Battery cell connection system signal acquisition processing method and device

Abstract

The application discloses a battery cell connection system (CCS) signal acquisition processing method and device, and relates to the technical field of signal processing. A plurality of multi-channel original acquisition signals of a plurality of battery cells acquired by a CCS are obtained; each first intrinsic modal function component is determined based on a current channel sampling curve in the multi-channel original acquisition signals; a second intrinsic modal function component is determined according to the correlation state between each first intrinsic modal function component and the current channel sampling curve; a normal target channel processing signal and an abnormal target channel processing signal are obtained based on the second intrinsic modal function component; the abnormal target channel processing signal is compensated by using the normal target channel processing signal, and compensated multi-channel sampling data is obtained; the compensated multi-channel sampling data is restored, and the acquisition values of the plurality of battery cells are output. The risk of local sampling link abnormality being transmitted to the final output result is reduced, and the stability of the CCS acquisition signal and the reliability of the battery management system state monitoring are improved.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and more specifically to a method and apparatus for acquiring and processing signals in a battery cell connection system. Background Technology

[0002] CCS cell acquisition signal processing refers to the process of preprocessing, noise reduction, anomaly identification, and output verification of multi-channel raw acquisition signals generated during cell voltage sampling, temperature sampling, etc. in the cell connection system. The processing results are usually used by the battery management system for cell status monitoring, equalization control, and safety judgment.

[0003] During the operation of power batteries or energy storage battery modules, the CCS acquisition link is easily affected by electromagnetic interference and local channel anomalies, resulting in non-stationary interference and abnormal channel deviations in the original acquisition signals of multiple channels. Existing methods are limited by single-stage filtering capabilities and independent processing logic of each channel, making it difficult to balance the accuracy of extracting the inherent mode function components inside the current channel sampling curve with the synergy of multi-channel signal anomaly compensation, thus leading to poor stability of the acquisition signals. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for acquiring and processing signals in a battery cell connection system, so as to solve the problems in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a signal processing method for a battery cell connection system, applied to a CCS signal processing device, the method comprising: Acquire multi-channel raw acquisition signals from multiple battery cells obtained by CCS. The multi-channel raw acquisition signals include at least the voltage and temperature signals of the battery cells. The first intrinsic mode function component is determined based on the current channel sampling curve in the multi-channel original acquisition signal. The first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The second intrinsic mode function component is determined based on each first intrinsic mode function component and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are associated with the current channel sampling curve. The abnormal target channel processing signal and the abnormal target channel processing signal are obtained based on the second intrinsic mode function component. After compensating the abnormal target channel processed signal by the non-abnormal target channel processed signal, the compensated multi-channel sampling data is obtained; After restoring and compensating the multi-channel sampling data, the collected values ​​of multiple battery cells are output.

[0006] Furthermore, based on the current channel sampling curve in the multi-channel raw acquired signal, the components of each first intrinsic mode function are determined, including: Perform synchronization consistency verification on multi-channel raw acquisition signals with synchronization time identifiers; For the multi-channel raw acquisition signal after synchronization consistency verification, sudden outliers are identified, and adjacent valid sampling points are used to complete the identified sudden outliers; Standardized multi-channel sampling data is obtained based on the completed multi-channel original acquisition signal; Select a single channel sampling sequence within the current sampling window from the multi-channel sampling data as the current channel sampling curve; Variational mode decomposition is performed on the current channel sampling curve to obtain the components of each first intrinsic mode function.

[0007] Furthermore, variational mode decomposition is performed on the current channel sampling curve to obtain the components of each first intrinsic mode function, including: Determine the initial parameters for mode decomposition based on the signal type, sampling rate, and current sampling window length corresponding to the current channel sampling curve; Based on the initial parameters of mode decomposition, the current channel sampling curve is decomposed into multiple bandwidth-limited sub-signals that vary around different center frequencies; During the decomposition iteration process, the center frequency and waveform of each bandwidth-limited sub-signal are updated, and the reconstruction deviation between the superimposed bandwidth-limited sub-signals and the current channel sampling curve is used to determine whether the iteration stop condition is met. Each bandwidth-limited sub-signal obtained when the iteration stopping condition is met is determined as a component of the first intrinsic mode function.

[0008] Furthermore, based on the signal type, sampling rate, and current sampling window length corresponding to the current channel sampling curve, the initial parameters for mode decomposition are determined, including: Based on the signal type corresponding to the voltage or temperature signal of the current channel sampling curve, determine the expected change amplitude and expected change frequency band of the current channel sampling curve; Based on the sampling rate and the current sampling window length, determine the range of signal changes that the current channel sampling curve can represent within the current sampling window; Based on the expected change amplitude, expected change frequency band, and signal change range, determine the initial parameters for mode decomposition to control the number of decompositions and the smoothness of components; If the decomposition result obtained based on the initial parameters of mode decomposition has component aliasing or reconstruction deviation that does not meet the preset conditions, adjust the initial parameters of mode decomposition and re-execute variational mode decomposition.

[0009] Furthermore, the bandwidth-limited sub-signals obtained when the iteration stopping condition is met are determined as components of each first intrinsic mode function, including: Sort the bandwidth-limited sub-signals from low to high or high to low according to their center frequencies; Record the channel identifier, sampling window identifier, and sampling time correspondence for each bandwidth-limited sub-signal, so that the sampling points of each bandwidth-limited sub-signal are aligned with the sampling curve of the current channel; Based on the changing trend of each bandwidth-limited sub-signal, each bandwidth-limited sub-signal is marked as at least one of low-frequency trend component, effective change candidate component, or high-frequency disturbance candidate component. Each bandwidth-limited sub-signal, after being sorted, recorded, and labeled, is taken as a component of each first intrinsic mode function.

[0010] Furthermore, the second intrinsic mode function components are determined based on each first intrinsic mode function component and the current channel sampling curve, including: Determine the degree of synchronization, information correlation, and energy contribution between each component of the first intrinsic mode function and the current channel sampling curve; Based on the degree of synchronization of changes, the degree of information correlation, and the degree of energy contribution, the component correlation evaluation results corresponding to each first intrinsic mode function component are generated. Based on the component correlation evaluation results, interference candidate components with low correlation to the current channel sampling curve are removed from each first intrinsic mode function component; The first intrinsic mode function component remaining after removing interfering candidate components is determined as the second intrinsic mode function component.

[0011] Furthermore, based on the component correlation evaluation results, interference candidate components with low correlation to the current channel sampling curve are removed from each first intrinsic mode function component, including: Based on the distribution of the component correlation evaluation results of each first intrinsic mode function component, determine the component screening conditions corresponding to the current sampling window; The first intrinsic mode function component whose component correlation evaluation results do not meet the component screening criteria and exhibit jitter, isolated abrupt changes, or inconsistent trends with the current channel sampling curve is identified as an interference candidate component. The first intrinsic mode function component whose component correlation evaluation results meet the component screening criteria and match the physical change trend of the voltage signal or temperature signal is identified as a valid candidate component. The second intrinsic mode function component is formed based on the effective candidate components, and the channel identifier and sampling time correspondence corresponding to the second intrinsic mode function component are retained.

[0012] Furthermore, based on the second intrinsic mode function component, the processing signals for the target channel without anomalies and the processing signals for the target channel with anomalies are obtained, including: Wavelet decomposition is performed on the components of the second intrinsic mode function to obtain the corresponding low-frequency approximation and high-frequency detail components; Based on the disturbance level of the high-frequency details and the change amplitude of the current channel sampling curve, determine the adaptive threshold processing conditions; According to the adaptive threshold processing conditions, the high-frequency detail part is shrunk, and the shrunk high-frequency detail part is reconstructed with the low-frequency approximation part to obtain the preliminary denoised channel signal; State tracking data is constructed based on the initial denoised channel signal, and baseline drift correction and state estimation are performed on the state tracking data by unscented Kalman filtering to obtain the target channel processed signal. Based on the channel signal type, the multiple target channel processing signals are divided into voltage channel processing signal group and temperature channel processing signal group; Within the channel processing signal group corresponding to the same channel signal type, determine the correlation between the target channel processing signals to form a multi-channel correlation evaluation result; Based on the multi-channel correlation evaluation results, the continuity of the processed signals of each target channel, and the reasonableness of the range corresponding to the processed signals of each target channel, abnormal target channel processed signals and non-abnormal target channel processed signals are identified.

[0013] Furthermore, after compensating the abnormal target channel processed signal with the non-abnormal target channel processed signal, the compensated multi-channel sampled data is obtained, including: Select a reference channel processing signal from the target channel processing signals without anomalies, whose correlation with the target channel processing signals with anomalies meets the preset compensation conditions; Based on the correspondence between the reference channel processed signal and the abnormal target channel processed signal within the historical synchronous sampling window, the weighted compensation relationship is determined, and the compensation signal corresponding to the abnormal target channel processed signal is generated according to the weighted compensation relationship. The abnormal target channel processing signal is replaced or fused with the compensation signal to obtain compensated multi-channel sampling data.

