CN-121980132-A - Magnetic field estimation method based on self-adaptive algorithm

CN121980132ACN 121980132 ACN121980132 ACN 121980132ACN-121980132-A

Abstract

The invention discloses a magnetic field estimation method based on an adaptive algorithm, which comprises the following steps of 1, establishing a model, establishing a spin evolution model, a process equation and a detection model in an atomic magnetometer, 2, establishing an extended Kalman filter, and establishing the adaptive extended Kalman filter on the basis of the extended Kalman filter to estimate a magnetic field. The conventional spin noise spectrum cannot solve the problem of time-varying magnetic field tracking, and the actual performance of the extended kalman filter is highly dependent on accurate system modeling and is sensitive to system noise variation. The invention obviously reduces the requirement of manual parameter tuning and improves the applicability and the operation robustness of the Kalman filtering in quantum metering application.

Inventors

KONG JIA
JIN XIAOFENG
WANG YIHAN
LU XIAOMING

Assignees

杭州电子科技大学

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (10)

1. The magnetic field estimation method based on the adaptive algorithm is characterized by comprising the following steps: S1, establishing a spin evolution model describing spin dynamics of an atomic ensemble in a magnetic field along a set direction; S2, parameterizing the intensity information of a magnetic field to be measured into Larmor frequency which is in direct proportion to the magnetic field intensity, and constructing the Larmor frequency and at least two transverse components of atomic spin together into an extended state vector; S3, based on Faraday rotation effect, establishing a function relation between the polarization rotation angle of the detection light and spin components in the state vector, and taking the function relation as an observation model to acquire a spin observation signal containing measurement noise through polarization change of the detection light; S4, based on a process equation, executing a time updating step of extended Kalman filtering to obtain a state vector and a priori estimation value of an error covariance matrix of the state vector; S5, dynamically estimating the measurement noise intensity at the current moment by analyzing the statistical characteristics of the prediction observation errors based on the observation model and the historical observation data; S6, combining the measured noise intensity of the dynamic estimation, executing a state updating step of the self-adaptive extended Kalman filtering, correcting the state vector of the prior estimation to obtain the state vector of the posterior estimation, and further calculating the Larmor frequency and the magnetic field estimation value from the state vector of the posterior estimation.
2. The method according to claim 1, wherein in step S1, the spin evolution model is constructed based on the bloch equation, wherein the evolution of spin vectors is linear dynamics and comprises relaxation effects characterized by transverse relaxation times, and noise introduced during the evolution is an increment subject to gaussian white noise statistics.
3. The method according to claim 1, wherein in the step S2, the process equation is a nonlinear equation whose state vector includes an x-component, a z-component, and larmor frequency of atomic spins, the process equation is iterated at discrete time intervals, and process noise is modeled as wiener increments.
4. The method according to claim 1, wherein in the step S3, the observation model is a linear model expressed as a linear function of the rotation angle of the polarization of the probe light and the spin x component in the state vector, the proportionality coefficient is a coupling constant of the probe light and the atomic spin, and measurement noise conforming to gaussian distribution is superimposed in the observation signal.
5. The method according to claim 1, wherein in the step S4, the time updating step of the extended Kalman filter comprises predicting a priori estimate of the current time state through a nonlinear state evolution function based on the posterior estimate of the previous time state, and predicting a priori estimate of the current time error covariance based on a jacobian matrix of the state transfer function, the previous time error covariance posterior estimate, and the process noise covariance.
6. The method according to claim 1, wherein the step S5 of dynamically estimating the measurement noise intensity comprises calculating a series of observation information at successive time instants, the observation information being a difference between an actual observation value and an observation value predicted based on an a priori state, and recursively estimating a covariance matrix of the measurement noise based on a sample covariance of the observation information, an observation model matrix, and a predetermined estimation window size.
7. The method according to claim 1, wherein the step S6 of updating the state of the adaptive extended Kalman filter includes calculating a Kalman gain matrix using a dynamically estimated measurement noise covariance matrix, an observation model matrix, and a priori estimate of an error covariance matrix, correcting the a priori estimate of the state vector using the Kalman gain matrix to obtain a posterior estimate, and updating the posterior estimate of the error covariance matrix.
8. The method of claim 5, wherein the state vector comprises an x-component, a z-component, and a larmor frequency of the atomic spins, and wherein the nonlinear state evolution function takes as input a state vector from a previous time instant and outputs a predicted value of the state vector at the current time instant.
9. The method of claim 5, wherein the jacobian of the state transfer function is derived from a state vector partial derivative of a nonlinear state evolution function, the matrix elements of which characterize the coupling and evolution sensitivity between state variables.
10. The method of claim 6, wherein the observation model matrix is a row vector whose only non-zero element is the coupling constant of the probe light to atomic spins, corresponding to the coefficients of the spin x component in the state vector.

