A lithium battery capacity degradation nonlinear trajectory analysis and prediction system

Abstract

The present application relates to the technical field of lithium battery, specifically to a lithium battery capacity attenuation nonlinear trajectory analysis and prediction system, which comprises a capacity anomaly identification module, an energy fluctuation grouping module, a life stage attribution module, a reverse behavior identification module and a prediction boundary voltage control module. The present application constructs a trajectory node structure through the turning point identification of the ratio trend direction, superimposes the power offset amplitude judgment, forms a classification identification method of the capacity anomaly period, uses the difference between the discharge power and the SOC boundary to construct the change mapping relationship of the energy behavior, combines the rate label to filter the continuous offset period, enhances the classification ability of the energy fluctuation section, uses the comparison judgment between the energy trend and the capacity stage boundary to replace the static capacity value, establishes a sectional life attribution structure, forms a period reverse combination criterion by the differential pressure rate and the capacity change direction, and widens the nonlinear trajectory identification dimension.

Description

Technical Field

[0001] This invention relates to the field of lithium battery technology, and in particular to a system for analyzing and predicting the nonlinear trajectory of lithium battery capacity decay. Background Technology

[0002] The field of lithium batteries mainly covers the research and application of lithium-ion batteries and related materials, structural design, electrochemical characteristics, charge and discharge control technology, life management, thermal management, safety protection and failure analysis. This field is widely used in scenarios with high energy density and long cycle life requirements, such as new energy vehicles, renewable energy storage systems, consumer electronics devices and aerospace. In recent years, with the expansion of battery application scale, the performance degradation law of lithium batteries, state of health assessment (SOH), remaining life prediction (RUL), multiphysics coupling modeling and intelligent management system (such as BMS) have become research hotspots, especially focusing on their reliability and predictive maintenance capabilities in complex operating environments.

[0003] Among them, the lithium battery capacity decay nonlinear trajectory analysis and prediction system is an intelligent analysis system developed to address the characteristics of lithium batteries' capacity gradually decreasing and exhibiting nonlinear changes during long-term use. This system utilizes data acquisition and modeling technology to analyze the trajectory of battery capacity changes over time and predicts its future performance trends through algorithms, thereby providing support for battery maintenance, replacement decisions, and vehicle management system optimization. Its purpose is to improve battery efficiency, extend lifespan, and prevent sudden failures, which is of great significance.

[0004] Existing technologies construct capacity trends based on time or cycles, failing to capture sudden jump cycles. Local changes are masked by average trends, and the impact of the multiplier is not separated at the cycle granularity, resulting in incorrect attribution of energy fluctuation characteristics. Discharge behavior lacks classification basis, lifetime stage division relies on capacity values, ignoring trend changes leads to error-prone stage labels, the prediction process relies on the continuation of historical residuals, and boundary settings are not adjusted to follow behavioral changes, causing the prediction range to expand and stability to decrease. Under dynamic operating conditions, nonlinear capacity changes cannot drive the model to respond, and abnormal trajectory cycles are judged with low sensitivity, resulting in limited prediction continuity and accuracy. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a nonlinear trajectory analysis and prediction system for lithium battery capacity decay.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a nonlinear trajectory analysis and prediction system for lithium battery capacity decay, the system comprising:

[0007] The capacity anomaly identification module analyzes the voltage and current ratio change trend based on lithium battery cycle data, calculates the slope change direction between adjacent cycles, determines whether a reverse trend occurs in three consecutive cycles, filters out power offset anomaly cycles, and generates slope turning trajectory nodes.

[0008] The energy fluctuation grouping module calls the periodic discharge power and the output power corresponding to the SOC boundary in the slope turning trajectory node, calculates the power difference for each cycle, filters the power deviation amplitude segment in the high-rate cycle, judges whether there is an abnormal change trend, and obtains the continuous segment of rate offset.

[0009] The lifetime stage attribution module determines whether there is a difference between the energy offset amplitude and the capacity trajectory baseline based on the continuous segment of the multiplier offset, filters out the period with prominent offset, reassigns it to the corresponding lifetime type, and obtains the capacity degradation stage sequence.

[0010] The reversal behavior recognition module calculates the pressure difference and duration of each cycle based on the capacity degradation stage sequence, determines whether the pressure difference rate continues to increase, compares whether the start and end capacities have reversed, identifies the cycle segments in the continuous cycle that satisfy the direction of change, and obtains the trajectory reversal joint cycle segments.

[0011] The present invention improves upon this invention by including the following: the slope inflection trajectory node includes a ratio change breakpoint, a power anomaly cycle marker, and a trajectory reverse trend label; the ratio offset continuous segment includes a ratio category grouping identifier, power difference quantification data, and fluctuation range number; the capacity degradation stage sequence includes a lifetime stage type identifier, a capacity reduction amplitude range, and a stage boundary positioning label; and the trajectory reversal joint cycle segment includes a capacity rebound marker cycle, a differential pressure change anomaly sequence, and a trend reversal identification number.

[0012] The present invention is improved in that the capacity anomaly identification module includes:

[0013] The periodic electrical parameter extraction submodule analyzes the original voltage and current data sequence in each cycle based on lithium battery cycle data, divides the data into segments according to the charging and discharging stages, calculates the periodic average values ​​of voltage and current in the charging segment and voltage and current in the discharging segment, and generates a periodic electrical parameter mean value group.

[0014] The ratio trend analysis submodule analyzes the comparison relationship between the charging ratio and the discharging ratio in each cycle based on the average value group of the periodic electrical parameters, constructs a cross-cycle ratio sequence, compares the rising and falling trend directions between adjacent cycles, and determines whether there are two slope trends with opposite directions of change in three consecutive cycles, forming a slope reversal trend.

[0015] The power offset filtering submodule filters out cycles with direction reversal characteristics based on the slope turning trend, calculates the offset difference between the average power of the cycle and the adjacent cycle, determines whether the degree of offset is higher than the normal range of offset change, compares the turning direction of its ratio trend again, and generates slope turning trajectory nodes.

[0016] The present invention is improved in that the energy fluctuation grouping module includes:

[0017] The periodic power calculation submodule calls the slope turning trajectory node to calculate the deviation between the discharge power contained in the node and the output power corresponding to SOC in each period, analyzes the difference distribution of the two in the time series, and constructs the periodic power difference sequence.

[0018] The ratio difference classification submodule analyzes the distribution pattern of the periodic power difference sequence under the ratio difference label, calculates the average amplitude and range of change of the difference in each category, compares the offset performance between each category, and filters the classification segments with prominent offset to obtain the concentrated interval of ratio deviation.

