A TIADC mismatch error digital calibration method and system applied to a radar data acquisition system

Abstract

This invention proposes a digital calibration method and system for TIADC mismatch error in radar data acquisition systems, belonging to the field of signal sampling and processing technology. It constructs a residual sequence containing all mismatch errors. Since the frequency of mismatch errors is finite and calculable, and the information in the mismatch is fixed, an adaptive signal decomposition method can be used to decompose the complex signal containing mismatch information into sub-signals with different information granularities. By combining neural network methods to explore the correlation of time-series data in the signal, and then utilizing the powerful regression capability of neural networks, TIADC calibration is achieved from a time-domain perspective. This invention solves the error problems caused by bias mismatch, gain mismatch, and time mismatch within the TIADC, thereby improving the performance indicators of the data acquisition system and enhancing the accuracy of data acquisition.

Description

Technical Field

[0001] This invention belongs to the field of radar signal sampling and processing technology, specifically, it relates to a TIADC mismatch error digital calibration method and system applied to radar data acquisition systems. Background Technology

[0002] Analog-to-digital converters (ADCs) are key components that convert analog signals into digital signals. Data acquisition systems with ADCs as their core components are widely used in modern electronic systems such as radar systems, communication systems, test instruments, and digitizers. Especially in radar systems, the signals received by radar are typically analog signals, which are continuous voltage waveforms. ADCs convert these analog signals into discrete digital samples. This allows the signals to be processed by digital systems, including storage, analysis, and display. The sampling rate of the ADC determines the system's response speed to the signal. Higher sampling rates can capture signal changes faster, thereby improving the response time of the radar system. Compared to the rapid increase in application requirements, the improvement rate of ADC manufacturing technology is much slower. It is now very difficult to manufacture ADCs that meet application requirements solely through technological advancements, causing the performance of data acquisition systems with ADCs as their core components to gradually become a performance bottleneck in modern electronic systems. Therefore, it is of great significance to explore how to achieve high-speed, high-resolution data acquisition systems using existing ADC chips.

[0003] Time-interleaved analog-to-digital converters (TIADCs) employ multiple identical ADCs to sequentially sample the same signal source, effectively improving the overall sampling rate of the system and overcoming the performance limitations of single-channel ADCs. However, due to differences in the manufacturing processes of ADC chips, it is impossible to ensure that each ADC has identical physical characteristics, which directly affects the performance of TIADCs. These imperfect physical characteristics can cause mismatch problems in TIADCs, leading to uneven signal sampling and spurious components in the output signal. These mismatches mainly include bias mismatch, gain mismatch, and time mismatch. Therefore, exploring and calibrating the factors that limit the performance of TIADCs is crucial for improving the performance of data acquisition systems and enhancing the capabilities of radar systems, thereby increasing their reliability and efficiency in various applications.

[0004] Existing TIADC calibration methods primarily focus on signal processing, studying different filter structures and combining them with signal characteristics to complete the calibration. Some calibration methods are limited by the number of TIADC channels, some cannot overcome the Nyquist band limitation, and some require designing separate calibration algorithms for each mismatch, leading to a complex calibration process. Summary of the Invention

[0005] This invention proposes a digital calibration method and system for TIADC mismatch error in radar data acquisition systems. It works on the TIADC in the radar data acquisition system and solves the error problems caused by bias mismatch, gain mismatch and time mismatch inside the TIADC, thereby improving the performance indicators of the data acquisition system, enhancing the accuracy of data acquisition, and improving the reliability and efficiency of the radar.

[0006] This invention is achieved through the following technical solution:

[0007] A digital calibration method for TIADC mismatch error applied to a radar data acquisition system, the method specifically includes the following steps:

[0008] Step 1: In offline mode, select a pair of radar input and output data as labels to train the model. Calculate the ideal sampling output of the radar signal according to the mathematical formula of the selected input analog signal, or connect a high-end digital oscilloscope directly to the signal generator to collect the desired signal. Combine the actual output data of TIADC to construct a residual sequence containing all mismatch errors.

[0009] Step 2: Standardize the residual sequence from Step 1, which contains all mismatch errors, to make the feature scales of the data similar.

