A high-precision battery voltage acquisition method based on dynamic bias compensation

Abstract

The application discloses a high-precision battery voltage collection method based on dynamic bias compensation, and relates to the technical field of battery management. The method comprises the following steps: acquiring sampling monitoring data of a battery, performing time alignment, generating a joint monitoring data set, performing feature extraction on the joint monitoring data set, and generating a dynamic bias feature set; performing first drift tracking on zero-point drift features in the dynamic bias feature set, generating a zero-point bias sequence, performing coupled calculation on gain drift features in the dynamic bias feature set, generating a gain bias sequence, performing component-type dynamic deduction on voltage sampling data in the joint monitoring data set according to the zero-point bias sequence and the gain bias sequence, and generating an initial compensation voltage sequence; and performing local constraint reconstruction on an abnormal sampling segment set through the zero-point bias sequence and the gain bias sequence, and obtaining high-precision battery voltage data. The high-precision battery voltage data is obtained by performing local constraint reconstruction, and the accuracy and reliability of the measurement data are ensured.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a high-precision battery voltage acquisition method based on dynamic bias compensation. Background Technology

[0002] With the rapid development of new energy vehicles, energy storage systems, and high-performance portable electronic devices, the accuracy requirements for battery voltage acquisition in battery management systems are constantly increasing. Existing technologies mainly measure battery port voltage through analog sampling circuits or digital samplers. To improve acquisition accuracy, some technologies introduce baseline correction, low-pass filtering, gain correction, and drift compensation methods to post-process voltage sampling data. In addition, some studies have also attempted to synchronously acquire and time-align voltage data from multiple sampling points, and compensate for factors such as temperature changes and electrochemical drift through feature extraction, providing basic data support for battery management, life prediction, and safety control.

[0003] However, existing technologies still have the following shortcomings: when dealing with zero drift and gain drift, existing methods rely on static parameters or empirical correction, which cannot dynamically track drift changes during continuous sampling, resulting in a decrease in measurement accuracy over a long period of time. For abnormal sampling points and local fluctuations, existing technologies lack dynamic identification and correction mechanisms, which can easily lead to misjudgment and data deviation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a high-precision battery voltage acquisition method based on dynamic bias compensation to solve the problem of being unable to dynamically track drift changes during continuous sampling.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a high-precision battery voltage acquisition method based on dynamic bias compensation, which includes,

[0008] Acquire battery sampling and monitoring data, perform time alignment, generate a joint monitoring dataset, extract features from the joint monitoring dataset, and generate a dynamic bias feature set;

[0009] Perform first drift tracking on the zero-point drift features in the dynamic bias feature set to generate a zero-point bias sequence. Perform coupled solution on the gain drift features in the dynamic bias feature set to generate a gain bias sequence. Perform component-based dynamic subtraction on the voltage sampling data in the joint monitoring dataset based on the zero-point bias sequence and the gain bias sequence to generate an initial compensation voltage sequence.

[0010] Residual analysis is performed on the initial compensation voltage sequence and the dynamic bias feature set to generate a filter state evaluation set. The initial compensation voltage sequence is then dynamically weighted using the filter state evaluation set to generate a smooth compensation voltage sequence.

[0011] Anomaly detection is performed on the smoothed compensated voltage sequence to generate an abnormal sampling segment set. Local constraint reconstruction is then performed on the abnormal sampling segment set using a zero-point bias sequence and a gain bias sequence to obtain high-precision battery voltage data.

[0012] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the joint monitoring dataset are as follows:

[0013] The voltage sampling data, temperature monitoring data, time reference monitoring data, and reference reference response data are acquired during the battery voltage sampling process to form sampling monitoring data;

[0014] The sampling time is marked on the sampled monitoring data to generate a monitoring time sequence;

[0015] By monitoring the time sequence, the sampled monitoring data is time-corresponded and ordered to generate a joint monitoring dataset.

[0016] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the dynamic bias feature set are as follows:

[0017] Baseline change analysis was performed on temperature monitoring data and voltage sampling data in the joint monitoring dataset to generate zero-point drift characteristics;

[0018] Amplitude variation analysis is performed on the time base monitoring data, reference base response data, and voltage sampling data in the joint monitoring dataset to generate gain drift characteristics;

[0019] The zero-point drift features and gain drift features are aggregated to generate a dynamic bias feature set.

[0020] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the zero-point bias sequence are as follows:

[0021] Arrange the zero-point drift features in the dynamic bias feature set in a temporal sequence to generate the zero-point drift trajectory;

[0022] Calculate the change amplitude between the zero-point drift corresponding to adjacent sampling times, and perform time-series aggregation to generate a drift change sequence;

[0023] By segmenting and fitting the continuous drift segment and fluctuating drift segment in the zero-point drift trajectory using the drift change sequence, a zero-point bias sequence is generated.