[0019] The continuous offset identification submodule determines the offset trend between consecutive periods within the concentrated range of the multiplier deviation, analyzes the continuous amplitude of the power difference, and calculates the change performance within each window in the sequence by combining the SOC boundary and the multiplier characteristics. It then filters out the continuous segments with key offset fluctuations to obtain the continuous segments of multiplier offset.

[0020] The present invention is improved in that the lifespan stage attribution module includes:

[0021] The multiplier offset segment identification submodule determines whether the amplitude difference between adjacent data in the multiplier sequence is continuous based on the continuous multiplier offset segment. If it is continuous, it is classified as a multiplier offset segment. By comparing the difference between the energy offset amplitude of the segment and the capacity trajectory reference, it determines whether it exceeds the allowable range. If it exceeds the allowable range, it marks the abnormal segment and obtains the offset amplitude abnormal range.

[0022] The offset mutation cycle screening submodule analyzes the capacity change trend corresponding to the start and end cycles of each segment in the abnormal offset amplitude interval, calculates the difference in capacity change rate between consecutive cycles, and screens out cycles with mutation characteristics by comparing with the capacity change trend baseline to obtain the offset mutation cycle sequence.

[0023] The lifetime type reconstruction and attribution submodule compares the capacity decrease magnitude of each cycle in the offset mutation cycle sequence with the segmentation criteria of the capacity degradation stage, determines the degradation stage boundary to which each cycle belongs, calculates the mean squared capacity matching deviation, reassigns the corresponding lifetime type, and obtains the capacity degradation stage sequence.

[0024] The present invention is improved in that the reversal behavior recognition module includes:

[0025] The differential pressure rate calculation submodule calculates the starting and ending pressure difference of each cycle based on the capacity degradation stage sequence. Combined with the start and end times within the corresponding time period, it analyzes the continuous change behavior within each cycle. Then, it converts the cycle pressure difference to time by ratio calculation to determine the change rate corresponding to the cycle and obtains the cycle pressure difference rate characteristics.

[0026] The rate trend judgment submodule analyzes the direction of change of the periodic differential pressure rate characteristics within a continuous period, determines whether there is a continuously increasing pattern sequence, filters out time periods that meet the continuous increase requirement, judges the trend structure by the direction of rate change, and obtains the differential pressure rate growth trend range.

[0027] The capacity reverse verification submodule calls the pressure difference rate growth trend interval, compares the capacity start and end states of each cycle, calculates the relative change between the capacity direction and the pressure difference rate, filters out the cycle segments where the capacity and rate directions are opposite, determines that they constitute trend reversal behavior, and obtains the trajectory reversal joint cycle segment.

[0028] The present invention has an improvement, wherein the system includes:

[0029] The prediction boundary pressure control module adjusts the proportion of historical residual interference in the cycle based on the trajectory reversal joint cycle segment, redefines the range of change of the allowable boundary of prediction error according to the reversal trend, limits the prediction interval to within the reference change range, and retains the boundary convergence operation in the continuous cycle to obtain the predicted trajectory compression interval.

[0030] The predicted trajectory compression interval includes the error boundary limit range, residual interference correction parameters, and continuous period compression record number.

[0031] The present invention is improved in that the predicted boundary pressure control module includes:

[0032] The trajectory reversal recognition submodule analyzes the trend fluctuation amplitude and directional changes of continuous cycles based on the trajectory reversal joint period segment, determines whether the trend has reversed direction between adjacent cycles, identifies segments with directional reversal characteristics by comparing the changes in the trend reversal amplitude, and generates reversal recognition mark content.

[0033] The historical interference weight adjustment submodule calls the reversal identification mark content, optimizes the interference structure composition of the current period, analyzes the changes in the proportion of trend offset interference and random disturbance in the prediction error, determines whether it is in a reversal state, and rebalances the role ratio between the difference interference items accordingly, filters the adjusted interference composition structure, and obtains the periodic interference weight combination.

[0034] The boundary convergence limiting submodule calculates the range of change of the prediction segment based on the combination of the periodic interference proportions, analyzes the historical prediction boundaries, judges the expansion segments that appear in the error interval, adjusts the boundary segments of the abnormal expansion parts, and obtains the prediction trajectory compression interval by redefining the range of error change amplitude.

[0035] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0036] In this invention, a trajectory node structure is constructed by identifying the turning point of the ratio trend, and a power offset amplitude judgment is superimposed to form a classification and identification method for capacity anomaly cycles. The difference between discharge power and SOC boundary is used to construct a mapping relationship of energy behavior changes. Combined with the rate label to screen continuous offset cycles, the ability to classify energy fluctuation segments is enhanced. The comparison judgment between energy trend and capacity stage boundary replaces static capacity value, and a segmented lifetime attribution structure is established. The pressure difference rate combined with the capacity change direction forms a cycle reversal combination criterion, which broadens the dimension of nonlinear trajectory identification. The residual weight dynamic adjustment strategy replaces the fixed residual weight, and an adaptive prediction boundary mechanism is constructed to make the prediction range of capacity mutation segment converge to the current cycle behavior. The processing flow forms a four-loop closure of trend tracking, mutation detection, stage attribution and boundary pressure control, so that the lithium battery degradation trajectory behavior has self-classification and prediction stability capabilities in the jump segment. Attached Figure Description

[0037] Figure 1 This is a system flowchart of the present invention;

[0038] Figure 2 This is a flowchart of the capacity anomaly identification module in this invention;

[0039] Figure 3 This is a flowchart of the energy fluctuation grouping module in this invention;

[0040] Figure 4 This is a flowchart of the lifespan stage attribution module in this invention;

[0041] Figure 5 This is a flowchart of the reversal behavior recognition module in this invention;

[0042] Figure 6 This is a flowchart of the predictive boundary pressure control module in this invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0044] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0045] Example

[0046] Please see Figure 1 This invention provides a technical solution: a lithium battery capacity decay nonlinear trajectory analysis and prediction system comprising:

[0047] The capacity anomaly identification module is based on lithium battery cycle data to obtain average charging voltage, average charging current, average discharging voltage and average discharging current, analyze the changing trend of voltage and current ratio within the cycle, calculate the slope change direction between adjacent cycles, determine whether there are two reverse trends in three consecutive cycles, and select cycles with abnormal power offset difference as trajectory breakpoints to generate slope turning trajectory nodes.

