36 results about "Least mean square algorithm" patented technology

The invention relates to the technical field of wireless communication anti-interference, in particular to a communication transceiving method and system for adaptive frequency hopping filtering, and the method comprises the following steps: S1, collecting and storing channel state information for channel quality evaluation and interference behavior prediction through continuously monitoring the interference level, signal-to-noise ratio and frequency spectrum time sequence characteristics of a current communication channel; according to the invention, dynamic optimization generation of the frequency hopping pattern is realized through interference trend prediction and weighted correction based on the attention mechanism-LSTM network, and channel interference is effectively suppressed by combining interference residual feature prediction and adaptive filtering parameter pre-configuration and real-time optimization of the least mean square algorithm. And high-precision blind synchronization is completed through joint detection and maximum likelihood estimation by using implicit pilot frequency and chaos frequency hopping sequence characteristics, and cooperative improvement of anti-interference performance, synchronization reliability and data recovery accuracy of a communication link in a complex electromagnetic environment is finally realized.

The invention discloses an optical fiber vibration and temperature measurement diagnosis method and system for drainage pipe leakage, and relates to the field of temperature measurement. Determining a temperature characteristic signal based on the vibration signal and a Raman scattering effect; performing wavelet packet transformation processing on the vibration signal to obtain a plurality of equal-broadband signals; eliminating noise in the plurality of equal-broadband signals, the similarity of which to a preset noise feature library is greater than a preset similarity threshold value, by adopting a minimum mean square algorithm to obtain a denoised vibration feature signal; an LSTM-GRU hybrid neural network model is constructed; splicing the output features of the LSTM layer and the GRU layer to obtain fusion features; and matching the fusion feature with a preset fault mode classification table to obtain a fault type of the drainage pipe, adding the fault type to a preset template, and outputting a diagnosis result of the drainage pipe. According to the method, the identification capability of tiny leakage in a complex environment is improved, the false alarm rate is reduced, and the accuracy of leakage detection of the drainage pipe network is improved.

This invention relates to a low-current harmonic direct torque control method for a dual three-phase permanent magnet synchronous motor (PMSM), belonging to the field of PMSM control technology. The method is as follows: S1: Establish mathematical models of the fundamental and harmonic subspaces of the dual three-phase PMSM under vector space decoupling, and clarify the voltage vector distribution characteristics of the two subspaces; S2: In the fundamental subspace, design a duty cycle modulation strategy based on the principle of minimizing the error vector, and an improved deadbeat flux linkage torque controller; S3: In the harmonic subspace, design an online learning harmonic canceller, using the least mean square algorithm to update the weight coefficients in real time; S4: Select the D1 vector set as the effective vector of the harmonic subspace and perform sector division; S5: Add the effective voltage vector action time of the fundamental and harmonic subspaces to generate the final switching sequence, which is output to the dual three-phase inverter to effectively suppress torque ripple and harmonic current of the dual three-phase PMSM.

The application discloses a kind of active control methods for containing unknown harmonic broadband microvibration, it is related to spacecraft microvibration control technical field, including the following steps: using acceleration sensor obtains the residual acceleration response at spacecraft load;Based on residual acceleration response, using PID controller obtains first control force;Using reference signal generator, harmonic frequency is roughly estimated, refined estimation and fine adjustment are carried out, and reference signal is generated;Based on reference signal, using parallel filter-based least mean square algorithm obtains second control force;According to first control force and second control force, total control force is obtained, and is output by executing mechanism, realizes the efficient suppression to containing multiple unknown harmonic broadband microvibration.The application solves the problem that traditional vibration control method is insufficient for harmonic disturbance control ability, and filter-based least mean square algorithm is difficult to obtain the reference signal meeting the requirements under the condition that disturbance characteristics are unknown.

The application relates to an electrical signal parameter estimation method and device, electronic equipment and a storage medium, and relates to the technical field of electrical signal processing. The method comprises the following steps: adopting a variable step size least mean square algorithm, performing cyclic estimation on a fundamental wave signal based on a received signal until the fundamental wave signal meets a preset condition; wherein the fundamental wave signal is a difference value of an error signal and the received signal, the error signal is an output signal of the received signal after a notch filter; a step size factor of the variable step size least mean square algorithm is adjustable, and the step size factor is adjusted based on a square of a time average value of an autocorrelation function of error signals at two adjacent moments; and a signal parameter of the fundamental wave signal is acquired. When the frequency of the electrical signal rapidly changes, good response speed and estimation accuracy can be simultaneously obtained, and the method is more suitable for parameter estimation of a non-stationary signal.