[0014] This application also provides a processing device for CCS acquired signals, the device comprising: The raw signal acquisition module is used to acquire multi-channel raw acquisition signals of multiple cells collected by CCS. The multi-channel raw acquisition signals include at least the voltage signal and temperature signal of the cells. The mode decomposition module is used to determine each first intrinsic mode function component based on the current channel sampling curve in the multi-channel original acquisition signal. Each first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The effective component determination module is used to determine the second intrinsic mode function component based on each first intrinsic mode function component and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are associated with the current channel sampling curve. The channel signal generation module is used to obtain the target channel processing signal based on the second intrinsic mode function component. The target channel processing signal includes the target channel processing signal without anomalies and the target channel processing signal with anomalies. The multi-channel compensation module is used to compensate for the abnormal target channel processing signal by using the target channel processing signal without abnormality, and then obtain the compensated multi-channel sampling data. The acquisition value output module is used to restore the compensated multi-channel sampling data and output the acquisition values ​​of multiple cells.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: Compared to existing CCS cell acquisition signal processing methods that primarily rely on single-stage filtering or independent processing of each channel, the cell connection system acquisition signal processing method provided in this application performs mode decomposition on the signals of each channel to be processed. It then filters effective mode components based on the information correlation between each mode component and the corresponding channel signal, thereby retaining effective components related to the actual acquisition changes within each channel and reducing the impact of non-stationary interference components on the signal reconstruction results. After obtaining the channel-processed signals of each channel, it identifies abnormal channels based on the correlation between processed signals of the same type of channel and uses processed signals of the same type of channel that were not identified as abnormal to perform redundancy compensation on the abnormal channel signals. This forms a continuous processing chain between the noise reduction results within a channel and the anomaly correction between channels. This reduces the risk of local sampling link anomalies being transmitted to the final output results, which is beneficial for improving the stability of CCS acquisition signals and the reliability of subsequent state monitoring of the battery management system. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a schematic diagram of the CCS cell acquisition and signal processing system architecture provided in an embodiment of this application. Figure 2 This is a schematic diagram of a CCS cell data acquisition application scenario provided in an embodiment of this application. Figure 3 This is a flowchart of the CCS signal acquisition and processing method provided in the embodiments of this application; Figure 4 This is a schematic diagram of multi-channel sampling preprocessing provided in an embodiment of this application; Figure 5 This is a schematic diagram of VMD decomposition provided in an embodiment of this application; Figure 6 A schematic diagram of intrinsic mode function component correlation filtering provided in an embodiment of this application; Figure 7 This is a schematic diagram of wavelet threshold denoising and UKF baseline correction provided in an embodiment of this application; Figure 8 This is a schematic diagram of the multi-channel correlation matrix and abnormal channel identification provided in the embodiments of this application; Figure 9 This is a schematic diagram of the abnormal channel redundancy compensation path provided in the embodiments of this application; Figure 10 This is a schematic diagram of the CCS signal acquisition and processing device module provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0019] In this application, CCS refers to Cell Contact System, which in a specific product can manifest as a cell connection assembly including a bus, acquisition line, insulation structure, temperature sensor, and connector; FPC refers to Flexible Printed Circuit; FFC refers to Flexible Flat Cable; PCB refers to Printed Circuit Board; BMS refers to Battery Management System; ADC refers to Analog-to-Digital Converter; and VMD refers to Variational Mode Decomposition. eDecomposition (VMD); UKF refers to Unscented Kalman Filter (UKF); CAN refers to Controller Area Network (CAN); CANFD refers to Controller Area Network with Flexible Data-Rate (CANFD); NTC refers to Negative Temperature Coefficient Thermistor (NTC); IMF refers to Intrinsic Mode Function (IMF); ESD refers to Electrostatic Discharge (ESD).

[0020] The first intrinsic mode function component referred to in this application is the bandwidth-limited mode component obtained after variational mode decomposition of the current channel sampling curve; the second intrinsic mode function component is the effective mode component that is screened and retained from the first intrinsic mode function component and is associated with the current channel sampling curve.

[0021] The CCS (Computer-Side Controller) may include conductive copper or aluminum busbars, FPC (Functional Circuit Board), FFC (Functional Circuit Board), or PCB (PCB) signal acquisition lines, insulating structural components, temperature sensors, connectors, and sampling nodes connected to the cell tabs. Multi-channel raw acquisition signals may include cell voltage and cell temperature signals. The CCS signal processing equipment can be located between the battery module acquisition link and the BMS (Battery Management System) host, and / or integrated into the CCS acquisition board, BMS slave board, or battery module acquisition controller.

[0022] Unless otherwise stated, Figures 1 to 10Solid arrows indicate the direction of transmission of sampling signals, digital data, or processing results; dashed boxes indicate hardware boundaries, software function groups, or processing stages, but do not indicate electrical or structural connections; multiple boxes arranged side-by-side and merged into the same processing module indicate parallel inputs or parallel components, and multiple boxes arranged sequentially along the arrows indicate the order of the corresponding processing steps.

[0023] Please refer to the instruction manual appendix. Figure 1 The CCS cell acquisition signal processing system includes, in sequence along the transmission direction of the sampling signal and digital data, the cell and CCS sampling node, FPC, FFC or PCB acquisition line, signal acquisition front end, synchronous sampling ADC, core processing control unit, isolated communication interface and BMS host.

[0024] The battery cell and CCS sampling nodes are used to generate voltage and temperature sampling signals corresponding to multiple battery cells; the FPC, FFC, or PCB acquisition lines are used to aggregate the sampling signals generated by different sampling nodes to the signal acquisition front end; the signal acquisition front end is used to perform at least one of the following processing on each sampling channel: channel selection, signal conditioning, filtering, and input protection, and outputs the sampling signal to be converted to the synchronous sampling ADC; the synchronous sampling ADC is used to convert the analog sampling signals of multiple channels into multi-channel digital sampling data with synchronization time identifiers under the same sampling trigger.

[0025] Figure 1 In the process, the analog switch or channel selection unit is used to select the voltage or temperature channel to be sampled according to the channel configuration; the signal conditioning unit is used to perform at least one of the following processing methods on the selected analog sampling signal: filtering, amplification, and impedance matching; the anti-aliasing filter unit is used to suppress high-frequency interference exceeding the effective sampling frequency band of the synchronous sampling ADC; the reference and bias management unit is used to provide a reference voltage or operating bias to the signal acquisition front end; and the protection circuit is used to limit electrostatic discharge, overvoltage, or transient impacts. The digital isolation unit in the isolated communication interface is used to achieve electrical isolation between the CCS signal processing equipment and the BMS host, and the communication transceiver unit is used to send the cell acquisition values ​​according to a preset communication protocol.

[0026] The core processing control unit is used to run the driver layer, preprocessing layer, core algorithm layer, and application output layer. The driver layer is used to perform sampling triggering, register configuration, and communication transmission and reception; the preprocessing layer is used to perform synchronization consistency verification, outlier completion, and normalization; the core algorithm layer is used to perform VMD decomposition, modal component correlation filtering, wavelet thresholding, UKF state estimation, and multi-channel anomaly compensation; the application output layer is used to perform range restoration, output verification, and BMS framing.

[0027] The isolated communication interface is used to send the cell acquisition values ​​output by the core processing control unit to the BMS host. The core processing control unit is a signal processing unit inside the CCS acquisition signal processing equipment, while the BMS host is an external acquisition result receiving and control device used to perform status monitoring, equalization control, and safety judgment based on the cell acquisition values. Therefore, the core processing control unit and the BMS host have different deployment locations and functions and are not duplicate modules.

[0028] Figure 1 The hardware link arrows indicate the transmission direction of cell voltage signals, temperature signals, or converted digital sampled data; the arrows between the various software functional layers within the core processing control unit indicate the data processing order, but do not indicate that there are independent physical connections between the software functional layers.

[0029] Please refer to the instruction manual appendix. Figure 2 The CCS bus can be set above the battery module and connected to the tab area of ​​multiple cells. The FPC acquisition line extends along the CCS bus. Voltage sampling points and temperature sampling points are arranged at positions corresponding to the cell tabs or temperature detection positions. The signal direction indicates that the voltage and temperature signals are collected by the FPC acquisition line and enter the CCS acquisition signal processing equipment. After the processing equipment completes synchronous sampling, cascaded noise reduction and channel compensation, it outputs the acquisition results to the BMS host through CAN and / or CANFD.

[0030] See the attached instruction manual for further details. Figure 2 , Figure 2 This describes the formation scenario of the multi-channel raw acquisition signal in step S101. A CCS connection component or bus is positioned above the battery module and connected to the tab areas of multiple battery cells. An FPC flexible acquisition line extends along the CCS connection component, with voltage and temperature sampling points arranged at positions corresponding to the battery cell tabs and temperature detection locations, respectively.

[0031] Figure 2 In the diagram, solid circles represent voltage sampling points, and hollow circles represent temperature sampling points. The connecting lines between each sampling point and its corresponding battery cell or temperature detection location indicate the correspondence between sampling locations. The FPC flexible acquisition circuit is used to collect the sampling signals generated by each sampling point to the CCS acquisition signal processing equipment. The multiple battery cells and multiple sampling points in the diagram are for illustrative purposes only and do not limit the number of battery cells or sampling channels in the battery module.

[0032] The voltage and temperature sampling signals are collected via the FPC flexible acquisition line and then input into the CCS signal processing device to form the multi-channel raw acquisition signal described in step S101. The CCS signal processing device performs subsequent synchronous sampling, cascaded noise reduction, anomaly identification, and channel compensation on the multi-channel raw acquisition signal, and outputs the verified cell acquisition value to the BMS host via CAN or CANFD. Therefore, Figure 2 This indicates the physical source of the multi-channel raw acquisition signals and the application objects of the processing results.

