A diamond color center quantum sensing differential signal wavelet denoising frequency tracking system

Abstract

The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing. The application discloses a diamond color center quantum sensing differential signal wavelet denoising frequency tracking system and relates to the technical field of quantum precision sensing.

Description

Technical Field

[0001] This invention relates to the field of quantum precision sensing technology, and in particular to a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing. Background Technology

[0002] Diamond NV center quantum sensors are used for precision detection of weak magnetic fields due to their ultra-high sensitivity. However, existing equipment suffers from several technical defects. First, the detection uses single-channel fluorescence acquisition, and optical path fluctuations, circuit temperature drift, and vibration introduce a large amount of common-mode noise. Conventional filtering cannot distinguish between effective signals and noise, resulting in a low signal-to-noise ratio. Second, noise processing often uses fixed wavelet thresholds, which cannot adaptively reduce noise in layers based on real-time noise characteristics. The noise reduction effect is significantly reduced in high-temperature and vehicle vibration scenarios. Third, the NV resonant frequency is susceptible to temperature and vibration drift. Traditional open-loop or simple closed-loop tracking has large steady-state errors and is prone to frequency lock-up. Fourth, the system lacks fault self-diagnosis and parameter iterative optimization capabilities. Parameters need to be manually adjusted for different operating conditions, resulting in poor equipment versatility. At the same time, the lack of a verification mechanism for data transmission between modules makes it prone to packet loss. It also lacks full-domain hardware overload protection, resulting in insufficient stability under complex industrial and vehicle operating conditions, making it difficult to meet the high-precision, long-term online quantum sensing requirements of multiple scenarios. Therefore, we propose a wavelet denoising frequency tracking system for differential signals from diamond color center quantum sensing. Summary of the Invention

[0003] The present invention proposes a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing, which solves the above-mentioned shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: A wavelet denoising frequency tracking system for differential signals from a diamond color center quantum sensor includes a quantum sensor differential signal acquisition module, a wavelet multi-layer decomposition denoising processing module, a resonant frequency real-time tracking control module, a signal timing buffer storage module, a working condition fault feature analysis module, an algorithm iterative learning optimization module, an overload safety protection module, a microwave drive execution output module, and a bus data security verification and interaction module. The quantum sensing differential signal acquisition module is electrically connected to the diamond double NV color center detection unit and outputs the original differential sensing signal to the wavelet multi-layer decomposition and denoising processing module. The wavelet multi-layer decomposition and denoising module outputs a clean differential signal to the resonant frequency real-time tracking control module and the signal timing cache storage module. The resonant frequency real-time tracking control module outputs a frequency adjustment command to the microwave drive execution output module; The signal timing cache storage module provides a timing data source for the operating condition fault feature analysis module; The fault feature analysis module outputs the fault feature dataset to the algorithm iterative learning and optimization module. The algorithm iterative learning and optimization module outputs updated parameters to the wavelet multi-layer decomposition denoising module and the resonant frequency real-time tracking and control module. The overload safety protection module is connected in parallel with all hardware modules; The bus data security verification and interaction module manages the entire module's data interaction link.

[0005] Furthermore, the quantum sensing differential signal acquisition module includes two photoelectric conversion subunits, a high-precision differential operational amplifier, and a 1MSps analog-to-digital conversion subunit; The two-way photoelectric conversion subunits receive fluorescent light signals output from two sets of symmetrically arranged diamond NV color centers one-to-one. The output analog electrical signals are connected to a high-precision differential operational amplifier to complete the difference calculation, eliminating common-mode interference signals between the optical path and the circuit. The differentially processed analog signals are converted into digital raw differential sensing signals by the analog-to-digital conversion subunit.

[0006] Furthermore, the wavelet multi-layer decomposition denoising module incorporates a fourth-order Daubechies wavelet basis function to perform hierarchical decomposition on the original differential sensing signal. The wavelet multi-layer decomposition denoising module is further configured with a noise entropy calculation subunit to calculate the high-frequency coefficient information entropy in real time. When the entropy value is lower than the preset lower limit, the current layer decomposition is terminated. The maximum number of decomposition layers does not exceed six. A fixed hard threshold is applied to all high-frequency wavelet coefficients from the first to the sixth layers obtained by decomposition to complete noise filtering. The low-frequency approximate wavelet coefficients are completely retained without any numerical correction. All wavelet coefficients are input into the inverse wavelet transform unit to reconstruct a pure differential sensing signal. The wavelet high-frequency coefficient information entropy (adaptive hierarchical denoising criterion) is used to control the number of wavelet decomposition levels and the adaptive generation of the denoising threshold. It is the core of the wavelet denoising module, and the formula is as follows: ; ; Layered termination rules: Stop wavelet decomposition; the maximum number of decomposition levels is limited to 6. in, The information entropy of the high-frequency wavelet coefficients of the j-th layer; For the i-th high-frequency wavelet coefficient of the j-th layer; These are normalized coefficient probabilities; The system presets a fixed critical entropy value as the wavelet layer termination entropy threshold. The entropy-inverse wavelet hard thresholding denoising formula is as follows: ; ; in, The high-frequency noise reduction threshold of the j-th layer wavelet; C is the inverse proportionality coefficient of the entropy threshold, which is 1.2 for standard working conditions and 1.8 for vehicle vibration working conditions. The information entropy of the high-frequency coefficients in the j-th layer; These are the original high-frequency wavelet coefficients before noise reduction; These are the noise-reduced high-frequency wavelet coefficients after hard thresholding.

