SERF quantum magnetometer sensitivity improving method based on random noise suppression

Abstract

The invention discloses an SERF quantum magnetometer sensitivity improving method based on random noise suppression, and belongs to the technical field of quantum precision measurement. The method comprises the following steps: firstly, constructing an equivalent optical rotation angle signal through polarized light splitting differential detection, and carrying out quadrature modulation and integral processing to obtain a noisy detection signal superposed with random noise; then, a chaos detection system based on a Duffing oscillator is introduced, and the system reaches a critical point of chaos and periodic states by adjusting the amplitude of a reference driving signal; a to-be-detected signal is introduced in a critical state, and the change of the amplitude of a weak signal triggers the transition of a system state. And enabling the system to return to a critical state by reversely adjusting the driving amplitude, and obtaining the real amplitude of the to-be-measured signal through inversion according to the variable quantity of the driving amplitude. The random noise of the quantum sensor is effectively suppressed, and the detection sensitivity and stability of the SERF atomic magnetometer under the strong noise background are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of quantum precision measurement technology, specifically relating to a method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression. Background Technology

[0002] Atomic magnetometers are highly sensitive quantum sensors that measure magnetic fields by utilizing the interaction between atomic spins and external magnetic fields. Among them, spin exchange relaxation free (SERF) atomic magnetometers exhibit extremely high theoretical sensitivity under extremely weak magnetic field conditions, showing significant application potential in fields such as biomagnetic signal detection, geomagnetic exploration, and fundamental physics research.

[0003] SERF atomic magnetometers typically operate based on the principles of optical pumping and optical detection. An external magnetic field induces the precession of atomic spins, which in turn modulates the polarization plane of the probe beam, generating an extremely small optical rotation angle signal. This signal is usually extracted using a polarization-splitting differential detection method, where a pair of photodetectors receives the orthogonally polarized components, and the results are differentially amplified to obtain an electrical signal proportional to the optical rotation angle.

[0004] However, in actual measurements, significant random noise inevitably exists in the detection signal due to various factors such as the thermal motion of atoms inside the atomic chamber, intensity fluctuations of pump and probe light, shot noise of the photodetector, and background noise of the electronic system. This noise cannot be effectively suppressed by traditional linear demodulation methods, often masking weak optical rotation angle signals and leading to a decrease in the signal-to-noise ratio of the final extracted signal, severely limiting the actual sensitivity of the magnetometer. Although existing studies have employed noise suppression methods based on statistical characteristics, such as lock-in amplification and digital averaging, these methods typically rely on specific assumptions about the noise distribution, and their performance degrades significantly in real-world environments where the noise statistics are unknown or non-stationary.

[0005] Therefore, the main problem facing existing technologies is how to achieve robust and highly sensitive detection of extremely weak optical rotation angle signals output by the SERF atomic magnetometer under strong random noise, thereby overcoming the limitation of noise on the final sensitivity of the magnetometer and realizing its theoretical performance. There is an urgent need for a novel signal processing method that does not rely on prior statistical characteristics of noise and has high detection sensitivity for weak signals. Summary of the Invention

[0006] This invention provides a method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression. By introducing a nonlinear dynamic mechanism with inherent robustness to random noise during signal processing, it overcomes the problem of insufficient random noise suppression capability of traditional linear demodulation methods, thereby achieving high-sensitivity detection of weak optical rotation angle signals and improving the measurement sensitivity of the SERF atomic magnetometer.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] Methods for improving the sensitivity of SERF quantum magnetometers based on random noise suppression include:

[0009] Step 1: Acquire two differential light intensity signals and construct a time-domain optical rotation angle signal;

[0010] Step 2: Perform orthogonal modulation and integration on the time-domain optical rotation angle signal to obtain two intermediate signals containing the target modulation component and superimposed random noise;

[0011] Step 3: Construct two orthogonal input signals to be tested based on the two intermediate signals. Perform steps 4-8 on the two orthogonal input signals to be tested respectively to obtain two orthogonal amplitude components.