[0010] Step 3: Using the adaptive signal decomposition method, the standardized residual signal constructed in Step 2 is decomposed into sub-signals with different information granularities, and then the LSTM model is used for training to obtain the correlation of time series data in the signal.

[0011] Step 4: Initialize the residual data for estimation in online mode, perform standardization, and then adaptively decompose the residual data to construct the input of the LSTM model;

[0012] Step 5: Use the model trained in Step 3 to perform residual estimation, reconstruction, and inverse standardization on the input data from Step 4. Simultaneously, combine the acquired real-time radar data stream to be processed to calculate the calibrated TIADC.

[0013] Output the data and finally verify the accuracy of the TIADC mismatch error digital calibration.

[0014] Furthermore, in step 1, the mismatch problem of TIADC is addressed from a time series perspective by constructing a residual sequence containing all mismatches:

[0015] y err =y out -y exp (1)

[0016] In the formula yerr (t){t=0,1,2,...,lk-1} is the constructed residual signal, y out For the TIADC system to output data, y exp It is the desired output signal, which does not contain any mismatch error.

[0017] Furthermore, in step 2, the constructed residual data is standardized:

[0018]

[0019] In the formula y e ′ rr ∈[0,1] represents the standardized residual data; y err For the original residual data, y max It is the maximum value of the original data, y min It is the minimum value of the original data; the constructed standardized residual signal provides input for subsequent signal decomposition and model training.

[0020] Furthermore, in step 3, the signal adaptive decomposition method is VMD decomposition: the normalized residual sequence y... e ′ rr Decomposition is performed, and multiple IMF components are obtained through VMD as y e ′ rr_IMF1 ~y e ′ rr_IMFn ;

[0021] Training using an LSTM model: The mapping function of the LSTM regression network is f LSTM (·),therefore:

[0022]

[0023] In the formula, X is the estimated value obtained from model training. train It is the reconstructed y′ err_IMF1 Each output value Corresponding to a set of corresponding X train The trained LSTM model is then used in the calibration phase.

[0024] Furthermore, in step 4, the initial residual data is y err The input to the LSTM model is X, which is (t){t=lk,...,l-1}. Preprocessing is required to satisfy the input conditions of the calibration model. VMD decomposition is performed on the residual data. test ,Y test The estimated values ​​are generated by the following model:

[0025]

[0026] In the formula, X is the estimated value output by the model. test This is the residual signal used for testing.

[0027] Furthermore, in step 5, Reconstruction yields a standardized residual sequence. The residual estimation results are obtained by inverse normalization. Through real-time measured signal y out and new residual signals The calibrated TIADC signal is calculated as follows:

[0028]

[0029] Finally, the digital calibration of the TIADC mismatch error was completed.

[0030] A TIADC mismatch error digital calibration system for radar data acquisition systems, the system comprising an offline training module and a calibration module;

[0031] The offline training module selects the input and output data of a pair of radars as labels to train the model in offline mode. It calculates the ideal sampling output of the radar signal according to the mathematical formula of the input analog signal, or directly connects a high-end digital oscilloscope to the signal generator to acquire the desired signal. Combined with the actual output data of TIADC, it constructs a residual sequence containing all mismatch errors. The residual sequence containing all mismatch errors is standardized to make the feature scales of the data similar. Using the signal adaptive decomposition method, the constructed standardized residual signal is decomposed into sub-signals with different information granularities. Then, the LSTM model is used for training to obtain the correlation of time series data in the signal.

[0032] The calibration module acquires the real-time radar data stream to be processed in an online state. First, it initializes the residual data for estimation, performs standardization, and then adaptively decomposes the residual data to construct the input of the LSTM model. Then, it uses the model trained by the offline training module to perform residual estimation, reconstruction, and inverse standardization on the input data of the LSTM model. Finally, it combines the real-time radar data stream to calculate the calibrated TIADC output data and verify the accuracy of the digital calibration of the TIADC mismatch error.

[0033] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0034] A computer-readable storage medium for storing computer instructions that, when executed by a processor, implement the steps of the above-described method.