The invention discloses a nonlinear secondary channel modeling method with extremely low additional operand, and belongs to the field of active control of vibration noise. The method aims at solving the problems that in active control over strong response objects such as a vibration source, secondary channel nonlinearity causes insufficient modeling precision, and an existing method is large in additional calculation amount. The method comprises the following steps: firstly, determining a target frequency, selecting N kinds of sinusoidal signals with gradual increase amplitudes to identify a secondary channel in an off-line manner, and obtaining N groups of pulse response sequences; constructing an M-row N-column two-dimensional array storage sequence; during control, a round instruction is built in a digital processor to round the output to obtain an index number, and a corresponding secondary channel coefficient is called from an array so as to provide a precise model for a self-adaptive algorithm. The method is high in modeling precision and low in additional operand, guarantees the real-time performance and control stability of the system, and is suitable for a vibration noise active control scene of a filtering least mean square (FxLMS) algorithm.

The invention discloses a self-adaptive narrowband interference suppression method and device for power line communication, and the method comprises the following steps: S1, receiving a mixed signal, and carrying out the decorrelation processing of the mixed signal, so as to obtain a decorrelation signal; s2, inputting the decorrelated signal into a filter, and estimating the narrowband interference signal at the current moment by the filter according to the filter coefficient at the current moment to obtain a narrowband interference estimation signal; s3, calculating the energy of the decorrelation signal; s4, subtracting the received signal at the current moment from the narrowband interference estimation signal to obtain an error signal; and S5, according to the filter coefficient at the current moment, the error signal and the energy of the decorrelation signal, calculating the filter coefficient at the next moment by using a power normalized least mean square algorithm, so that the algorithm maintains the characteristic of insensitivity to the power change of the input signal, the calculation resource consumption is greatly reduced, and the calculation efficiency is improved. And the feasibility and efficiency of the algorithm in practical application are improved.

The invention relates to the technical field of signal processing, and particularly discloses an adaptive graph signal temperature online reconstruction method based on a maximum dustpan line criterion, which comprises the following steps: taking a meteorological station as a node of a graph, taking temperature as a signal value of the node on the graph, and constructing a graph signal model according to an observation signal influenced by impulse noise; based on the graph signal model, establishing an optimization model by using a maximum dustpan line function criterion and a minimum mean square algorithm and using an L2 norm and an L1 norm; constructing an MGLMS robust algorithm based on the optimization model, and converting the MGLMS robust algorithm into a time domain update equation by using inverse Fourier transform of a graph; and performing on-line reconstruction on the signal values of the nodes on the graph by using the time domain update equation to obtain reconstructed graph signals, thereby obtaining the temperature on the unknown node. According to the adaptive graph signal temperature online reconstruction method based on the maximum dustpan line criterion, data reconstruction and processing of a GSP in an impulse noise environment are improved.

The invention discloses a feed-forward equalizer manufacturing optimization method for a clock data recovery circuit, which relates to the field of high-speed serial communication, and comprises the following steps: based on a short-tap feed-forward equalizer, setting a tap coefficient, and outputting a normalized output discrete signal amplitude; taking the normalized output discrete signal amplitude of the short-tap feed-forward equalizer as an input signal in a constant modulus algorithm updating process, correcting a tap coefficient based on a constant modulus criterion, and optimizing the short-tap feed-forward equalizer under the constant modulus algorithm; based on a short-tap feed-forward equalizer under an optimized constant modulus algorithm, a normalized expected signal amplitude judged by a receiving end judgment device and an instantaneous error between a normalized output discrete signal amplitude and the normalized expected signal amplitude are calculated, and a tap coefficient is further corrected through a minimum mean square algorithm; and optimizing the short-tap feed-forward equalizer under the minimum mean square algorithm.

The application discloses a high-precision torque control method for a permanent magnet motor of a new energy vehicle, constructs a neural network torque observer by taking motor voltage, current, temperature and electric angular velocity as input, performs online parameter identification on the motor by introducing historical data of the motor under different working conditions, and calculates partial derivative information of torque to current by using the identified parameters. The application substitutes the partial derivative information of torque to current and the observed torque of the torque observer into a Taylor formula of the historical data to determine whether the network output is correct, and adjusts the network interior by using a least mean square algorithm, which improves the torque observation precision of the network, solves the shortcoming that the fixed parameters in the traditional network cannot adapt to time-varying working conditions, improves the accuracy and robustness of the network, and is suitable for working conditions of parameter change caused by motor temperature change, torque control precision reduction and MTPA trajectory deviation.