[0007] Furthermore, the resonant frequency real-time tracking control module incorporates a fast Fourier transform subunit, a peak frequency locking subunit, and a closed-loop step adjustment subunit. The Fast Fourier Transform subunit completes the time-domain to frequency-domain conversion of the pure differential signal; The peak frequency locking subunit locks the inherent resonant frequency corresponding to the maximum spectral amplitude. The closed-loop step adjustment subunit calculates the difference between the current microwave output frequency and the predicted frequency and generates an adjustment command. The resonant frequency real-time tracking control module is further configured with a frequency deviation trend extrapolation subunit, which continuously records the resonant frequency point displacement over the past N sampling cycles, uses linear extrapolation to predict the target frequency point for the next cycle, and uses the predicted frequency point as the initial reference value for closed-loop step adjustment. The formula for linear extrapolation prediction of resonance frequency is shown below: ; in, Predict the resonance frequency for the next sampling period; This is the measured NV resonant frequency point for the current frame; The frequency drift slope is obtained by least-squares fitting of N historical frequency points; This is the system sampling period.

[0008] Furthermore, the signal timing cache storage module adopts a cyclic first-in-first-out storage architecture, and the hardware partitions are divided into an original signal storage area, a noise reduction signal storage area, and a frequency parameter storage area. The system completes a time-series data write every millisecond. When the storage space is full, it automatically overwrites the earliest historical time-series data, retains 30 minutes of continuous complete running data, passively receives the fixed-point read command from the working condition fault characteristic analysis module, and completes the directional data output.

[0009] Furthermore, the operating condition fault feature analysis module pre-stores three types of fixed threshold databases: a signal distortion threshold database, a frequency offset threshold database, and an environmental noise amplitude threshold database. The cached timing data is compared with the corresponding threshold library standard parameters frame by frame. When the signal distortion value exceeds the signal distortion threshold, it is judged as a noise abnormality fault. When the difference between the real-time tracking frequency point and the standard resonant frequency point exceeds the frequency offset threshold, it is judged as a frequency lock-out fault. When the difference between the amplitudes of the two front-end acquired signals exceeds the preset optical path deviation threshold, it is judged as a detection optical path offset fault. Each type of fault is matched with a unique fixed numerical code, and all fault data are uniformly summarized to generate a fault feature dataset.

[0010] Furthermore, the algorithm iterative learning optimization module is equipped with a gradient descent optimization algorithm to perform global parameter fitting calculations on the structured fault feature dataset; Based on noise anomaly fault data, the noise reduction threshold of wavelet high-frequency coefficients is individually corrected; based on frequency lock-out fault data, the frequency tracking closed-loop adjustment step size is individually corrected; based on probe optical path offset fault data, the front-end differential signal acquisition gain is individually corrected. The algorithm iterative learning optimization module is further configured with a cross-device feature mapping subunit to map the local fault feature dataset to a unified feature vector space, generate a standardized parameter correction model that is independent of individual devices, and export the generated correction model through the bus data security verification and interaction module, and directly load and solidify it in other devices of the same model. The gradient descent parameter iterative optimization formula is as follows: ; in, The system parameters to be optimized; The learning rate is used for gradient descent. The loss function is constructed based on the fault dataset; This represents the number of algorithm iterations.

[0011] Furthermore, the overload safety protection module is equipped with three independent voltage acquisition circuits, a bus transmission rate monitoring subunit, and a microwave output power sampling subunit. The three voltage acquisition circuits monitor the real-time operating voltage of the signal acquisition module, the signal processing module, and the microwave drive module, respectively. The bus transmission rate monitoring subunit counts the number of valid bytes transmitted on the bus per unit time. The microwave output power sampling subunit collects the instantaneous output power of the microwave signal in real time; The overload safety protection module is further configured with a periodic fault injection self-test subunit, which injects known standard disturbance signals into the voltage acquisition circuit, bus monitoring subunit and power sampling subunit at fixed time intervals, and compares the consistency of the protection response action with the preset action sequence. If any of the three monitoring parameters exceeds the preset upper limit threshold, the system will trigger hardware current limiting protection, control command lockout protection, and laser detection optical path power-off protection in a fixed order.

[0012] Furthermore, the microwave drive execution output module includes a numerically controlled frequency source subunit, a fixed gain power amplifier subunit, and a unidirectional directional coupling output subunit; The numerically controlled frequency source subunit accurately responds to external frequency adjustment commands and outputs a sinusoidal excitation electrical signal at a specified frequency. The fixed-gain power amplifier subunit amplifies the sinusoidal electrical signal to the system's rated operating power. The unidirectional directional coupling output subunit isolates the reverse microwave signal reflected by the rear resonant cavity and outputs a stable microwave excitation signal without reflection interference to the diamond color center resonant cavity.

[0013] Furthermore, the bus data security verification and interaction module is equipped with a 32-bit cyclic redundancy check encoding unit, a device anomaly alarm unit, and an automatic data retransmission control unit; All data packets exchanged between modules in the system are appended with a 32-bit cyclic redundancy check code. The receiving end verifies the integrity of the data packets through the check encoding unit. When the verification fails, the abnormal alarm unit sends a fixed-format alarm message to the host computer, and at the same time, the automatic data retransmission control unit instructs the data sending end to retransmit the complete sensor data of the current frame.