[0048] The energy fluctuation grouping module calls the total discharge power contained in the cycle of the slope turning trajectory node and the output power corresponding to the SOC boundary, calculates the power difference in each cycle, divides the difference rate category according to the rate label, filters the key segments of power deviation in high rate cycles, judges whether they have a continuous abnormal change trend, and obtains the continuous segment of rate offset.

[0049] The lifespan stage attribution module determines whether there is a segment difference between the energy offset amplitude and the capacity trajectory division benchmark based on the continuous segment of the rate offset. It selects the period with the most significant offset as the stage classification adjustment target and reassigns it to the corresponding lifespan type according to the battery capacity degradation segmentation standard to obtain the capacity degradation stage sequence.

[0050] The reversal behavior recognition module is based on the capacity degradation stage sequence. It calculates the pressure difference rate formed by the start and end pressure difference and the duration of each cycle, determines whether the rate continues to increase in multiple cycles, compares whether the start and end capacity constitutes a reverse change, identifies the cycle segments in continuous cycles that simultaneously satisfy the direction of change, and obtains the trajectory reversal joint cycle segments.

[0051] The prediction boundary pressure control module is based on the trajectory reversal joint period segment. It adjusts the proportion of historical residual interference in the period, redefines the range of change of the allowable boundary of prediction error according to the reversal trend, limits the prediction interval to within the reference change range, and retains the boundary convergence operation in the continuous period to obtain the predicted trajectory compression interval.

[0052] The slope inflection trajectory nodes include ratio change breakpoints, power anomaly cycle markers, and trajectory reverse trend labels. The continuous segment of the ratio offset includes ratio category grouping identifiers, power difference quantification data, and fluctuation range number. The capacity degradation stage sequence includes lifetime stage type identifiers, capacity reduction amplitude range, and stage boundary positioning labels. The trajectory reversal joint cycle segment includes capacity rebound marker cycle, pressure difference change anomaly sequence, and trend reversal identification number. The predicted trajectory compression interval includes error boundary limit range, residual interference correction parameters, and continuous cycle compression record number.

[0053] The slope change direction refers to the increasing or decreasing trend of the difference sequence of the average voltage and current ratio on the time axis in two adjacent cycles, reflecting whether the ratio trend shows an upward or downward state. A reverse trend refers to a reversal of the slope change direction twice in three consecutive cycles, such as an initial increase followed by a decrease followed by an increase. This structure is used to identify the existence of trajectory abrupt inflection points. The power offset difference refers to the magnitude of the offset in power performance corresponding to the current cycle ratio compared to the previous cycle, often manifested as output power deviation caused by a sudden change in the voltage / current ratio. The trajectory breakpoint refers to the location where the ratio trend changes abruptly in the time series, used to define the discontinuous starting point of the capacity decay trajectory. The SOC boundary refers to the starting and ending values ​​of the battery's state of charge within each discharge cycle, i.e., the two endpoints used to calculate the theoretically required output energy, often used to evaluate the discharge performance per unit capacity. The critical amplitude segment refers to the set of cycles in the power difference fluctuation data where the power change is large, the trend is significant, and continuous, used to identify typical abnormal energy behavior during rate discharge. The capacity trajectory refers to the trend line of the battery's capacity changing over time in continuous charge and discharge cycles, usually showing a non-linear decrease with staged decline characteristics. As a crucial basis for lifespan attribution; the differential pressure rate refers to the change in internal resistance voltage drop corresponding to the change in battery terminal voltage within a single prediction cycle, divided by the cycle duration, reflecting the trend of internal resistance growth and serving as one of the indicators for judging degradation activity; the start and end capacity refers to the capacity values ​​corresponding to the start and end of the cycle, used to determine whether the capacity has experienced unexpected growth within the cycle, i.e., "rebound" behavior, serving as a reversal identification condition; historical residual interference refers to the offset effect caused by the reference of past error data in the current cycle prediction process. If not corrected, it will cause the model results to continuously deviate from the true trajectory. The boundary convergence operation refers to the compression processing performed on the upper and lower limits of the current prediction, so that the range of the prediction result converges to the allowable error range cycle by cycle, preventing the prediction from going out of control due to abnormal segments. An abnormal segment refers to a range in which the capacity or related parameters (such as ratio, power, voltage difference, SOC) show fluctuation behavior that is significantly inconsistent with the overall degradation trend within several consecutive cycles in the time series. If the lithium battery capacity shows a stable decline in 200 to 230 charge-discharge cycles, but suddenly rebounds in the 215th to 220th cycle and is accompanied by a rapid increase in voltage difference, these 6 cycles constitute an abnormal segment.

[0054] Please see Figure 2 The capacity anomaly detection module includes:

[0055] The periodic electrical parameter extraction submodule analyzes the original voltage and current data sequence in each cycle based on lithium battery cycle data, divides the data into segments according to the charging and discharging stages, calculates the periodic average values ​​of voltage and current in the charging segment and voltage and current in the discharging segment, and generates a periodic electrical parameter mean value group.

[0056] The system reads the raw voltage and current sequence data continuously recorded during a complete charge and discharge cycle of the battery. Based on changes in current polarity or charge / discharge control signals, it divides the system into charging and discharging stages. For example, it determines whether the current changes from negative to positive and remains positive for a certain period of time, thus entering the charging stage; otherwise, it enters the discharging stage. Voltage and current data are accumulated within each stage, and then divided by the total number of data points to obtain the average voltage and average current for that stage. For example, in the charging stage, the total voltage sampling data is 1260, and the number of sampling points is 300, resulting in an average charging voltage of 4.2 ohms, a total current of 600 ohms, and an average charging current of 2 ohms. In the discharging stage, the total voltage is 1480, the number of sampling points is 400, the average discharging voltage is 3.7 ohms, the total current is -800 ohms, and the average discharging current is -2 ohms. Finally, the average charging voltage, average charging current, average discharging voltage, and average discharging current for each cycle are compiled into a periodic electrical parameter mean group, which is used for subsequent ratio calculations and trend analysis.

[0057] The ratio trend analysis submodule analyzes the comparison relationship between the charging ratio and the discharging ratio in each cycle based on the average value group of periodic electrical parameters, constructs a cross-cycle ratio sequence, compares the rising and falling trend directions between adjacent cycles, and determines whether there are two slope trends with opposite directions of change in three consecutive cycles, forming a slope reversal trend.

