Analog-to-digital converter dynamic calibration method and system based on statistical form constraint

Abstract

The invention discloses an analog-to-digital converter dynamic calibration method and system based on statistical form constraint. According to the method, the statistical form subspace of the integral nonlinear error of the analog-to-digital converter is constructed, and the original high-dimensional and complex nonlinear error change is compressed into a small number of identifiable statistical forms, so that effective modeling and reconstruction of the complex nonlinear error are realized; on the basis, a consistency verification mechanism is introduced in the affine aging compensation process, untrusted deformation updating caused by noise, abnormity or faults can be effectively restrained while long-term aging drifting of a device is compensated, and the calibration reliability and stability of the system in the full life cycle operation process are remarkably improved; in addition, real-time calibration output is carried out by adopting a full-code-domain calibration mapping table and combining a direct addressing mode of a tight coupling memory, the operation calculation complexity is remarkably reduced while high precision is guaranteed, and zero-delay calibration processing is achieved.

Description

Technical Field

[0001] This invention relates to the field of precision electronic measurement and signal processing technology, specifically to a dynamic calibration method and system for the entire lifecycle of an analog-to-digital converter based on statistical morphological subspace constraints and consistency verification. Background Technology

[0002] With the rapid development of industrial automation, power monitoring, and medical electronics, the accuracy of data acquisition systems (DAQ) directly determines the upper limit of equipment performance. As a core component, the analog-to-digital converter (ADC) is inevitably subject to drift in acquisition accuracy due to changes in ambient temperature and device aging during long-term operation. Traditional maintenance methods often rely on replacing modules or returning the device to the factory for recalibration, which not only interrupts the production process but also incurs extremely high maintenance costs and application inconvenience.

[0003] Most existing ADC online calibration techniques are based on linear model assumptions, compensating only for gain (slope) and zero-point (intercept) drift. However, in high-precision applications, the physical mismatch of the capacitor / resistor array inside the ADC chip introduces inherent integral nonlinearity (INL), causing the transmission characteristic curve to exhibit "S-shaped," "W-shaped," or irregular high-order distortions that vary with the code value. Traditional linear compensation methods essentially only rotate and translate the transmission curve in coordinate space, failing to fundamentally eliminate this nonlinear waveform distortion. This results in significant residuals after calibration, making it difficult to meet the stringent requirements of high-precision measurements. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a dynamic calibration method and system for analog-to-digital converters (ADCs) based on statistical morphological constraints. By modeling the statistical morphology of the integral nonlinearity error of the ADC and performing constrained dynamic updates during operation in conjunction with temperature information and aging characteristics, the invention achieves long-term high-precision operation of the ADC under multiple temperature and aging stages while reducing production calibration costs.

[0005] The technical solution of the present invention is as follows:

[0006] In a first aspect, embodiments of this application provide a dynamic calibration method for analog-to-digital converters based on statistical morphological constraints, comprising the following steps:

[0007] Construction of statistical morphological space: Obtain integral nonlinearity (INL) characteristic curve data of several analog-to-digital converters of the same model under multiple temperature conditions, perform statistical decomposition on the integral nonlinearity characteristic curve data, extract several mutually orthogonal dominant error morphological basis functions, and construct a statistical morphological subspace to characterize the inherent nonlinear characteristics of the analog-to-digital converter of this model.

[0008] Nonlinear reference reconstruction at current temperature: During online operation, the current temperature value of the analog-to-digital converter is acquired in real time. Based on the constructed statistical morphological subspace and combined with the modal weighting coefficient, the dominant error morphological basis functions are weighted and combined to reconstruct the nonlinear reference curve under the current temperature condition.

[0009] Sparse aging feature anchor point sampling: Based on a preset sparse sampling strategy, the actual error measurement value of the analog-to-digital converter is obtained at at least two key code value points through a reference signal or a known reference input, which serve as aging feature anchor points characterizing the aging state of the device.

