Intelligent robot battery cell consistency evaluation method and system

Abstract

The application relates to the technical field of battery pack fault monitoring, and discloses a kind of intelligent robot battery pack cell consistency evaluation method and system, wherein, method includes: capture pulse waveform and direction identifier;Reconstruct transient impedance spectrum;Decoupling identification ohmic resistance and polarization parameter;Locate fault cell;Quantify unit pulse additional heat production.Compared with the long steady state condition required by the traditional direct current resistance method, especially in the random pulse working condition of the alternating appearance of the braking energy recovery and acceleration discharge of the distribution robot, the technical problem that the ohmic resistance and polarization resistance difference cannot be obtained from the millisecond wideband pulse response in real time and leads to the imbalance of heat distribution is solved, since the application is coupled by transient triggering and wavelet reconstruction and decoupled by direction identification mechanism, realizes the consistency of alternating current resistance spectrum online fast identification and heat imbalance quantification, improves the timeliness of fault location and heat management prediction ability.

Description

Technical Field

[0001] This invention relates to the field of battery pack fault monitoring technology, and in particular to a method and system for evaluating the consistency of battery pack cells using an intelligent robot. Background Technology

[0002] Currently, mobile platforms such as intelligent delivery robots widely use power battery packs composed of multiple strings of cells. To ensure range performance and thermal safety, the industry generally requires online consistency monitoring of cell internal resistance and capacity. Existing technologies typically utilize the DC internal resistance method, applying a current pulse of a specific rate under steady-state charge and discharge conditions and measuring the voltage response to obtain the ohmic internal resistance of the cell; or employing electrochemical impedance spectroscopy for periodic offline frequency sweep detection, evaluating the cell state by analyzing impedance spectrum characteristics. These methods can provide some reference in energy storage scenarios with fixed operating conditions or where static testing can be arranged, but in the dynamic operating environment of delivery robots with frequent start-stop cycles and random overlaps of energy recovery and accelerated discharge, their detection timeliness and applicability are significantly insufficient.

[0003] For example, when a delivery robot performs last-mile delivery tasks, the regenerative braking process generates a large-current charging pulse lasting tens to hundreds of milliseconds. The subsequent acceleration generates a large-current discharging pulse of the same magnitude. These two asymmetrical short-duration pulses alternate randomly. Traditional DC internal resistance methods require the current and voltage responses to reach a quasi-steady state before internal resistance can be calculated. They cannot obtain stable readings within millisecond-level pulse durations and cannot separate the different contributions of ohmic and polarization resistance to the voltage response. Offline electrochemical impedance spectroscopy relies on external excitation signals and static conditions, making it impossible to synchronize with the robot's real-time operation. Under such random pulse conditions, even milliohm-level differences in ohmic internal resistance or polarization resistance shifts between cells will gradually accumulate local temperature differences during frequent charge-discharge commutation, leading to thermal imbalance. However, existing methods cannot quickly identify resistive inconsistencies from the wideband response of the robot's own operating pulses, resulting in delayed fault diagnosis and blind thermal management responses.

[0004] Therefore, there is an urgent need for a method that can reconstruct the equivalent circuit parameter distribution of each cell online using only the millisecond-level wideband response generated by its own operating current pulse under the pulsed operating conditions of random energy recovery and accelerated discharge of delivery robots. This would allow for rapid diagnosis of inconsistencies between the ohmic internal resistance and polarization internal resistance of cells without external excitation or steady-state conditions, to locate abnormal cells and quantify their degree of thermal imbalance, thereby improving the real-time performance and pre-diagnostic capabilities of battery pack thermal safety management. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the present invention aims to propose a method for evaluating the consistency of battery cells in intelligent robot battery packs. This method addresses the technical problem that the traditional DC internal resistance method requires long steady-state conditions, especially under the random pulse conditions of alternating braking energy recovery and acceleration discharge in delivery robots. It cannot obtain the difference between ohmic internal resistance and polarization internal resistance in real time from the millisecond-level wideband pulse response, which leads to the imbalance of thermal distribution.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for evaluating the consistency of battery cells in an intelligent robot battery pack.