[0014] Compared with existing technologies, the beneficial effects of this invention are: 1. This invention employs dual NV color center symmetrical differential acquisition to eliminate common-mode interference, combined with 4th-order Daubechies wavelet combined with information entropy adaptive hierarchical noise reduction. The number of decomposition layers and noise reduction threshold are dynamically adjusted according to noise entropy. Laboratory steady-state noise suppression is improved by 42%, and noise filtering efficiency under vehicle vibration and high-temperature conditions is improved by 32dB and 35% respectively. The effective sensing signal is completely preserved. The noise interference problem is solved from both the front end and the algorithm. The use of differential acquisition + adaptive wavelet noise reduction greatly improves the noise suppression capability. 2. This invention integrates FFT spectrum peak locking and historical frequency point linear extrapolation prediction to construct a closed-loop stepping microwave adjustment mechanism, which can predict frequency drift in advance and correct the microwave output frequency. At room temperature, the steady-state error of frequency tracking is ≤2kHz, and in high-temperature and vehicle-mounted dynamic vibration scenarios, drift can be controlled within 5kHz, with no frequency loss throughout the process, ensuring continuous and stable control of the NV color center quantum state. Through linear extrapolation closed-loop frequency tracking, stable locking of the resonant frequency is achieved. 3. This invention establishes a multi-threshold fault identification system that accurately distinguishes noise, frequency offset, and optical path faults and generates a feature dataset. It utilizes the gradient descent algorithm to directionally correct the noise reduction threshold, adjust the step size, and optimize the acquisition gain. A standardized, universal model is generated through cross-device feature mapping. The fault identification accuracy exceeds 96%, and the model can be reused in batches on the same model of equipment, eliminating the need for repeated manual parameter tuning. It employs fault self-diagnosis combined with gradient descent iterative optimization, enabling the equipment to adapt to various operating conditions. 4. The present invention features a global bus equipped with 32-bit CRC check, which automatically retransmits and alarms when data is corrupted. The independent protection module monitors voltage, bus speed, and microwave power in real time, and triggers current limiting, frequency locking, and laser power-off protection when over-limits are exceeded. It also has a periodic self-test function. The data packet loss rate in vehicle scenarios is less than 0.01%, and single probe array faults can be isolated to avoid hardware damage and data loss. In summary, this system solves the noise problem in multiple scenarios through differential acquisition and entropy adaptive wavelet algorithm, combines linear extrapolation closed-loop control to stably lock the NV resonance frequency point, achieves parameter adaptation by relying on gradient descent iteration of fault dataset, and is equipped with CRC data verification and graded overload protection to ensure long-term stable operation. It is suitable for laboratory, high-temperature industrial, vehicle-mounted mapping, and multi-probe array scenarios, and takes into account detection accuracy, working condition adaptability and equipment safety. Attached Figure Description

[0015] Figure 1 This is an overall system block diagram of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention; Figure 2 This is an internal block diagram of the quantum sensing differential signal acquisition module of a wavelet denoising frequency tracking system for diamond color center quantum sensing differential signals proposed in this invention. Figure 3 This is an internal block diagram of the wavelet multi-layer decomposition denoising processing module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 4 This is an internal block diagram of the resonant frequency real-time tracking control module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 5 This is an internal block diagram of the signal timing cache storage module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention; Figure 6 This is an internal block diagram of the fault characteristic analysis module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 7 This is an internal block diagram of the algorithm iterative learning and optimization module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 8 This is an internal block diagram of the overload safety protection module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 9 This is an internal block diagram of the microwave drive execution output module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Figure 10 This is an internal block diagram of the bus data security verification and interaction module of a wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing proposed in this invention. Detailed Implementation

[0016] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0017] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0018] Example 1 Reference Figure 1 - Figure 10 Basic Standard Diamond NV Color Center Magnetic Field Quantum Sensing System This embodiment is suitable for conventional room temperature diamond double NV center magnetic field detection scenarios. The hardware modules are fully equipped according to claims 1 to 10, without additional expansion units, and are used for laboratory steady-state weak magnetic field measurement.

[0019] Hardware connection and parameter configuration: The quantum sensing differential signal acquisition module is equipped with two symmetrical silicon-based photoelectric conversion sub-units, matched with 0.5mm dual NV diamond probes, the differential operational amplifier adopts instrument-grade low-noise chips, and the analog-to-digital conversion sub-unit has a fixed sampling rate of 1MSps; The wavelet multi-layer decomposition denoising module uses a 4th-order Daubechies wavelet basis, with a maximum decomposition layer of 6 layers. The noise entropy threshold is set to 0.35, and the hard threshold for high-frequency coefficients is inversely proportional to the entropy value by a coefficient of 1.2. The resonant frequency real-time tracking control module linearly extrapolates the number of historical sampling points N=20, and the closed-loop step adjustment base step size is 10kHz. The signal timing buffer storage module uses a FIFO loop to store and retain 30 minutes of millisecond-level timing data completely. The microwave drive execution output module digitally controlled frequency source outputs microwaves in the 2.87GHz~2.92GHz NV color center resonant frequency band, and the directional coupler isolates 25dB of reverse reflection signals; The overload safety protection module has a voltage monitoring threshold of 12V, a bus speed limit of 10MB / s, and a microwave power limit of 200mW. The bus interaction module uniformly adopts 32-bit CRC check.