[0058] The ratios between voltage and current in the charging and discharging phases of each cycle are calculated and arranged sequentially to form a ratio list. Discharging phase ratios are uniformly processed using absolute values ​​to avoid directional interference. By calculating the ratio difference between the current and previous cycles, it is recorded whether the ratio is increasing or decreasing. For example, if the ratio is 2.0 in the first cycle and 2.3 in the second, it is considered an increase; if it is 2.0 in the third cycle, it is considered a decrease. If a combination of increase followed by decrease or decrease followed by increase occurs in three consecutive cycles, it is identified as a directional reversal, defined as a slope inflection point. If the ratio change during this reversal exceeds a preset threshold (e.g., a threshold of 0.1), the change must be greater than this value to be considered a valid slope change. This threshold is set based on the battery platform voltage fluctuation range and the accuracy assessment of the sampling system. Historical data shows that the normal fluctuation range in this system is 0.05 to 0.1, therefore 0.1 is used as the lower limit for judgment. When a change pattern of more than two valid reversals is detected, a slope inflection trend is determined.

[0059] The power offset filtering submodule filters cycles with direction reversal characteristics based on the slope turning trend, calculates the offset difference between the average power of the cycle and the adjacent cycle, determines whether its offset degree is higher than the normal range of offset change, and compares the turning direction of its ratio trend again to generate slope turning trajectory nodes.

[0060] For the slope inflection trend cycle, the average voltage and current values ​​of this cycle, the previous cycle, and the next cycle are extracted, and the average power of the three cycles is calculated accordingly. By comparing the power difference between the current cycle and the previous and next cycles, it is determined whether there is a significant jump in the power of the current cycle. For example, if the power of the current cycle is 8.4, the previous cycle is 6.8, and the next cycle is 9.0, then the difference with the previous cycle is 1.6, and the difference with the next cycle is 0.6. The average of the absolute values ​​of the two is 1.1. Then it is compared with the normal range of system power deviation. The normal range is constructed by referring to the average value and fluctuation amplitude calculated from the power change data of 100 historical cycles, and the upper and lower boundary values ​​are set to 0.2 to 1.0. If the average deviation of the current cycle exceeds the upper limit of 1.0, it is determined to be an abnormal power deviation cycle. Then, it is compared again to see if there is a reversal of the ratio direction in the current cycle. If it still meets the trend inflection, the cycle is marked as a slope inflection trajectory node. This node will serve as a key identification point for the nonlinear characteristics of the capacity trajectory.

[0061] Please see Figure 3 The energy fluctuation grouping module includes:

[0062] The periodic power calculation submodule calls the slope turning trajectory node to calculate the deviation between the discharge power contained in the node and the output power corresponding to SOC in each period, analyzes the difference distribution of the two in the time series, and constructs the periodic power difference sequence.

[0063] For each identified cycle, the actual output power value of the discharge phase within that cycle is extracted one by one. The instantaneous power at each moment is obtained by multiplying the voltage and current sequences collected during the discharge phase point by point. Then, the instantaneous power at all moments is summed and divided by the number of points to obtain the average power of the discharge segment within that cycle. At the same time, based on the SOC change range within that cycle, the mapping relationship curve between SOC and output power is found, and the reference output power sequence between the initial and final SOC values ​​is extracted and averaged to obtain the theoretical output power corresponding to the SOC interval. Then, the power deviation is obtained by subtracting the theoretical power from the average discharge power of the current cycle. For example, if the SOC drops from 0.8 to 0.3 in a certain cycle, the theoretical output power is 6.5, and the measured discharge power is 5.8, then the deviation is -0.7. The deviation is calculated for each slope transition cycle in this way, and a power difference sequence arranged in time order is constructed. The number of each cycle is used as an index, and the corresponding power deviation is used as the value to form a cycle power difference sequence with time labeling, which is used for subsequent classification analysis and anomaly identification.

[0064] The ratio difference classification submodule analyzes the distribution pattern of the periodic power difference sequence under the ratio difference label, calculates the average amplitude and range of the difference in each category, compares the offset performance between each category, and filters the classification segments with prominent offset to obtain the concentrated interval of ratio deviation.

[0065] The process involves calling the corresponding rate identifier for each cycle and grouping all cycles by rate, for example, labeling the rates as 1C, 2C, and 3C. The power deviations within each rate group are then categorized according to the group. Next, the deviation values ​​within each rate group are averaged to obtain the average deviation amplitude for each rate group. The range of variation is then calculated using the difference between the minimum and maximum deviation values ​​within each group. For example, in group 1C, the minimum is -0.4 and the maximum is 1.1, so the range of variation is 1.5; in group 2C, the minimum is -1.0 and the maximum is 0.6, so the range of variation is 1.6. Finally, the average deviation amplitude is compared between each rate group. The deviation range is also considered. If the average deviation of a group significantly deviates from other groups, and its internal range exceeds the set boundary, then the group is judged to have a prominent deviation. The average deviation threshold is set at 0.8, which is derived from the statistical mean plus one standard deviation of 500 historical multiples. For example, if the mean is 0.5 and the standard deviation is 0.3, then the threshold is defined as 0.8. When the average deviation of a multiple group is higher than this threshold, and its internal range exceeds 1.2, then the group is selected as the deviation concentration segment. For example, if the average deviation of the 3C multiple group is 1.2 and the range is 1.6, then it is identified as the multiple deviation concentration interval.

[0066] The continuous offset identification submodule determines the offset trend between consecutive periods within the concentrated range of rate deviation, analyzes the continuous amplitude of power difference, and combines the SOC boundary and rate characteristics, using the following formula:

[0067] ;

[0068] The changes within each window of the sequence are calculated, and continuous segments with key offset fluctuations are selected to obtain continuous segments of the multiplier offset. This represents the magnitude shift, reflecting the cumulative change level of the shift characteristics in each window. Representing the Cycle power difference, which is the difference between the discharge power and the output power corresponding to the state of charge (SOC) in a given cycle. Indicates the first The power difference over the period is used to compare the offset amplitude with that of the current period. Representing the The periodic SOC boundary corresponds to several terms, representing the limit parameters under the energy state change of that period. Representing the The output power of the cycle is used to measure the discharge capability under SOC conditions. Representing the The average amplitude of the difference sequence corresponding to the periodic multiple label is a typical offset used to classify multiple types. Indicates the first The degree of skewness in the difference sequence corresponding to the period multiple label reflects the symmetry of the data distribution. Indicates the first The fluctuation range of the difference sequence within the period multiple label measures the stability of the offset change within that category. It refers to the number of cycles;