[0007] The method for evaluating the consistency of battery cells in a smart robot battery pack includes: Step S10: Obtain battery pack operating data, and based on the battery pack operating data, use a pulse event triggering and segment association caching mechanism to execute an effective pulse event capture task and output a pulse event data packet; Step S20: Based on the pulse event data packet, the instantaneous impedance spectrum construction task is performed using the Morlet wavelet transform mechanism with adaptive center frequency of charge state, and the frequency domain impedance data packet is output. Step S30: Based on the frequency domain impedance data packet, the fractional-order equivalent circuit model and the direction decoupling identification mechanism are used to perform the cell equivalent circuit parameter extraction task, and the equivalent circuit parameter data packet is output. Step S40: Based on the equivalent circuit parameter data package, a robust statistical mechanism with median absolute deviation is used to perform the cell consistency assessment task, and the assessment results of abnormal cells are output. Step S50: Based on the abnormal cell evaluation results, the thermal loss quantification task is performed using a pulse event correlation and integral heat generation mechanism, and the thermal loss quantification results are output.

[0008] Preferably, step S10, which involves acquiring battery pack operating data, executing a valid pulse event capture task based on the battery pack operating data using a pulse event triggering and segment association caching mechanism, and outputting a pulse event data packet, specifically includes: Step S101: Synchronously collect the terminal voltage of each cell, the state of charge of each cell, and the bus current of the battery pack at a preset sampling frequency to form the battery pack operating data; subtract the bus current at the current sampling moment from the bus current at the adjacent previous sampling moment to obtain the bus current difference; calculate the sampling time interval between the current sampling moment and the adjacent previous sampling moment; calculate the absolute value of the bus current change rate based on the bus current difference and the sampling time interval; Step S102: Calculate the root mean square value for the absolute value of the bus current change rate within a preset time window before the current sampling time; multiply the root mean square value by a preset dynamic threshold coefficient to obtain a dynamic trigger determination value; when the absolute value of the bus current change rate is greater than the dynamic trigger determination value, and the sign of the bus current remains unchanged within a preset duration, determine that a valid pulse event has occurred, and determine the direction identifier based on the sign of the bus current relative to the preset current reference direction; Step S103: Write the terminal voltages of each cell within a preset time period before and after the triggering time of the valid pulse event into the associated buffer to obtain a voltage waveform segment; write the bus current within the corresponding time period into the associated buffer to obtain a bus current waveform segment; calculate the average state of charge of each cell within a preset time window before the triggering time to obtain the average state of charge; assign a cell identifier to each cell and generate a pulse event associated identifier for the valid pulse event; encapsulate the voltage waveform segment, the bus current waveform segment, the direction identifier, the average state of charge, the cell identifier, and the pulse event associated identifier to obtain the pulse event data packet.

[0009] Preferably, step S20, which involves using the Morlet wavelet transform mechanism with adaptive center frequency of charge state to perform the instantaneous impedance spectrum construction task based on the pulse event data packet and outputting the frequency domain impedance data packet, specifically includes: Step S201: Detrend the voltage waveform segment and bus current waveform segment in the pulse event data packet; determine the adaptive center frequency based on the average state of charge, the preset reference center frequency, and the preset center frequency adjustment coefficient; perform Morlet wavelet transform on the detrendened voltage waveform segment and bus current waveform segment based on the adaptive center frequency to obtain the Morlet wavelet transform result; Step S202: Based on the Morlet wavelet transform results, obtain the voltage wavelet coefficients and the current wavelet coefficients, and map the voltage wavelet coefficients and the current wavelet coefficients to the corresponding frequency points; Step S203: Select effective frequency points from the frequency points that simultaneously meet the preset spectral energy ratio determination condition, the preset signal-to-noise ratio determination condition, and the preset effective frequency range determination condition that matches the average state of charge; calculate the instantaneous impedance value based on the ratio of the voltage wavelet coefficient to the current wavelet coefficient at each effective frequency point; construct an instantaneous impedance spectrum based on the instantaneous impedance value and the effective frequency points; and encapsulate the instantaneous impedance spectrum, the effective frequency points, the voltage wavelet coefficient, the current wavelet coefficient, the direction identifier, the cell identifier, and the pulse event association identifier to obtain the frequency domain impedance data packet.

[0010] Preferably, step S201, the step of determining the adaptive center frequency based on the average state of charge, the preset reference center frequency, and the preset center frequency adjustment coefficient, includes: Add 1 to the adjustment increment and calculate the square root of the sum to obtain the frequency adjustment term; multiply the preset reference center frequency by the frequency adjustment term to obtain the adaptive center frequency; The adjustment increment is obtained by multiplying the preset center frequency adjustment coefficient by the square of the deviation of the average state of charge from the preset normalized state of charge reference value; The preset normalized state of charge reference value is 0.5, and the preset center frequency adjustment coefficient is a preset dimensionless coefficient ranging from 0.1 to 0.5. When the deviation of the average state of charge from the preset normalized state of charge reference value increases, the adaptive center frequency is increased to improve the impedance characteristic resolution capability in the transition region between the high-frequency and mid-frequency bands.