[0033] This application provides a method for processing CCS acquired signals, which is applied to CCS acquired signal processing equipment. The specific steps of the method are S101 to S106: S101, acquire the multi-channel raw acquisition signals of multiple battery cells collected by CCS. The multi-channel raw acquisition signals include at least the voltage signal and temperature signal of the battery cells. The CCS acquisition signal processing device can receive the voltage and temperature sample values ​​from multiple battery cells through the sampling front end, synchronous sampling ADC, isolated communication interface and / or battery module acquisition controller, and write the channel number, battery cell position, sampling frame number and sampling time into the same data record, so that subsequent mode decomposition and channel compensation can be traced back to the corresponding battery cell and the corresponding sampling window.

[0034] S102, based on the current channel sampling curve in the multi-channel original acquisition signal, determine each first intrinsic mode function component. The first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The current channel sampling curve can be a continuous sampling sequence of a voltage channel and / or a temperature channel within the current sampling window. The processing device uses this sequence as the variational mode decomposition object and maintains the channel position correspondence between the sequence and the original multi-channel sampling matrix.

[0035] Please refer to the instruction manual appendix. Figure 4 The multi-channel data frames with synchronization time identifiers are sequentially processed through synchronization consistency verification, burst outlier identification, adjacent valid sampling point completion, and standardization to form standardized multi-channel sampling data. The processing device simultaneously records the range boundaries and standardization parameters corresponding to each channel for range restoration in step S106. Figure 4 The multi-channel raw sampling matrix in the image represents the preprocessing input, and the standardized multi-channel sampling data represents the preprocessing output.

[0036] Figure 4In the multi-channel original sampling matrix, ti represents the i-th sampling time within the current sampling window, where i = 1, 2, ..., M, and M represents the total number of sampling points in the current sampling window; channel j represents the j-th sampling channel participating in the current processing, where j = 1, 2, ..., N, and N represents the total number of sampling channels participating in the processing within the current sampling window. The data cell located in the i-th row and i-th column of the multi-channel original sampling matrix represents the original sampled value corresponding to the j-th sampling channel at sampling time ti; the standardized multi-channel sampling data maintains the same row and column correspondence, and each data cell represents the standardized sampled value obtained after standardization processing of the corresponding original sampled value.

[0037] In one possible example, the specific implementation of step S102, which determines each first intrinsic mode function component based on the current channel sampling curve in the multi-channel raw acquisition signal, includes steps S1021 to S1025: S1021, Perform synchronization consistency verification on the multi-channel raw acquisition signal with synchronization time identifier; wherein, the synchronization time identifier can be the sampling trigger number, hardware count value, data frame timestamp and / or ADC sampling sequence number, the processing device can compare the deviation between the time identifier of each channel in the same sampling frame and the reference time identifier, and mark the sampling frame with the deviation exceeding the synchronization consistency condition as the frame to be verified.

[0038] S1022, identify sudden outliers in the multi-channel raw acquisition signal after synchronization consistency verification, and use adjacent valid sampling points to complete the identified sudden outliers; wherein, sudden outliers can be identified by window mean, window median value, the direction of change of neighboring sampling points and / or the synchronous change relationship of adjacent channels. When completing, linear interpolation of adjacent valid sampling points can be used to maintain the previous valid value, and the change amount of the same type of channel can be corrected and / or the sampling point can be marked as a value to be verified.

[0039] S1023: Obtain standardized multi-channel sampling data based on the completed multi-channel original acquisition signal; wherein, standardization can be performed according to the channel range boundary, historical effective window boundary, voltage sampling range, NTC effective temperature measurement range and / or range parameters issued by BMS. The standardization result retains the original channel number and range restoration record for subsequent output recovery.

[0040] S1024, Select a single channel sampling sequence within the current sampling window from the multi-channel sampling data as the current channel sampling curve; wherein, the current sampling window can be a fixed-length sliding window, an adaptive-length window and / or a window truncated according to the operating condition event, and the single channel sampling sequence can be selected from the voltage channel or the temperature channel, and is saved together with the sampling window identifier and the sampling time sequence.

[0041] S1025, Perform variational mode decomposition on the current channel sampling curve to obtain each first intrinsic mode function component. The variational mode decomposition can be performed by setting initial parameters based on the current channel's sampling rate, window length, signal type, and / or expected frequency band variation, and decomposes the current channel sampling curve into multiple sub-signal components that vary around different center frequencies.

[0042] Please refer to the instruction manual appendix. Figure 5 The current channel sampling curve serves as the input for VMD decomposition. The processing device decomposes the current channel sampling curve according to the number of modes K, the bandwidth constraint parameter α, and the iteration stopping condition, obtaining multiple parallel first intrinsic mode function components. Each first intrinsic mode function component belongs to the parallel results generated by the same VMD decomposition and there is no sequential generation relationship between them.

[0043] Figure 5 In this context, K represents the expected number of modal components obtained after VMD decomposition of the current channel sampling curve, where K is a positive integer; a represents the bandwidth constraint parameter, used to control the bandwidth concentration and smoothness of each modal component; IMFi represents the i-th first intrinsic mode function component, where = 1, 2, ..., K. Figure 5 The IMF1 to IMF4 shown are only for illustrating modal components with different center frequencies and trends of change. They do not imply that the actual decomposition must yield four components, nor do they indicate that IMF1 to IMF4 have a fixed category in all sampling windows.

[0044] In one possible example, step S1025, which performs variational mode decomposition on the current channel sampling curve to obtain the components of each first intrinsic mode function, includes steps S10251 to S102514: S10251, determine the initial parameters of mode decomposition based on the signal type, sampling rate and current sampling window length corresponding to the current channel sampling curve; wherein, the signal type can be a voltage signal or a temperature signal, the sampling rate can be determined by the driver layer sampling configuration, the current sampling window length can be determined by the BMS host and / or the local configuration of the processing device, and the initial parameters of mode decomposition can include the number of decompositions, the degree of bandwidth constraint, the initial value of the center frequency and / or the iteration stopping condition.

[0045] In one possible example, the initial parameters for mode decomposition are determined based on the signal type, sampling rate, and current sampling window length corresponding to the current channel sampling curve. The specific steps are as follows: S102511, based on the signal type corresponding to the voltage signal or temperature signal of the current channel sampling curve, determine the expected change amplitude and expected change frequency band of the current channel sampling curve; wherein, the expected change of the voltage signal may be related to the charging and discharging state, the equalization state and / or the transient change of the sampling circuit, and the expected change of the temperature signal may be related to the thermal conduction hysteresis, the NTC installation position and / or the heat dissipation state of the battery module.

[0046] S102512, based on the sampling rate and the current sampling window length, determine the range of signal changes that the current channel sampling curve can represent within the current sampling window; wherein, the range of signal changes can be determined by the window duration, the number of sampling points, the resolvable frequency band and / or the sampling interval. If the window is too short, the processing device can reduce the number of decompositions or extend the window.

[0047] S102513, Based on the expected change amplitude, expected change frequency band, and signal change range, determine the initial parameters for mode decomposition to control the number of decompositions and the smoothness of components; wherein, the parameter controlling the number of decompositions can correspond to the number of modes, and the parameter controlling the smoothness of components can correspond to the degree of bandwidth constraint, and both can be adjusted according to the channel type and / or historical window decomposition records.

[0048] S102514, when the decomposition result obtained based on the initial parameters of mode decomposition has component aliasing or reconstruction deviation that does not meet the preset conditions, the initial parameters of mode decomposition are adjusted, and variational mode decomposition is re-executed. Among them, component aliasing can be manifested as two components having similar center frequencies, component waveform repetition, or effective variations and high-frequency disturbances being indistinguishable. The processing device can increase or decrease the number of decompositions, adjust the smoothing constraints, and / or change the initial value of the center frequency before re-decomposing.

[0049] As can be seen, in this example, by binding the initial parameters of mode decomposition with the signal type, sampling rate and window length, the processing device can use different parameter combinations when dealing with voltage channels and temperature channels, avoiding the use of fixed parameters to process sampling curves with different physical meanings.

[0050] S10252, according to the initial parameters of mode decomposition, the current channel sampling curve is decomposed into multiple bandwidth-limited sub-signals that vary around different center frequencies; wherein, the bandwidth-limited sub-signals can be used to express the low-frequency trend, effective changes, high-frequency disturbances and / or local transient changes in the current channel sampling curve, and each bandwidth-limited sub-signal is consistent with the current channel sampling curve in terms of the number of sampling points and time sequence.

[0051] S10253, during the decomposition iteration process, the center frequency and waveform of each bandwidth-limited sub-signal are updated, and the reconstruction deviation between the superimposed bandwidth-limited sub-signals and the current channel sampling curve is used to determine whether the iteration stop condition is met; wherein, the iteration stop condition may be related to the reconstruction deviation, the center frequency change amplitude, the upper limit of the number of iterations and / or the component stability state. If the reconstruction deviation continues to be high, the processing device may retain the decomposition anomaly mark.

[0052] S10254, the bandwidth-limited sub-signals obtained when the iteration stopping condition is met are identified as the components of the first intrinsic mode function. These components can be subsequently recorded according to their center frequency order, energy percentage order, and / or degree of correlation with the original curve, for later use in the component selection stage.

[0053] In one possible example, the specific implementation of step S10254, which determines each bandwidth-limited sub-signal obtained when the iteration stopping condition is met, as each first intrinsic mode function component, includes steps S102541 to S102544: S102541, sort the components according to their center frequencies from low to high or from high to low; the sorting direction can be determined by system configuration, channel type and / or the calling habits of subsequent filtering algorithms, and the sorting result is used to maintain the consistency between component number and center frequency.