[0069] Call the range of concentrated rate deviations, for example, this range is from period 2 to period 5, corresponding to power differences of 1.3, 1.5, 1.6, and 1.7 respectively. Set the SOC boundary parameters to 0.58, 0.60, 0.62, and 0.61, with output powers of 27.8, 28.5, 27.9, and 27.1 respectively. Set the average amplitude of the rate deviation corresponding to the rate label to 1.4, 1.5, 1.6, and 1.5, the skewness μ to 0.1, 0.15, 0.12, and 0.08 respectively, and the fluctuation range σ to 0.2, 0.18, 0.16, and 0.19. Substitute the above parameters into the formula in sequence:

[0070] ;

[0071] Substituting periods 2 through 5 sequentially, the first set of parameters is the difference between period 2 and period 1. The corresponding SOC is 0.60, the output power is 28.5, the average rate difference is 1.5, the skewness is 0.15, and the fluctuation range is 0.18. Substituting these values ​​into the calculation, we get:

[0072] ;

[0073] calculate ,but , The product of the first term is approximately:

[0074] ;

[0075] The second set of parameters is the difference between period 3 and period 2. Given: SOC 0.62, output power 27.9, rate assist 1.6, skew 0.12, fluctuation range 0.16, calculate:

[0076] ;

[0077] ,but , The product is:

[0078] ;

[0079] The difference between period 4 and period 3 in the third set of parameters is: SOC is 0.61, power is 27.1, rate capability is 1.5, skew is 0.08, and fluctuation is 0.19.

[0080] ;

[0081] , , The product is:

[0082] ;

[0083] Summing yields the overall change:

[0084] ;

[0085] This result represents a change of 0.136 in the continuous segment of the magnification offset. If the set judgment benchmark is 0.1, this result exceeds the judgment benchmark, indicating that the continuous period has a continuous offset tendency. This segment is used as the key segment for subsequent diagnostic analysis, and the output result is the continuous segment of the magnification offset.

[0086] Please see Figure 4 The lifespan stage attribution module includes:

[0087] The multiplier offset segment identification submodule determines whether the amplitude difference between adjacent data in the multiplier sequence is continuous based on the continuous multiplier offset segment. If it is continuous, it is classified as a multiplier offset segment. By comparing the difference between the energy offset amplitude of this segment and the capacity trajectory baseline, it determines whether it exceeds the allowable range. If it exceeds the allowable range, it marks the abnormal segment and obtains the offset amplitude abnormal range.

[0088] Extract all period numbers contained in each continuous segment of the multiplier offset, arrange them in ascending order of period number to form a multiplier sequence, and call the multiplier values ​​of adjacent periods one by one and perform difference calculations. Determine whether the difference between any two adjacent multipliers is within the set continuous change tolerance range. This tolerance range can be set to 0.2 to 0.5. If the difference between adjacent multipliers is always within this range and the direction of change is consistent, it is marked as continuous. If the number of continuous periods is not less than 3, it is considered a valid multiplier offset segment. For example, if the multipliers of periods 21, 22, 23, and 24 are 1.0, 1.3, 1.5, and 1.8 respectively, then the adjacent differences are 0.3, 0.2, and 0.3, all within the tolerance range, and the number of continuous periods is 4, which satisfies the condition. Under certain conditions, the actual discharge energy output values ​​of each cycle in the rate offset segment are extracted and the difference between them and the expected energy values ​​of the corresponding cycle of the reference capacity trajectory baseline curve is used to obtain the energy offset amplitude. The baseline curve is constructed based on the standard decay model of each stage of the battery life, from the initial stage to the middle stage and the end stage. If the expected energy in the reference curve is 5.0, 4.8, 4.5, and 4.3, and the measured values ​​are 4.0, 3.9, 3.5, and 3.1, then the difference values ​​are calculated as 1.0, 0.9, 1.0, and 1.2, respectively. The average offset is taken as 1.03. After comparing this value with the baseline allowable difference value of 0.5, it is determined that the energy offset exceeds the upper limit because it exceeds the upper limit. Finally, the segment from cycle 21 to 24 is marked as the offset amplitude abnormal range.

[0089] The offset mutation cycle screening submodule analyzes the capacity change trend corresponding to the start and end cycles of each segment in the abnormal offset amplitude interval, calculates the difference in capacity change rate between consecutive cycles, and screens out cycles with mutation characteristics by comparing with the capacity change trend baseline to obtain the offset mutation cycle sequence.

[0090] Capacity data for each consecutive period in the abnormal segment is extracted, and the change between the current period's capacity and the previous period's capacity is calculated to form a capacity change rate sequence. Then, the differences between the terms within this rate sequence are calculated to obtain the rate difference, i.e., the difference between adjacent rates is processed. Further, it is determined whether the rate difference is greater than a preset mutation judgment value. This value is set to 0.3 based on the rate fluctuation standard in historical stable operating cycles. For example, if the capacity values ​​for cycles 21 to 24 are 96%, 94%, 89%, and 82% respectively, and the capacity change rates are -2%, -5%, and -7% respectively, then the rate difference is 3% and 2%. If the rate difference exceeds the set threshold of 0.3, it is determined that there is a rate mutation feature. Then, the difference between this rate and the baseline capacity change trend is calculated. The baseline trend is obtained by statistically analyzing the normal operating range in the initial stage. For example, if the baseline capacity stable decrease rate is between -1.5% and -2.5%, then the above rates of -5% and -7% are far beyond the baseline range and are determined to be mutations. Finally, cycles 22 to 24 are marked as the offset mutation cycle sequence.

[0091] The lifetime type reconstruction submodule compares the capacity decrease magnitude of each cycle in the offset mutation cycle sequence with the segmentation criteria of the capacity degradation stage to determine the boundary of the degradation stage to which each cycle belongs, using the formula:

[0092] ;

[0093] Calculate the square mean of capacity matching deviation This is used to measure the difference between the magnitude of capacity decline in a mutation cycle and the combination of dynamometer shift and capacity trajectory trend, and to reassign the corresponding lifetime type to obtain a capacity degradation stage sequence, where, Indicates the first The rate of capacity decrease per cycle reflects the actual change in capacity during that cycle. Indicates the first The rate offset amplitude over each period is used to describe the degree of shift in the rate data within that period. Indicates the first The trend of capacity trajectory changes over a period of time indicates the direction and magnitude of capacity evolution over that period. This represents the total number of mutation cycles and is used to normalize the deviation calculation results across all cycles.