[0024] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the gain bias sequence are as follows:

[0025] The gain drift characteristics in the dynamic bias feature set are processed to organize the amplitude distribution and generate the gain drift distribution result;

[0026] The gain correlation results are generated by mapping the reference change segment in the reference response data with the gain drift distribution results.

[0027] The gain correlation results are used to compare the amplitude response and quantify the deviation of the voltage sampling data in the joint monitoring dataset to generate a gain bias sequence.

[0028] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the initial compensation voltage sequence are as follows:

[0029] Baseline correction is performed on the voltage sampling data using a zero-point bias sequence to generate a zero-point compensated voltage sequence.

[0030] The zero-point compensation voltage sequence is modified by applying a gain bias sequence to generate an initial compensation voltage sequence.

[0031] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the filtered state evaluation set are as follows:

[0032] Calculate the voltage difference between adjacent sampling points in the initial compensation voltage sequence, and arrange them in order of sampling time to obtain the voltage residual sequence;

[0033] By correlating the zero-point drift characteristics and gain drift characteristics through the voltage residual sequence, residual correlation results are generated.

[0034] The residual correlation results are evaluated to generate a filter state evaluation set.

[0035] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the generation of a smooth compensation voltage sequence refers to adjusting the weight allocation of the initial compensation voltage sequence by adjusting the filter state evaluation set, performing multi-level filtering on the adjusted initial compensation voltage sequence, and generating a smooth compensation voltage sequence.

[0036] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for generating the abnormal sampling segment set are as follows:

[0037] Continuity detection is performed on all sampling points in the smoothed compensation voltage sequence to generate continuity detection results;

[0038] Based on the continuous detection results, abnormal segmentation is performed on the fluctuating abrupt segments in the smoothed compensation voltage sequence to generate a set of abnormal sampling segments.

[0039] As a preferred embodiment of the high-precision battery voltage acquisition method based on dynamic bias compensation described in this invention, the specific steps for obtaining high-precision battery voltage data are as follows:

[0040] The abnormal sampling points in the abnormal sampling segment set are biased and corrected using the zero-point bias sequence and the gain bias sequence to generate reconstructed sampling segments;

[0041] High-precision battery voltage data is obtained by replacing the corresponding aberrant sampling segments in the smoothed compensation voltage sequence with reconstructed sampling segments.

[0042] The beneficial effects of this invention are as follows: By extracting features, zero-point drift features and gain drift features are generated, significantly improving the accuracy and reliability of battery voltage acquisition; by performing piecewise fitting, a zero-point bias sequence is generated, improving the accuracy of zero-point compensation and providing a precise basis for the subsequent generation of the initial compensation voltage sequence; by performing amplitude response comparison and deviation quantization, a gain bias sequence is generated, improving the accuracy of voltage amplitude correction; by performing amplitude correction, an initial compensation voltage sequence is generated, eliminating the superposition effect of zero-point drift and gain drift; by performing multi-level filtering processing, a smooth compensation voltage sequence is generated, improving the stability and usability of the voltage sequence; by performing local constraint reconstruction, high-precision battery voltage data is obtained, ensuring the accuracy and reliability of the measurement data. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of a high-precision battery voltage acquisition method based on dynamic bias compensation.

[0045] Figure 2 A flowchart for generating dynamic bias feature sets.

[0046] Figure 3 The flowchart for generating the initial compensation voltage sequence.

[0047] Figure 4 The flowchart for generating a set of abnormal sampling fragments. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0051] Reference Figures 1-4 This is one embodiment of the present invention, which provides a high-precision battery voltage acquisition method based on dynamic bias compensation, comprising the following steps:

[0052] S1: Acquire battery sampling monitoring data, perform time alignment, generate a joint monitoring dataset, extract features from the joint monitoring dataset, and generate a dynamic bias feature set.

[0053] S1.1: Acquire voltage sampling data, temperature monitoring data, time reference monitoring data, and reference reference response data during the battery voltage sampling process to form sampling monitoring data.