[0010] Constrained Affine Update: Based on the deviation relationship between the aging feature anchor point and the nonlinear reference curve, the affine perturbation parameter used to characterize the device aging drift is calculated, and the nonlinear reference curve is updated affinely according to the affine perturbation parameter to generate a candidate calibration curve.

[0011] Subspace consistency verification: Project the candidate calibration curve into the statistical morphology subspace and calculate its projection residual energy; when the projection residual energy is equal to or lower than a preset confidence threshold, confirm that the candidate calibration curve is consistent with the statistical morphology prior and update the calibration mapping relationship to generate a new calibration mapping table; when the projection residual energy is higher than the confidence threshold, trigger the anomaly handling mechanism, maintain the calibration mapping relationship of the previous moment, and do not update the calibration mapping table.

[0012] Real-time calibration output: The calibration mapping table that has passed the consistency check is stored. When the analog-to-digital converter outputs the real-time raw code value, the raw code value is used as the memory address offset to directly address the calibration mapping table, obtain the corresponding error calibration data, and output a high-precision voltage value after error calibration data compensation.

[0013] In one possible implementation, the statistical morphological subspace is obtained by performing principal component analysis (PCA) or singular value decomposition (SVD) on the integral nonlinearity (INL) characteristic curve data, and the subspace is selected based on the cumulative contribution rate exceeding a preset proportion. Each eigenvector serves as a dominant error morphology basis function. The statistical morphology subspace is spanned by a linear combination of the dominant error morphology basis functions and is used to characterize the inherent nonlinear error morphology range of this type of analog-to-digital converter.

[0014] In one possible implementation, the modal weighting coefficients are modeled as a quadratic polynomial function of temperature, the coefficients of which are determined and stored in memory during the construction of the statistical morphological space.

[0015] In one possible implementation, the preset sparse sampling strategy is based on a sampling value function, which is used to evaluate the effective information contribution of different code values ​​to the estimation of affine perturbation parameters. The sparse sampling strategy evaluates and filters candidate code values ​​across the entire code domain according to the sampling value function, prioritizing the selection of at least two code values ​​that contribute the most to the identifiability of the affine perturbation parameters, and generating a non-uniform sampling index table to guide the analog-to-digital converter to perform measurements only at the sparse code values ​​indicated by the non-uniform sampling index table.

[0016] In one possible implementation, the affine perturbation parameter is used to characterize the overall aging drift characteristics of the digital-to-analog converter during long-term operation, including at least the gain drift coefficient. and bias drift coefficient The gain drift coefficient mentioned above and bias drift coefficient The calculation method is as follows:

[0017]

[0018]

[0019] in, Large value points in aging characteristic anchor points, The minimum value point in the aging characteristic anchor point,

[0020] These are the measured output values ​​of the large-value points in the aging characteristic anchor points. This represents the measured output value of the small value point in the aging characteristic anchor point.

[0021] The process of generating candidate calibration curves follows the following model:

[0022]

[0023] in, For calibration curves, It is a nonlinear reference curve.

[0024] In one possible implementation, the projection residual energy is defined as the square of the Euclidean distance between the candidate calibration curve and its orthogonal projection in the statistical morphological subspace; the anomaly handling mechanism is to reject the current update and maintain the calibration mapping relationship of the previous time step.

[0025] In one possible implementation, the calibration mapping is stored in memory as a lookup table (LUT), and the lookup table is only atomically updated after a consistency check.

[0026] In one possible implementation, the confidence threshold is determined by analyzing the morphological deviation distribution statistical characteristics of a large number of ADCs of the same model during normal aging, and is used to distinguish between normal aging deformation and abnormal deformation caused by circuit faults.

[0027] Secondly, embodiments of this application also provide a dynamic calibration system for an analog-to-digital converter based on statistical morphological constraints, including a non-volatile memory, analog front-end hardware, and a processor unit.