[0054] S102542 records the channel identifier, sampling window identifier, and sampling time correspondence for each bandwidth-limited sub-signal, ensuring that the sampling points of each bandwidth-limited sub-signal are aligned with the current channel sampling curve. The channel identifier may include the channel number, cell serial number, and / or sensor position; the sampling window identifier may include the window number and / or window start and end time; and the sampling time correspondence is used to ensure that sampling point misalignment does not occur during subsequent reconstruction.

[0055] S102543, based on the changing trend of each bandwidth-limited sub-signal, each bandwidth-limited sub-signal is marked as at least one of low-frequency trend component, effective change candidate component, or high-frequency disturbance candidate component; wherein, the changing trend can be determined based on the center frequency, amplitude change, continuity, peak-valley distribution and / or the relationship with the changing direction of the original curve, and a single component can be assigned different labels in different windows.

[0056] S102544: After sorting, recording, and labeling, each bandwidth-limited sub-signal is used as a first intrinsic mode function component. The first intrinsic mode function component includes not only the component waveform but also sorting information, channel information, window information, sampling time information, and / or candidate type labels.

[0057] As can be seen in this application example, since the identifier and tag are saved simultaneously when the bandwidth-limited sub-signal is converted into the first intrinsic mode function component, the subsequent correlation evaluation stage can determine the source and physical meaning of the component, avoiding mechanical screening based solely on frequency.

[0058] As can be seen, in this example, due to the synchronous verification, outlier completion, standardization, window selection and VMD decomposition steps being presented in the same implementation link, the current channel sampling curve already has time alignment and range record before entering mode decomposition, and the first intrinsic mode function component obtained subsequently can continue to maintain channel traceability.

[0059] S103, determine the second intrinsic mode function component based on each first intrinsic mode function component and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are related to the current channel sampling curve. The processing device can judge the correlation state between each first intrinsic mode function component and the current channel sampling curve from the perspectives of change synchronization degree, information correlation degree and / or energy contribution degree, and remove candidate components that exhibit isolated jitter, component aliasing, non-sampling condition related drift and / or transient spikes from the reconstruction link.

[0060] Please refer to the instruction manual appendix. Figure 6 Each first intrinsic mode function component is used as a parallel input and is evaluated based on its degree of synchronization of changes, degree of information correlation, and degree of energy contribution. First intrinsic mode function components that meet the component selection criteria are identified as valid candidate components and used to form second intrinsic mode function components; first intrinsic mode function components that do not meet the component selection criteria are identified as interference candidate components and are either eliminated or not involved in subsequent reconstruction.

[0061] Figure 6 In this context, 'n' represents the total number of first intrinsic mode function components obtained after VMD decomposition of the current sampling window. Without component deletion or merging, 'n' can be equal to... Figure 5 The number of modes K in the model is equal.

[0062] In one possible example, the second intrinsic mode function components are determined based on each first intrinsic mode function component and the current channel sampling curve. The specific steps are as follows: S1031, determine the degree of synchronization of changes, degree of information correlation, and degree of energy contribution between each first intrinsic mode function component and the current channel sampling curve; wherein, the degree of synchronization of changes can reflect the consistency between the component and the current channel sampling curve at rising, falling and / or turning points, the degree of information correlation can reflect the degree to which the component carries the effective information of the original curve, and the degree of energy contribution can reflect the contribution of the component to the change of the current window amplitude.

[0063] S1032, based on the degree of change synchronization, the degree of information correlation, and the degree of energy contribution, generate component correlation evaluation results corresponding to each first intrinsic mode function component; wherein, the component correlation evaluation results can be grade results, ranking results, confidence labels, and / or numerical evaluation results, and the processing device can select different expression methods according to the window state.

[0064] S1033, Based on the component correlation evaluation results, remove interference candidate components with low correlation to the current channel sampling curve from each first intrinsic mode function component; wherein, interference candidate components may include high-frequency isolated spike components, local jitter components unrelated to charging and discharging conditions, low-frequency abnormal drift components and / or components that do not match the temperature change trend.

[0065] In one possible example, based on the component correlation evaluation results, interference candidate components with low correlation to the current channel sampling curve are removed from each first intrinsic mode function component. The specific steps are as follows: S10331, Based on the distribution state of the component correlation evaluation results of each first intrinsic mode function component, determine the component screening conditions corresponding to the current sampling window; wherein, the distribution state can be the relative high-low relationship, dispersion, window ranking relationship and / or historical window comparison relationship between multiple component correlation evaluation results, and the component screening conditions can be determined according to fixed conditions and / or adaptive conditions.

[0066] S10332, the first intrinsic mode function component whose component correlation evaluation result does not meet the component screening conditions and exhibits jitter, isolated abrupt change, or inconsistent with the change trend of the current channel sampling curve is identified as an interference candidate component; among them, jitter can be manifested as short-term high-frequency back-and-forth changes, isolated abrupt change can be manifested as a single point or a few sampling point spikes, and inconsistent with the change trend can be manifested as the component continuously decreasing when the original curve rises and / or the component fluctuating abnormally when the original curve is stable.

[0067] S10333, the first intrinsic mode function component that meets the component screening conditions and matches the physical change trend of the voltage signal or temperature signal is determined as a valid candidate component; wherein, the physical change trend of the voltage signal may include changes caused by charging and discharging switching, equalization action and / or changes in cell load, and the physical change trend of the temperature signal may include slow heating, slow cooling and / or changes in thermal diffusion hysteresis.

[0068] S10334, a second intrinsic mode function component is formed based on the valid candidate components, and the channel identifier and sampling time correspondence corresponding to the second intrinsic mode function component are retained. The second intrinsic mode function component can be used as input for subsequent wavelet processing and continues to carry the channel identifier, sampling window identifier, sampling time, and / or valid candidate marker.

[0069] As can be seen, in this example, by combining the component selection criteria with the current window distribution state, the processing device can determine the retained objects based on the actual component performance in the window, thus avoiding misjudging real changes in voltage or temperature as noise.

[0070] S1034, the first intrinsic mode function component remaining after removing interfering candidate components is determined as the second intrinsic mode function component. The second intrinsic mode function component can consist of one or more valid candidate components and can vary with the sampling window, channel type, and / or operating conditions.

[0071] As can be seen, in this example, since the interference candidate components are screened out by the correlation evaluation results, the second intrinsic mode function component can retain the mode components that are actually related to the current channel sampling curve, providing a more stable input for subsequent wavelet thresholding and UKF correction.

[0072] S104, Based on the second intrinsic mode function component, obtain the non-abnormal target channel processing signal and the abnormal target channel processing signal; wherein, the processing device can continue to perform wavelet threshold denoising and / or UKF baseline correction on the second intrinsic mode function component to obtain the target channel processing signal with channel identification, and then distinguish the non-abnormal target channel processing signal and the abnormal target channel processing signal according to the correlation between channels of the same type, continuity and / or range rationality.

[0073] Please refer to the instruction manual appendix. Figure 7 The second intrinsic mode function component is first decomposed by wavelet to obtain a low-frequency approximation part and a high-frequency detail part. The processing device performs adaptive thresholding on the high-frequency detail part and reconstructs the processed high-frequency detail part and the low-frequency approximation part to obtain a preliminary denoised channel signal. Then, the processing device constructs state tracking data based on the preliminary denoised channel signal and performs state estimation and baseline drift correction through UKF to obtain the target channel processed signal.

[0074] Please refer to the instruction manual appendix. Figure 8 Multiple target channel processing signals are divided into voltage channel processing signal groups and temperature channel processing signal groups according to the channel signal type. The processing equipment forms a correlation matrix within the channel processing signal group corresponding to the same channel signal type, and identifies abnormal target channel processing signals and non-abnormal target channel processing signals by combining the continuity and range rationality of each target channel processing signal.

[0075] In one possible example, the processing signals for the target channel without anomalies and the target channel with anomalies are obtained based on the second intrinsic mode function components. The specific steps are as follows: S1041, perform wavelet decomposition on the second intrinsic mode function components to obtain the corresponding low-frequency approximation part and high-frequency detail part; wherein, the wavelet decomposition level can be determined according to the channel type, sampling rate, window length and / or number of effective candidate components, the low-frequency approximation part is used to preserve the slow changes of the channel, and the high-frequency detail part is used to represent short-time disturbances and edge changes.

[0076] S1042, determine the adaptive threshold processing conditions based on the perturbation level of the high-frequency detail and the change amplitude of the current channel sampling curve; wherein, the perturbation level can be determined based on the amplitude distribution, continuity, peak density and / or difference from the historical window of the high-frequency detail, and the adaptive threshold processing conditions can be changed according to the voltage channel and temperature channel.

[0077] S1043, according to the adaptive threshold processing conditions, the high-frequency detail part is shrunk, and the shrunk high-frequency detail part is reconstructed with the low-frequency approximation part to obtain the preliminary denoised channel signal; wherein, the shrunk processing can be soft threshold shrunk, hard threshold truncation, segmented threshold processing and / or local shrunk that retains edge changes, and the channel identifier and sampling time correspondence are maintained during reconstruction.

[0078] S1044, construct state tracking data based on the preliminary denoised channel signal, and perform baseline drift correction and state estimation on the state tracking data through unscented Kalman filtering to obtain the target channel processing signal; wherein, the state tracking data may include the current sampling point value, the estimated value of the previous sampling point, the sampling time interval, the channel change trend, the baseline drift term and / or the range boundary, and the target channel processing signal is the channel signal result after baseline correction.