[0012] Step 4: Establish a chaotic oscillator system under a reference driving signal, wherein the frequency of the reference driving signal is the same as the frequency of the target modulation component;

[0013] Step 5: By adjusting the amplitude of the reference driving signal, the chaotic oscillator system is brought to a critical state between the chaotic state and the periodic state.

[0014] Step 6: Introduce one of the orthogonal input signals to the chaotic oscillator system in a critical state, and record the state changes of the chaotic oscillator system.

[0015] Step 7: Adjust the amplitude of the reference driving signal in reverse based on the state change until the chaotic oscillator system returns to the critical state, and calculate the change in the amplitude of the driving force before and after the adjustment.

[0016] Step 8: Based on the change in the amplitude of the driving force, invert the amplitude of the target modulation component contained in one of the input signals to be measured, and take it as an orthogonal amplitude component;

[0017] Step 9: Based on the two orthogonal amplitude components, the final amplitude information of the input signal to be tested is obtained through combination operation.

[0018] Furthermore, the quadrature modulation and integration processing in step 2 specifically includes:

[0019] The time-domain optical rotation angle signal is multiplied by a sine signal and a cosine signal with the target modulation frequency, respectively, to obtain the two-way multiplication result;

[0020] The product of the two signals is integrated and averaged within a preset time window, the length of which is an integer multiple of the period corresponding to the target modulation frequency, in order to suppress non-target frequency components and obtain the two intermediate signals.

[0021] Furthermore, the specific steps in step 3 for constructing two orthogonal input signals to be tested are as follows:

[0022] The two intermediate signals are multiplied by a cosine reference signal of the same frequency to form two input signals to be tested, which are sinusoidal carrier signals superimposed with random noise, and their carrier frequencies are the same as the target modulation frequency.

[0023] Furthermore, the chaotic oscillator system in step 4 is a Duffing oscillator system, whose dynamic equations include linear damping terms, negative linear restoring force terms, nonlinear cubic restoring force terms, and periodic driving terms; the reference driving signal is a cosine signal with the same frequency as the target modulation component.

[0024] Furthermore, in step 5, the maximum disturbance characteristic index of the chaotic oscillator system under the current reference driving signal amplitude is calculated by numerical methods, and the maximum disturbance characteristic index approaching zero is used as the criterion for the system to be in a critical state between a chaotic state and a periodic state.

[0025] Furthermore, the calculation process of the maximum disturbance characteristic index includes:

[0026] Numerical integration is performed on the system disturbance equations to solve for the system state.

[0027] During the integration process, the perturbation matrix representing the two linearly independent perturbation directions is subjected to periodic QR decomposition and orthogonalization;

[0028] The maximum perturbation characteristic index is obtained by summing the logarithmic values ​​of the diagonal elements of the upper triangular matrix obtained from QR decomposition and calculating the time average.

[0029] Furthermore, in step 7, if the system state changes from a critical chaotic state to a periodic state after the introduction of the input signal to be measured, the system returns to the critical state by gradually reducing the amplitude of the reference driving signal; the change in the driving force amplitude is the absolute value of the difference between the amplitude of the reference driving signal when the system returns to the critical state and the initial critical driving force amplitude.

[0030] Furthermore, the combination operation in step 9 specifically involves performing a square root operation on the two orthogonal amplitude components to reconstruct the amplitude of the original target modulation component.

[0031] In a second aspect, the present invention provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs; wherein, when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the aforementioned method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression.

[0032] Thirdly, the present invention provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enable the processor to implement the aforementioned method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression.

[0033] The beneficial effects of this invention are as follows:

[0034] Effectively suppressing random noise in quantum sensors and improving detection sensitivity: This invention utilizes the characteristic that chaotic systems are highly sensitive to small disturbances in critical states to transform weak signal amplitude information into observable system state transitions, while random noise does not change the macroscopic dynamic behavior of the system. Thus, it achieves effective extraction of target signals in the context of strong noise, significantly improving the measurement sensitivity of the SERF atomic magnetometer.