[0035] Beneficial effects of the invention

[0036] To eliminate the impact of mismatch errors, this invention constructs a residual sequence containing all mismatch errors. Since the frequency of mismatch errors is finite and calculable, and the information in the mismatch is fixed, an adaptive signal decomposition method can be used to decompose the complex signal containing mismatch information into sub-signals with different information granularities. By combining neural network methods to explore the correlation of time-series data in the signal, and then utilizing the powerful regression capability of neural networks, TIADC calibration is achieved from a time-domain perspective.

[0037] This invention enables TIADC calibration for any channel, the input frequency of the calibration signal is not limited by the Nyquist band, and it can complete the calibration of three mismatch errors at once. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the method framework of the present invention;

[0039] Figure 2 This is a calibration flowchart of the present invention;

[0040] Figure 3 The signal spectrum before calibration;

[0041] Figure 4 This is a normalized residual signal diagram;

[0042] Figure 5 The VMD decomposition diagram of the residual signal;

[0043] Figure 6 The calibrated signal spectrum;

[0044] Figure 7 This is a schematic diagram of the TIADC structure based on the AD9467 design;

[0045] Figure 8 This is a spectrum diagram of the system when mismatch exists;

[0046] Figure 9 This is the TIADC spectrum after calibration. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Combination Figures 1 to 9 .

[0049] A digital calibration method for TIADC mismatch error applied to radar data acquisition systems:

[0050] The method specifically includes the following steps, mainly divided into two parts: model training and calibration;

[0051] During the calibration model training phase, a neural network is used to learn the mapping relationship between the input and output signals. The neural network model is trained using the residual signal containing all mismatch errors to construct the input signal and the corresponding output signal. The neural network will learn to capture the mismatch information in the input signal and the correlation between the preceding and following data, and output an estimated residual signal for calibration.

[0052] Model training phase:

[0053] Step 1: In offline mode, select a pair of radar input and output data as labels to train the model. Calculate the ideal sampling output of the radar signal according to the mathematical formula of the input analog signal, or connect a high-end digital oscilloscope directly to the signal generator to collect the desired signal. Combine the actual output data of TIADC to construct a residual sequence containing all mismatch errors.

[0054] In step 1, the mismatch problem of TIADC is addressed from a time series perspective. To eliminate the impact of mismatch errors, a residual sequence containing all mismatch errors is constructed:

[0055] y err =y out -y exp (1)

[0056] In the formula y err (t){t=0,1,2,...,lk-1} is the constructed residual signal, y out For the TIADC system to output data, y exp This is the desired output signal, containing no mismatch error. In reality, y exp The signal can be calculated by inputting a mathematical formula from an analog signal, or acquired by directly connecting a high-end digital oscilloscope to a signal generator. The desired signal is used only once during model training and is not present during calibration.

[0057] Since the frequency of mismatch error is finite and calculable, and the information contained therein is fixed, the signal adaptive decomposition method can be used to decompose complex signals containing mismatch information into sub-signals with different information granularities.

[0058] Step 2: Standardize (normalize) the residual sequence from Step 1, which contains all mismatch errors, to make the feature scales of the data similar.

[0059] In step 2, normalizing the input signals ensures that their features have similar scales, thereby enhancing the model's generalization ability and performance. Normalization can also accelerate model convergence when data values ​​are large. The constructed residual data is then standardized.

[0060]

[0061] In the formula y′ err ∈[0,1] represents the standardized residual data; y err For the original residual data, y max It is the maximum value of the original data, y min It is the minimum value of the original data; the constructed standardized residual signal provides input for subsequent signal decomposition and model training.

[0062] To achieve TIADC calibration, a neural network approach is used to explore the correlation of time-series data in the signal, and the powerful regression capability of the neural network is used to achieve calibration from a time-domain perspective.

[0063] Step 3: Using the adaptive signal decomposition method, the standardized residual signal constructed in Step 2 is decomposed into sub-signals with different information granularities, and then the LSTM model is used for training to obtain the correlation of time series data in the signal.