[0011] Preferably, step S30, which involves using a fractional-order equivalent circuit model and a direction decoupling identification mechanism based on the frequency domain impedance data packet to perform the cell equivalent circuit parameter extraction task and output the equivalent circuit parameter data packet, specifically includes: Step S301: Establish a fractional-order equivalent circuit model of the battery cell. The fractional-order equivalent circuit model includes an ohmic internal resistance and a polarization branch composed of a charge transfer resistor and a constant-phase element connected in parallel. The ohmic internal resistance is connected in series with the polarization branch. The constant-phase element is characterized by the constant-phase element parameter and the fractional-order exponent. The fractional-order exponent is a dimensionless parameter that is greater than zero and less than or equal to one. Step S302: Divide the instantaneous impedance spectrum into a charging direction impedance spectrum and a discharging direction impedance spectrum according to the direction identifier; for each direction impedance spectrum, select high-frequency effective frequency points with frequencies greater than the preset high-frequency boundary from the effective frequency points; use the square of the amplitude of the current wavelet coefficient corresponding to each high-frequency effective frequency point as the weight to perform a weighted average of the real part of the instantaneous impedance at each high-frequency effective frequency point to obtain the ohmic internal resistance. Step S303: Under the condition of fixing the ohmic internal resistance, for each direction impedance spectrum, select the mid-frequency effective frequency point located between the preset low-frequency boundary and the preset high-frequency boundary from the effective frequency points; use a weighted nonlinear least squares fitting method to identify the charge transfer resistance, constant phase element parameters and fractional exponent; determine the polarization branch characteristic relaxation time based on the identified charge transfer resistance, constant phase element parameters and fractional exponent; and encapsulate the ohmic internal resistance, the charge transfer resistance, the constant phase element parameters, the fractional exponent, the polarization branch characteristic relaxation time, the direction identifier, the cell identifier and the pulse event association identifier to obtain the equivalent circuit parameter data package.

[0012] Preferably, step S40, which involves performing a cell consistency assessment task based on the equivalent circuit parameter data package using a robust statistical mechanism of median absolute deviation and outputting the assessment results for abnormal cells, specifically includes: Step S401: Group the equivalent circuit parameter data packets corresponding to each cell according to the direction identifier, and obtain the ohmic internal resistance sequence and charge transfer resistance sequence under the same direction identifier respectively; Step S402: For each of the ohmic internal resistance sequences and each of the charge transfer resistance sequences, calculate the median and the absolute deviation of the median of the corresponding parameter sequences, wherein the absolute deviation of the median satisfies:

[0013] Where MAD(X) represents the absolute deviation of the median, X represents the sequence of ohmic internal resistances or charge transfer resistances under the same directional identifier, median(·) represents the median operation, |·| represents the absolute value operation, and 1.4826 represents the preset robust scale normalization coefficient. Step S403: For any given cell, calculate the deviation of the cell's internal resistance from the median of the internal resistance under the same direction identifier, and the deviation of the cell's charge transfer resistance from the median of the charge transfer resistance under the same direction identifier; when the deviation of the internal resistance or the deviation of the charge transfer resistance is greater than three times the absolute deviation of the median of the corresponding parameter sequence, the cell is determined to be a faulty cell; when the deviation of the internal resistance or the deviation of the charge transfer resistance is greater than twice the absolute deviation of the median of the corresponding parameter sequence and less than or equal to three times the absolute deviation of the median of the corresponding parameter sequence, the cell is determined to be an aging concern cell; use the cell identifier of the faulty cell or the aging concern cell as abnormal cell location information, and encapsulate the abnormal category, the direction identifier, the pulse event association identifier, the internal resistance, the charge transfer resistance, the polarization branch characteristic relaxation time, the median internal resistance, and the median charge transfer resistance to obtain the abnormal cell evaluation result.