[0079] S1045, according to the channel signal type, the multiple target channel processing signals are divided into voltage channel processing signal group and temperature channel processing signal group; wherein, the channel signal type can be determined according to the channel basic file, sampling interface configuration, sensor type and / or BMS issued configuration, and the same or similar physical judgment rules are used for the correlation evaluation of channels of the same type.

[0080] S1046. Within the channel processing signal group corresponding to the same channel signal type, determine the correlation between the target channel processing signals to form a multi-channel correlation evaluation result. The correlation can come from the current window, historical synchronous sampling window, physical adjacent channel set and / or the same group of temperature sampling points. The multi-channel correlation evaluation result can be represented as a correlation matrix, correlation level or reference channel sorting.

[0081] S1047. Based on the multi-channel correlation evaluation results, the continuity of the processed signals of each target channel, and the reasonableness of the range corresponding to the processed signals of each target channel, identify abnormal target channel processed signals and non-abnormal target channel processed signals. Among them, the abnormality identification can simultaneously refer to correlation deviation, continuity interruption, range out-of-bounds, low-frequency drift and / or short-term spikes, while non-abnormal target channel processed signals can be used as a reference for subsequent compensation.

[0082] As can be seen, in the example of this application, since the second intrinsic mode function component is further processed by wavelet threshold denoising, UKF state estimation and multi-channel correlation identification, the target channel processing signal of a single channel not only completes internal denoising, but can also enter the same type of channel group for anomaly judgment.

[0083] S105, after compensating the abnormal target channel processing signal with the abnormal target channel processing signal, the compensated multi-channel sampling data is obtained; wherein, the abnormal target channel processing signal can come from physically adjacent channels, stable channels of the same type and / or reference channels in the historical synchronous sampling window whose correlation meets the conditions, and the compensation method can be to replace the abnormal channel data, to fuse the abnormal channel data and reference channel data according to weight, or to generate the compensation data to be reviewed while retaining the abnormal marker.

[0084] Please refer to the instruction manual appendix. Figure 9 The abnormal target channel processing signal, the historical synchronous sampling window, and the reference channel set consisting of channels corresponding to channels without abnormal target channel processing signals are used as compensation basis information. Based on this compensation basis information, the processing device determines the weighted compensation relationship, generates the compensation signal corresponding to the abnormal target channel processing signal, and forms compensated multi-channel sampling data through replacement or fusion methods.

[0085] In one possible example, after compensating the abnormal target channel processed signal with the non-abnormal target channel processed signal, the compensated multi-channel sampled data is obtained. The specific steps are as follows: S1051, Select a reference channel processing signal from the target channel processing signals without anomalies, whose correlation with the target channel processing signals with anomalies meets the preset compensation conditions; wherein, the reference channel processing signal may come from physically adjacent channels, channels of the same type with high historical correlation, current window stable channels and / or backup reference channels configured by BMS, and the preset compensation conditions may include correlation conditions, reasonable range conditions and / or continuity conditions.

[0086] S1052, based on the correspondence between the reference channel processed signal and the abnormal target channel processed signal within the historical synchronous sampling window, determine the weighted compensation relationship, and generate the compensation signal corresponding to the abnormal target channel processed signal according to the weighted compensation relationship; wherein, the historical synchronous sampling window can record the proportional relationship, difference relationship, trend relationship and / or fitting residual between the abnormal channel and the reference channel under normal conditions, and the weight can be determined according to the historical correlation degree, physical adjacency degree, current window stability degree and / or reference channel reliability.

[0087] S1053, the target channel processing signal with the compensation signal is replaced or fused to obtain the compensated multi-channel sampling data. The replacement method can be used for short-term sudden anomalies, and the fusion method can be used for anomalies with slight offsets or where some effective trends still exist. The compensated multi-channel sampling data can continue to carry compensation labels, anomaly type labels, and / or confidence labels.

[0088] As can be seen, in this example, since the compensation relationship between the non-abnormal target channel processing signal and the abnormal target channel processing signal is established on the basis of historical synchronous sampling window and correlation degree, the compensation signal is not an isolated estimation result, but is generated by the redundancy relationship of the same type of channel, which enables the abnormal channel to continue to form verifiable acquisition values ​​in short-term abnormal scenarios.

[0089] S106, after restoring and compensating the multi-channel sampling data, outputs the collected values ​​of multiple cells.

[0090] The processing equipment can perform inverse standardization according to the standardized boundaries, channel ranges and / or sensor conversion relationships saved in the preprocessing stage, so that the voltage channel is restored to the actual voltage value and the temperature channel is restored to the actual temperature value, and outputs it to the BMS host after passing the range verification and / or continuity verification.

[0091] Through steps S101 to S106 above, this application first performs modal decomposition on the current channel sampling curve within a single channel, and then filters effective modal components based on the correlation between each modal component and the current channel sampling curve. Next, it identifies abnormal target channel processing signals among multiple channels of the same type, and uses the processing signals of channels without abnormal targets to perform redundancy compensation on the abnormal target channel processing signals, thus forming a continuous processing chain between intra-channel noise reduction and inter-channel anomaly correction. This reduces the risk of local sampling link anomalies being transmitted to the final cell acquisition value, improves the stability of the CCS acquisition signal, and enhances the reliability of subsequent BMS status monitoring.

[0092] As can be seen, compared to the existing CCS cell acquisition signal processing methods that mainly rely on single-stage filtering or independent processing of each channel, the cell connection system acquisition signal processing method provided in this application performs mode decomposition on the signals of each channel to be processed, and filters effective mode components based on the information correlation between each mode component and the corresponding channel signal to be processed. This retains effective components related to the actual acquisition changes within the channel, reducing the impact of non-stationary interference components on the signal reconstruction results. After obtaining the channel processed signals of each channel, abnormal channels are identified based on the correlation between processed signals of the same type of channel, and redundant compensation is performed on the abnormal channel signals using processed signals of the same type of channel that are not identified as abnormal. This forms a continuous processing chain between the noise reduction results within the channel and the anomaly correction between channels. This reduces the risk of local sampling link anomalies being transmitted to the final output results, which is beneficial to improving the stability of CCS acquisition signals and the reliability of subsequent state monitoring of the battery management system.

[0093] Please refer to the instruction manual appendix. Figure 10 The CCS signal processing device includes a raw signal acquisition module, a mode decomposition module, an effective component determination module, a channel signal generation module, a multi-channel compensation module, and a acquired value output module, all located within the same device boundary. Each module sequentially transmits the processing results. The external CCS multi-channel raw acquisition signal is input to the raw signal acquisition module, and the acquired value output module outputs the restored and verified cell acquisition value to the BMS host. The CCS signal processing device in this application includes: The raw signal acquisition module is used to acquire multi-channel raw acquisition signals of multiple cells collected by CCS. The multi-channel raw acquisition signals include at least the voltage signal and temperature signal of the cells. The mode decomposition module is used to determine each first intrinsic mode function component based on the current channel sampling curve in the multi-channel original acquisition signal. Each first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The effective component determination module is used to determine the second intrinsic mode function component based on each of the first intrinsic mode function components and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are associated with the current channel sampling curve. The channel signal generation module is used to obtain the target channel processing signal based on the second intrinsic mode function component, wherein the target channel processing signal includes a target channel processing signal without abnormalities and a target channel processing signal with abnormalities. The multi-channel compensation module is used to compensate for the abnormal target channel processing signal by using the target channel processing signal without abnormality, and then obtain the compensated multi-channel sampling data. The acquisition value output module is used to restore the compensated multi-channel sampling data and output the acquisition values ​​of multiple cells.

[0094] As can be seen, in this embodiment, compared to the existing CCS cell acquisition signal processing methods that mainly rely on single-stage filtering or independent processing of each channel, the device in this application performs mode decomposition on the signals of each channel to be processed, and filters effective mode components based on the information correlation between each mode component and the corresponding signal of the channel to be processed. This retains effective components related to the actual acquisition changes within the channel, reducing the impact of non-stationary interference components on the signal reconstruction results. After obtaining the channel processing signals of each channel, abnormal channels are identified based on the correlation between the processed signals of the same type of channel, and redundant compensation is performed on the abnormal channel signals using the processed signals of the same type of channel that are not identified as abnormal. This forms a continuous processing chain between the noise reduction results within the channel and the anomaly correction between channels. This reduces the risk of local sampling link anomalies being transmitted to the final output results, which is beneficial to improving the stability of CCS acquisition signals and the reliability of subsequent state monitoring of the battery management system.

[0095] Example 1 Please refer to the instruction manual appendix. Figure 3 The CCS signal acquisition and processing device executes the CCS signal acquisition and processing method in the order of steps S101 to S106.

[0096] Step S101 involves acquiring multi-channel raw acquisition signals from multiple battery cells. The processing device receives voltage and temperature signals from the CCS acquisition link. Step S102 involves determining the first intrinsic mode function component based on the current channel sampling curve. The processing device performs synchronization consistency verification, outlier completion, and standardization on the multi-channel raw acquisition signals. The current channel sampling curve is selected from the standardized multi-channel sampling data, and VMD decomposition is performed on the current channel sampling curve. Step S103 involves determining the second intrinsic mode function component based on correlation evaluation. The processing device filters out interfering candidate components based on the degree of synchronization of changes, information correlation, and energy contribution between the first intrinsic mode function component and the current channel sampling curve.