[0028] The non-volatile memory is used to store statistical morphological subspace information obtained by the processor unit during offline analysis, as well as basic parameters to support the processor unit in performing nonlinear benchmark reconstruction and calibration operations during online operation.

[0029] The analog front-end hardware is used to provide the processor unit with real-time temperature information and sampling data from the analog-to-digital converter (ADC). It includes a temperature sensor, a reference source module, and the ADC itself. The temperature sensor is mounted on the ADC and is used to acquire real-time temperature information from the ADC and provide it to the processor unit. The reference source module provides the ADC with a reference signal or reference input for sampling sparse aging feature anchor points. The ADC itself acquires aging feature anchor points according to a preset sparse sampling strategy using the reference signal or reference input provided by the reference source module. The ADC transmits the aging feature anchor points and the sampling data acquired in real-time to the processor unit.

[0030] The processor unit processes real-time temperature information and sampled data acquired by the analog front-end hardware, generates corresponding calibration mapping relationships, and outputs calibrated high-precision voltage values. It includes a statistical morphology analysis module, a nonlinear model reconstruction engine, an affine transformation update module, a consistency verification module, and a direct addressing compensation interface. Specifically, the statistical morphology analysis module constructs and analyzes statistical morphology subspace information offline; the nonlinear model reconstruction engine reconstructs the nonlinear baseline curve at the current temperature based on the dominant error morphology basis function set in the non-volatile memory and real-time temperature information; the affine transformation update module calculates affine perturbation parameters based on aging feature anchor point data and performs affine updates on the nonlinear baseline curve accordingly; the consistency verification module performs consistency verification of the statistical morphology subspace and updates the calibration mapping table; and the direct addressing compensation interface performs direct addressing compensation on the real-time raw code value output by the analog-to-digital converter after the calibration mapping table update, outputting a calibrated high-precision voltage value.

[0031] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory;

[0032] The memory is used to store computer programs.

[0033] When the processor executes the program stored in the memory, it implements any of the analog-to-digital converter dynamic calibration methods described in this application.

[0034] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the analog-to-digital converter dynamic calibration methods described in this application.

[0035] Fifthly, embodiments of this application provide a computer program product containing instructions that, when run on a computer, cause the computer to execute any of the analog-to-digital converter dynamic calibration methods described in this application.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] This invention constructs a statistical morphological subspace for the integral nonlinearity error of an analog-to-digital converter, compressing the originally high-dimensional and complex nonlinear error changes into a small number of identifiable statistical forms, thus achieving effective modeling and reconstruction of complex nonlinear errors. Based on this, a consistency verification mechanism is introduced during the affine aging compensation process, which can effectively suppress unreliable deformation updates caused by noise, anomalies, or faults while compensating for long-term aging drift of the device, significantly improving the calibration reliability and stability of the system throughout its entire lifecycle. Furthermore, this invention employs a full-code-domain calibration mapping table combined with a direct addressing method using tightly coupled memory for real-time calibration output, significantly reducing computational complexity while ensuring high accuracy, and achieving zero-latency calibration processing. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other embodiments can be obtained based on these drawings.

[0039] Figure 1 This is a schematic diagram of the overall process of the method of the present invention.

[0040] Figure 2 This is a block diagram illustrating the system architecture and data flow of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art based on this application are within the scope of protection of this application.

[0042] like Figure 1 As shown in the figure, this embodiment proposes a dynamic calibration method for the entire lifecycle of an analog-to-digital converter (ADC) based on statistical morphological subspace constraints and consistency verification. This method aims to address the nonlinear error drift problem caused by temperature variations and device aging during long-term operation of the ADC. The method mainly consists of an offline feature extraction stage (S1) and an online operation and full lifecycle calibration stage (S2).

[0043] Step S1: Construct a statistical morphological subspace that characterizes the inherent nonlinear features of the ADC.