[0064] In step 3, the signal adaptive decomposition method is VMD decomposition: the normalized residual sequence y′ err The decomposition is performed, and multiple IMF components y′ are obtained through VMD. err_IMF1 ~y′ err_IMFn ;

[0065] Training using an LSTM model: The mapping function of the LSTM regression network is f LSTM (·),therefore:

[0066]

[0067] In the formula, X is the estimated value obtained from model training. train It is the reconstructed y′ err_IMF1Each output value Corresponding to a set of corresponding X train The trained LSTM model is then used in the calibration phase.

[0068] During calibration, a pre-trained neural network model is used to calibrate the input signal. By inputting the residual input signal used for estimation into the neural network, the estimated residual output signal can be obtained. Combined with the real-time TIADC output signal, the influence of mismatch error on the input signal is calculated and eliminated, resulting in a more accurate output signal.

[0069] Step 4, Calibration process as follows Figure 2 As shown, this is also equivalent to testing the model; radar data for real-time use is obtained in an online state, and residual data for estimation is obtained through initialization (data preprocessing). After standardization, the residual data is adaptively decomposed (VMD decomposition) to construct the input of the LSTM model.

[0070] In step 4, the initial residual data is y err The input to the LSTM model is X, which is (t){t=lk,...,l-1}. Preprocessing is required to satisfy the input conditions of the calibration model. VMD decomposition is performed on the residual data. test ,Y test The estimated values ​​are generated by the following model:

[0071]

[0072] In the formula, X is the estimated value output by the model. test This is the residual signal used for testing.

[0073] Step 5: Use the model trained in Step 3 to perform residual estimation, reconstruction, and inverse standardization on the input data in Step 4. Combined with the collected real-time radar data stream to be processed, calculate the calibrated TIADC output data to verify the accuracy of the digital calibration of TIADC mismatch error.

[0074] In step 5, Reconstruction yields a standardized residual sequence. The residual estimation results are obtained by inverse normalization. Through real-time measured signal y out and new residual signals The calibrated TIADC signal is calculated as follows:

[0075]

[0076] Finally, the digital calibration of the TIADC mismatch error was completed.

[0077] A TIADC mismatch error digital calibration system for use in a data acquisition system, the system comprising an offline training module and a calibration module;

[0078] The offline training module selects the input and output data of a pair of radars as labels to train the model in offline mode. It calculates the ideal sampling output of the radar signal according to the mathematical formula of the input analog signal, or directly connects a high-end digital oscilloscope to the signal generator to acquire the desired signal. Combined with the actual output data of TIADC, it constructs a residual sequence containing all mismatch errors. The residual sequence containing all mismatch errors is standardized to make the feature scales of the data similar. Using the signal adaptive decomposition method, the constructed standardized residual signal is decomposed into sub-signals with different information granularities. Then, the LSTM model is used for training to obtain the correlation of time series data in the signal.

[0079] The calibration module acquires real-time radar data in an online state, initializes residual data for estimation, performs standardization processing, adaptively decomposes the residual data, and constructs the input of the LSTM model. The model trained by the offline training module performs residual estimation, reconstruction, and inverse standardization on the input data of the LSTM model. Combined with the collected real-time radar data stream to be processed, the calibrated TIADC output data is calculated to verify the accuracy of the digital calibration of TIADC mismatch error.

[0080] The method of this invention is applied to radar systems to acquire and process echo signals to achieve functions such as target detection, tracking, and measurement. Using the TIADC calibration method can improve the measurement accuracy and target recognition capability of the radar system.

[0081] This invention can also be applied to communication systems, medical imaging, industrial automation, and various scientific research projects that require high-speed, high-precision data acquisition.

[0082] In communication systems, data such as high-speed data transmission, modulation / demodulation, and channel estimation are collected. By using the TIADC calibration method, the performance of the communication system can be improved, and the impact of mismatch errors on data transmission quality can be reduced.

[0083] In medical imaging fields such as magnetic resonance imaging (MRI) and positron emission tomography (PET), high-precision signal acquisition and reconstruction are required to obtain accurate image information. The TIADC calibration method can help improve the resolution and accuracy of medical imaging systems.

