P+Q;

[0038] B) Update: calculate Kalman gain K=P / (P+R);

[0039] Data update: X=X+K·(measured value-X);

[0040] Estimation covariance update: P=(1-K)·P.

[0041] II. Second-order low-pass filter:

[0042] The second-order low-pass filter is designed by using the bilinear transformation method, and the main design parameters are: cut-off frequency 20Hz, sampling frequency 100Hz, and the implementation process is as follows:

[0043] Calculate the pre-twist factor: K=tan(π·f_c / f_s)=0.726543;

[0044] Calculate the normalization factor: norm=1 / (1+√2·K+K2)=0.391359;

[0045] Calculate the coefficient: b0=K 2 ·norm=0.206601;

[0046] b1=2·b0=0.413203;

[0047] b2=b0=0.206601;

[0048] a1=2·(K 2 -1)·norm=-0.369549;

[0049] a2=(1-√2·K+K​2 )·norm = 0.195888.

[0050] The corresponding difference equation is:

[0051] y[n] = b0x[n] + b1x[n-1] + b2x[n-2] - a1y[n-1] - a2y[n-2];

[0052] III. Moving average filter

[0053] The moving average filter is to remove random noise and short-term fluctuations in the signal by calculating the arithmetic mean of data points in the sliding window. The core is to regard the 4 data points collected continuously as a queue, remove the oldest data, insert new data, and perform arithmetic mean on all data in the window every time the window slides;

[0054] After the measurement value is subjected to Kalman filtering, second-order low-pass filtering and mean smoothing filtering, it is verified through experiments that the anti-interference ability of the product is greatly improved. When the source side direct current is 0A, 10A and 30A, 20-200Hz / 2A, 5A and 8A AC current interference is superimposed, the measurement value of the product is good in consistency during the whole test process, and the maximum fluctuation amplitude is about ±450mA.

[0055] Figure 2 For measuring 5A direct current, a 50Hz frequency and 8A effective amplitude AC interference source is superimposed on the source side current end through the signal generator and power amplifier, and a comparison chart is formed according to the recorded measurement values before and after filtering.

[0056] Figure 3 For measuring 5A direct current, a 100Hz frequency and 8A effective amplitude AC interference source is superimposed on the source side current end through the signal generator and power amplifier, and a comparison chart is formed according to the recorded measurement values before and after filtering.

[0057] Figure 4 For measuring 5A direct current, a 300Hz frequency and 8A effective amplitude AC interference source is superimposed on the source side current end through the signal generator and power amplifier, and a comparison chart is formed according to the recorded measurement values before and after filtering.

[0058] The technical means disclosed in the scheme of the present application is not limited to the technical means disclosed in the above-mentioned embodiments, but also includes the technical solutions composed of any combination of the above technical features. It should be noted that for ordinary skilled persons in the art, without departing from the principle of the present application, a number of improvements and refinements can be made, and these improvements and refinements are also considered to be within the protection scope of the present application.

Claims

1. A method of improving the interference immunity of a Hall current sensor, characterized in that, Adopt three-stage filter, including: S1, Kalman filter: processing linear Gaussian noise; S2, second-order low-pass filter: cut-off frequency 20Hz, specific to inhibit high-frequency noise; S3, moving average filter: smooth transient fluctuations.

2. The method for improving the anti-interference ability of a Hall current sensor according to claim 1, characterized in that, In step S1, under laboratory conditions, the noise of complex environment is simulated by using signal generator plus power amplifier, and the direct current is measured with different parameters to confirm the optimal filter parameters; Kalman filter is designed with four parameters, including estimated value X, estimated covariance P, process noise covariance Q and measurement noise covariance R; the lower the value of process noise covariance Q and the higher the value of measurement noise covariance R, the more obvious the suppression effect of noise and the better the filtering effect.

3. The method of claim 2, wherein the Hall current sensor is a Hall effect sensor. Step S1 includes the following steps: S11, make a prediction: P=P+Q; S12, update: calculate Kalman gain K=P / (P+R); data update: X=X+K·(measured value-X); estimated covariance update: P=(1-K)·P.

4. The method of claim 1, wherein, In step S2, in order to improve the frequency response characteristics of the product and suppress nonlinear or non-Gaussian noise, a second-order low-pass filter designed by using second-order bilinear transformation method is superimposed on the Kalman filter, and bilinear transformation is a method of converting continuous-time system into discrete-time system, and the design process is as follows: The analog transfer function of the second-order Butterworth low-pass filter is: H(s) = ωo2 / (s 2 + (ωo / Q)s + ωo2) Where ω0=2πf0, f0 is the cut-off frequency; Q is the quality factor, Q=1 / √2≈0.7071, to achieve the maximum flat response; Calculate the pre-distortion factor to compensate for the frequency nonlinear distortion caused by bilinear transformation: K=tan(π·f_c / f_s); Where f_c is the cut-off frequency; f_s is the sampling frequency; Calculate the normalization factor: norm=1 / (1+√2·K+K2); Compute coefficients: b0 = K 2 • norm; b1 = 2 · b0; b2 = b0; a1 = 2 · (K 2 - 1) · norm; a2 = (1 - V2 · K + K 2 ) · norm; After bilinear transformation, the difference equation of the digital filter is obtained: y[n]=b0x[n]+b1x[n-1]+b2x[n-2]-a1y[n-1]-a2y[n-2] 5. The method of claim 1, wherein, In step S3, the N data points collected continuously are regarded as a window, the oldest data is removed and new data is inserted every time the window slides, and arithmetic average is performed on all data in the window, and the random noise and short-term fluctuations in the signal are removed by calculating the arithmetic average of the data points in the sliding window.

Images