The invention provides a mechanical arm visual servo control method and system based on a self-learning disturbance observer, and relates to the technical field of robot control. Extracting an image feature vector; performing state estimation through extended Kalman filtering to obtain a filtered image feature state; carrying out online learning through a kernel least mean square algorithm to obtain a disturbance estimation value; generating a nominal control sequence and a nominal state trajectory based on the undisturbed ideal model; and generating a final control instruction according to the deviation and the disturbance estimation value and driving the robotic arm to move. According to the method, the extended Kalman filtering and the kernel minimum mean square algorithm are combined to construct the composite disturbance observer, adaptive online learning and high-precision estimation of unknown disturbance are realized, and a double-layer robust control structure is formed in combination with tubular model predictive control. And the control precision and robust stability of the visual servo system of the mechanical arm in a complex disturbance environment are remarkably improved.

The invention relates to a fault judgment method and system for a wind turbine generator. The method comprises the steps that core electrical parameters and auxiliary environment parameters in the wind turbine generator are collected to serve as an initial data set; preprocessing data in the initial data set by adopting a data abnormal value judgment and data alignment method to obtain a target data set; performing noise reduction on the target data set by adopting an improved minimum mean square algorithm to obtain noise-reduced data; a time-frequency domain feature extraction algorithm is adopted to extract time-frequency domain features of the noise reduction data, and aligned time-frequency features are obtained; a dynamic weight generator based on a long short-term memory network carries out fault discrimination on the time-frequency characteristics; and performing fault judgment processing based on the overall feature data of the wind turbine generator to obtain a predicted fault type corresponding to a power grid where the target wind turbine generator is located. According to the method, the influence of factors such as data acquisition, noise reduction processing and time domain feature fusion is fully considered, the reliability degree of the result is determined, and the effect of accurately reflecting the fault state of the wind turbine generator is achieved.

Wander magnitude measurement by inputting a first input wander frequency into a timing phase locked loop circuit to produce a first output wander frequency, determining two parameters based on the first input wander frequency and the first output wander frequency, and using a least mean square (LMS) algorithm to estimate a wander transfer function based on the two parameters.

Aiming at the problem of signal nonlinear distortion caused by nonlinear characteristics of a power amplifier in satellite communication, the invention provides a generalized linear dual-mode switching blind equalization algorithm based on a third-order truncated Volterra structure, and the technical scheme and the application effect are expanded according to the following steps S1 to S3: S1: initializing parameters of a generalized linear blind equalizer based on the third-order Volterra structure; s2, taking a modified constant modulus algorithm and a decision-oriented least mean square algorithm in the linear blind equalization field as theoretical basis, combining signal processing characteristics of a third-order truncated Volterra nonlinear model, deriving to obtain two generalized linear equalizer tap updating formulas suitable for the nonlinear model, and providing algorithm support for a dual-mode equalization mechanism; s3, adopting a staged adaptive equalization strategy to construct a switching mechanism: in an equalization initial stage, using a generalized linear correction constant modulus algorithm of a nonlinear channel, and utilizing a rapid convergence characteristic of the algorithm to realize rapid reduction of a mean square error; and when the equalization process reaches a preset judgment condition, automatically switching to a generalized linear judgment guide algorithm of a nonlinear channel, further reducing the steady-state error through judgment feedback, and considering both the convergence speed and the equalization precision.

The invention relates to the technical field of voice processing, in particular to a voice signal enhancement method and system, and the method comprises the steps: obtaining a real-time voice signal and a historical voice signal of a target region, and carrying out the filtering of the real-time voice signal of the target region through an improved least mean square algorithm, so as to enhance the voice quality; a concept of self-adaptive initial step length is introduced, and the step length is dynamically adjusted according to the maximum characteristic value of an autocorrelation matrix of a real-time signal and the step length adjustment degree so as to adapt to the change of a voice signal; the step length adjustment reflects the convergence trend of the voice signal filtering process; the difference degree is used for quantifying the difference between the real-time voice signal and the historical voice signal, the characteristics of the voice signal can be effectively evaluated, and the filtering process is dynamically adjusted. According to the invention, the problem that the existing algorithm is difficult to realize ideal convergence efficiency and filtering effect in different scenes is solved.