[0044] Specifically, this embodiment selects 10 ADC chips of the same model as samples, and sets 10 test temperature points in a constant temperature chamber, with an interval of 5℃ between the test temperature points and a temperature range of -5℃ to 45℃. At each temperature point, the integral nonlinearity (INL) characteristic curve data of the ADC is measured using a high-precision digital multimeter or a standard source. After acquiring a large amount of INL characteristic curve data, it is delinearized, retaining only the nonlinear components. Subsequently, principal component analysis (PCA) algorithm is used to statistically decompose the delinearized INL data matrix. Based on the eigenvalues ​​corresponding to the dominant components, the components with a cumulative contribution rate exceeding a preset proportion are extracted. Using 10 eigenvectors as the dominant error morphological basis functions, we obtain the set of dominant error morphological basis functions. The dominant error morphological basis functions are normalized according to a preset code value order and a basis function matrix is ​​constructed. The column space of the basis function matrix is ​​defined as a statistical morphological subspace characterizing the inherent nonlinear characteristics of this type of analog-to-digital converter. Based on this statistical morphological subspace, nonlinear error data under multiple temperature conditions are projected and reconstructed. Simultaneously, a mapping relationship between temperature and modal weighting coefficients is established, modeling the modal weighting coefficients as a quadratic polynomial function of temperature.

[0045]

[0046] in, Represents the first in the statistical morphological subspace One modality, ; , , These are the temperature correlation coefficients determined through offline fitting.

[0047] Step S2: Perform dynamic calibration periodically, which includes the following sub-steps:

[0048] Step S201: Based on the offline stored temperature correlation coefficient and the current temperature value Calculate the mode weight coefficients of each morphological basis function at the current temperature. Based on this, using the dominant error morphological basis function as the basis of the statistical morphological subspace, and performing a linear weighted combination of the dominant error morphological basis function according to the modal weighting coefficients, a nonlinear reference curve at the current temperature is reconstructed within the statistical morphological subspace. :

[0049]

[0050] in, This is the original digital code value output by the analog-to-digital converter, and its value range is the full code domain of the analog-to-digital converter.

[0051] Step S202: Sampling sparse aging feature anchor points based on a preset sparse sampling strategy.

[0052] To capture the specific aging processes that occur in individual devices over time, this embodiment does not perform full code scanning but instead employs a sparse sampling strategy. This sparse sampling strategy evaluates different code values ​​(or combinations of code values) using a sampling value function, measuring their effective information contribution to the estimation of aging affine parameters.

[0053] Based on the evaluation results of the sampling value function, the two key code value points that contribute the most to the identifiability of the aging affine parameter are selected from the full code domain to generate a non-uniform sampling index table.

[0054] In one possible implementation, the sampling value function is constructed based on the amplitude discrimination capability of a nonlinear reference curve at different code value points at the current temperature, and its expression is, for example, as follows:

[0055]

[0056] in, , The outputs of the analog-to-digital converters are respectively the first The and the first One original numeric code value, and .

[0057] Using a reference signal or a known reference input, only at these key code value locations (e.g.) , (etc.) to obtain the actual error measurements of the ADC, these measurements are called "aging feature anchors".

[0058] Step S203: Restricted Affine Update

[0059] Based on the measured aging characteristic anchor points and the nonlinear reference curve reconstructed in step S201 Based on the numerical differences at corresponding code values, an affine matching relationship is established to describe the overall trend of the nonlinear reference curve under aging. Based on this, affine perturbation parameters characterizing device aging drift are estimated, including the gain drift coefficient. and bias drift coefficient The aforementioned gain drift coefficient and bias drift coefficient The calculation method is as follows:

[0060]

[0061]

[0062] in, Large value points in aging characteristic anchor points, The minimum value point in the aging characteristic anchor point,

[0063] These are the measured output values ​​of the large-value points in the aging characteristic anchor points. This represents the measured output value of the small value point in the aging characteristic anchor point.