[0020] The overall workflow of the machine is as follows: Two sets of diamond NV color centers are excited by laser and output fluorescence. Two photoelectric conversion units collect the fluorescence light signal and convert it into analog voltage. A differential amplifier eliminates common-mode noise in the optical path and circuit. The signal is converted into a raw differential digital signal by a 1MSps ADC and sent to a wavelet denoising module. The wavelet module calculates the high-frequency information entropy in layers, filters out noise layer by layer, and reconstructs the pure sensing signal. One pure signal is sent to the FFT frequency domain analysis to lock the NV resonance peak frequency. The frequency offset extrapolation unit predicts the target frequency at the next moment. The closed-loop module outputs adjustment commands to control the numerically controlled microwave source. The other pure signal, along with the original signal and microwave frequency parameters, is synchronously stored in the loop buffer. The fault module compares the time-series data with three types of threshold libraries frame by frame to identify noise anomalies, frequency lock-out, and optical path offset faults and generate coded datasets. The gradient descent algorithm corrects the wavelet noise reduction threshold, frequency adjustment step size, and front-end acquisition gain based on the fault datasets. The overload protection module monitors the three-channel voltage, bus speed, and microwave power in real time. If the limit is exceeded, it will trigger current limiting, command lockout, and laser power-off protection in sequence. All data packets exchanged between modules are subject to 32-bit CRC check. If the check fails, the data will be automatically retransmitted and an alarm will be sent to the host computer.

[0021] Implementation results: For laboratory steady-state weak magnetic field measurement scenarios, common-mode noise suppression is improved by 42%, signal-to-noise ratio is improved by 28dB after wavelet denoising, steady-state error of resonant frequency tracking is ≤2kHz, there are no data packet loss or microwave reflection interference issues, and the basic fault identification accuracy is 96%, meeting the needs of conventional quantum precision magnetic field detection.

[0022] Example 2 Industrial High-Temperature Operating Condition Extended Diamond Color Center Quantum Sensing System: This embodiment is an industrial high-temperature environment adaptation modification based on Embodiment 1. It is designed for online monitoring of weak magnetic fields inside metallurgical and high-temperature equipment, adds high-temperature environment compensation logic, optimizes fault identification and algorithm iteration capabilities, and adapts to temperature variation detection conditions from 0 to 180℃.

[0023] Differentiated hardware and parameter optimization: The photoelectric conversion unit of the quantum sensing differential signal acquisition module is replaced with a high-temperature resistant photodiode, and a temperature compensation resistor is added to dynamically correct the fluorescence acquisition gain under high temperature. The wavelet multi-layer decomposition denoising module adds a temperature correlation entropy correction coefficient. For every 30°C increase in ambient temperature, the lower limit of entropy judgment is simultaneously lowered by 0.05 to adapt to the noise distortion characteristics under high temperature. The resonant frequency real-time tracking control module increases the number of linear extrapolation sampling points N to 50, adds a temperature-resonant frequency offset compensation lookup table, and pre-compensates the inherent frequency drift of the NV color center based on the real-time temperature of the probe. The threshold library of the working condition fault feature analysis module adds a temperature coupling offset threshold to distinguish between simple optical path faults and frequency drift faults induced by high temperature. The algorithm iterative learning optimization module enables a cross-device feature mapping subunit, collects high-temperature operating condition fault datasets to generate standardized correction models, which can be batch distributed to industrial sensing devices of the same model. The overload safety protection module adds a probe temperature acquisition channel, triggers an early warning when the temperature exceeds 180℃, and reduces microwave output power. The hardware architecture of the remaining buses, microwave drivers, and storage modules remains consistent with that of Example 1.

[0024] The overall machine differentiation workflow is as follows: Under high temperature conditions, the fluorescence signal of the dual NV diamond probe is attenuated by temperature. The two photoelectric units output analog signals through the temperature compensation circuit. After the differential circuit eliminates high temperature common-mode interference, it is converted into the original differential signal. The wavelet module combines the real-time probe temperature to dynamically adjust the information entropy judgment threshold and filter out high temperature thermal noise in layers. The pure sensing signal is analyzed by FFT spectrum and the temperature frequency drift compensation is added to lock the true resonant frequency point. Long sequence linear extrapolation reduces the frequency jump caused by high temperature noise. The cache module continuously stores three-dimensional time series data of temperature, signal and frequency. The fault analysis module compares four types of thresholds: signal distortion, frequency offset, optical path deviation and temperature coupling, distinguishes between high temperature-derived faults and inherent hardware faults, and generates a fault feature dataset with temperature label. The gradient descent algorithm, combined with temperature tags, optimizes the noise reduction threshold, closed-loop step size, and front-end gain to generate a standardized parameter model for general high-temperature operating conditions. This model is then synchronized to other industrial terminals via a CRC check bus. If any parameter, such as high temperature, voltage, power, or bus, exceeds the limit, the protection module will reduce power, lock the frequency, and cut off the laser source in stages, while simultaneously uploading a high-temperature fault alarm message.

[0025] Implementation results: Under continuous high temperature conditions of 180℃, the system's resonant frequency tracking drift is controlled within 5kHz, the high temperature thermal noise filtering efficiency is improved by 35%, the false fault identification rate induced by high temperature is reduced by 90%, and the standardized algorithm model can be directly reused on multiple industrial quantum sensing probes to achieve synchronous self-optimization of multiple devices on site, and adapt to long-term online monitoring of industrial equipment.

[0026] Example 3 High-dynamic vibration vehicle-mounted diamond color center quantum sensing system: This embodiment is designed for vehicle-mounted mobile detection scenarios (rail transit, vehicle-mounted geomagnetic mapping). It is optimized for scenarios involving high-frequency vibration, severe optical path disturbance, and instantaneous high bus load caused by vehicle movement, and enhances the anti-vibration capability of differential acquisition, wavelet dynamic denoising, fast frequency locking, and secure bus transmission performance.