Description

Magnetic field estimation method based on self-adaptive algorithm Technical Field The invention relates to the technical field of magnetic field measurement, and mainly applies adaptive extended Kalman filtering, in particular to a magnetic field estimation method based on an adaptive algorithm. Background The atomic magnetometer is widely used as a high-sensitivity magnetic field measuring device in the fields of medical diagnosis (such as a brain magnetic chart and a heart magnetic chart), biological magnetic measurement, basic physical research and the like. Atomic magnetometers not only possess comparable sensitivity to superconducting quantum interferometers (SQUIDs), but also do not require cryogenic cooling, and have achieved chip-scale miniaturization, thus exhibiting significant advantages in portable, low-power magnetic field sensing applications. In practical applications of atomic magnetometers, spin noise based magnetic field estimation methods are of great interest because of their non-invasiveness and high resolution. However, this approach faces core problems in several ways: the signal-to-noise ratio is extremely low, the signal extraction is difficult, namely the amplitude of the spin noise signal is weak, and the spin noise signal is extremely easy to be submerged by environmental noise and detection system noise in the detection process, so that the direct extraction of effective magnetic field information is extremely difficult. The traditional spectrum analysis method can not adapt to a dynamic magnetic field, and the current commonly used spin noise spectrum analysis method essentially belongs to a spectrum analysis technology, and the output of the method is static or time-averaged magnetic field information. This approach cannot capture the time-varying dynamics of the magnetic field and is therefore severely limited in applicability in time-varying magnetic field scenarios. The noise environment is complex and non-stationary, the model dependence is strong, in the actual working environment, the noise sources are various (such as optical noise, electronic noise, environmental magnetic noise and the like), the statistical characteristics of the noise tend to change along with time, and the noise presents non-stationary characteristics. Traditional Kalman filtering methods rely on accurate prior information of a noise covariance matrix, and in an actual system, the matrix is difficult to accurately characterize or acquire in real time, so that filtering performance is reduced or even diverged. The method lacks the self-adaptive tracking capability for the unknown waveform magnetic field, and the existing method generally assumes that the magnetic field change mode is known or slowly changed, cannot adapt to the real-time tracking of the sudden or complex waveform magnetic field, and limits the practicability of the method in a dynamic scene. Therefore, how to realize high-precision and self-adaptive real-time magnetic field estimation without prior waveform information in an actual environment with low signal-to-noise ratio, uncertain noise statistical characteristics and dynamic magnetic field change becomes a key problem to be solved in the technical field. Disclosure of Invention The invention provides a magnetic field estimation method based on a self-adaptive algorithm according to the defects of the prior art. The scheme adopts the self-adaptive extended Kalman filtering technology and combines parameter estimation to successfully realize self-adaptive estimation of the magnetic field under the condition that the waveform of the prior experimental magnetic field is unknown, and carries out self-adaptive estimation on the observed noise parameter. In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: a magnetic field estimation method based on an adaptive algorithm comprises the following steps: S1, establishing an evolution model describing spin dynamics of an atomic ensemble in a magnetic field along a set direction; S2, parameterizing the intensity information of a magnetic field to be measured into Larmor frequency which is in direct proportion to the magnetic field intensity, and constructing the Larmor frequency and at least two transverse components of atomic spin together into an extended state vector; S3, based on Faraday rotation effect, establishing a function relation between a polarization rotation angle of the detection light and spin components in the state vector, and taking the function relation as an observation model to acquire a spin observation signal containing measurement noise through polarization change of the detection light; S4, based on the process equation, executing a time updating step of extended Kalman filtering to obtain a state vector and a priori estimation value of an error covariance matrix of the state vector; S5, dynamically estimating the measured noise intensity at the current moment by analy