[0035] Independent of noise statistical characteristics, with strong robustness: Unlike traditional methods based on lock-in amplification or statistical averaging, this invention does not require prior assumptions about the probability distribution or statistical characteristics of noise, avoiding performance degradation caused by noise model mismatch, and has stronger adaptability and stability in complex and non-stationary noise environments.

[0036] Transforming amplitude detection into state discrimination reduces hardware dependence: This method does not directly amplify and estimate the signal linearly, but indirectly inverts the signal amplitude by judging the transition between the chaotic system and the periodic state. This reduces the dependence on high-precision, low-noise analog front-end circuits and is beneficial for system integration and miniaturization.

[0037] Quantization inversion based on critical driving amplitude yields stable and repeatable results: By using the critical driving amplitude of the chaotic system as a unified criterion, the process of solving the signal amplitude has clear physical meaning and good repeatability, overcoming the problems of subjective and difficult-to-unify threshold setting in traditional chaotic detection methods.

[0038] Facilitates digital implementation and system integration: The core steps of the method can be implemented in the form of algorithms in digital signal processors or embedded systems. The process is clear and the calculation is stable, making it suitable for real-time online signal processing and integrated applications of quantum sensing systems such as atomic magnetometers. Attached Figure Description

[0039] Figure 1 This is a flowchart of the method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression, as described in this invention.

[0040] Figure 2 This is a schematic diagram of QR decomposition;

[0041] Figure 3 Flowchart for solving the system disturbance characteristic index;

[0042] Figure 4 System model diagram;

[0043] Figure 5 This is a graph showing the variation of the disturbance characteristic index of a chaotic system with the amplitude of the driving force. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] like Figure 1 As shown, this invention provides a method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression, the method comprising:

[0046] Step 1: Acquire two differential light intensity signals and construct a time-domain optical rotation angle signal.

[0047] The light intensity signals output from two photodetectors are acquired, and summation and subtraction operations are performed on the two signals respectively to construct an equivalent photoelectric signal proportional to the optical rotation angle, which is then regarded as a time-domain optical rotation angle signal. The light intensity signals output from the two photodetectors... , They can be represented as:

[0048] (1)

[0049] In the formula, This represents the sum of the light intensities of the two laser beams before they are split. This is the optical rotation angle signal;

[0050] By summing and subtracting the two light intensity signals, an equivalent photoelectric signal proportional to the optical rotation angle can be obtained, and its expression is:

[0051] (2)

[0052] In actual measurement, the optical rotation angle signal It varies with time and can be regarded as a time-domain signal, which can be represented by a Fourier series expansion as follows:

[0053] (3)

[0054] In the formula, This represents the DC component of the optical rotation angle signal. , and The first The amplitude, angular frequency, and initial phase of the first harmonic component, where t represents time.

[0055] Furthermore, the target modulation frequency component can be separated from the total signal, and the optical rotation angle signal can be rewritten as:

[0056] (4)

[0057] in The modulated signal component of the target to be detected. It's still the same optical rotation angle signal, just expressed in a different form. The amplitude of the modulated signal to be detected. The frequency of the modulated signal of the target to be detected. is the phase of the target modulation signal to be detected, and m is used to distinguish the target signal from the other signals.

[0058] Step 2: Perform orthogonal modulation and integration on the time-domain optical rotation angle signal to obtain two intermediate signals containing the target modulation component and superimposed random noise.