[0027] The dedicated hardware and parameter adjustments are as follows: The quantum sensing differential signal acquisition module adds an optical vibration damping bracket to the two photoelectric conversion units, improves the dynamic differential gain adjustment range of the differential amplifier, and uses an ADC sampling and holding synchronous trigger circuit to suppress sampling timing offset caused by vibration. The wavelet multi-layer decomposition denoising module enables a dynamic layer termination mechanism, automatically increasing the decomposition layer to 6 layers when vibration noise is severe, and increasing the high-frequency hard threshold inverse proportional coefficient to 1.8 to enhance the filtering of high-frequency vibration noise. The resonant frequency real-time tracking control module shortens the closed-loop step adjustment cycle to 0.5ms, the peak frequency locking unit adds multi-frame spectrum averaging filtering to eliminate spectrum glitches caused by vibration, the signal timing cache storage module expands the partition storage capacity to retain 40 minutes of continuous vehicle data and supports high-speed fixed-point batch reading by the host computer, and the bus data security verification and interaction module adds a multi-frame cache retransmission queue to batch cache data packets in case of instantaneous transmission abnormalities to avoid data loss caused by vehicle bumps; The bus rate monitoring threshold of the overload safety protection module has been increased to 20MB / s. A vibration and shock instantaneous power buffer logic has been added to prevent damage to the probe from sudden changes in microwave power under bumps. The directional coupler of the microwave drive module has been upgraded to a high-isolation model to resist electromagnetic noise reflection interference from the vehicle.

[0028] The specific operating logic for in-vehicle scenarios is as follows: Vehicle vibration causes a slight shift in the optical path of the diamond probe. Two vibration-damping photoelectric units synchronously collect fluorescence signals. A differential circuit cancels out the intensity fluctuations of the two synchronous lights caused by vibration in real time. A synchronous ADC latches the sampling timing to eliminate jitter. The original differential signal is sent to a wavelet module. The decomposition depth and noise reduction threshold are dynamically adjusted according to the real-time high-frequency entropy value to filter out broadband noise generated by vehicle vibration. The clean signal after noise reduction is smoothed by multi-frame average FFT to quickly lock the resonant frequency point. Short-cycle closed-loop adjustment corrects the microwave output frequency in real time to suppress frequency jump caused by vibration. The circular cache continuously stores the time-series data of the entire vehicle process. The fault analysis module focuses on identifying vibration-induced optical path offset faults and instantaneous noise distortion faults, generating a high-dynamic vibration-specific fault dataset. The algorithm iteration module optimizes the front-end differential gain, wavelet dynamic threshold, and frequency closed-loop adjustment step size based on vehicle fault samples to adapt to mobile vibration environments. When the vehicle bus experiences instantaneous fluctuations, the 32-bit CRC check identifies erroneous data packets, automatically starts a batch buffer retransmission queue, and continuously pushes alarms to the host computer. If any indicator such as vibration, voltage, microwave power, or bus load is overloaded, the protection module quickly reduces the microwave output power and stabilizes the laser output power to ensure uninterrupted vehicle detection.

[0029] Implementation results: Under dynamic vibration conditions at vehicle speeds of 0-120 km / h, the common-mode interference caused by optical path vibration is suppressed by 91%, the vibration noise signal-to-noise ratio is improved by 32 dB, there is no loss of resonance frequency, and the bus data packet loss rate is less than 0.01%. It can stably complete the mobile mapping of vehicle-mounted geomagnetic fields and underground pipeline magnetic fields.

[0030] Example 4 Multi-probe array distributed quantum sensing extension system This embodiment extends the distributed detection function of the multi-NV probe array based on the architecture of the basic embodiment 1, adapts to the distributed mapping scenario of large-area spatial magnetic field, realizes multi-probe collaborative control by relying on the bus verification and interaction module, and realizes multi-probe joint parameter optimization by the algorithm iteration module.

[0031] Array expansion hardware configuration: The bus data security verification and interaction module expands multiple distributed bus interfaces. Each bus corresponds to a set of independent quantum sensing differential signal acquisition modules and dual NV color center probes. Each acquisition unit is independently configured with wavelet denoising and frequency tracking sub-units. The signal timing buffer storage module divides each probe into an independent storage partition and synchronously stores the timing data of multiple probes. The fault characteristic analysis module adds multi-probe consistency comparison logic to compare the resonant frequency points and signal amplitudes of multiple probes and identify optical path offset and noise anomalies of a single probe in the array. The gradient descent algorithm in the algorithm iterative learning optimization module supports joint fitting of multiple probes and uniformly optimizes the noise reduction threshold, adjustment step size, and acquisition gain of all probes in the array. The overload safety protection module monitors the independent voltage and power of each probe acquisition and microwave drive branch. Overload of a single branch does not affect the normal operation of the other array probes. The microwave drive execution output module uses multiple independent digitally controlled frequency sources to output matching resonant microwaves for each group of NV probes.

[0032] The array collaboration workflow is as follows: Multiple sets of symmetrical diamond dual NV probes synchronously acquire fluorescence signals from multiple points in space. Each independent differential acquisition unit completes photoelectric conversion and differential noise reduction. The ADC outputs multiple raw differential signals. Each wavelet denoising module independently filters out noise in layers and outputs multiple clean sensing signals. Each resonance tracking module locks the corresponding probe NV resonance frequency point and generates independent frequency adjustment commands to control multiple digitally controlled microwave sources. Multiple time-series data are partitioned and stored in a circular buffer. The fault analysis module not only compares thresholds for a single channel, but also compares the signals and frequency consistency of all probes in the array horizontally to identify single-point faults and overall environmental noise faults in the array, generating a multi-dimensional array fault dataset. The algorithm iteration module fits the fault data of all arrays globally and outputs unified optimized parameters to all branch wavelet modules and frequency tracking modules. Each data packet of the distributed bus is independently appended with a 32-bit CRC check. If a single channel transmission is abnormal, only the corresponding branch data is retransmitted, while the rest of the array acquires data normally. If the voltage or microwave power of a single probe branch exceeds the limit, the protection module only cuts off the laser and microwave of that branch, while the other array probes continue to work without interrupting the full-domain magnetic field mapping.