[0094] By extracting the initial and final capacities from the capacity data sequence of each mutation cycle, denoted as the cycle start capacity and cycle end capacity respectively, the difference between the two is calculated as the capacity decrease magnitude. For example, if the initial capacity of the first cycle is 2.10 Ah and the end capacity is 1.98 Ah, then the capacity decrease magnitude is 0.12 Ah. Next, continuous measurement points of the dynamometer data within each cycle are extracted to obtain the difference between the maximum and minimum dynamometer values, which is input as the dynamometer offset magnitude into the RE term. For example, if the maximum dynamometer value in a certain cycle is 2.4 and the minimum is 0.6, then the dynamometer offset magnitude is 1.8. Subsequently, the capacity trajectory within that cycle is acquired. The trend data is calculated by taking the average capacity reduction in each cycle as the capacity trajectory trend term QR. For example, if the capacity decreases by 0.02, 0.025, 0.024, 0.026, and 0.028 Ah in five cycles, the trajectory trend is the average of the five cycles, which is 0.0246 Ah. Substituting these three data points into the formula, the calculation is performed cycle by cycle. The capacity decrease magnitude QD, the ratio offset magnitude RE, and the capacity trajectory trend QR in each cycle are multiplied, squared, and then the difference is squared to reflect the matching difference between the theoretical offset capacity and the actual decrease. The multiplication is performed first. Then, take the square root of the product, and then the difference between this and the capacity decrease rate, square that difference, and average it over all periods to generate the squared mean of capacity matching deviation (RS). If there are 5 periods, with the corresponding parameters: QD [0.12, 0.15, 0.21, 0.26, 0.19], RE [0.85, 1.05, 1.40, 1.80, 1.10], and QR [0.75, 0.90, 1.10, 1.30, 0.95], then the formula expands to:

[0095] ;

[0096] ;

[0097] ;

[0098] The value 0.901 represents the average matching deviation under the current five mutation cycles. If it corresponds to the second stage interval in the lifespan stage division standard, then the current cycle group can be determined to belong to the second stage, thus completing the acquisition of the capacity degradation stage sequence.

[0099] Please see Figure 5 The reversal behavior recognition module includes:

[0100] The differential pressure rate calculation submodule calculates the starting and ending pressure difference of each cycle based on the capacity degradation stage sequence. Combined with the start and end times within the corresponding time period, it analyzes the continuous change behavior within each cycle. Then, it converts the cycle pressure difference to time by ratio conversion to determine the change rate corresponding to the cycle and obtains the cycle pressure difference rate characteristics.

[0101] The differential voltage rate calculation submodule is based on the capacity degradation stage sequence. First, it reads the original recorded data of the start and end voltages of each cycle in each degradation stage, and calculates the voltage difference between the two time points to obtain the start and end voltage difference for that cycle. Then, it obtains the timestamps of the start and end of the cycle, and calculates the difference to obtain the duration of the voltage difference. While ensuring that the time data within the cycle is accurately marked at the millisecond level, it calculates the rate of change of the voltage difference over time within that cycle. For example, in cycle 36, the start voltage is 4.15 and the end voltage is 3.60. The start time is... If the time is 540 seconds and the termination time is 1060 seconds, then the pressure difference is 0.55 and the duration is 520 seconds. After conversion, the pressure difference rate is 0.00106. Repeating this calculation process to traverse all degradation stage cycles forms a complete cycle pressure difference rate sequence. During the execution, the corresponding cycle number is recorded for each rate value as a sequence index. The final cycle pressure difference rate characteristics are as follows: cycle number 36 corresponds to a rate of 0.00106, cycle number 37 corresponds to a rate of 0.00123, cycle number 38 corresponds to a rate of 0.00142, etc.

[0102] The rate trend judgment submodule analyzes the direction of change of the periodic differential pressure rate characteristics within a continuous period, determines whether there is a continuously increasing pattern sequence, filters out time periods that meet the continuous increase requirement, judges the trend structure by the direction of rate change, and obtains the differential pressure rate growth trend range.

[0103] The rate trend judgment submodule analyzes the direction of change of the differential pressure rate characteristics within a continuous cycle. First, it reads the differential pressure rate values ​​sequentially by cycle number. It then calculates the difference between the current cycle rate and the previous cycle rate, determining if the difference is positive, i.e., whether the current cycle shows a rate increase relative to the previous cycle. If the rate difference for three or more consecutive cycles is greater than the set minimum increment threshold of 0.0001, it is considered to form an increasing pattern sequence. For example, if the rate in cycle 40 is 0.00102, in cycle 41 it is 0.00113, and in cycle 42 it is 0.00127, the differences are 0.00011 and 0.00014 respectively, both exceeding the threshold, thus indicating a valid increasing trend. The system sequentially compares the rate differences formed by all cycles and accumulates adjacent consecutive increasing segments. If the number of consecutive cycles is not less than 3 and all of them satisfy the condition that the difference direction is consistent, it is marked as a continuous upward time period. During this process, all cycle segments with a rate difference value of negative or less than 0.0001 are removed and the current segment recording is terminated. At the same time, the start and end cycle numbers of the continuous upward segment are output to describe the trend structure. For example, the rates corresponding to cycles 40 to 44 are 0.00102, 0.00113, 0.00127, 0.00138, and 0.00149, respectively. Each of their differences is greater than 0.0001 and increases. Therefore, this segment is marked as an effective growth trend segment, and the pressure difference rate growth trend interval is output.

[0104] The capacity reverse verification submodule calls the differential pressure rate growth trend interval, compares the start and end states of capacity for each cycle, and calculates the relative change of capacity direction with differential pressure rate using the formula:

[0105] ;

[0106] By filtering out periodic segments where the capacity and rate directions are opposite, and determining whether they constitute trend-reversing behavior, we obtain joint periodic segments with trajectory reversal. Indicates the first The ratio of capacity-rate reverse coupling degree corresponding to each period of the differential pressure rate growth trend. Indicates the first The difference between the ending capacity and the starting capacity in each cycle reflects the amount of capacity change within that cycle. Indicates the first The difference between the termination differential pressure rate and the starting differential pressure rate in each cycle reflects the change in differential pressure rate within that cycle. Indicates the first The number of consecutive cycles contained in a growth trend cycle segment;

[0107] The differential pressure rate growth trend range is called, and each included period is processed sequentially. First, the start and end difference of the content in each period is calculated, let the first period be... The initial capacity of the cycle is The termination capacity is Its capacity change is expressed as Simultaneously, the change in differential pressure rate corresponding to that cycle is recorded, defined as... ,in The differential pressure rate represents the degree of change in pressure difference per unit time. Then, the capacity changes over all cycles are combined with the differential pressure rate changes and substituted into the formula to calculate the reverse coupling ratio. Using three consecutive cycles as an example, the capacity changes are from 90 to 87, from 87 to 83, and from 83 to 80, respectively. The corresponding capacity changes are as follows: , , Meanwhile, the corresponding differential pressure rate changes from 0.10 to 0.14, from 0.14 to 0.18, and from 0.18 to 0.22. Therefore, the change in differential pressure rate is... , , Substitute the values ​​into the formula:

[0108] ;

[0109] ;

[0110] This result indicates that during this growth trend period, capacity continuously decreases while the differential pressure rate continuously increases, with the two trends moving in opposite directions. The calculated reverse coupling ratio is [value missing]. This ratio can be used for subsequent comparison with the set reference standard. For example, when the set judgment standard is 0.03, since this value is significantly lower than the judgment benchmark, it can be identified that the period constitutes a valid trend reversal behavior.