[0054] Under normal battery operation, the voltage at the battery port is continuously measured at a fixed sampling frequency, e.g., 100Hz. The voltage value is recorded every fixed time interval, e.g., 0.01s, and these values ​​are arranged sequentially to form voltage sampling data. Each voltage sampling data entry includes the sampling time and the corresponding voltage value; for example, 0.01 seconds corresponds to a voltage of 3.72V. At the same fixed time intervals, the battery surface temperature is measured using a temperature sensor, and the battery surface temperatures are arranged sequentially to form temperature monitoring data; for example, 0.01 seconds corresponds to a temperature of 28.5℃. At the same fixed time intervals, the timestamp information of each sampling is obtained using a clock. Time reference monitoring data is recorded sequentially. For example, the first sampling point corresponds to a time of 0.01 seconds, the second sampling point corresponds to a time of 0.02 seconds, and so on. This provides a time reference for subsequent alignment and organization of the sampling monitoring data. At the same fixed time interval, a standard voltage pulse with known amplitude and frequency is applied to the battery sampling link through a reference signal injection method, and the response output on the sampling link is measured to form reference reference response data. For example, the response voltage of the first sampling point is 0.02V, and the response voltage of the second sampling point is 0.021V. The voltage sampling data, temperature monitoring data, time reference monitoring data, and reference reference response data are collected to form the sampling monitoring data.

[0055] S1.2: Mark the sampling time of the sampled monitoring data to generate a monitoring time sequence; use the monitoring time sequence to perform time correspondence and ordering of the sampled monitoring data to generate a joint monitoring dataset.

[0056] Sampling times are assigned to all data in the sampling monitoring data. Each voltage sampling data, temperature monitoring data, time base monitoring data, and reference base response data is mapped to a sampling time. For example, the first sampling point corresponds to a time of 0.01s, the second sampling point corresponds to a time of 0.02s, forming a monitoring time sequence. Based on the monitoring time sequence, the various types of data in the sampling monitoring data are organized in the sampling order. The voltage sampling data, temperature monitoring data, time base monitoring data, and reference base response data are matched one-to-one according to the sampling time, so that the various types of data are completely aligned in time, generating a joint monitoring dataset.

[0057] S1.3: Perform baseline change analysis on the temperature monitoring data and voltage sampling data in the joint monitoring dataset to generate zero-point drift characteristics.

[0058] In the joint monitoring dataset, an arbitrary sampling point is selected as the current sampling point. The difference between the temperature monitoring data of the current sampling point and the temperature monitoring data of the previous sampling point is calculated to obtain the temperature difference value of the current sampling point. The temperature difference values ​​of all sampling points are arranged in the sampling order to form a temperature difference sequence. The voltage sampling data of each sampling point is calculated with the voltage sampling data of the previous sampling point to obtain the voltage change of each sampling point. Before battery sampling, a short-term low load or no-load state is applied to the battery port, and the voltage change is collected to obtain a zero voltage reference value. The zero voltage reference value is used to correct the voltage change of subsequent sampling points. The component caused by the actual battery state change is removed, and only the offset caused by temperature or measurement link is retained to obtain the zero drift value of each sampling point. The zero drift values ​​of all sampling points are arranged in the sampling order to form a zero drift feature.

[0059] S1.4: Perform amplitude variation analysis on the time reference monitoring data, reference reference response data and voltage sampling data in the joint monitoring dataset to generate gain drift characteristics; aggregate the zero drift characteristics and gain drift characteristics to generate a dynamic bias characteristic set.

[0060] In the joint monitoring dataset, for any sampling point, the time reference monitoring data, reference reference response data, and voltage sampling data of this sampling point are read. The difference between the amplitude of the reference reference response data and the amplitude of the voltage sampling data is calculated to obtain the amplitude deviation. The amplitude deviation is compared with the sampling time interval between adjacent sampling points to obtain the standardized amplitude change, and recorded as the gain drift feature of this sampling point. The gain drift features of all sampling points in the joint monitoring dataset are arranged in sequence to form the gain drift feature. The zero-point drift feature and the gain drift feature are collected in the sampling order to generate a complete dynamic bias feature set.

[0061] S2: Perform the first drift tracking on the zero-point drift feature in the dynamic bias feature set to generate a zero-point bias sequence. Perform coupled solution on the gain drift feature in the dynamic bias feature set to generate a gain bias sequence. Perform component-based dynamic subtraction on the voltage sampling data in the joint monitoring dataset based on the zero-point bias sequence and the gain bias sequence to generate an initial compensation voltage sequence.

[0062] S2.1: Arrange the zero-point drift features in the dynamic bias feature set in time sequence to generate the zero-point drift trajectory; calculate the change amplitude between the zero-point drift amounts corresponding to adjacent sampling times and perform time sequence aggregation to generate the drift change sequence; perform segmented fitting of the continuous drift segment and fluctuating drift segment in the zero-point drift trajectory through the drift change sequence to generate the zero-point bias sequence.