Wander magnitude measurement by inputting a first input wander frequency into a timing phase locked loop circuit to produce a first output wander frequency, determining two parameters based on the first input wander frequency and the first output wander frequency, and using a least mean square (LMS) algorithm to estimate a wander transfer function based on the two parameters.

The application provides a spacecraft on-orbit cooperative positioning method and system, and relates to the field of spacecraft navigation, and comprises the following steps: a cooperative positioning random dynamic observation model is established; a distributed differential privacy adaptive estimation algorithm is constructed; interference noise is added to the parameter estimation value of each observation station at the time point, so that the parameter estimation value with noise is obtained and is transmitted to all neighbor observation stations; the observation station i utilizes the parameter estimation value at the time point, the parameter estimation value with noise of the neighbor observation station j at the time point k, the random regression vector of the observation station at the time point, and the observation vector, and combines the distributed least square algorithm to calculate the local parameter estimation value at the next time point. The application is used for joint positioning of multiple observation stations or multiple sensor nodes, realizes high-precision position estimation while protecting the privacy of sensitive data, and effectively guarantees the high safety, high precision and real-time requirement of the on-orbit autonomous positioning and navigation task.

The invention discloses a pipeline successive approximation analog-to-digital converter calibration method, device and equipment, relates to the technical field of analog-to-digital converters, and aims to solve the problem that calibration is inaccurate due to the fact that large residual error swing is introduced when pseudo-random sequence injection is carried out in an existing Piped SAR ADC calibration method. The method comprises the steps of obtaining a to-be-compared signal; successive approximation comparison is carried out based on the to-be-compared signal and the two comparators, a jitter injection flag bit corresponding to each cycle is determined, pseudo-random sequence injection is carried out according to the jitter injection flag bit, and residual error voltage is obtained; determining a finely quantized digital code based on the residual voltage; carrying out iterative calculation on the bit weight by adopting a minimum mean square algorithm until a convergence condition is met, and obtaining a target bit weight; and performing calibration based on the target bit weight. The streamline successive approximation analog-to-digital converter calibration method provided by the invention is used for reducing residual error swing caused by pseudo-random sequence injection and improving calibration accuracy.

The application provides a kind of mechanical arm visual servo control method and system based on self-learning disturbance observer, it is related to robot control technical field, the method comprises: obtaining real-time image data;Extract image feature vector;State estimation is carried out by extended Kalman filtering, and the filtered image feature state is obtained;Disturbance estimation value is obtained by kernel least mean square algorithm online learning;Based on the ideal model without disturbance, nominal control sequence and nominal state trajectory are generated;According to the deviation and disturbance estimation value, the final control command is generated and the robot arm is driven to move.The application combines extended Kalman filtering and kernel least mean square algorithm to construct a composite disturbance observer, realizes adaptive online learning and high-precision estimation of unknown disturbance, and forms a double-layer robust control structure combined with tube model predictive control, which significantly improves the control accuracy and robust stability of the mechanical arm visual servo system in complex disturbance environment.

The application provides a precipitation estimation method, system, device and storage medium applied in low-altitude economic services, and the method comprises the following steps: acquiring dual-polarization radar echo data and environmental parameters in a low-altitude atmosphere layer and taking the data and the parameters as inputs of a least mean square algorithm, combining a fuzzy logic control algorithm to dynamically adjust weight coefficients of an adaptive filter, using the adaptive filter with the adjusted weight to suppress noise of the dual-polarization radar echo data, and obtaining filtered radar echo data; extracting key precipitation characteristic parameters from the filtered radar echo data and calculating radar original precipitation estimation values; acquiring observation data of a ground meteorological station and performing fusion processing to generate regional precipitation reference data; using the regional precipitation reference data, determining correction coefficients by using a regularized linear regression method, correcting the radar original precipitation estimation values by using the correction coefficients, and outputting final precipitation estimation values. The application effectively improves the precipitation estimation precision in a low-altitude scene.