[0111] Please see Figure 6 The predicted boundary pressure control module includes:

[0112] The trajectory reversal recognition submodule analyzes the trend fluctuation amplitude and directional changes of continuous cycles based on trajectory reversal combined period segments, determines whether the trend has reversed direction between adjacent periods, identifies segments with directional reversal characteristics by comparing the changes in the trend reversal amplitude, and generates reversal recognition mark content.

[0113] The voltage, current, and capacity trend data are read sequentially for each cycle. The end value of each cycle is extracted and continuously compared with the starting value of the next cycle. The direction and fluctuation amplitude of the current cycle trend are calculated. If the capacity decreases in the previous cycle and increases in the next cycle, it is recorded as a direction reversal. The fluctuation amplitude is obtained by the difference between the end value and the starting value of the cycle. The change amplitude of such difference is compared with that of adjacent cycles to determine whether the direction shows a reverse trend. For example, if the capacity at the end of cycle 58 is 85.2%, the capacity at the beginning of cycle 59 is 85.3%, and the capacity at the end of cycle is 86.1%, then there is an increase relative to cycle 58, and the direction is opposite, forming a trend reversal. If three consecutive cycles show the opposite trend of first decreasing and then increasing or first increasing and then decreasing, and the change amplitude of each cycle is greater than the set threshold of 0.3%, it is determined to be a direction reversal sequence. The trend recognition threshold of 0.3% is set based on the average deviation of the natural fluctuation of historical capacity. This value ensures that the recognition change is not caused by sampling error. The cycle segment with the characteristics of trend reversal direction and excessive fluctuation amplitude is recorded as the reversal recognition mark.

[0114] The historical interference weight adjustment submodule calls the reversal identification mark content, optimizes the interference structure composition of the current period, analyzes the changes in the proportion of trend offset interference and random disturbance in the prediction error, determines whether it is in a reversal state, and rebalances the role ratio between the difference interference items accordingly, filters the adjusted interference composition structure, and obtains the periodic interference weight combination.

[0115] Within the marked period, the error composition data for each period is extracted in chronological order. The trend offset error term and the non-trend random disturbance error term are retrieved separately, and their respective percentages of the total prediction error are calculated as proportion indicators. Then, the difference between the trend offset proportion of the current period and the previous period is used to determine the situation. If the trend proportion gradually decreases while the disturbance proportion continues to increase over three consecutive periods, it is determined that the period is in a state dominated by reversal disturbances. If the decrease in the offset proportion exceeds 15% and the increase in the disturbance proportion exceeds 20%, the disturbance structure of the current period is marked as needing adjustment. Based on this judgment, the weights of the two types of interference items are reset. The weight of the trend item is adjusted from 70% to 50%, and the weight of the disturbance item is adjusted from 30% to 50%. The adjusted interference ratios are then applied to the prediction error structure of the same period. The resulting interference weight structure is 50% for the trend item and 50% for the disturbance item. If the error value of the disturbance item is 0.45 and the error value of the trend item is 0.45 in a specific period, and the total error is 0.9, then the weights meet the rebalancing condition. This structure is selected as the adjusted interference composition structure, and the periodic interference weight combination is obtained.

[0116] The boundary convergence limiting submodule calculates the range of change of the prediction segment based on the combination of periodic interference proportions, analyzes the historical prediction boundary, judges the expansion segment that appears in the error interval, adjusts the boundary segment of the abnormal expansion part, and obtains the prediction trajectory compression interval by redefining the range of error change amplitude.

[0117] The start and end period numbers of the target prediction segment in the time series are determined. The error between the predicted value and the actual value of each period is extracted. The upper and lower bounds of the error are statistically analyzed and marked as the current prediction boundary. Then, the upper and lower fluctuation range of the error in each period is calculated sequentially within the segment. It is determined whether the rate of increase of the upper bound of the error exceeds the set threshold for two or more consecutive periods. If the upper bound of the error increases by more than 5% in a certain segment, it is recorded as an expansion segment. Then, the disturbance weight structure is re-invoked for the expansion segment, and the error adjustment range is redefined according to the ratio of disturbance to trend. For example, the prediction error is allowed to be ±0.5 in the trend-dominant period and reduced to ±0.3 in the disturbance-dominant period. If the upper bound of the prediction error expands from 0.6 to 0.8 in periods 63 to 65, it is an abnormal expansion. After readjusting the upper and lower bounds of the prediction to ±0.3, the prediction boundary range is shrunk to [-0.3, 0.3]. The adjustment results are integrated into the entire prediction trajectory data. The changes in the data before and after the adjustment are compared, and the time period and interval index information corresponding to the adjustment operation are retained to obtain the compressed prediction trajectory interval.