[0014] Preferably, step S50, which involves performing thermal runaway quantification based on the abnormal cell evaluation results using a pulse event correlation and integral heat generation mechanism, and outputting the thermal runaway quantification results, specifically includes: Step S501: Based on the direction identifier and the pulse event association identifier in the abnormal cell evaluation result, read the corresponding bus current waveform segment from the pulse event data packet to obtain the pulse current; Step S502: Compare the ohmic internal resistance of the abnormal cell with the median ohmic internal resistance. If the ohmic internal resistance of the abnormal cell is greater than the median ohmic internal resistance, the difference between the two is determined as the ohmic internal resistance increment; otherwise, the ohmic internal resistance increment is determined as zero. Compare the charge transfer resistance of the abnormal cell with the median charge transfer resistance. If the charge transfer resistance of the abnormal cell is greater than the median charge transfer resistance, the difference between the two is determined as the charge transfer resistance increment; otherwise, the charge transfer resistance increment is determined as zero. During the duration of the pulse current, perform integral calculation based on the square of the pulse current, the ohmic internal resistance increment, the charge transfer resistance increment, and the characteristic relaxation time of the polarization branch. Then, apply time-decay weighting to the polarization loss according to the characteristic relaxation time of the polarization branch to obtain the additional heat generated by ohmic loss and the additional heat generated by polarization loss, respectively. Add the additional heat generated by ohmic loss and the additional heat generated by polarization loss to obtain the additional heat generated by a single pulse. Step S503: Compare the single-pulse additional heat generation with a preset heat generation warning threshold; when the single-pulse additional heat generation is greater than the preset heat generation warning threshold, generate thermal management control information, the thermal management control information including at least one of active heat dissipation control command, power derating control command and abnormal warning information; associate and encapsulate the single-pulse additional heat generation and the thermal management control information to obtain the thermal loss quantification result.

[0015] This invention also provides an intelligent robot battery pack cell consistency evaluation system, comprising: The pulse event data construction module is used to acquire battery pack operating data, and based on the battery pack operating data, it uses a pulse event triggering and segment association caching mechanism to execute an effective pulse event capture task and output a pulse event data packet. The instantaneous impedance spectrum construction module is used to perform the instantaneous impedance spectrum construction task based on the pulse event data packet using the Morlet wavelet transform mechanism with adaptive center frequency of charge state, and output frequency domain impedance data packet; The equivalent circuit parameter identification module is used to perform the cell equivalent circuit parameter extraction task based on the frequency domain impedance data packet using a fractional-order equivalent circuit model and a direction decoupling identification mechanism, and output the equivalent circuit parameter data packet. The consistency evaluation module is used to perform a cell consistency evaluation task based on the equivalent circuit parameter data package using a robust statistical mechanism of median absolute deviation, and output the evaluation results of abnormal cells. The thermal loss quantification and early warning module is used to perform thermal loss quantification tasks based on the abnormal cell evaluation results using a pulse event correlation and integral heat generation mechanism, and output the thermal loss quantification results.

[0016] The present invention also provides a smart robot battery pack cell consistency evaluation device, the smart robot battery pack cell consistency evaluation device comprising: a memory, a processor, and a smart robot battery pack cell consistency evaluation program stored in the memory and executable on the processor, wherein the smart robot battery pack cell consistency evaluation program implements the above method when executed by the processor.

[0017] The present invention also provides a computer program product, the computer program product including an intelligent robot battery pack cell consistency evaluation program, which implements the above method when executed by a processor.

[0018] The beneficial effects of this invention are as follows: First, by using a pulse event triggering and segment association caching mechanism to output pulse event data packets, the cell terminal voltage, bus current, average state of charge, cell identifier, and pulse event association identifier are bound under the same event dimension, reducing the impact of false triggering, missed triggering, and cross-event waveform mismatch on subsequent impedance analysis under random operating conditions.

[0019] Second, by using the Morlet wavelet transform mechanism with adaptive center frequency of charge state to output frequency domain impedance data packets, the voltage and current responses in short-time random pulses are mapped to effective frequency points, enabling the acquisition of an instantaneous impedance spectrum that matches the current charge state without introducing external frequency sweep excitation.

[0020] Third, by using a fractional-order equivalent circuit model and a direction decoupling identification mechanism to output equivalent circuit parameter data packets, the ohmic internal resistance, charge transfer resistance, constant phase element parameters, fractional-order exponent, and polarization branch characteristic relaxation time are identified separately according to the charging or discharging direction, thereby reducing parameter offset caused by the mixing of asymmetric charging and discharging responses.

[0021] Fourth, the abnormal cell evaluation results are output through the robust statistical mechanism of median absolute deviation, so that faulty cells and aging cells of concern can be identified according to the parameter distribution within the group under the same directional identifier, avoiding the masking of other abnormal cells by skewing the mean of individual extreme cells.