[0084] In industrial automation, accurate data acquisition and analysis of various sensor signals are required to monitor and control the production process. Using the TIADC calibration method can improve the measurement accuracy and response speed of industrial automation systems.

[0085] In scientific research, such as physics experiments, astronomical observations, and biological signal acquisition, high-precision data acquisition is required to obtain accurate experimental results and signal analysis. The TIADC calibration method can help researchers eliminate the impact of mismatch errors on data, thereby improving the reliability and accuracy of research.

[0086] Example:

[0087] (1) Simulation Test: Taking the TIADC simulation model as an example, a data acquisition system was established to calibrate and verify three types of mismatch errors. The TIADC is a four-channel system with a total sampling rate of 1 GHz. The test signal frequency is 25.125 MHz, and the offset mismatch magnitude is Δo. i = [0.000, 0.030, -0.010, 0.030], the gain mismatch is Δg i = [0.000, -0.020, -0.013, 0.017], the time mismatch is Δt. i =[0.000,0.050,-0.010,0.010]·T s .

[0088] To simulate the actual sampling process, AWGN noise was added to the signal. In formula (1), y... err The total data length is 10000, i.e., l is 10000. During the calibration model training process, k is 2000, i.e., y err Of the data, 8000 were used for training, and the remaining 2000 were used for model testing, i.e., during the calibration process. err The length of the component. Based on the average instantaneous frequency of the component, the number of VMD decompositions is n = 3. Figure 3 This is the signal spectrum before calibration. Figure 4 The normalized residual signal diagram. Figure 5 These are the signal components after VMD decomposition. Figure 6 This is the calibrated signal spectrum.

[0089] By comparison Figure 4 and Figure 6 It can be seen that the spurious spectrum caused by the mismatch error has been largely eliminated by the calibration algorithm. Specifically, the signal's SFDR increased from 39.32 dB before calibration to 64.89 dB, and the SNDR improved from 34.26 dB before calibration to 55.94 dB.

[0090] (2) Verification with measured data: In order to further verify the effectiveness of the calibration method proposed in this invention using measurement data, in accordance with Figure 7The structure shown illustrates a TIADC data acquisition system designed based on four Analog Devices (ADI) AD9467 chips. This system simulates the mismatch problem between the AD9467 chips. Measurement data acquired from this TIADC system is used to verify the calibration method. The AD9467 is a 16-bit 250MSPS chip, and the designed TIADC is a 16-bit 1GSPS system. Figure 8 This is the spectrum diagram when mismatch exists in the system. Figure 9 This is a calibrated TIADC spectrum based on the AD9467 chip. (Comparison is needed.) Figure 8 and Figure 9 It can be seen that, through the calibration algorithm proposed in this invention, the system SFDR is improved by 54.52dB and SNDR is improved by 42.12dB.

[0091] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0092] A computer-readable storage medium for storing computer instructions that, when executed by a processor, implement the steps of the above-described method.

[0093] The memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory of the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0094] 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 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 means such as coaxial cable, optical fiber, digital subscriber line, DSL, or wireless means such as infrared, wireless, microwave, etc. 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 such as a floppy disk, hard disk, magnetic tape; an optical medium such as a high-density digital video disc, DVD; or a semiconductor medium such as a solid-state disk, SSD, etc.