[0059] To effectively extract the target modulation signal component to be detected, the optical rotation angle signal is multiplied and modulated with a co-frequency orthogonal reference signal, and then integrated and averaged within a preset time window to obtain two intermediate signals containing the target modulation component and superimposed with random noise. Specifically, the optical rotation angle signal is multiplied and modulated with a co-frequency orthogonal reference signal. and Multiply them to obtain the intermediate signals of the two paths. and :

[0060] (5)

[0061] (6)

[0062] Substituting equations (5) and (6) into equation (4) and expanding using trigonometric identities, we can obtain the component corresponding to the target modulation term as follows:

[0063] (7)

[0064] (8)

[0065] Using trigonometric identities to product and sum, we can obtain:

[0066] (9)

[0067] (10)

[0068] (11)

[0069] (12)

[0070] The preset time window is selected as the period corresponding to the lowest modulation frequency, so that non-target frequency components are suppressed during the integration process. and Within the preset time window By performing an integral averaging process within the range, we obtain:

[0071] (13)

[0072] (14)

[0073] Step 3: Construct two orthogonal input signals to be tested based on the two intermediate signals. Perform steps 4-8 on the two orthogonal input signals to be tested respectively to obtain two orthogonal amplitude components.

[0074] The two intermediate signals are respectively constructed as sinusoidal modulated signals superimposed with random noise, so as to serve as orthogonal input signals to be tested, so as to satisfy the subsequent chaotic oscillator system.

[0075] Under actual measurement conditions, random noise is inevitably superimposed on the optical rotation angle signal. Its actual expression should be:

[0076] (15)

[0077] When the noisy optical rotation angle signal is multiplied by the reference signal and within the time window During internal integration, the statistical properties of random noise do not disappear, nor does the randomness of the random noise change. The final actual output signal can be expressed as:

[0078] (16)

[0079] (17)

[0080] in, and These represent the equivalent noise terms after random noise has undergone multiplication modulation and integration processing, respectively.

[0081] Further and Respectively with reference signal Multiplying them together, we get:

[0082] (18)

[0083] (19)

[0084] in and These are the equivalent random noises after multiplying with the reference signal.

[0085] Therefore, the processed signal appears as a sinusoidal modulated signal superimposed with random noise, satisfying the requirements of a chaotic system for the input signal structure. Thus, to address the noise present, this paper uses a chaotic system to process the aforementioned actual signal. and Recorded as:

[0086] (20)

[0087] (twenty one)

[0088] in , .

[0089] For ease of subsequent explanation, and The unified representation is the input signal to be measured. :

[0090] (twenty two)

[0091] Step 4: Establish the chaotic oscillator system under the reference driving signal.

[0092] A Duffing chaotic oscillator system containing nonlinear terms is constructed as the subsequent detection system, and parameters are selected. , , , Where k is the damping parameter of the chaotic system, and a and b are the nonlinear restoring force parameters. To determine the angular frequency of the driving signal, a reference driving signal with the same driving frequency as the input signal under test is introduced. Its amplitude is adjustable, and it takes the following form:

[0093] (twenty three)

[0094] in This represents the amplitude of the reference drive signal, which is a controllable parameter. The frequency of the reference driving signal is equal to the frequency of the signal under test.

[0095] The structural model of the chaotic oscillator system is as follows: Figure 4 As shown.

[0096] Step 5: Adjust the chaotic oscillator system to the critical state between chaos and periodicity.

[0097] By adjusting the amplitude of the reference driving signal, the chaotic oscillator system is brought to a critical state between chaotic and periodic states, and the amplitude of the corresponding reference driving signal is recorded as the critical driving force amplitude. :

[0098] (twenty four)

[0099] Where x is the displacement of the system. Let be the derivative of displacement with respect to time, and let represent velocity. Let be the second derivative of displacement with respect to time, and let represent acceleration.

[0100] Consider the aforementioned dimensionless chaotic system:

[0101] (25)

[0102] Introduce state variables The system can then be written as a two-dimensional non-autonomous system:

[0103] Suppose the solution to this system is: However, when a very small perturbation exists in the vicinity of the system solution, the solution will become a perturbation solution: ,in and These are all extremely small quantities, and when combined, they form the perturbation vector. This disturbance vector is the object of study for the system disturbance characteristic index.