[0097] Step S104 involves generating the target channel processing signal through wavelet threshold denoising and UKF correction, and identifying abnormal target channel processing signals and non-abnormal target channel processing signals within the same type of channel group; Step S105 involves using the non-abnormal target channel processing signal to perform redundancy compensation on the abnormal target channel processing signal to form compensated multi-channel sampling data; Step S106 involves performing range restoration and output verification on the compensated multi-channel sampling data, and outputting the collected values ​​of multiple cells.

[0098] Figure 3 Step S101 in the text corresponds to Figure 10 The original signal acquisition module, step S102 corresponds to the mode decomposition module, step S103 corresponds to the effective component determination module, step S104 corresponds to the channel signal generation module, step S105 corresponds to the multi-channel compensation module, and step S106 corresponds to the acquired value output module. Therefore, Figure 3 The method flow shown is the same as Figure 10 The device modules shown form a one-to-one correspondence.

[0099] Example 2 Please refer to the instruction manual appendix. Figure 2 Multi-channel synchronous sampling can use an FPC flexible acquisition line as the signal input basis. The FPC flexible acquisition line is connected to the CCS bus and sampling pads. Multiple cells are arranged inside the battery module, and temperature sensors are placed at the locations where the temperature needs to be detected. The sampling signal travels along... Figure 2 The signal direction indicated enters the signal processing and system interaction area. After receiving multi-channel input, the CCS signal processing device performs synchronous sampling, cascaded noise reduction and channel compensation. The BMS host receives the verified acquisition value and uses it for status monitoring, equalization control and safety judgment.

[0100] During the synchronous sampling phase, the synchronous sampling ADC can read multiple voltage and temperature channels under the same sampling trigger. The driver layer writes a sampling frame number, channel number, and synchronization time identifier to each sampling frame. When the deviation between the channel time identifier and the reference time identifier of a certain sampling frame does not meet the synchronization consistency condition, the processing device marks the sampling frame as a frame to be reviewed. If the sampling frame affects the integrity of the current sampling window, the processing device can supplement or discard it by combining the valid sampling frames before and after it, so as to avoid data at different sampling times being mistakenly regarded as multiple channels within the same sampling window.

[0101] In the FPC flexible acquisition line, voltage sampling points, temperature sampling points, and CCS bus should be established in a corresponding relationship in the channel basic file. The channel basic file records the channel type, cell serial number, NTC installation location, sampling line location, nearest neighbor channel set, and range boundary. Subsequent preprocessing, VMD decomposition, correlation identification, and redundancy compensation all refer to this file. If the channel basic file lacks a nearest neighbor channel set for a certain channel, the processing equipment can first establish a temporary reference set according to the historical valid windows of the same type of channel and mark the corresponding sampled values ​​as output to be verified.

[0102] Example 3 Please refer to the instruction manual appendix. Figure 4 After completing multi-channel ADC synchronous sampling, the processing equipment first follows... Figure 4The upper process sequentially performs synchronization consistency verification, outlier completion, and standardization to form a preprocessed multi-channel sampling matrix. The original multi-channel sampling matrix can represent sampling time by row and sampling channel by column. The preprocessed multi-channel sampling matrix retains the same channel arrangement relationship and saves the channel identifier, sampling time, and standardized sampling value in the record corresponding to each matrix unit.

[0103] Synchronous consistency check is used to confirm whether multi-channel sampled data comes from the same sampling trigger. Outlier completion is used to correct single-point spikes, pulse interference, or abnormal points caused by sampling interruptions. Standardization is used to map voltage and temperature signals of different ranges to a unified numerical range. If there are not enough valid sampling points in the current window, the processing device can use the valid maximum and minimum values ​​in the historical sliding window as the standardization boundary. If the current channel is a temperature channel, the processing device can determine the standardization boundary in combination with the effective temperature measurement range of NTC.

[0104] The preprocessed data is not only used as input for subsequent algorithms, but also as a basis for anomaly tracing. If a channel is marked with a completion flag during the outlier completion stage, the flag will enter the VMD decomposition, wavelet thresholding, UKF baseline correction, and correlation compensation stages along with the sampling window. When the processing device outputs the collected values, it can convert the flag into verification information and send it to the BMS host. If the number of consecutive outliers in a window exceeds the preset number, the processing device can stop performing normal compensation for that window and list it as an abnormal window for confirmation by subsequent windows.

[0105] Example 4 Please refer to the instruction manual appendix. Figure 5 The processing device selects the current channel sampling curve from the preprocessed multi-channel sampling data and uses the curve as the input for VMD decomposition. The VMD decomposition parameters may include the number of modes K and the constraint parameter α. The number of modes K is used to control the number of components obtained by decomposition, and the constraint parameter α is used to control the degree of bandwidth constraint. The processing device determines the initial parameters based on whether the current channel is a voltage channel or a temperature channel, the sampling rate, the sampling window length, and the expected frequency band change.

[0106] The VMD decomposition module decomposes the current channel sampling curve into multiple first intrinsic mode function components. Figure 5 The example shows that IMF1 is a high-frequency jitter component, IMF2 and IMF3 are effective variation components, and IMF4 is a trend component. In actual processing, the number of components can be changed according to the value of K. The processing device updates the center frequency and waveform of each component in each iteration and stops the iteration when the reconstruction error and the center frequency change meet the preset conditions.

[0107] The processing device does not use frequency as the sole criterion for deleting components. If a high-frequency component keeps pace with the rapid changes in the voltage channel during charging and discharging, the high-frequency component can be retained as a valid candidate component. If a low-frequency component exhibits a slow drift that is inconsistent with the neighboring channel, the low-frequency component can be used as an interference candidate component for subsequent screening. If the decomposition result shows mode aliasing, the center frequency is unstable for a long time, or the reconstruction deviation does not meet the conditions, the processing device can adjust K or α and re-execute the VMD decomposition of the current window.

[0108] Example 5 Please refer to the instruction manual appendix. Figure 6 After obtaining the first intrinsic mode function components 1 to n, the processing device performs correlation evaluation on each component one by one. The evaluation content includes the degree of change synchronization, the degree of information correlation, and the degree of energy contribution. The degree of change synchronization is used to reflect whether the change direction of the component is consistent with the change direction of the current channel sampling curve. The degree of information correlation is used to reflect whether the component carries the effective change information in the current channel sampling curve. The degree of energy contribution is used to reflect the energy ratio of the component in the current sampling window.

[0109] The processing device determines the screening conditions based on the distribution of multiple components in the current window. If the component correlation evaluation result meets the screening conditions, the component is determined as a valid candidate component and forms the second intrinsic mode function component. If the component correlation evaluation result does not meet the screening conditions, the component is determined as an interference candidate component and is eliminated or does not participate in the reconstruction. After the screening is completed, the processing device retains the channel identifier, sampling window identifier and sampling time correspondence, so that the second intrinsic mode function component can still be traced back to the original sampling channel.

[0110] The second intrinsic mode function component is not a single fixed frequency band component, but is composed of valid candidate components that meet the current window screening conditions. Its number can change with the sampling window and channel type. Therefore, the processing device can retain signal components related to the actual voltage or temperature changes when strong electromagnetic interference, baseline drift and normal operating condition changes coexist. If the number of valid candidate components is lower than the number required to maintain reconstruction, the processing device can retain the original window and decomposition record and output a low confidence mark.

[0111] Example 6 Please refer to the instruction manual appendix. Figure 7 The processing device performs wavelet thresholding denoising and UKF baseline correction on the second intrinsic mode function component. Figure 7The upper signal-level noise reduction path includes the second intrinsic mode function component, adaptive wavelet decomposition, wavelet detail threshold shrinkage, and signal reconstruction. The adaptive wavelet decomposition divides the second intrinsic mode function component into a low-frequency approximation part and a high-frequency detail part. The threshold shrinkage step determines the threshold processing conditions based on the perturbation level of the high-frequency detail part and the change amplitude of the current channel sampling curve.

[0112] When the processing device shrinks the high-frequency details, it retains edge variations related to voltage or temperature changes, and suppresses isolated jitter and spike details. Then, it reconstructs the shrunken high-frequency details with the low-frequency approximation to obtain the preliminary denoised channel signal. The preliminary denoised channel signal is not directly output as the final acquisition value, but enters the signal tracking stage. If the current channel is a temperature channel, the threshold processing conditions can be more biased towards retaining the low-frequency approximation. If the current channel is a voltage channel, the threshold processing conditions can retain the rapid changes synchronized with the charging and discharging conditions.

[0113] The signal tracking path includes signal tracking data, UKF state estimation, and target channel processed signal. The signal tracking data may include the current sampling point value, the estimated value of the previous sampling point, the sampling time interval, the channel change trend, and the baseline drift term. The UKF state estimation is predicted and updated based on the state at the previous moment and the current observation value to obtain the target channel processed signal after baseline correction. If the UKF output deviates from the reasonable range within the continuous window, the processing device can fall back to the initial denoised channel signal and add a baseline correction anomaly mark.

[0114] Example 7 Please refer to the instruction manual appendix. Figure 8 The processing equipment takes the target channel processing signal of each channel as input and constructs target channel processing signal sets according to voltage channel and temperature channel respectively. Figure 8 The correlation matrix R is used to represent the correlation between channels. R is a multi-channel correlation matrix. Each element in the matrix corresponds to the correlation state between one channel and another channel in the current window or historical synchronous sampling window. The correlation state can be represented by high, medium, low or numerical correlation degree.