[0022] Fifth, by using pulse event correlation and integral heat generation mechanism to output thermal loss quantification results, the incremental ohmic internal resistance and charge transfer resistance of abnormal cells are further converted into additional heat generation by single pulses, and thermal management control information is generated when the preset heat generation warning threshold is exceeded, thereby linking the consistency assessment results with thermal safety control. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the first embodiment of the intelligent robot battery pack cell consistency evaluation method of the present invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0026] Example 1: As Figure 1The diagram shown is a flowchart of the first embodiment of the intelligent robot battery pack cell consistency evaluation method of the present invention, which presents the first embodiment of the intelligent robot battery pack cell consistency evaluation method of the present invention.

[0027] In the first embodiment, the method for evaluating the consistency of battery cells in a smart robot battery pack includes: Step S10: Obtain battery pack operating data, and based on the battery pack operating data, use a pulse event triggering and segment association caching mechanism to execute an effective pulse event capture task and output a pulse event data packet; Battery pack operation data refers to multi-source time-series data synchronously collected by the battery management system during the actual operation of the intelligent robot's battery pack, including the terminal voltage of each cell, the state of charge of each cell, and the bus current of the battery pack. The pulse event triggering and segment association caching mechanism is a processing mechanism that identifies effective current pulses caused by operating conditions such as regenerative braking, acceleration during start-up, emergency stop, or hill climbing in the continuously sampled data stream, and encapsulates the voltage, current, state of charge, cell identifier, and event identifier associated with the pulse into a single data packet. This mechanism focuses not on a single sampling point, but on a traceable pulse event and its preceding and following response segments.

[0028] During execution, the system synchronously collects the terminal voltage of each cell, the state of charge of each cell, and the bus current at a preset sampling frequency. It subtracts the bus current at the current sampling moment from the bus current at the previous sampling moment, and then combines this with the sampling time interval to obtain the absolute value of the bus current change rate. Within a preset time window before the current sampling moment, the system calculates the root mean square value of this absolute value of change and multiplies it by a preset dynamic threshold coefficient to obtain a dynamic trigger judgment value. Only when the current absolute value of the change rate is higher than this judgment value, and the sign of the bus current remains unchanged within a preset duration, is a valid pulse event confirmed. Subsequently, the system extracts the terminal voltage of each cell and the bus current for a preset duration before and after the trigger moment, calculates the average state of charge within the window before triggering, and encapsulates the direction identifier, cell identifier, and pulse event association identifier together into a pulse event data packet.

[0029] Through the above processing, the subsequent impedance spectrum construction yields a data source with well-defined boundaries, directions, and charge state context. The dynamic trigger determination value changes with the robot's operating noise level, reducing false triggers caused by road vibrations, motor commutation interference, or sampling spikes. Current sign persistence constraints eliminate short-term back-and-forth jitter, clearly distinguishing pulse events in the charging and discharging directions. Pulse event association identifiers bind voltage segments, current segments, and cell numbers within the same event together, preventing cross-event mismatches when the S20 reads voltage and current waveforms.

[0030] Traditional DC internal resistance methods typically require the current to enter a relatively stable step state before calculating voltage changes, making it difficult to utilize the random pulses with durations of only tens to hundreds of milliseconds during intelligent robot operation. If triggered solely by a fixed current threshold, motor control noise is easily mistaken for valid pulses, or real braking and recovery pulses with small amplitudes but sufficient durations are missed. Step S10 uses the rate of change, dynamic threshold, directional stability, and fragment buffering to constrain pulse events, allowing natural operating pulses in random conditions to be converted into data packets usable for impedance analysis.

[0031] Taking a delivery robot performing a last-mile delivery task as an example, the battery management system monitors 10 battery cells at a sampling rate of 2.5 kHz. When the robot transitions from constant speed to regenerative braking, the bus current rapidly changes from a discharge direction of 12 A to a charging direction of -32 A. The rate of current change significantly exceeds the dynamic trigger determination value calculated from the previous operating noise, and the current sign remains in the charging direction for the following 10 ms. The system writes the voltage and current waveforms from 50 ms before and 150 ms after the trigger moment into a buffer, calculates the average state of charge of each cell in the 5 seconds before the trigger to be approximately 68%, and records the direction identifier as the charging direction, ultimately forming a pulse event data packet containing the identifiers of 10 battery cells.