The invention provides a spacecraft on-orbit cooperative positioning method and system, and relates to the field of spacecraft navigation, and the method comprises the steps: building a cooperative positioning random dynamic observation model; constructing a distributed differential privacy adaptive estimation algorithm: adding interference noise into a parameter estimation value of each observation station at a moment to obtain a parameter estimation value with noise, and transmitting the parameter estimation value with noise to all neighbor observation stations; and the observation station i utilizes the parameter estimation value at the moment, the noise estimation value of the neighbor observation station j at the moment k, and the random regression vector and the observation vector of the observation station at the moment to calculate and obtain a local parameter estimation value at the next moment in combination with a distributed least mean square algorithm. The method is used for combined positioning of multiple observation stations or multiple sensor nodes, high-precision position estimation is achieved while sensitive data privacy is protected, and the requirements for high safety, high precision and real-time performance of an on-orbit autonomous positioning navigation task are effectively guaranteed.

The invention provides an MMC submodule voltage-sharing method and device based on a parameter updating voltage prediction model, and the method comprises the steps: measuring the bridge arm current of each phase of upper and lower bridge arms of an MMC and the submodule capacitor voltage of each half-bridge submodule in real time, and carrying out the online updating of a parameter vector of a voltage prediction model based on a minimum mean square algorithm; inputting the bridge arm current, the capacitor voltage of the sub-module and the parameter vector into a voltage prediction model, and calculating to obtain a capacitor voltage prediction value of the sub-module at the next moment; and determining the number of newly added input or cut-off sub-modules at the next moment, and selecting the sub-module with the maximum or minimum capacitor voltage predicted value to perform input or cut-off operation according to the current direction of the bridge arm. According to the method, the parameters of the voltage prediction model are updated on line through the minimum mean square algorithm, the sub-module capacitor voltage prediction precision is improved, the influence of parameter mismatch on the voltage-sharing strategy is effectively eliminated, the steady-state voltage-sharing degree is further improved, and the capacitor voltage standard deviation and the voltage ripple are reduced.

The invention relates to a low-noise double-path digital auxiliary phase-locked loop circuit for a large-area-array image sensor, and belongs to the technical field of integrated circuit clock circuits. The problems that a traditional charge pump phase-locked loop cannot achieve fractional frequency division, the resolution ratio is low, the frequency modulation range is limited, the locking speed is low, and frequency control is inaccurate are solved. According to the technical scheme, a double-path structure is included, a proportional path adopts a third-order passive filter to optimize dynamic response and suppress in-band phase noise, and an integral path eliminates charge pump nonlinearity to lock frequency; the digital-to-time converter compensates the quantization error of the sigma modulator in a feedback path; and the DTC calibration module dynamically adjusts the gain through a symbol least mean square algorithm. The method has the technical effects that the phase noise is obviously reduced, the output frequency precision is improved, the environmental adaptability is enhanced, the system stability and the locking speed are optimized, and wide-range application such as a large-area-array image sensor is supported.

The invention discloses a thin-wall structure vibration control method based on an inerter type nonlinear energy trap, and relates to the field of civil aviation aircraft thin-wall structure dynamics. Aiming at the problems of complex broadband vibration characteristics, contradiction between light weight and vibration control, insufficient multi-working-condition adaptability and the like of an aerospace thin-wall structure, the invention provides a distributed inerter type nonlinear energy trap system, a coupling dynamic model is established, and related parameters and connection modes are determined to realize broadband fine control; adopting a harmonic balance method to solve steady-state response, and combining short-time Fourier transform, a subspace identification principle and a minimum mean square algorithm to realize vibration response frequency tracking and parameter adaptive adjustment; and with minimization of system response as a target, key parameters are optimized by using a particle swarm method, and a multi-working-condition adaptive control algorithm is developed. According to the invention, the nonlinear rigidity is dynamically adjusted to match the excitation frequency, and efficient, broadband and lightweight vibration control under multiple working conditions is realized.

The application relates to a co-sited interference cancellation method for integrated sensing and communication, and is characterized by comprising the following steps: step 1, processing a transmission signal x(t) after radio frequency modulation of a transmission end of an integrated sensing and communication base station; and step 2, reconstructing interference by constructing a multi-tap circuit at a local transceiver end of the integrated sensing and communication base station. The co-sited interference cancellation method for integrated sensing and communication has the superior technical effect that a variable step size least mean square algorithm is adopted to effectively solve the restrictive relationship between a step size factor and a steady-state error in the case of a fixed step size, co-sited interference can be cancelled in an integrated sensing and communication system, and sensing performance is improved; interference cancellation in a radio frequency domain is designed, the complexity of a hardware structure is lower; and a gradient descent method is used to optimize a root mean square value of a signal, so that an optimized value of a downlink communication system is realized.