[0115] Figure 8In the diagram, Ci represents the i-th target channel processed signal in the channel processed signal group corresponding to the same channel signal type, i=1,2,...,Nc, where Nc represents the number of channels contained in the corresponding channel processed signal group. The examples C1 to C6 shown in the figure are only six channels, and the actual number of channels is not limited. The correlation matrix is ​​denoted as R=[rij], where rij represents the correlation between channel Ci and channel Cj within the current sampling window or historical synchronous sampling window; when using the normalized correlation coefficient, −1≤rij≤1, and the correlation coefficient rii=1 between the same channel and itself; when using the symmetric correlation evaluation method, rij=rji.

[0116] When performing anomaly identification, the processing device not only relies on the correlation relationships in the correlation matrix R, but also considers the range rationality, continuity, and synchronization window effectiveness of the target channel processed signal. If the correlation relationship between a certain channel and its reference channel set is lower than the normal distribution, and the channel exhibits at least one of the following characteristics: short-term spikes, continuity interruptions, abnormal range boundaries, or low-frequency drift, then the processing device identifies the channel as an abnormal target channel processed signal. If the correlation relationship of a certain channel is stable and the range and continuity meet the conditions, then the channel is identified as a target channel processed signal without anomalies.

[0117] The processing device outputs abnormal target channel processing signals and non-abnormal target channel processing signals based on the identification results. The correlation matrix R can also be used to determine the reference channel set. If a physically adjacent channel maintains a high degree of correlation with the abnormal channel in the historical synchronization window, the adjacent channel can be given priority as a reference channel. If the physically adjacent channel itself has an abnormal mark, the processing device selects other stable channels of the same type as compensation references. If there are not enough stable channels of the same type to form compensation references, the processing device can retain the target channel processing signal of the abnormal channel and output a mark to be reviewed.

[0118] Example 8 Please refer to the instruction manual appendix. Figure 9 The abnormal channel redundancy compensation path includes compensation basis information, weight and relationship determination stage, generation of compensation signal and compensation result. The compensation basis information includes abnormal channel processing signal, historical synchronous sampling window and reference channel set. The abnormal channel processing signal can be the channel to be compensated Cx, where Cx represents the channel to be compensated corresponding to the xth target channel processing signal identified as abnormal, where x is the channel index of the abnormal channel in the channel basic file. The historical synchronous sampling window is used to record the correspondence between Cx and reference channels. The reference channel set consists of channels corresponding to target channel processing signals without abnormalities.

[0119] The weight and relationship determination stage is used to determine the weighted compensation relationship. The weights in this relationship are related to the historical correspondence and the degree of correlation. The processing device can determine the compensation weights based on the historical correlation between the reference channel and the abnormal channel, the degree of physical adjacency, the stability of the current window, and the fitting residual. After the compensation weights are determined, the processing device generates a compensation signal and uses the compensation signal to replace or fuse the abnormal target channel processing signal.

[0120] The compensation result is the multi-channel sampled data after compensation. For short-term sudden anomalies, the processing device can replace the abnormal target channel processing signal with the compensation signal. For slightly offset channels that still retain some effective change trends, the processing device can fuse the compensation signal with the abnormal target channel processing signal. The fusion ratio changes with the degree of anomaly, correlation, and continuous deviation. The multi-channel sampled data after compensation then enters the range restoration and output verification stage. If there is a continuous directional conflict between the compensation signal and the abnormal target channel processing signal, the processing device will not output the frame as a normal acquisition value.

[0121] Example 9 Please refer to the instruction manual appendix. Figure 3 and Figure 10 During the output recovery phase, the processing equipment performs inverse normalization on the compensated multi-channel sampling data based on the standardized parameter records saved in the preprocessing phase, so that the voltage channel is restored to the actual voltage value and the temperature channel is restored to the actual temperature value. Subsequently, the processing equipment performs range verification, continuity verification and channel consistency verification on the restored physical quantities. If the verification is qualified, the data is framed according to the BMS communication protocol and sent to the BMS host. If the verification is unqualified, the corresponding sampling frame is marked as an abnormal frame or a frame to be reviewed.

[0122] The output of the acquired values ​​is not a simple data forwarding, but is linked with the aforementioned preprocessing markers, component filtering markers, abnormal channel markers, and compensation confidence markers. The processing device can carry the channel number, sampling window number, cell acquisition value, abnormal type marker, and confidence marker in the output message, so that the BMS host can distinguish between the original normal channel, the compensated channel, and the channel to be verified when performing status monitoring, equalization control, and safety judgment, and avoid mixing the compensation results with the original acquisition results.

[0123] Example 10 Please refer to the instruction manual appendix. Figure 10 The CCS signal processing device of this application includes a raw signal acquisition module, a mode decomposition module, an effective component determination module, a channel signal generation module, a multi-channel compensation module, and an acquisition value output module, all located within the same device boundary.

[0124] The raw signal acquisition module receives the raw acquisition signals from the CCS multi-channel; the mode decomposition module determines the first intrinsic mode function components based on the current channel sampling curve; the effective component determination module determines the second intrinsic mode function components based on the correlation between the first intrinsic mode function components and the current channel sampling curve; the channel signal generation module performs wavelet threshold denoising and UKF baseline correction on the second intrinsic mode function components, and identifies abnormal target channel processing signals and non-abnormal target channel processing signals; the multi-channel compensation module compensates for abnormal target channel processing signals using non-abnormal target channel processing signals; and the acquisition value output module performs range restoration and output verification on the compensated multi-channel sampling data, and outputs the acquisition values ​​of multiple cells to the BMS host.

[0125] Figure 10 The six modules are connected in the order of result transmission. Each module is a component of the CCS signal acquisition and processing device and is no longer divided into upper-level or lower-level module groups located outside the device. The modules can be deployed in the same processor or executed in conjunction with the sampling controller, signal processor, and communication processor.

[0126] Example 11 In one operational record example, a battery module is configured with 12 voltage sampling channels and 4 temperature sampling channels, a sampling rate of 200Hz, and a current sampling window length of 128 sampling points. The processing device first follows... Figure 4 Complete synchronization consistency verification, outlier completion, and standardization, and follow the instructions. Figure 5 VMD decomposition is performed on the current channel sampling curve. If the synchronization time deviation of all 16 channels is no greater than 1μs, the sampling frame group enters VMD decomposition. If the 7th voltage channel has an isolated spike at a certain sampling point, and the direction of change of the spike relative to the previous and next sampling points is discontinuous, the processing device uses the previous and next valid sampling points to complete it, and writes the outlier completion mark in the channel status record.

[0127] The 7th voltage channel enters. Figure 5 The VMD decomposition process shown involves the processing device determining VMD decomposition parameters based on the voltage channel type and sampling rate. If the decomposed high-frequency component exhibits isolated jitter and its trend is inconsistent with the current channel's sampling curve, then this component... Figure 6 In the correlation filtering process shown, components identified as interference candidates are removed, and the remaining valid components proceed to the next step. Figure 7 The wavelet thresholding denoising and UKF baseline correction process is shown.

[0128] If the 7th voltage channel is in Figure 8If the correlation matrix shown indicates a significant decrease in the correlation with adjacent voltage channels and a discontinuity interruption, then the processing device will identify channel 7 as an abnormal target channel signal and process it accordingly. Figure 9 The compensation path shown selects the channel corresponding to the target channel processing signal without abnormalities as the reference channel to generate the compensation signal.

[0129] Example 12 In one parameter and recording implementation, the processing device can store in memory a sampling configuration table, an algorithm parameter table, a channel basic profile, a historical valid window set, and an abnormal event record table. The sampling configuration table records the sampling rate, window length, synchronization time deviation conditions, and range boundaries. The algorithm parameter table records the number of modes, component filtering conditions, wavelet threshold processing conditions, and UKF state tracking parameters. The channel basic profile records the cell serial number, channel type, NTC location, FPC line location, and neighboring channel set. The abnormal event record table records the abnormal channel number, abnormal type, reference channel number, compensation method, output reliability, and verification results.

[0130] The numerical ranges and parameter settings in the above embodiments are used to illustrate the possible implementations of this application. Actual products can be adjusted according to cell type, module series-parallel structure, CCS line length, sampling front-end capability, and BMS communication cycle, as long as the processing equipment can meet the requirements. Figures 1 to 10 The relationship shown is that the current channel sampling curve is formed on the basis of synchronous sampling. Modal decomposition and effective component screening are performed on the current channel sampling curve. Then, wavelet threshold denoising, UKF baseline correction, multi-channel correlation anomaly identification and non-abnormal channel redundancy compensation are used to output the cell acquisition value. All of these can fall within the technical concept disclosed in this application.

[0131] In actual deployment, the processing equipment can be configured according to... Figure 1 The software architecture shown is for Figure 3 Each step in the process sets a running status flag. After step S101 is completed, a flag indicating the original signal acquisition is complete is written; after step S102 is completed, a flag indicating mode decomposition is complete is written; after step S103 is completed, a flag indicating effective component filtering is complete is written; after step S104 is completed, a flag indicating target channel processing signal generation and anomaly identification is complete is written; after step S105 is completed, a flag indicating anomaly channel compensation is complete is written; after step S106 is completed, a flag indicating range restoration and output verification is complete is written. These flags are respectively associated with… Figure 10 The output results of the original signal acquisition module, mode decomposition module, effective component determination module, channel signal generation module, multi-channel compensation module, and acquisition value output module correspond to each other.