[0033] Implementation results: It supports synchronous distributed measurement of up to 8 diamond NV color center probe arrays, with 100% isolation rate for single-point array faults, and the consistency error of the whole-area magnetic field mapping data is ≤3nT. After the multi-probe joint algorithm optimization, the overall detection signal-to-noise ratio is improved by an average of 25dB, and the stability of continuous mapping of large-area spatial magnetic fields is greatly improved.

[0034] This system relies on nine core modules to form a fully closed-loop automated workflow: signal acquisition, layered noise reduction, frequency tracking, timing storage, fault diagnosis, algorithm self-optimization, security protection, microwave drive, and bus verification. The complete timing workflow operates synchronously and in parallel on three levels: signal flow, control feedback, and security inspection. The specific overall work steps are as follows: Step 1: Synchronous acquisition of differential fluorescence signals from dual NV color centers at the front end: The diamond dual NV color center detection unit is continuously excited by a fixed wavelength laser. The two symmetrically arranged color centers generate fluorescence signals respectively. The two photoelectric conversion sub-units in the quantum sensing differential signal acquisition module receive the two sets of fluorescence one-to-one and convert the light intensity signal into two analog voltage signals. The high-precision differential operational amplifier performs differential calculations on the two analog voltages, directly eliminating common-mode interference caused by optical path loss, laser power fluctuations, and circuit temperature drift. The differential analog signal is sent to the 1MSps analog-to-digital conversion subunit, which completes the analog-to-digital conversion at a millisecond sampling rate and outputs the original differential sensing digital signal. The original data of a single frame is accompanied by a collection timestamp and is synchronously transmitted to the wavelet multi-layer decomposition and denoising processing module.

[0035] Step 2: Wavelet adaptive hierarchical decomposition and noise filtering reconstruction: The wavelet multi-level decomposition denoising module calls the 4th-order Daubechies wavelet basis function to start multi-level wavelet decomposition on the input original differential sensing signal. During the decomposition process, the noise entropy calculation subunit solves the information entropy of each layer of high-frequency wavelet coefficients in real time and compares the calculated entropy value with the preset lower limit threshold. If the entropy value is lower than the threshold, it is determined that there is no effective noise in the current layer, and the hierarchical decomposition is terminated immediately. The maximum number of decomposition layers in the system is forcibly limited to six layers to avoid excessive decomposition and distortion of the effective signal. A hard threshold, inversely proportional to the entropy of the layer, is applied to all high-frequency wavelet coefficients from layers 1 to 6 to zero out the noise coefficients. Low-frequency approximate wavelet coefficients are preserved without any correction. After decomposition and threshold denoising, all high- and low-frequency wavelet coefficients are fed into the inverse wavelet transform unit to reconstruct and output a pure differential sensing signal free from noise interference.

[0036] The pure differential signal is output in two parallel paths: one path is sent to the resonant frequency real-time tracking control module for frequency locking, and the other path is sent to the signal timing buffer storage module to complete timing archiving.

[0037] Step 3: Resonant frequency FFT transformation, peak locking, and closed-loop microwave tuning: After receiving the pure differential signal, the resonant frequency real-time tracking control module converts the time-domain signal to the frequency-domain spectrum using the fast Fourier transform subunit. The peak frequency locking subunit traverses all spectrum amplitudes and locks the NV color center inherent resonant frequency corresponding to the maximum amplitude as the target reference frequency. The frequency deviation trend extrapolation subunit retrieves historical resonant frequency displacement data for N consecutive sampling periods in the buffer, uses a linear extrapolation algorithm to predict the theoretical target frequency for the next period, and uses the predicted frequency as the initial reference value for closed-loop adjustment. The closed-loop step adjustment subunit compares the current actual microwave output frequency with the predicted target frequency, calculates the frequency deviation value, converts it into a standardized frequency adjustment command, and sends it to the microwave drive execution output module.

[0038] Step 4: Time-series data cyclic partitioning and caching archiving: The signal timing buffer storage module adopts a cyclic first-in-first-out (FIFO) hardware architecture. The hardware is divided into three independent partitions: the original signal storage area, the noise reduction signal storage area, and the frequency parameter storage area. The system strictly follows the data writing timing every millisecond, storing the original differential signal with timestamps, the wavelet reconstructed pure signal, and the real-time resonant frequency parameters into the corresponding partitions respectively. When any partition storage space is full, the oldest historical data is automatically overwritten. The hardware ensures that no data is lost during continuous operation for 30 minutes. When the working condition fault feature analysis module initiates a fixed-point read request, the storage module retrieves time-series data from a specified time period and a specified partition, and outputs the time-series data source in batches.

[0039] Step 5: Multi-dimensional threshold comparison, fault identification and dataset generation: The fault characteristic analysis module preloads four types of standard parameters: signal distortion threshold library, frequency offset threshold library, environmental noise amplitude threshold library, and optical path deviation threshold library. It then reads the buffered time-series data frame by frame for parallel comparison. If the signal distortion value exceeds the distortion threshold, a noise abnormality fault is determined. If the difference between the real-time tracking frequency point and the standard resonant frequency point exceeds the offset threshold, a frequency lock-out fault is determined. If the amplitude difference between the two front-end acquired signals exceeds the optical path deviation threshold, it is determined that there is a fault in the detection optical path offset. Each type of fault is matched with a unique fixed numerical code, and the fault occurrence time, fault amplitude, current temperature, microwave power and other auxiliary operating condition information are recorded synchronously. All fault records are uniformly summarized and structured to generate a standardized fault feature dataset, which is then transmitted to the algorithm iterative learning and optimization module.