[0063] Zero-point drift features are read from the dynamic bias feature set and sorted from earliest to latest sampling time to form a time-series zero-point drift trajectory. The difference between the zero-point drift value of each sampling point and the zero-point drift value of the previous sampling point is calculated to obtain the drift amplitude of each sampling point. The drift amplitudes of all sampling points are arranged in time sequence according to the zero-point drift trajectory to form a drift change sequence. The drift amplitude segment threshold is set according to the measurement accuracy and noise level of the battery voltage acquisition system. The value range of the drift amplitude segment threshold is 0.5mV-2mV. The value range of the drift amplitude segment threshold is set according to the typical fluctuation range of historical battery voltage sampling data. The value range of 0.5mV-2mV can effectively distinguish between continuous drift and abnormal fluctuation, and avoid misjudgment caused by noise interference. The segment in the drift change sequence where the continuous drift amplitude is less than the drift amplitude segment threshold is recorded as a continuous drift segment. A segment with a continuous drift amplitude greater than or equal to the drift amplitude segment threshold is called a fluctuating drift segment. In the continuous drift segment or fluctuating drift segment where each sampling point is located, the zero-point drift value of this sampling point is accumulated to the zero-point offset value of the previous sampling point. At the end of each zeroing cycle, the accumulated zero-point offset value is corrected with the short-time zero input calibration reference value to obtain the calibration zero-point offset value of the current sampling point. For the sampling points of the fluctuating drift segment, the moving average of the zero-point offset value is calculated within the segment to suppress the influence of instantaneous abnormal fluctuations. It should be noted that the initial zero-point offset value is 0. The zero-point offset value of this sampling point is obtained. The zero-point offset zeroing cycle is set according to the battery sampling frequency and the battery voltage change rate, for example, zeroing once every 1000 consecutive sampling points or every 1 second. The zero-point offset zeroing cycle can balance the zero-point cumulative error and the real-time response requirements. The zero-point offset values ​​of all sampling points are repeatedly calculated and arranged in the order of sampling time to form a zero-point offset sequence.

[0064] S2.2: The gain drift features in the dynamic bias feature set are processed to generate a gain drift distribution result; the gain drift distribution result is used to perform a correlation mapping on the reference change segment in the reference response data to generate a gain correlation result.

[0065] Gain drift features are read from the dynamic bias feature set and sorted according to sampling time to form a gain drift time series. Each sampling point in the gain drift time series is selected as the center sampling point. The gain drift value of the center sampling point is combined with the gain drift values ​​of the three adjacent sampling points before and after it to form a local dataset. The mean and standard deviation of the local dataset are calculated and used as the local amplitude distribution information of the center sampling point. This calculation of the local amplitude distribution information of each sampling point in the gain drift time series is repeated and arranged according to sampling time to form the gain drift distribution result. Based on the measurement accuracy of battery voltage acquisition and the fluctuation characteristics of historical sampling data, amplitude matching rules for the reference change segment are set, and phase-locked demodulation or... The reference pulse peak identification method separates the response of the reference pulse signal in the sampling link from the battery voltage sampling signal. Based on the amplitude matching rule, the estimated local gain drift of each sampling point is compared with the amplitude of the corresponding reference pulse response at that sampling point. When the mean of the local amplitude distribution falls within the allowable deviation range of the reference response data, this sampling point is marked as matched; otherwise, it is marked as mismatched. The allowable deviation range refers to the theoretical fluctuation range of the reference pulse amplitude. All matched sampling points are arranged sequentially according to the sampling time order to form a gain correlation result. The allowable deviation range of the reference response data refers to the amplitude fluctuation range of the reference change segment, which is obtained through statistical analysis of historical sampling data.

[0066] S2.3: The gain correlation results are used to compare the amplitude response and quantize the deviation of the voltage sampling data in the joint monitoring dataset to generate a gain bias sequence.

[0067] The voltage sampling data of each sampling point is read from the joint monitoring dataset. Based on the gain correlation result, the sampling points are selected and marked as matching sampling points. The ratio of the reference pulse response amplitude of the marked matching sampling point to the reference pulse standard amplitude is calculated to obtain the gain correction coefficient of this sampling point. The gain correction coefficients are arranged in the order of sampling time to form a gain bias sequence.

[0068] S2.4: Perform baseline correction on the voltage sampling data using the zero-point bias sequence to generate a zero-point compensation voltage sequence; perform amplitude correction on the zero-point compensation voltage sequence using the gain bias sequence to generate an initial compensation voltage sequence.

[0069] The voltage sampling data of each sampling point is read from the joint monitoring dataset. According to the sampling time sequence, the difference between the voltage sampling data of each sampling point and the zero-point bias value of that sampling point is calculated to obtain the zero-point compensation voltage value. The zero-point compensation voltage value is calculated repeatedly for all sampling points. All zero-point compensation voltage values ​​are arranged to form a zero-point compensation voltage sequence. The gain correction coefficient is set according to the amplitude variation law of historical battery sampling data and measurement accuracy. The gain correction coefficient is a dimensionless ratio, for example, 1.002, which means that the zero-point compensation voltage value of each sampling point needs to be amplified by 0.2% for amplitude correction. The product of the gain bias sequence and the zero-point compensation voltage value of the corresponding sampling point is calculated to obtain the gain compensation voltage value of the sampling point. The gain compensation voltage value is calculated repeatedly for all sampling points and arranged according to the sampling time sequence to form an initial compensation voltage sequence.