[0064] Based on the affine perturbation parameters, the nonlinear reference curve is updated affinely to generate candidate calibration curves. The calculation model is as follows:

[0065]

[0066] in, For calibration curves, It is a nonlinear reference curve.

[0067] Step S204: Subspace Consistency Verification

[0068] To prevent erroneous calibration due to sampling noise, circuit failure, or external interference, a consistency verification mechanism is introduced. The candidate calibration curves generated in step S203 are then used... Project back into the statistical morphological subspace constructed in step S1, and calculate its projection residual energy. :

[0069]

[0070] Set a confidence threshold This threshold is determined based on the statistical distribution of morphological deviations observed during the normal aging process of a large number of devices. If The candidate curve is determined to conform to the physical characteristics of this ADC model. After verification, a new calibration mapping table is generated based on the updated calibration mapping relationship. If the result is deemed an abnormal deformation, the update is rejected, the calibration mapping relationship from the previous moment is maintained, and the calibration mapping table is not updated. If the calibration ejection table is updated, it is rewritten into memory.

[0071] Step S205: Real-time calibration output

[0072] During real-time acquisition by the ADC, the system directly uses the raw code value output by the ADC as the memory address offset to directly address the calibration mapping table in the DTCM. The read data, i.e., the corresponding error calibration data, is superimposed with the actual sampled data to obtain the calibrated high-precision voltage value. Due to the extremely high access speed of the DTCM, this process does not consume additional CPU computing power, achieving zero-latency real-time calibration.

[0073] like Figure 2 As shown in the figure, this application embodiment also provides a dynamic calibration system for analog-to-digital converters based on statistical morphological constraints. The system includes non-volatile memory, analog front-end hardware, and a processor unit.

[0074] The non-volatile memory stores statistical morphological subspace information obtained by the processor unit during offline analysis, as well as fundamental parameters to support nonlinear benchmark reconstruction and calibration operations performed by the processor unit during online operation. Specifically, it includes a set of dominant error morphological basis functions obtained through statistical analysis and decomposition of integral nonlinear characteristic data acquired under multiple temperature conditions, temperature correlation coefficients corresponding to each error morphological mode, and a non-uniform sampling index table to guide the sampling of sparse aging feature anchor points. The parameter information stored in the non-volatile memory provides prior support for nonlinear benchmark curve reconstruction, temperature compensation, and aging calibration.

[0075] The analog front-end hardware is used to provide the processor unit with real-time temperature information and sampling data from the analog-to-digital converter (ADC). It includes a temperature sensor, a reference source module, and the ADC itself. The temperature sensor is mounted on the ADC and is used to acquire real-time temperature information from the ADC and provide it to the processor unit. The reference source module provides the ADC with a reference signal or reference input for sampling sparse aging feature anchor points. The ADC itself acquires aging feature anchor points according to a preset sparse sampling strategy using the reference signal or reference input provided by the reference source module. The ADC transmits the aging feature anchor points and the sampling data acquired in real-time to the processor unit.

[0076] The processor unit processes real-time temperature information and sampled data acquired by the analog front-end hardware, generates corresponding calibration mapping relationships, and outputs calibrated high-precision voltage values. It includes a statistical morphology analysis module, a nonlinear model reconstruction engine, an affine transformation update module, a consistency verification module, and a direct addressing compensation interface. Specifically, the statistical morphology analysis module constructs and analyzes statistical morphology subspace information offline; the nonlinear model reconstruction engine reconstructs the nonlinear baseline curve at the current temperature based on the dominant error morphology basis function set in the non-volatile memory and real-time temperature information; the affine transformation update module calculates affine perturbation parameters based on aging feature anchor point data and performs affine updates on the nonlinear baseline curve accordingly; the consistency verification module performs consistency verification of the statistical morphology subspace and updates the calibration mapping table; and the direct addressing compensation interface performs direct addressing compensation on the real-time raw code value output by the analog-to-digital converter after the calibration mapping table update, outputting a calibrated high-precision voltage value.