[0104] The solution of the system in equation (25) Replace with perturbation solution We can obtain:

[0105] (26)

[0106] because and It is an infinitesimal, indicated by a superscript. Let denote the first derivative, where all higher-order terms are negligible, retaining only the first-order terms. Expanding the cubic terms in the equation and using equations (25) and (26), we can obtain the perturbation equation:

[0107] (27)

[0108] To facilitate subsequent calculations, the above formula is rewritten in matrix form:

[0109] (28)

[0110] Because a chaotic system is two-dimensional in phase space, its system state is determined by displacement. With speed The system is jointly determined. The system disturbance characteristic index is used to characterize the sensitivity of adjacent orbits in the system's phase space to small disturbances in initial conditions, while the disturbance itself refers to the small shift of the system state in different directions in the phase space.

[0111] For a two-dimensional system, there are at least two linearly independent perturbation directions in the local phase space, each corresponding to an independent change in the state variable in a different direction. Therefore, it is necessary to consider two linearly independent perturbation directions in order to fully characterize the stability structure of the system.

[0112] (29)

[0113] in , These represent the components of the initial disturbance evolving over time along the x1 and x2 directions, respectively. , This refers to the generalized displacement perturbation corresponding to two linearly independent modes; , This corresponds to the generalized velocity perturbation.

[0114] Therefore, perturbations from both directions are added simultaneously to obtain the perturbation matrix. Record it as :

[0115] (30)

[0116] Choose an initial perturbation basis, and select two linearly independent perturbations at time t=0:

[0117] (31)

[0118] but I represents the identity matrix, and combining with equation (28), we can obtain:

[0119] in For Jacobian matrices, Each column is a perturbation vector, representing a linearly independent perturbation direction.

[0120] Since the above differential equations are generally difficult to solve analytically, this paper uses numerical integration to solve them.

[0121] (32)

[0122] Let the time step be Then the evolution relationship of the perturbation matrix between adjacent time steps can be approximately expressed as:

[0123] To improve the accuracy and stability of numerical integration, this paper employs the fourth-order Runge-Kutta (RK4) method to discretize and solve the evolution equation of the perturbation matrix, thereby obtaining the discrete evolution form of the perturbation matrix:

[0124] (33)

[0125] The slope of the matrix at each stage is defined as follows:

[0126] (34)

[0127] in This represents the value of the perturbation matrix at the nth time step, while t n This corresponds to the physical time at that moment.

[0128] Since matrix multiplication can be viewed as an independent operation on each column vector of the perturbation matrix, the above-described matrix-form fourth-order Runge-Kutta algorithm can also be equivalently expanded into a column-by-column update of each perturbation vector in numerical implementation.

[0129] Since the system studied in this paper is a two-dimensional continuous dynamical system, it has two system perturbation characteristic indices, which describe the average growth characteristics of the perturbation in two linearly independent directions. In the numerical calculation process, if the perturbation matrix is ​​directly integrated over a long period of time, the perturbation vector will gradually collapse towards the direction corresponding to the largest system perturbation characteristic index, causing the information in the second perturbation direction to be submerged by the numerical values, making it difficult to accurately extract the complete system perturbation characteristic index spectrum.

[0130] To avoid the aforementioned numerical instability issues, it is necessary to introduce an orthogonalization mechanism during the evolution of the perturbation matrix, performing periodic orthogonal processing on the perturbation vector. Based on QR decomposition theory, numerical correlations between different perturbation directions can be eliminated while preserving perturbation growth information, thereby achieving stable calculation of the system's perturbation characteristic index.

[0131] On the other hand, since the initial conditions of the perturbation matrix are taken as Furthermore, the system variational equation is a continuous linear differential equation, and the perturbation matrix within a finite time interval... It always maintains non-singularity. Therefore, as Figure 2 As shown, at any orthogonalization time, the perturbation matrix can be decomposed into QR, thus ensuring the feasibility and numerical stability of the algorithm.