[0070] Furthermore, the zero-point drift feature obtains the zero-point bias sequence through piecewise fitting of the zero-point drift trajectory, drift change sequence, and continuous drift segment and fluctuating drift segment. This avoids accumulating short-term fluctuations directly as long-term zero-point offsets. The gain drift feature is coupled and solved by the gain drift distribution result, the reference change segment in the reference response data, and the amplitude response relationship of the voltage sampling data. This can distinguish between the amplitude amplification deviation of the measurement link and the battery's own voltage change. By performing component-based dynamic subtraction on the voltage sampling data through the zero-point bias sequence and the gain bias sequence, the zero-point error and gain error are corrected separately under different physical sources. This avoids the over-compensation or under-compensation caused by mixing zero-point offset, gain offset, and abnormal fluctuations in traditional single drift compensation. In the long-term continuous sampling process, a more stable initial compensation voltage sequence that is closer to the real battery voltage is obtained.

[0071] It should be noted that the first drift tracking and coupled solution are designed for different types of sampling error sources. The first drift tracking is used to handle baseline offset caused by temperature changes, zero-point drift of the sampling link, and short-term fluctuations, generating a zero-point bias sequence with temporal continuity. The coupled solution is used to handle gain offset reflected by changes in the reference response and voltage sampling amplitude, generating a gain bias sequence for amplitude correction. The zero-point bias sequence is used to correct the baseline position of the voltage sampling data, and the gain bias sequence is used to correct the amplitude ratio after zero-point compensation. In the component dynamic subtraction, these are applied sequentially to the voltage sampling data, so that the baseline error and amplitude error are corrected respectively. This reduces the risk of cumulative amplification of zero-point drift, incorrect compensation of gain drift, and incorrect participation of instantaneous fluctuations in compensation during long-term continuous sampling, so that the initial compensated voltage sequence has both baseline stability and amplitude accuracy.

[0072] S3: Perform residual analysis on the initial compensation voltage sequence and dynamic bias feature set to generate a filter state evaluation set. Then, dynamically switch the weights of the initial compensation voltage sequence using the filter state evaluation set to generate a smooth compensation voltage sequence.

[0073] S3.1: Calculate the voltage difference between adjacent sampling points in the initial compensation voltage sequence and arrange them in the order of sampling time to obtain the voltage residual sequence.

[0074] The voltage value of each sampling point in the initial compensation voltage sequence is read from the joint monitoring dataset. The difference between the voltage value of each sampling point and the voltage value of the previous sampling point is calculated according to the sampling time order to obtain the voltage difference of each sampling point. The voltage differences of all sampling points are arranged in the order of sampling time to form a voltage residual sequence.

[0075] S3.2: Correlate the zero-point drift characteristics and gain drift characteristics through the voltage residual sequence to generate residual correlation results; perform state evaluation on the residual correlation results to generate a filter state evaluation set.

[0076] The initial compensation voltage value of each sampling point is read from the joint monitoring dataset, and the voltage difference between adjacent sampling points is calculated to obtain the voltage residual value. The difference between the voltage offset of the zero-point drift characteristic and the voltage residual value is calculated to obtain the zero-point residual deviation. The product of the gain drift characteristic (dimensionless gain deviation) and the initial compensation voltage value of the corresponding sampling point is calculated. The product is converted to voltage dimension and then multiplied with the voltage residual value to obtain the gain residual deviation. The absolute values ​​of the zero-point residual deviation and the gain residual deviation are calculated respectively.