[0077] In one possible implementation, the preset sparse sampling strategy is based on a sampling value function, which is used to evaluate the effective information contribution of different code values ​​to the estimation of affine perturbation parameters. The sparse sampling strategy evaluates and filters candidate code values ​​across the entire code domain according to the sampling value function, prioritizing the selection of at least two code values ​​that contribute the most to the identifiability of the affine perturbation parameters, and generating a non-uniform sampling index table to guide the analog-to-digital converter to perform measurements only at the sparse code values ​​indicated by the non-uniform sampling index table.

[0078] In one possible implementation, the statistical morphological analysis module operates as follows: acquiring integral nonlinearity (INL) characteristic curve data of several analog-to-digital converters of the same model under multiple temperature conditions, performing statistical decomposition on the integral nonlinearity characteristic curve data, extracting several mutually orthogonal dominant error morphological basis functions, and constructing a statistical morphological subspace to characterize the inherent nonlinear characteristics of the analog-to-digital converter of that model.

[0079] The aforementioned statistical morphological subspace is obtained by performing principal component analysis (PCA) or singular value decomposition (SVD) on the integral nonlinearity (INL) characteristic curve data, and the subspace with a cumulative contribution rate exceeding a preset proportion is selected. Each eigenvector serves as a dominant error morphology basis function. The statistical morphology subspace is spanned by a linear combination of the dominant error morphology basis functions and is used to characterize the inherent nonlinear error morphology range of this type of analog-to-digital converter.

[0080] In one possible implementation, the nonlinear model reconstruction engine operates as follows: during online operation, the current temperature value of the analog-to-digital converter is acquired in real time; based on the constructed statistical morphological subspace and combined with the modal weighting coefficients, the dominant error morphological basis functions are weighted and combined to reconstruct the nonlinear baseline curve under the current temperature conditions.

[0081] The modal weighting coefficients are modeled as a quadratic polynomial function of temperature. The coefficients of the quadratic polynomial function, i.e. the temperature correlation coefficients, are determined and stored in memory during the construction of the statistical morphological space.

[0082] In one possible implementation, the affine transformation update module operates as follows: based on the deviation relationship between the aging feature anchor point and the nonlinear reference curve, it calculates the affine perturbation parameters used to characterize the device aging drift, and performs affine update on the nonlinear reference curve according to the affine perturbation parameters to generate a candidate calibration curve.

[0083] The affine perturbation parameters are used to characterize the overall aging drift characteristics of the digital-to-analog converter during long-term operation, including at least the gain drift coefficient. and bias drift coefficient The gain drift coefficient mentioned above and bias drift coefficient The calculation method is as follows:

[0084]

[0085]

[0086] in, Large value points in aging characteristic anchor points, The minimum value point in the aging characteristic anchor point,

[0087] These are the measured output values ​​of the large-value points in the aging characteristic anchor points. This represents the measured output value of the small value point in the aging characteristic anchor point.

[0088] The process of generating candidate calibration curves follows the following model:

[0089]

[0090] in, For calibration curves, It is a nonlinear reference curve.

[0091] In one possible implementation, the consistency verification module operates as follows: projecting the candidate calibration curve into the statistical morphological subspace and calculating its projection residual energy; when the projection residual energy is equal to or lower than a preset confidence threshold, confirming that the candidate calibration curve is consistent with the statistical morphological prior, and updating the calibration mapping relationship to generate a new calibration mapping table; when the projection residual energy is higher than the confidence threshold, triggering an anomaly handling mechanism, maintaining the calibration mapping relationship of the previous moment, and not updating the calibration mapping table.

[0092] The projection residual energy is defined as the square of the Euclidean distance between the candidate calibration curve and its orthogonal projection in the statistical morphological subspace; the anomaly handling mechanism is to reject the current update and maintain the calibration mapping relationship of the previous time step.