[0132] In the At the time of orthogonalization, let the perturbation matrix be expressed as:

[0133] (35)

[0134] in , Let be two sets of linearly independent perturbation vectors.

[0135] The specific process of QR decomposition can be regarded as Gram-Schmidt orthogonalization of the perturbation vector. The result can be expressed as:

[0136] (36)

[0137] in:

[0138] ;

[0139] in , This represents the instantaneous scaling factor of the disturbance in the first and second orthogonal directions. , These are the first and second orthogonal basis vectors after normalization. It is the projected coupling coefficient between the two disturbance directions, reflecting the cross-transmission intensity of disturbance energy in the non-principal axis direction.

[0140] After completing the QR decomposition, use orthogonal matrices The original perturbation matrix is ​​replaced as the initial condition for the perturbation evolution in the next time period to ensure the stability of subsequent numerical integration.

[0141] To calculate the system disturbance characteristic index, after each orthogonalization operation, the upper triangular matrix is... Logarithmic summation of the diagonal elements, defining the cumulant:

[0142] (37)

[0143] (38)

[0144] Where the initial cumulative amount , .

[0145] Let the time interval for orthogonalization be . The total number of orthogonalizations is Then the system's first The perturbation characteristic index can be expressed as:

[0146] (39)

[0147] Thus, the characteristic index spectrum of the system disturbance is obtained. The maximum system disturbance characteristic index is defined as:

[0148] ,

[0149] The complete solution process for the system disturbance characteristic index is as follows: Figure 3 As shown.

[0150] Step 6: Introduce one of the orthogonal input signals to the chaotic oscillator system in a critical state, and record the state changes of the chaotic oscillator system.

[0151] When the system is in a critical state, the input signal S to be measured is introduced, causing the system state to transition from chaotic to periodic. The complete equation of the chaotic system at this time is as follows:

[0152] (40)

[0153] Since noise has no effect on the system state of a chaotic system, this term can be ignored, and the driving term can be denoted as... :

[0154] (41)

[0155] Step 7: Adjust the amplitude of the reference drive signal in reverse based on the state change, and calculate the change in the amplitude of the drive force before and after the adjustment.

[0156] Based on the direction of the state change of the chaotic oscillator system, the amplitude of the reference driving signal is adjusted in the opposite direction until the chaotic oscillator system returns to the critical state between chaos and periodicity. The driving amplitude at this time is recorded as the driving force amplitude after adjustment, and the change in driving force amplitude before and after adjustment is calculated.

[0157] A chaotic system will transition from a critical state between chaos and periodicity to a stable periodic state by continuously decreasing the amplitude of the reference driving signal. until the magnitude of the driving term. Return to critical value ,at this time Then we can find out .

[0158] The curve of the chaotic system disturbance characteristic index as a function of driving force amplitude used in this invention was obtained experimentally, such as... Figure 5 As shown, the difference in the amplitude of the driving force before and after adjustment is the amplitude of the input signal to be measured.

[0159] Step 8: Based on the change in the amplitude of the driving force, invert the orthogonal amplitude component of one of the input signals to be measured.

[0160] The preliminary amplitude information of the input signal to be measured is calculated by using the difference between the critical driving force amplitude and the adjusted driving force amplitude at the return to the critical state.

[0161] Step 9: Based on the two orthogonal amplitude components, the final amplitude information of the input signal to be tested is obtained through combination operation.

[0162] For the two signals mentioned above respectively and Solving for each will yield their corresponding results. and Then through and The signal to be measured was then solved. .

[0163] In a second aspect, the present invention provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs; wherein, when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the aforementioned method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression.

[0164] Thirdly, the present invention provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enable the processor to implement the aforementioned method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression.