[0077] A zero-point deviation threshold is set based on the typical fluctuation range of historical battery voltage sampling data and the measurement accuracy of the battery voltage acquisition system. The value range of the zero-point deviation threshold is 0.5mV-2mV. The value range of the zero-point deviation threshold is also set based on the drift characteristics of the battery voltage sampling data. Using a value range of 0.5mV-2mV effectively distinguishes actual drift from noise fluctuations and avoids misjudgment. A gain deviation threshold is set based on the fluctuation characteristics of historical sampling data and the measurement accuracy. The value range of the gain deviation threshold is 0.5%-2%. The value range of the gain deviation threshold is also set based on the gain drift variation law. Using a value range of 0.5%-2% effectively determines the gain offset adjustment requirement. If the absolute value of the zero-point residual deviation is greater than the zero-point deviation... If the zero-point drift characteristic of a sampling point is determined to be below the gain deviation threshold, then the zero-point drift characteristic of that sampling point needs adjustment; otherwise, the original value is maintained. If the absolute value of the gain residual deviation is greater than the gain deviation threshold, then the gain drift characteristic of that sampling point needs adjustment; otherwise, the original value is maintained. The zero-point drift adjustment requirements and gain drift adjustment requirements of each sampling point are merged and recorded to form the residual association mark of that sampling point. All sampling points in the joint monitoring dataset are processed repeatedly, and all residual association marks are arranged in the order of sampling time to form a complete residual association result. The state of the residual association mark of each sampling point in the residual association result is evaluated, including whether the bias adjustment needs to be applied or kept at its original value. The state evaluation results of each sampling point are arranged in sequence to generate a filter state evaluation set.

[0078] S3.3: Adjust the weight allocation of the initial compensation voltage sequence by using the filtered state evaluation set, and perform multi-level filtering on the adjusted initial compensation voltage sequence to generate a smooth compensation voltage sequence.

[0079] The voltage value of each sampling point in the initial compensation voltage sequence is read from the joint monitoring dataset. Based on the state assessment results of the corresponding sampling point in the filter state evaluation set, the weight allocation coefficient of the sampling point in the multi-stage filtering process is determined. The weight allocation coefficient is set according to the filter level and the residual correlation result of the sampling point. For example, a weight allocation coefficient between 0.1 and 1 indicates the degree of participation of this sampling point in each filter level. Specifically, if the residual correlation result value is less than 0.5mV, the weight of the sampling point is set to 0.9 in the low-pass filter level and 0.2 in the high-pass filter level. If the residual correlation result value is greater than or equal to 0.5mV, the weight of the sampling point is set to 0.2 in the low-pass filter level and 0.9 in the high-pass filter level. The output of each filter level is passed to the next filter level in the sampling order, and the weights of all sampling points are allocated sequentially to form a weighted average. A weighted initial compensation voltage sequence is generated, and each sampling point in the weighted initial compensation voltage sequence is sequentially input into a filter stage sequence for processing. The filter stage sequence is set according to the measurement accuracy of the battery voltage acquisition and the fluctuation characteristics of historical sampling data. For example, it consists of three consecutive filter stages. Each filter stage uses a recursive moving average algorithm, with filter window lengths of 3, 5, and 7 sampling points, respectively, to balance noise suppression effect and signal response speed. In each filter stage, the filtering calculation result of each sampling point is weighted with the corresponding sampling point weight to obtain the output value of each sampling point. The compensation voltage value of this sampling point is replaced by the output value of each sampling point. All sampling points are processed repeatedly, and the output values ​​of all sampling points are sequentially accumulated to obtain the final filtering result. The filtering results of all sampling points are arranged in the order of sampling time to form a smooth compensation voltage sequence.

[0080] It should be noted that in the weighted initial compensation voltage sequence, the filtered output of each sampling point is weighted according to the corresponding weight and updated to the compensation voltage value of that sampling point. In order to match the filtering with the typical time constant of battery voltage change, three consecutive filtering stages are set. The recursive moving average algorithm is applied to each filtering stage. The window lengths of each filtering stage are 200, 400 and 600 sampling points, respectively, corresponding to time periods of 2s, 4s and 6s. This can effectively filter out high-frequency sampling noise and low-frequency power frequency interference, while maintaining the true characteristics of battery voltage change.

[0081] S4: Perform anomaly detection on the smoothed compensated voltage sequence to generate an abnormal sampling segment set. Perform local constraint reconstruction on the abnormal sampling segment set through the zero-point bias sequence and gain bias sequence to obtain high-precision battery voltage data.

[0082] S4.1: Perform continuity detection on all sampling points in the smoothed compensation voltage sequence and generate continuity detection results; perform anomaly segmentation on the fluctuation and sudden change segments in the smoothed compensation voltage sequence based on the continuity detection results and generate a set of abnormal sampling segments.