[0093] The calibration mapping relationship is stored in memory in the form of a lookup table (LUT), and the lookup table is only atomically updated after passing the consistency check.

[0094] The confidence threshold is determined by analyzing the morphological deviation distribution statistical characteristics of a large number of ADCs of the same model during the normal aging process, and is used to distinguish between normal aging deformation and abnormal deformation caused by circuit faults.

[0095] In one possible implementation, the direct addressing compensation interface operates as follows: the calibration mapping table that has passed the consistency check is stored; when the analog-to-digital converter outputs the real-time raw code value, the raw code value is used as the memory address offset to directly address the calibration mapping table, obtain the corresponding error calibration data, and output a high-precision voltage value after compensation by the error calibration data.

[0096] This application also provides an electronic device, including a processor and a memory.

[0097] The memory is used to store computer programs.

[0098] When the processor executes a program stored in the memory, it implements any of the methods described in this application.

[0099] In one possible implementation, the electronic device of this application embodiment further includes a communication interface and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.

[0100] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0101] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0102] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0103] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0104] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, implements any of the methods described in this application.

[0105] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform any of the methods described in this application.

[0106] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0107] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0108] The various embodiments in this specification are described in a related manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other.

[0109] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application are included within the scope of protection of this application.

Claims

1. A dynamic calibration method for analog-to-digital converters based on statistical morphological constraints, characterized in that, Includes the following steps: Construction of statistical morphological space: Obtain the dominant error morphological basis functions of the analog-to-digital converter and construct a statistical morphological subspace to characterize the inherent nonlinear characteristics of the analog-to-digital converter; Nonlinear benchmark reconstruction at current temperature: Based on the constructed statistical morphological subspace and combined with modal weighting coefficients, the nonlinear benchmark curve under the current temperature condition is reconstructed; Sparse aging feature anchor point sampling: Based on a preset sparse sampling strategy, the actual error measurement value of the analog-to-digital converter is obtained at at least two key code value points through a reference signal or a known reference input, which serve as aging feature anchor points characterizing the aging state of the device. Constrained affine update: Calculate the affine perturbation parameters used to characterize device aging drift, and perform affine update on the nonlinear reference curve based on the affine perturbation parameters to generate candidate calibration curves; Subspace consistency verification: Project the candidate calibration curve into the statistical morphology subspace and calculate its projection residual energy; when the projection residual energy is equal to or lower than a preset confidence threshold, confirm that the candidate calibration curve is consistent with the statistical morphology prior, and update the calibration mapping relationship to generate a new calibration mapping table; when the projection residual energy is higher than the confidence threshold, trigger the anomaly handling mechanism, maintain the calibration mapping relationship of the previous moment, and do not update the calibration mapping table; Real-time calibration output: The calibration mapping table that has passed the consistency check is stored. When the analog-to-digital converter outputs the real-time raw code value, the raw code value is used as the memory address offset to directly address the calibration mapping table, obtain the corresponding error calibration data, and output a high-precision voltage value after error calibration data compensation.

2. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The statistical morphological subspace is obtained by performing principal component analysis or singular value decomposition on the integral nonlinear characteristic curve data, and the subspace with a cumulative contribution rate exceeding a preset proportion is selected. Each eigenvector serves as a dominant error morphology basis function. The statistical morphology subspace is spanned by a linear combination of the dominant error morphology basis functions and is used to characterize the inherent nonlinear error morphology range of this type of analog-to-digital converter.

3. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The modal weighting coefficients are modeled as a quadratic polynomial function of temperature, and the coefficients of the quadratic polynomial function are determined and stored in memory during the construction of the statistical morphological space.

4. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The preset sparse sampling strategy is based on a sampling value function, which is used to evaluate the effective information contribution of different code values ​​to the estimation of affine perturbation parameters. The sparse sampling strategy evaluates and filters candidate code points in the entire code domain based on the sampling value function, and prioritizes the selection of at least two code points that contribute the most to the identifiability of the affine perturbation parameter, generating a non-uniform sampling index table to guide the analog-to-digital converter to perform measurements only at the sparse code points indicated by the non-uniform sampling index table.

5. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The affine perturbation parameters are used to characterize the overall aging drift characteristics of the digital-to-analog converter during long-term operation, including the gain drift coefficient. and bias drift coefficient The gain drift coefficient mentioned above and bias drift coefficient The calculation method is as follows: in, Large value points in aging characteristic anchor points, The minimum value point in the aging characteristic anchor point, These are the measured output values ​​of the large-value points in the aging characteristic anchor points. The measured output value of the small value point in the aging characteristic anchor point; The process of generating candidate calibration curves follows the following model: in, For calibration curves, It is a nonlinear reference curve.

6. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The projection residual energy is defined as the square of the Euclidean distance between the candidate calibration curve and its orthogonal projection in the statistical morphological subspace; the anomaly handling mechanism is to reject the current update and maintain the calibration mapping relationship of the previous time step.

7. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The calibration mapping relationship is stored in memory in the form of a lookup table (LUT), and the lookup table is only atomically updated after passing the consistency check.

8. The dynamic calibration method for an analog-to-digital converter based on statistical morphological constraints according to claim 1, characterized in that, The confidence threshold is determined by analyzing the morphological deviation distribution statistical characteristics of a large number of ADCs of the same model during the normal aging process, and is used to distinguish between normal aging deformation and abnormal deformation caused by circuit faults.

9. A dynamic calibration system for an analog-to-digital converter based on statistical morphological constraints, comprising non-volatile memory, analog front-end hardware, and a processor unit; The non-volatile memory is used to store statistical morphological subspace information obtained by the processor unit in the offline stage, as well as basic parameters to support the processor unit in performing nonlinear benchmark reconstruction and calibration operations in the online operation stage. The analog front-end hardware is used to provide the processor unit with real-time temperature information and sampling data from the analog-to-digital converter (ADC). It includes a temperature sensor, a reference source module, and the ADC itself. The temperature sensor is mounted on the ADC itself and is used to acquire real-time temperature information from the ADC and provide it to the processor unit. The reference source module provides the ADC with a reference signal or reference input for sparse aging feature anchor point sampling. The ADC itself acquires aging feature anchor points according to a preset sparse sampling strategy using the reference signal or reference input provided by the reference source module. The analog-to-digital converter transmits the aging feature anchor points and the sampled data acquired in real time to the processor unit; The processor unit is used to process the real-time temperature information and sampled data acquired by the analog front-end hardware, generate corresponding calibration mapping relationships, and output calibrated high-precision voltage values. It includes a statistical morphological analysis module, a nonlinear model reconstruction engine, an affine transformation update module, a consistency verification module, and a direct addressing compensation interface. The statistical morphology analysis module is used to construct and analyze statistical morphology subspace information in the offline stage; the nonlinear model reconstruction engine is used to reconstruct the nonlinear baseline curve at the current temperature based on the set of dominant error morphology basis functions in the non-transmissive memory and real-time temperature information. The affine transformation update module calculates affine perturbation parameters based on aging feature anchor data and performs affine updates on the nonlinear reference curve based on the affine perturbation parameters; the consistency verification module is used to complete the consistency verification of the statistical morphological subspace and the update of the calibration mapping table; the direct addressing compensation interface is used to perform direct addressing compensation on the real-time raw code value output by the analog-to-digital converter after the calibration mapping table update is completed, and output a calibrated high-precision voltage value.

10. An electronic device, characterized in that, Including processor and memory; The memory is used to store computer programs; When the processor executes the program stored in the memory, it implements the analog-to-digital converter dynamic calibration method according to any one of claims 1-8.

Images