[0165] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression, characterized in that, include: Step 1: Acquire two differential light intensity signals and construct a time-domain optical rotation angle signal; Step 2: Perform orthogonal modulation and integration on the time-domain optical rotation angle signal to obtain two intermediate signals containing the target modulation component and superimposed random noise; Step 3: Construct two orthogonal input signals to be tested based on the two intermediate signals. Perform steps 4-8 on the two orthogonal input signals to be tested respectively to obtain two orthogonal amplitude components. Step 4: Establish a chaotic oscillator system under a reference driving signal, wherein the frequency of the reference driving signal is the same as the frequency of the target modulation component; Step 5: By adjusting the amplitude of the reference driving signal, the chaotic oscillator system is brought to a critical state between the chaotic state and the periodic state. Step 6: Introduce one of the orthogonal input signals to the chaotic oscillator system in a critical state, and record the state changes of the chaotic oscillator system. Step 7: Adjust the amplitude of the reference driving signal in reverse based on the state change until the chaotic oscillator system returns to the critical state, and calculate the change in the amplitude of the driving force before and after the adjustment. Step 8: Based on the change in the amplitude of the driving force, invert the amplitude of the target modulation component contained in one of the input signals to be measured, and take it as an orthogonal amplitude component; Step 9: Based on the two orthogonal amplitude components, the final amplitude information of the input signal to be tested is obtained through combination operation.

2. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 1, characterized in that, The quadrature modulation and integration processing in step 2 specifically includes: The time-domain optical rotation angle signal is multiplied by a sine signal and a cosine signal with the target modulation frequency, respectively, to obtain the two-way multiplication result; The product of the two signals is integrated and averaged within a preset time window, the length of which is an integer multiple of the period corresponding to the target modulation frequency, in order to suppress non-target frequency components and obtain the two intermediate signals.

3. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 2, characterized in that, The specific steps for constructing two orthogonal input signals to be tested in step 3 are as follows: The two intermediate signals are multiplied by a cosine reference signal of the same frequency to form two input signals to be tested, which are sinusoidal carrier signals superimposed with random noise, and their carrier frequencies are the same as the target modulation frequency.

4. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 1, characterized in that, The chaotic oscillator system in step 4 is a Duffing oscillator system, whose dynamic equations include linear damping terms, negative linear restoring force terms, nonlinear cubic restoring force terms, and periodic driving terms; the reference driving signal is a cosine signal with the same frequency as the target modulation component.

5. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 1, characterized in that, In step 5, the maximum disturbance characteristic index of the chaotic oscillator system under the current reference driving signal amplitude is calculated by numerical method, and the maximum disturbance characteristic index approaching zero is used as the criterion for the system to be in the critical state between the chaotic state and the periodic state.

6. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 5, characterized in that, The calculation process of the maximum disturbance characteristic index includes: Numerical integration is performed on the system disturbance equations to solve for the system state. During the integration process, the perturbation matrix representing the two linearly independent perturbation directions is subjected to periodic QR decomposition and orthogonalization; The maximum perturbation characteristic index is obtained by summing the logarithmic values ​​of the diagonal elements of the upper triangular matrix obtained from QR decomposition and calculating the time average.

7. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 1, characterized in that, In step 7, if the system state changes from a critical chaotic state to a periodic state after the input signal to be measured is introduced, the system returns to the critical state by gradually reducing the amplitude of the reference driving signal; the change in the driving force amplitude is the absolute value of the difference between the amplitude of the reference driving signal when the system returns to the critical state and the initial critical driving force amplitude.

8. The method for improving the sensitivity of a SERF quantum magnetometer based on random noise suppression according to claim 1, characterized in that, The combination operation in step 9 specifically involves performing a square root operation on the two orthogonal amplitude components to reconstruct the amplitude of the original target modulation component.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When one or more programs are executed by the one or more processors, the one or more processors implement the SERF quantum magnetometer sensitivity enhancement method based on random noise suppression as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed by a processor, enable the processor to implement the SERF quantum magnetometer sensitivity enhancement method based on random noise suppression as described in any one of claims 1-8.

Images