The application relates to the technical field of voice processing, in particular to a voice signal enhancement method and system, which comprises the following steps: obtaining real-time voice signals and historical voice signals of a target area; filtering the real-time voice signals of the target area by using an improved least mean square algorithm to enhance voice quality; introducing the concept of an adaptive initial step, dynamically adjusting the step according to the maximum eigenvalue of the autocorrelation matrix of the real-time signals and the step adjustment degree to adapt to the change of the voice signals; the step adjustment degree reflects the convergence trend of the voice signal filtering process; the difference degree is used for quantifying the difference between the real-time voice signals and the historical voice signals, which can effectively evaluate the characteristics of the voice signals and dynamically adjust the filtering process. The application solves the problem that the existing algorithm is difficult to realize ideal convergence efficiency and filtering effect under different scenes.

The invention discloses an efficient feed-forward narrowband active noise control system and method, and belongs to the technical field of active noise control. The system comprises a reference signal synthesis subsystem and a feed-forward narrowband secondary sound source synthesis subsystem, the reference signal synthesis subsystem comprises a cosine signal synthesizer and an adder, and provides a single reference input signal for the feed-forward narrowband secondary sound source synthesis subsystem; the feed-forward narrowband secondary sound source synthesis subsystem comprises a low-order control filter and the single filtering-X minimum mean square algorithm module, and is used for synthesizing a narrowband secondary sound source. According to the invention, the system can effectively deal with narrow-band target noise with harmonic wave or non-harmonic wave; a low-order control filter and a single-channel filtering-X minimum mean square algorithm module are adopted, so that the structure and calculation complexity of the system are effectively reduced, the convergence and noise reduction performance of the system can be improved, and the miniaturization and low-cost noise reduction occasion application of the feed-forward narrowband active noise control system is further widened.

The application provides a precipitation estimation method, system, device and storage medium applied in low-altitude economic services, and the method comprises the following steps: acquiring dual-polarization radar echo data and environmental parameters in a low-altitude atmosphere layer and taking the data and the parameters as inputs of a least mean square algorithm, combining a fuzzy logic control algorithm to dynamically adjust weight coefficients of an adaptive filter, using the adaptive filter with the adjusted weight to suppress noise of the dual-polarization radar echo data, and obtaining filtered radar echo data; extracting key precipitation characteristic parameters from the filtered radar echo data and calculating radar original precipitation estimation values; acquiring observation data of a ground meteorological station and performing fusion processing to generate regional precipitation reference data; using the regional precipitation reference data, determining correction coefficients by using a regularized linear regression method, correcting the radar original precipitation estimation values by using the correction coefficients, and outputting final precipitation estimation values. The application effectively improves the precipitation estimation precision in a low-altitude scene.

A first calibrator detects mismatches in the capacitor array and updates a look-up table (LUT) with calibrated weights that are copied to a positive LUT and a negative LUT, which are then adjusted for non-linear errors by a second calibrator using a least mean square (LMS) algorithm. The binary code in a successive approximation register (SAR) is complemented to generate a complement with a sign bit. When the sign bit is positive, entries from the positive LUT with complemented data bits = 1 are read and summed, a first offset is added, and the sum is normalized to obtain a corrected code. When the sign bit is negative, entries from the negative LUT with complemented data bits = 0 are read and summed, a second offset is added, and the sum is normalized to obtain a corrected code. A multivariate stochastic gradient descent generates polynomial coefficients that further correct the corrected code.

The invention relates to a dynamic noise balance LMS (Least Mean Square) calibration method applied to an SAR (Synthetic Aperture Radar) ADC (Analog to Digital Converter) chip, which comprises the following steps of: configuring a capacitor array comprising a weight bit capacitor, a dynamic noise balance capacitor, a bridging capacitor and a disturbance injection capacitor, and injecting disturbance voltage by controlling the switching state of the disturbance injection capacitor, after sampling and holding, a weight bit digital code value is quantized by a weight bit capacitor. After the weight bit capacitor is quantized, the residual voltage is quantized for multiple times by the dynamic noise balance capacitor. And calculating a quantization error based on the related code value and the disturbance voltage theoretical offset, and iteratively updating the capacitance weight value to error convergence by adopting a minimum mean square algorithm. According to the method, dynamic noise balance and LMS calibration are combined, error changes can be tracked in real time, on-chip automatic calibration is achieved, on-chip noise and capacitor mismatch influences are effectively improved, the resolution of an SAR ADC chip is improved, and the application cost is reduced.