[0040] Step 6: Gradient descent iterative self-optimization, adaptive update of system parameters: The algorithm iterative learning optimization module loads the structured fault feature dataset and performs global parameter fitting calculations based on the gradient descent optimization algorithm to achieve targeted correction of the system's core parameters according to fault type. Based on noise-abnormal fault samples, the high-frequency coefficient noise reduction hard threshold of the wavelet multi-layer decomposition denoising module is individually adjusted. Based on frequency lockout fault samples, the closed-loop step adjustment step size of the resonant frequency real-time tracking control module is individually corrected. Based on the fault samples of the probe optical path offset, the front-end differential acquisition gain of the quantum sensing differential signal acquisition module is individually corrected; Meanwhile, the cross-device feature mapping subunit within the module maps the local fault dataset to a unified feature vector space, generating a device-independent standardized parameter correction model. The updated noise reduction threshold, closed-loop step size, and acquisition gain parameters are synchronously sent to the wavelet multi-layer decomposition denoising processing module and the resonant frequency real-time tracking control module to complete parameter refresh. After the standardized correction model is verified by the bus data security verification interaction module, it can be exported and directly solidified to other sensing devices of the same model.

[0041] Step 7: Microwave excitation signal generation, power amplification, and anti-reflection output: The microwave drive execution output module receives the frequency adjustment command issued by the resonant frequency real-time tracking control module. The numerically controlled frequency source subunit accurately outputs the sinusoidal excitation signal at the specified frequency point. The fixed gain power amplifier subunit amplifies the sinusoidal signal to the rated operating power of the system. The unidirectional directional coupling output subunit isolates the reverse microwave generated by the reflection of the diamond resonant cavity at the back end, eliminates the frequency drift and power disturbance caused by the reflected signal, and outputs a stable microwave excitation signal without reflection interference to the diamond color center resonant cavity to complete the quantum state control of the NV color center, forming a closed-loop control of "signal acquisition - frequency tracking - microwave excitation".

[0042] Step 8: End-to-end bus data security verification, retransmission, and alarm: The bus data security verification and interaction module centrally manages the data interaction links between all modules in the system. Each data packet transmitted between modules is appended with a 32-bit cyclic redundancy check (CRC) code. The receiving end uses a 32-bit CRC encoding unit to verify the integrity and correctness of the data packets. Verification passed: Data was received normally, and the corresponding calculation / storage / control logic was executed. Verification failed: The device anomaly alarm unit pushes a standardized fault alarm message to the host computer, and at the same time the automatic data retransmission control unit issues a retransmission command, requiring the sending end to retransmit the current complete frame of sensor data; Configuration parameters, control commands, and model files issued by the host computer are also verified by CRC before being distributed to each hardware module to prevent system malfunctions caused by transmission distortion.

[0043] Step 9: Full hardware parallel real-time safety inspection and graded overload protection: The overload safety protection module operates in parallel with all hardware modules of the system, running synchronously and in parallel with steps 1-8 above, continuously performing triple monitoring and periodic self-checks: The three independent voltage acquisition circuits monitor the real-time operating voltage of the signal acquisition, signal processing, and microwave drive modules respectively. The bus transmission rate monitoring subunit counts the effective bytes transmitted on the bus per unit time, and the microwave output power sampling subunit collects the instantaneous microwave output power in real time. The periodic fault injection self-test subunit injects standard disturbance signals into the voltage, bus, and power monitoring units at fixed time intervals, compares the protection response actions with the preset action sequence, and immediately reports an alarm if the self-test is abnormal. If any monitored parameter, such as voltage, bus speed, or microwave power, exceeds the preset upper limit threshold, the system will trigger protection in a fixed hierarchical order: hardware current limiting protection → control command lockout protection → laser detection optical path power failure protection, to prevent hardware overload burnout and probe failure. Any module failure, transmission anomaly, or parameter exceeding limits in the entire machine are simultaneously uploaded to the host computer, enabling visualized monitoring of the fault.

[0044] Step 10: The entire machine operates in a closed-loop cycle. After a single complete signal acquisition, noise reduction, frequency adjustment, storage, diagnosis, and parameter optimization process is completed, the system automatically returns to step 1 and continues to acquire the next millisecond of the diamond dual NV color center differential fluorescence signal, cyclically executing the entire process; When the system receives a shutdown command from the host computer, the overload safety protection module prioritizes cutting off the laser and microwave outputs, and the cache module completes the storage of the current time sequence data before powering off the entire machine, ensuring data integrity and hardware security.

[0045] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A wavelet denoising frequency tracking system for differential signals from diamond color center quantum sensing, characterized in that, It includes a quantum sensing differential signal acquisition module, a wavelet multi-layer decomposition and denoising processing module, a resonant frequency real-time tracking and control module, a signal timing buffer storage module, a working condition fault feature analysis module, an algorithm iterative learning and optimization module, an overload safety protection module, a microwave drive execution output module, and a bus data security verification and interaction module. The quantum sensing differential signal acquisition module is electrically connected to the diamond double NV color center detection unit and outputs the original differential sensing signal to the wavelet multi-layer decomposition and denoising processing module. The wavelet multi-layer decomposition and denoising module outputs a clean differential signal to the resonant frequency real-time tracking control module and the signal timing cache storage module. The resonant frequency real-time tracking control module outputs a frequency adjustment command to the microwave drive execution output module; The signal timing cache storage module provides a timing data source for the operating condition fault feature analysis module; The fault feature analysis module outputs the fault feature dataset to the algorithm iterative learning and optimization module. The algorithm iterative learning and optimization module outputs updated parameters to the wavelet multi-layer decomposition denoising module and the resonant frequency real-time tracking and control module. The overload safety protection module is connected in parallel with all hardware modules; The bus data security verification and interaction module manages the entire module's data interaction link.

2. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The quantum sensing differential signal acquisition module includes two photoelectric conversion subunits, a high-precision differential operational amplifier, and a 1MSps analog-to-digital conversion subunit. The two-way photoelectric conversion subunits receive the fluorescent light signals output by two sets of symmetrically arranged diamond NV color centers one-to-one. The output analog electrical signals are connected to a high-precision differential operational amplifier to complete the difference calculation. The differentially processed analog signals are converted into digital raw differential sensing signals by an analog-to-digital conversion subunit.

3. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The wavelet multi-level decomposition and denoising module has a built-in fourth-order Daubechies wavelet basis function to perform hierarchical decomposition on the original differential sensing signal. The wavelet multi-layer decomposition denoising module is further configured with a noise entropy calculation subunit to calculate the high-frequency coefficient information entropy in real time. When the entropy value is lower than the preset lower limit, the current layer decomposition is terminated and the maximum number of decomposition layers does not exceed six. A fixed hard threshold, inversely proportional to the entropy of that layer, is applied to the high-frequency wavelet coefficients of each layer to remove noise, while the low-frequency approximate wavelet coefficients are fully preserved. All wavelet coefficients are then input into the inverse wavelet transform unit to reconstruct a pure differential sensing signal.

4. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The resonant frequency real-time tracking control module incorporates a fast Fourier transform subunit, a peak frequency locking subunit, and a closed-loop step adjustment subunit. The Fast Fourier Transform subunit performs time-domain to frequency-domain conversion of the pure differential signal; The peak frequency locking subunit locks the inherent resonant frequency corresponding to the maximum spectral amplitude. The closed-loop step adjustment subunit calculates the difference between the current microwave output frequency and the predicted frequency and generates an adjustment command.

5. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The signal timing cache storage module adopts a circular first-in-first-out storage architecture, and the hardware partition is divided into the original signal storage area, the noise reduction signal storage area and the frequency parameter storage area. The system completes a time-series data write every millisecond. When the storage space is full, it automatically overwrites the oldest historical time-series data, retains 30 minutes of continuous complete running data, passively receives the fixed-point read command from the working condition fault characteristic analysis module, and completes the directional data output.

6. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The fault characteristic analysis module for operating conditions pre-stores three types of fixed threshold databases: a signal distortion threshold database, a frequency offset threshold database, and an environmental noise amplitude threshold database. The cached timing data is compared with the corresponding threshold library standard parameters frame by frame. When the signal distortion value exceeds the signal distortion threshold, it is judged as a noise abnormality fault. When the difference between the real-time tracking frequency point and the standard resonant frequency point exceeds the frequency offset threshold, it is judged as a frequency lock-out fault. When the difference between the amplitudes of the two front-end acquired signals exceeds the preset optical path deviation threshold, it is judged as a detection optical path offset fault. Each type of fault is matched with a unique fixed numerical code, and all fault data are uniformly summarized to generate a fault feature dataset.

7. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The algorithm iterative learning optimization module is equipped with a gradient descent optimization algorithm to perform global parameter fitting calculations on the structured fault feature dataset; The noise reduction threshold of wavelet high-frequency coefficients is adjusted independently based on noise anomaly fault data; the frequency tracking closed-loop adjustment step size is adjusted independently based on frequency lock-out fault data; and the front-end differential signal acquisition gain is adjusted independently based on probe optical path offset fault data.

8. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The overload safety protection module is equipped with three independent voltage acquisition circuits, a bus transmission rate monitoring subunit, and a microwave output power sampling subunit. The three voltage acquisition circuits monitor the real-time operating voltage of the signal acquisition module, the signal processing module, and the microwave drive module, respectively. The bus transmission rate monitoring subunit counts the number of valid bytes transmitted on the bus per unit time. The microwave output power sampling subunit collects the instantaneous output power of the microwave signal in real time; The overload safety protection module is further configured with a periodic fault injection self-test subunit, which injects known standard disturbance signals into the voltage acquisition circuit, bus monitoring subunit and power sampling subunit at fixed time intervals, and compares the consistency of the protection response action with the preset action sequence.

9. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The microwave drive execution output module includes a numerically controlled frequency source subunit, a fixed gain power amplifier subunit, and a unidirectional directional coupling output subunit. The numerically controlled frequency source subunit accurately responds to external frequency adjustment commands and outputs a sinusoidal excitation electrical signal at a specified frequency. The fixed-gain power amplifier subunit amplifies the sinusoidal electrical signal to the system's rated operating power. The unidirectional directional coupling output subunit isolates the reverse microwave signal reflected by the rear resonant cavity and outputs a stable microwave excitation signal without reflection interference to the diamond color center resonant cavity.

10. The wavelet denoising frequency tracking system for differential signals of diamond color center quantum sensing according to claim 1, characterized in that, The bus data security verification and interaction module is equipped with a 32-bit cyclic redundancy check encoding unit, a device anomaly alarm unit, and an automatic data retransmission control unit. All data packets exchanged between modules in the system are appended with a 32-bit cyclic redundancy check code. The receiving end verifies the integrity of the data packets through the check encoding unit. When the verification fails, the abnormal alarm unit sends a fixed-format alarm message to the host computer, and at the same time, the automatic data retransmission control unit instructs the data sending end to retransmit the complete sensor data of the current frame.

Images