[0083] The voltage value of each sampling point is read from the smoothed compensation voltage sequence, and the voltage difference between adjacent sampling points is calculated. A continuity judgment threshold is set based on battery manufacturing tolerances and operating temperature variation ranges, with a value range of 0.5mV-2mV. The range of the continuity judgment threshold is also set based on the typical fluctuation range of historical battery voltage sampling data; a range of 0.5mV-2mV effectively distinguishes between normal continuous sampling points and abnormal fluctuation points, while avoiding misjudgments caused by noise interference. A reference value for the battery charge / discharge current change rate is set based on the battery's rated capacity, typical charge / discharge curve, and the statistical characteristics of continuous current changes in historical sampling data. For example, for a 10Ah battery, the battery charge / discharge current change rate... The reference value for the rate of change can be set to about 0.1C / s, which can reflect the voltage change rate under normal charging and discharging conditions, thereby avoiding misjudging normal voltage changes as abnormal fluctuations. If the voltage difference exceeds the threshold and the current change rate is lower than the reference value, the sampling point is marked as abnormal; otherwise, the sampling point is determined to be continuous. All sampling points in the smoothed compensation voltage sequence are compared with the continuity judgment threshold and the reference value of the battery charging and discharging current change rate in turn to form a continuity detection result. Based on the continuity detection result, the smoothed compensation voltage sequence is scanned, and the interval where the voltage difference exceeds the threshold and the current change rate is lower than the reference value is identified as a fluctuation mutation segment. The starting sampling point and the ending sampling point are extracted from each fluctuation mutation segment and arranged in the order of sampling time to form a set of abnormal sampling segments.

[0084] S4.2: The abnormal sampling points in the abnormal sampling segment set are biased and corrected by the zero-point bias sequence and the gain bias sequence to generate the reconstructed sampling segment; the corresponding abnormal sampling segment in the smooth compensation voltage sequence is replaced by the reconstructed sampling segment to obtain high-precision battery voltage data.

[0085] The voltage sampling data of each abnormal sampling point is read from the abnormal sampling segment set. The zero-point bias value of the corresponding sampling point is read from the zero-point bias sequence, and the gain correction coefficient of the corresponding sampling point is read from the gain bias sequence. The difference between the voltage sampling data and the zero-point bias value of each abnormal sampling point is calculated, and the difference is multiplied by the gain bias value to obtain the reconstructed sampling point voltage value. All sampling points in the abnormal sampling segment set are processed in sequence to form a complete reconstructed sampling segment.

[0086] Based on the short-term fluctuation amplitude and signal noise level of the reference battery port voltage under static calibration conditions, a tolerance is set, for example, ±2mV. Within a fixed time window before and after each abnormal sampling point, for example, ±5 sampling points, if the voltage value of the reconstructed sampling point deviates from the voltage of the adjacent sampling point beyond the set tolerance, the excess is linearly smoothed to suppress abnormal spikes or abrupt changes. Each sampling point in the reconstructed sampling segment replaces the corresponding abnormal sampling point voltage value in the smoothed compensation voltage sequence according to the sampling time sequence. After all abnormal sampling points are replaced, the updated voltage values ​​are arranged according to the sampling time sequence to form complete high-precision battery voltage data.

[0087] This embodiment also provides a computer device applicable to the high-precision battery voltage acquisition method based on dynamic bias compensation, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the high-precision battery voltage acquisition method based on dynamic bias compensation as proposed in the above embodiment.

[0088] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0089] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the high-precision battery voltage acquisition method based on dynamic bias compensation as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0090] In summary, this invention significantly improves the accuracy and reliability of battery voltage acquisition by generating zero-point drift and gain drift features through feature extraction. By performing piecewise fitting, a zero-point bias sequence is generated, improving the accuracy of zero-point compensation and providing a precise foundation for the subsequent generation of the initial compensation voltage sequence. By performing amplitude response comparison and deviation quantization, a gain bias sequence is generated, improving the accuracy of voltage amplitude correction. By performing amplitude correction, an initial compensation voltage sequence is generated, eliminating the superposition effect of zero-point drift and gain drift. By performing multi-level filtering, a smooth compensation voltage sequence is generated, improving the stability and usability of the voltage sequence. By performing local constraint reconstruction, high-precision battery voltage data is obtained, ensuring the accuracy and reliability of the measurement data.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A high-precision battery voltage acquisition method based on dynamic bias compensation, characterized in that: include, S1 acquires battery sampling monitoring data, performs time alignment, generates a joint monitoring dataset, extracts features from the joint monitoring dataset, and generates a dynamic bias feature set. S2 performs the first drift tracking on the zero-point drift feature in the dynamic bias feature set to generate a zero-point bias sequence, performs coupled solution on the gain drift feature in the dynamic bias feature set to generate a gain bias sequence, and performs component-based dynamic subtraction on the voltage sampling data in the joint monitoring dataset based on the zero-point bias sequence and the gain bias sequence to generate an initial compensation voltage sequence. The component-based dynamic subtraction of voltage sampling data in the joint monitoring dataset based on the zero-point bias sequence and the gain bias sequence includes: performing baseline correction on the voltage sampling data using the zero-point bias sequence to generate a zero-point compensation voltage sequence; performing amplitude correction on the zero-point compensation voltage sequence using the gain bias sequence to generate an initial compensation voltage sequence; reading the voltage sampling data of each sampling point in the joint monitoring dataset, calculating the difference between the voltage sampling data of each sampling point and the zero-point bias value of that sampling point according to the sampling time sequence to obtain the zero-point compensation voltage value, repeating the calculation of the zero-point compensation voltage value for all sampling points in sequence, arranging all the zero-point compensation voltage values ​​to form a zero-point compensation voltage sequence, setting the gain correction coefficient according to the amplitude variation law and measurement accuracy of the historical battery sampling data, calculating the product of the gain bias sequence and the zero-point compensation voltage value of the corresponding sampling point to obtain the gain compensation voltage value of the sampling point, repeating the calculation of the gain compensation voltage value for all sampling points in sequence, and arranging them according to the sampling time sequence to form an initial compensation voltage sequence; S3 performs residual analysis on the initial compensation voltage sequence and dynamic bias feature set to generate a filter state evaluation set. The initial compensation voltage sequence is then dynamically weighted using the filter state evaluation set to generate a smooth compensation voltage sequence. S4 performs anomaly detection on the smoothed compensated voltage sequence, generates an abnormal sampling segment set, and performs local constraint reconstruction on the abnormal sampling segment set through zero-point bias sequence and gain bias sequence to obtain high-precision battery voltage data.

2. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 1, characterized in that: The specific steps for generating the joint monitoring dataset are as follows: The voltage sampling data, temperature monitoring data, time reference monitoring data, and reference reference response data are acquired during the battery voltage sampling process to form sampling monitoring data; The sampling time is marked on the sampled monitoring data to generate a monitoring time sequence; By monitoring the time sequence, the sampled monitoring data is time-corresponded and ordered to generate a joint monitoring dataset.

3. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 2, characterized in that: The specific steps for generating the dynamic bias feature set are as follows: Baseline change analysis was performed on temperature monitoring data and voltage sampling data in the joint monitoring dataset to generate zero-point drift characteristics; Amplitude variation analysis is performed on the time base monitoring data, reference base response data, and voltage sampling data in the joint monitoring dataset to generate gain drift characteristics; The zero-point drift features and gain drift features are aggregated to generate a dynamic bias feature set.

4. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 3, characterized in that: The specific steps for generating the zero-point bias sequence are as follows: Arrange the zero-point drift features in the dynamic bias feature set in a temporal sequence to generate the zero-point drift trajectory; Calculate the change amplitude between the zero-point drift corresponding to adjacent sampling times, and perform time-series aggregation to generate a drift change sequence; By segmenting and fitting the continuous drift segment and fluctuating drift segment in the zero-point drift trajectory using the drift change sequence, a zero-point bias sequence is generated.

5. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 4, characterized in that: The specific steps for generating the gain bias sequence are as follows. The gain drift characteristics in the dynamic bias feature set are processed to organize the amplitude distribution and generate the gain drift distribution result; The gain correlation results are generated by mapping the reference change segment in the reference response data with the gain drift distribution results. The gain correlation results are used to compare the amplitude response and quantify the deviation of the voltage sampling data in the joint monitoring dataset to generate a gain bias sequence.

6. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 5, characterized in that: The specific steps for generating the filter state evaluation set are as follows: Calculate the voltage difference between adjacent sampling points in the initial compensation voltage sequence, and arrange them in order of sampling time to obtain the voltage residual sequence; By correlating the zero-point drift characteristics and gain drift characteristics through the voltage residual sequence, residual correlation results are generated. The residual correlation results are evaluated to generate a filter state evaluation set.

7. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 6, characterized in that: The generation of a smooth compensation voltage sequence refers to adjusting the weight allocation of the initial compensation voltage sequence by using a filtered state evaluation set, and then performing multi-level filtering on the adjusted initial compensation voltage sequence to generate a smooth compensation voltage sequence.

8. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 7, characterized in that: The specific steps for generating the set of abnormal sampling fragments are as follows: Continuity detection is performed on all sampling points in the smoothed compensation voltage sequence to generate continuity detection results; Based on the continuous detection results, abnormal segmentation is performed on the fluctuating abrupt segments in the smoothed compensation voltage sequence to generate a set of abnormal sampling segments.

9. The high-precision battery voltage acquisition method based on dynamic bias compensation as described in claim 8, characterized in that: The specific steps to obtain high-precision battery voltage data are as follows. The abnormal sampling points in the abnormal sampling segment set are biased and corrected using the zero-point bias sequence and the gain bias sequence to generate reconstructed sampling segments; High-precision battery voltage data is obtained by replacing the corresponding aberrant sampling segments in the smoothed compensation voltage sequence with reconstructed sampling segments.

Images