[0095] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0096] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as execution by a hardware decoding processor, or as execution by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0097] The foregoing has provided a detailed description of the TIADC mismatch error digital calibration method and system proposed for radar data acquisition systems, and has elucidated the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A digital calibration method for TIADC mismatch error applied to a radar data acquisition system, characterized in that: The method specifically includes the following steps: Step 1: In offline mode, select a pair of radar input and output data as labels to train the model. Calculate the ideal sampling output of the radar signal according to the mathematical formula of the input analog signal, or connect a high-end digital oscilloscope directly to the signal generator to collect the desired signal. Combine the actual output data of TIADC to construct a residual sequence containing all mismatch errors. In step 1, the mismatch problem of TIADC is addressed from a time series perspective by constructing a residual sequence containing all mismatches: （1） In the formula , For the constructed residual signal, Output data for the TIADC system. It is the desired output signal, which does not contain any mismatch error; Step 2: Standardize the residual sequence from Step 1, which contains all mismatch errors, to make the feature scales of the data similar. Step 3: Using the adaptive signal decomposition method, the standardized residual signal constructed in Step 2 is decomposed into sub-signals with different information granularities, and then the LSTM model is used for training to obtain the correlation of time series data in the signal. In step 3, the signal adaptive decomposition method is VMD decomposition: the normalized residual sequence... The decomposition process is performed, and multiple IMF components are obtained through VMD. ~ ; n The number of VMD decompositions; Training using an LSTM model: The mapping function of the LSTM regression network is... ,therefore: （3） In the formula, The estimated value obtained from model training. It is the reconstructed version Each output value Corresponding to a set of corresponding The trained LSTM model is then used in the calibration phase. Step 4: Obtain radar data for real-time use while online, initialize residual data for estimation, perform standardization processing, adaptively decompose the residual data, and construct the input of the LSTM model. Step 5: Use the model trained in Step 3 to perform residual estimation, reconstruction, and inverse standardization on the input data in Step 4. Combined with the collected real-time radar data stream to be processed, calculate the calibrated TIADC output data to verify the accuracy of the digital calibration of TIADC mismatch error.

2. The calibration method according to claim 1, characterized in that: In step 2, the constructed residual data is standardized: （2） In the formula The data is the standardized residual data; The original residual data, It is the maximum value of the original data. It is the minimum value of the original data; the constructed standardized residual signal provides input for subsequent signal decomposition and model training.

3. The calibration method according to claim 2, characterized in that: In step 4, the initial residual data is , Preprocessing is required to meet the input conditions of the calibration model. VMD decomposition is performed on the residual data to construct the input for the LSTM model. The estimated values ​​are generated by the following model: （4） In the formula, The estimated value output by the model. This is the residual signal used for testing.

4. The calibration method according to claim 3, characterized in that: In step 5, Reconstruction yields a standardized residual sequence. The residual estimation results are obtained by inverse normalization. ; through real-time measured signals and new residual signals The calibrated TIADC signal is calculated as follows: （5） Finally, the digital calibration of the TIADC mismatch error was completed.

5. A TIADC mismatch error digital calibration system applied to a radar data acquisition system, characterized in that: The system is used to perform the TIADC mismatch error digital calibration method for radar data acquisition systems as described in any one of claims 1 to 4; The system includes an offline training module and a calibration module; The offline training module selects a pair of radar input and output data as labels to train the model in offline mode. It calculates the ideal sampling output of the radar signal according to the mathematical formula of the input analog signal, or directly connects a high-end digital oscilloscope to the signal generator to collect the desired signal. Combined with the actual output data of TIADC, it constructs a residual sequence containing all mismatch errors. To address the mismatch problem of TIADC from a time series perspective, a residual sequence containing all mismatches is constructed: （1） In the formula , For the constructed residual signal, Output data for the TIADC system. It is the desired output signal, which does not contain any mismatch error; The residual sequence containing all mismatch errors is standardized to make the feature scales of the data similar. The standardized residual signal is decomposed into sub-signals with different information granularities using the adaptive signal decomposition method. Then, the LSTM model is used for training to obtain the correlation of time series data in the signal. The adaptive signal decomposition method is VMD decomposition: the normalized residual sequence... The decomposition process is performed, and multiple IMF components are obtained through VMD. ~ ; n The number of VMD decompositions; Training using an LSTM model: The mapping function of the LSTM regression network is... ,therefore: （3） In the formula, The estimated value obtained from model training. It is the reconstructed version Each output value Corresponding to a set of corresponding The trained LSTM model is then used in the calibration phase. The calibration module obtains radar data for real-time use in an online state, initializes residual data for estimation, performs standardization processing, adaptively decomposes the residual data, and constructs the input of the LSTM model. The model trained using the offline training module performs residual estimation, reconstruction, and inverse standardization on the input data of the LSTM model. Combined with the collected real-time radar data stream to be processed, the calibrated TIADC output data is calculated to verify the accuracy of the digital calibration of TIADC mismatch error.

6. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 4.

Images