[0118] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A system for analyzing and predicting the nonlinear trajectory of lithium battery capacity decay, characterized in that, The system includes: The capacity anomaly identification module analyzes the voltage and current ratio change trend based on lithium battery cycle data, calculates the slope change direction between adjacent cycles, determines whether a reverse trend occurs in three consecutive cycles, filters out power offset anomaly cycles, and generates slope turning trajectory nodes. The capacity anomaly identification module includes: The periodic electrical parameter extraction submodule analyzes the original voltage and current data sequence in each cycle based on lithium battery cycle data, divides the data into segments according to the charging and discharging stages, calculates the periodic average values ​​of voltage and current in the charging segment and voltage and current in the discharging segment, and generates a periodic electrical parameter mean value group. The ratio trend analysis submodule analyzes the comparison relationship between the charging ratio and the discharging ratio in each cycle based on the average value group of the periodic electrical parameters, constructs a cross-cycle ratio sequence, compares the rising and falling trend directions between adjacent cycles, and determines whether there are two slope trends with opposite directions of change in three consecutive cycles, forming a slope reversal trend. The power offset filtering submodule filters out cycles with direction reversal characteristics based on the slope turning trend, calculates the offset difference between the average power of the cycle and the adjacent cycle, determines whether its offset degree is higher than the normal range of offset change, and compares the turning direction of its ratio trend again to generate slope turning trajectory nodes. The energy fluctuation grouping module calls the periodic discharge power and the output power corresponding to the SOC boundary in the slope turning trajectory node, calculates the power difference for each cycle, filters the power deviation amplitude segment in the high-rate cycle, judges whether there is an abnormal change trend, and obtains the continuous segment of rate offset. The lifetime stage attribution module determines whether there is a difference between the energy offset amplitude and the capacity trajectory baseline based on the continuous segment of the multiplier offset, filters out the period with prominent offset, reassigns it to the corresponding lifetime type, and obtains the capacity degradation stage sequence. The reversal behavior recognition module calculates the pressure difference and duration of each cycle based on the capacity degradation stage sequence, determines whether the pressure difference rate continues to increase, compares whether the start and end capacities have reversed, identifies the cycle segments in the continuous cycle that satisfy the direction of change, and obtains the trajectory reversal joint cycle segments.

2. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 1, characterized in that, The slope inflection trajectory node includes a ratio change breakpoint, a power anomaly cycle marker, and a trajectory reverse trend label. The ratio offset continuous segment includes a ratio category grouping identifier, power difference quantification data, and fluctuation range number. The capacity degradation stage sequence includes a lifetime stage type identifier, a capacity reduction amplitude range, and a stage boundary positioning label. The trajectory reversal joint cycle segment includes a capacity rebound marker cycle, a differential pressure change anomaly sequence, and a trend reversal identification number.

3. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 1, characterized in that, The energy fluctuation grouping module includes: The periodic power calculation submodule calls the slope turning trajectory node to calculate the deviation between the discharge power contained in the node and the output power corresponding to SOC in each period, analyzes the difference distribution of the two in the time series, and constructs the periodic power difference sequence. The ratio difference classification submodule analyzes the distribution pattern of the periodic power difference sequence under the ratio difference label, calculates the average amplitude and range of change of the difference in each category, compares the offset performance between each category, and filters the classification segments with prominent offset to obtain the concentrated interval of ratio deviation. The continuous offset identification submodule determines the offset trend between consecutive periods within the concentrated range of the multiplier deviation, analyzes the continuous amplitude of the power difference, and calculates the change performance within each window in the sequence by combining the SOC boundary and the multiplier characteristics. It then filters out the continuous segments with key offset fluctuations to obtain the continuous segments of multiplier offset.

4. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 1, characterized in that, The lifespan stage attribution module includes: The multiplier offset segment identification submodule determines whether the amplitude difference between adjacent data in the multiplier sequence is continuous based on the continuous multiplier offset segment. If it is continuous, it is classified as a multiplier offset segment. By comparing the difference between the energy offset amplitude of the segment and the capacity trajectory reference, it determines whether it exceeds the allowable range. If it exceeds the allowable range, it marks the abnormal segment and obtains the offset amplitude abnormal range. The offset mutation cycle screening submodule analyzes the capacity change trend corresponding to the start and end cycles of each segment in the abnormal offset amplitude interval, calculates the difference in capacity change rate between consecutive cycles, and screens out cycles with mutation characteristics by comparing with the capacity change trend baseline to obtain the offset mutation cycle sequence. The lifetime type reconstruction and attribution submodule compares the capacity decrease magnitude of each cycle in the offset mutation cycle sequence with the segmentation criteria of the capacity degradation stage, determines the degradation stage boundary to which each cycle belongs, calculates the mean squared capacity matching deviation, reassigns the corresponding lifetime type, and obtains the capacity degradation stage sequence.

5. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 1, characterized in that, The reversal behavior recognition module includes: The differential pressure rate calculation submodule calculates the starting and ending pressure difference of each cycle based on the capacity degradation stage sequence. Combined with the start and end times within the corresponding time period, it analyzes the continuous change behavior within each cycle. Then, it converts the cycle pressure difference to time by ratio calculation to determine the change rate corresponding to the cycle and obtains the cycle pressure difference rate characteristics. The rate trend judgment submodule analyzes the direction of change of the periodic differential pressure rate characteristics within a continuous period, determines whether there is a continuously increasing pattern sequence, filters out time periods that meet the continuous increase requirement, judges the trend structure by the direction of rate change, and obtains the differential pressure rate growth trend range. The capacity reverse verification submodule calls the pressure difference rate growth trend interval, compares the capacity start and end states of each cycle, calculates the relative change between the capacity direction and the pressure difference rate, filters out the cycle segments where the capacity and rate directions are opposite, determines that they constitute trend reversal behavior, and obtains the trajectory reversal joint cycle segment.

6. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 1, characterized in that, The system includes: The prediction boundary pressure control module adjusts the proportion of historical residual interference in the cycle based on the trajectory reversal joint cycle segment, redefines the range of change of the allowable boundary of prediction error according to the reversal trend, limits the prediction interval to within the reference change range, and retains the boundary convergence operation in the continuous cycle to obtain the predicted trajectory compression interval. The predicted trajectory compression interval includes the error boundary limit range, residual interference correction parameters, and continuous period compression record number.

7. The lithium battery capacity decay nonlinear trajectory analysis and prediction system according to claim 6, characterized in that, The predicted boundary pressure control module includes: The trajectory reversal recognition submodule analyzes the trend fluctuation amplitude and directional changes of continuous cycles based on the trajectory reversal joint period segment, determines whether the trend has reversed direction between adjacent cycles, identifies segments with directional reversal characteristics by comparing the changes in the trend reversal amplitude, and generates reversal recognition mark content. The historical interference weight adjustment submodule calls the reversal identification mark content, optimizes the interference structure composition of the current period, analyzes the changes in the proportion of trend offset interference and random disturbance in the prediction error, determines whether it is in a reversal state, and rebalances the role ratio between the difference interference items accordingly, filters the adjusted interference composition structure, and obtains the periodic interference weight combination. The boundary convergence limiting submodule calculates the range of change of the prediction segment based on the combination of the periodic interference proportions, analyzes the historical prediction boundaries, judges the expansion segments that appear in the error interval, adjusts the boundary segments of the abnormal expansion parts, and obtains the prediction trajectory compression interval by redefining the range of error change amplitude.

Images