[0132] when Figure 7 After the wavelet thresholding denoising and UKF baseline correction processes shown generate the target channel processed signal, the target channel processed signal enters... Figure 8 The multi-channel correlation matrix and abnormal channel identification process are shown. The processing device will... Figure 8 The output abnormal target channel processing signal and the non-abnormal target channel processing signal are transmitted to Figure 9 The abnormal channel redundancy compensation path is shown, and is... Figure 10 The multi-channel compensation module generates compensated multi-channel sampling data. If the same channel repeatedly shows an anomaly marker within multiple consecutive sampling windows, the processing device can retain the compensation output and write the number of consecutive anomalies and the reference channel number into the anomaly event log table so that the BMS host can verify the sampling link status.

[0133] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for acquiring and processing signals in a battery cell connection system, characterized in that, The method, applied to CCS signal acquisition and processing equipment, includes: Acquire multi-channel raw acquisition signals of multiple battery cells from CCS, wherein the multi-channel raw acquisition signals include at least the voltage signal and temperature signal of the battery cell; Each first intrinsic mode function component is determined based on the current channel sampling curve in the multi-channel original acquisition signal. The first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The second intrinsic mode function component is determined based on each of the first intrinsic mode function components and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are associated with the current channel sampling curve. The target channel processing signal is obtained based on the second intrinsic mode function component; According to the channel signal type, the multiple target channel processing signals are divided into voltage channel processing signal group and temperature channel processing signal group; Within the channel processing signal group corresponding to the same channel signal type, determine the correlation between the target channel processing signals to form a multi-channel correlation evaluation result; Based on the multi-channel correlation evaluation results, the continuity of the processed signals of each target channel, and the reasonableness of the range corresponding to the processed signals of each target channel, abnormal target channel processed signals and non-abnormal target channel processed signals are identified. After compensating the abnormal target channel processing signal with the non-abnormal target channel processing signal, the compensated multi-channel sampling data is obtained. After restoring the compensated multi-channel sampling data, the collected values ​​of multiple battery cells are output.

2. The method as described in claim 1, characterized in that, The determination of each first intrinsic mode function component based on the current channel sampling curve in the multi-channel original acquired signal includes: The synchronization consistency of the multi-channel raw acquisition signals with synchronization time identifiers is checked. After the synchronization consistency check, identify sudden outliers in the multi-channel raw acquisition signal and use adjacent valid sampling points to complete the identified sudden outliers; Based on the completed multi-channel original acquisition signal, standardized multi-channel sampling data is obtained; Select a single channel sampling sequence within the current sampling window from the multi-channel sampling data as the current channel sampling curve; Variational mode decomposition is performed on the current channel sampling curve to obtain the components of each first intrinsic mode function.

3. The method as described in claim 2, characterized in that, The step of performing variational mode decomposition on the current channel sampling curve to obtain the components of each first intrinsic mode function includes: The initial parameters for mode decomposition are determined based on the signal type, sampling rate, and current sampling window length corresponding to the current channel sampling curve. According to the initial parameters of the modal decomposition, the current channel sampling curve is decomposed into multiple bandwidth-limited sub-signals that vary around different center frequencies; During the decomposition iteration process, the center frequency and waveform of each bandwidth-limited sub-signal are updated, and the reconstruction deviation between the superimposed bandwidth-limited sub-signals and the current channel sampling curve is used to determine whether the iteration stop condition is met. Each bandwidth-limited sub-signal obtained when the iteration stopping condition is met is determined as each first intrinsic mode function component.

4. The method as described in claim 3, characterized in that, The step of determining the initial parameters for mode decomposition based on the signal type, sampling rate, and current sampling window length corresponding to the current channel sampling curve includes: Based on the signal type corresponding to the voltage or temperature signal of the current channel sampling curve, determine the expected change amplitude and expected change frequency band of the current channel sampling curve; Based on the sampling rate and the current sampling window length, determine the range of signal changes that the current channel sampling curve can represent within the current sampling window; Based on the expected change amplitude, the expected change frequency band, and the signal change range, determine the initial parameters for mode decomposition to control the number of decompositions and the smoothness of components; If the decomposition result obtained based on the initial parameters of the mode decomposition has component aliasing or reconstruction deviation that does not meet the preset conditions, the initial parameters of the mode decomposition are adjusted and variational mode decomposition is re-executed.

5. The method as described in claim 4, characterized in that, The step of determining each bandwidth-limited sub-signal obtained when the iteration stopping condition is satisfied as each first intrinsic mode function component includes: Sort the bandwidth-limited sub-signals from low to high or high to low according to their center frequencies; Record the channel identifier, sampling window identifier, and sampling time correspondence for each bandwidth-limited sub-signal to ensure that the sampling points of each bandwidth-limited sub-signal are aligned with the current channel sampling curve. Based on the changing trend of each bandwidth-limited sub-signal, each bandwidth-limited sub-signal is marked as at least one of low-frequency trend component, effective change candidate component, or high-frequency disturbance candidate component; Each bandwidth-limited sub-signal after sorting, recording, and labeling is used as a component of the first intrinsic mode function.

6. The method according to any one of claims 1 to 5, characterized in that, The step of determining the second intrinsic mode function component based on each of the first intrinsic mode function components and the current channel sampling curve includes: Determine the degree of synchronization, information correlation, and energy contribution between each of the first intrinsic mode function components and the current channel sampling curve; Based on the degree of change synchronization, the degree of information correlation, and the degree of energy contribution, component correlation evaluation results are generated for each of the first intrinsic mode function components. Based on the component correlation evaluation results, interference candidate components with low correlation to the current channel sampling curve are removed from each of the first intrinsic mode function components; The first intrinsic mode function component remaining after removing the interference candidate components is determined as the second intrinsic mode function component.

7. The method as described in claim 6, characterized in that, The step of removing interference candidate components with low correlation to the current channel sampling curve from each of the first intrinsic mode function components based on the component correlation evaluation results includes: Based on the distribution of the component correlation evaluation results of each of the first intrinsic mode function components, the component screening conditions corresponding to the current sampling window are determined. The first intrinsic mode function component whose component correlation evaluation result does not meet the component screening conditions and exhibits jitter, isolated abrupt change, or inconsistent with the trend of the current channel sampling curve is identified as the interference candidate component. The first intrinsic mode function component whose component correlation evaluation result meets the component screening conditions and matches the physical change trend of the voltage signal or temperature signal is determined as a valid candidate component. The second intrinsic mode function component is formed based on the effective candidate components, and the channel identifier and sampling time correspondence corresponding to the second intrinsic mode function component are retained.

8. The method as described in claim 7, characterized in that, The step of obtaining the target channel processing signal based on the second intrinsic mode function component includes: Wavelet decomposition is performed on the second intrinsic mode function component to obtain the corresponding low-frequency approximation part and high-frequency detail part; Based on the perturbation level of the high-frequency detail portion and the change amplitude of the current channel sampling curve, the adaptive threshold processing conditions are determined; According to the adaptive threshold processing conditions, the high-frequency detail part is shrunk, and the shrunk high-frequency detail part is reconstructed with the low-frequency approximation part to obtain the preliminary denoised channel signal; State tracking data is constructed based on the initial denoised channel signal, and baseline drift correction and state estimation are performed on the state tracking data using unscented Kalman filtering to obtain the target channel processed signal.

9. The method as described in claim 8, characterized in that, After compensating the abnormal target channel processing signal with the abnormal target channel processing signal, the compensated multi-channel sampling data is obtained, including: Select a reference channel processing signal from the target channel processing signals without anomalies, whose correlation with the target channel processing signals with anomalies meets the preset compensation conditions; Based on the correspondence between the reference channel processing signal and the abnormal target channel processing signal within the historical synchronous sampling window, a weighted compensation relationship is determined, and a compensation signal corresponding to the abnormal target channel processing signal is generated according to the weighted compensation relationship. The compensated target channel processing signal is replaced or fused with the compensated signal to obtain the compensated multi-channel sampling data.

10. A signal acquisition and processing device for a battery cell connection system, characterized in that, The apparatus for implementing the method according to any one of claims 1 to 9, the apparatus comprising: The raw signal acquisition module is used to acquire multi-channel raw acquisition signals of multiple cells collected by CCS. The multi-channel raw acquisition signals include at least the voltage signal and temperature signal of the cells. The mode decomposition module is used to determine each first intrinsic mode function component based on the current channel sampling curve in the multi-channel original acquisition signal. Each first intrinsic mode function component is used to characterize the sub-signal components formed after the current channel sampling curve is decomposed. The current channel sampling curve is used to represent the sampling sequence selected from the multi-channel original acquisition signal within the current sampling window. The effective component determination module is used to determine the second intrinsic mode function component based on each of the first intrinsic mode function components and the current channel sampling curve. The second intrinsic mode function component is used to retain the modal components that are associated with the current channel sampling curve. The channel signal generation module is used to obtain the target channel processing signal based on the second intrinsic mode function component, divide the multiple target channel processing signals into voltage channel processing signal group and temperature channel processing signal group according to the channel signal type, determine the correlation between each target channel processing signal within the channel processing signal group corresponding to the same channel signal type, form a multi-channel correlation evaluation result, and identify abnormal target channel processing signals and non-abnormal target channel processing signals based on the multi-channel correlation evaluation result, the continuity of each target channel processing signal, and the reasonableness of the range corresponding to each target channel processing signal. A multi-channel compensation module is used to compensate the abnormal target channel processing signal with the non-abnormal target channel processing signal to obtain compensated multi-channel sampling data; The acquisition value output module is used to restore the compensated multi-channel sampling data and output the acquisition values ​​of multiple cells.

Images