[0032] Step S20: Based on the pulse event data packet, the instantaneous impedance spectrum construction task is performed using the Morlet wavelet transform mechanism with adaptive center frequency of charge state, and the frequency domain impedance data packet is output. Frequency domain impedance data packets refer to the impedance spectrum data set formed by time-frequency analysis of voltage waveform segments and bus current waveform segments from pulse event data packets. It includes at least the instantaneous impedance spectrum, effective frequency points, voltage wavelet coefficients, current wavelet coefficients, direction identifiers, cell identifiers, and pulse event association identifiers. The state-of-charge adaptive center frequency Morlet wavelet transform mechanism adjusts the center frequency using the average state of charge before the wavelet transform, matching the wavelet time-frequency window with the impedance response characteristics of the cell under different state of charge.

[0033] The system first detrends the voltage and bus current waveform segments in the pulse event data packets to reduce the impact of sampling baseline drift and slow load changes on the time-frequency coefficients. Then, the system determines an adaptive center frequency based on the average state of charge, a preset reference center frequency, and a preset center frequency adjustment coefficient. This center frequency is then used to perform Morlet wavelet transform on the detrendized voltage and current segments. The transform results are split into voltage and current wavelet coefficients and mapped to corresponding frequency points. Only frequency points that simultaneously satisfy the spectral energy ratio, signal-to-noise ratio, and the effective frequency range matching the average state of charge are used to calculate the instantaneous impedance value and construct the instantaneous impedance spectrum.

[0034] After the above processing, the time-domain impulse response obtained by S10 is converted into frequency-domain impedance data that can be used for subsequent parameter identification. Detrending processing prevents slow changes in the voltage plateau from being mistaken for low-frequency impedance characteristics; adaptive center frequency allows for more suitable frequency resolution of interface responses that change more rapidly under high or low charge states; effective frequency point screening removes frequency points with insufficient current energy, excessive noise, or unsuitability for the current charge state, preventing S30 from being influenced by low-quality frequency points when fitting equivalent circuit parameters.

[0035] Traditional offline electrochemical impedance spectroscopy relies on external sinusoidal frequency sweep excitation, which typically cannot be embedded in the continuous operation of a robot. Directly performing ordinary spectral transformation on short pulses is prone to spectral leakage due to pulse instability and short duration. Step S20 utilizes the time-frequency localization capability of Morlet wavelets to observe the high-frequency abrupt response and mid-frequency relaxation response that coexist in random running pulses separately, and adjusts the observation window using the state of charge, so that natural working pulses can serve as the data source for impedance spectrum construction.

[0036] Using the aforementioned regenerative braking event, the system reads pulse event data packets with the direction identifier indicating the charging direction and an average state of charge of approximately 68%. After detrending the voltage and current segments, Morlet wavelet transform generates multiple candidate frequency points in the range of approximately 0.5 Hz to 800 Hz. Among these, low-energy points with excessively weak current components, points with high noise levels, and points not within the effective range of the current state of charge are filtered out. The remaining frequency points reflect ohmic transfer characteristics in the high-frequency portion and charge transfer and polarization relaxation characteristics in the mid-frequency portion. The system encapsulates the instantaneous impedance values, corresponding wavelet coefficients, direction identifiers, and event identifiers of these frequency points into frequency domain impedance data packets.

[0037] Step S30: Based on the frequency domain impedance data packet, the fractional-order equivalent circuit model and the direction decoupling identification mechanism are used to perform the cell equivalent circuit parameter extraction task, and the equivalent circuit parameter data packet is output. The fractional-order equivalent circuit model refers to an equivalent model that uses an ohmic internal resistance series polarization branch to describe the cell's pulse response. The polarization branch is composed of a charge transfer resistor and a constant-phase element connected in parallel. The constant-phase element is characterized by its parameters and a fractional-order exponent. The directional decoupling identification mechanism refers to processing the charging directional impedance spectrum and the discharging directional impedance spectrum separately based on the directional identifier in the frequency domain impedance data packet, ensuring that differences in interface responses along the charging and discharging paths are not mixed into the same set of parameters.

[0038] After establishing a fractional-order equivalent circuit model, the system first divides the instantaneous impedance spectrum according to the direction identifier. For each direction impedance spectrum, the system selects high-frequency effective frequencies with frequencies higher than a preset high-frequency boundary from the effective frequency points, and uses the square of the amplitude of the corresponding current wavelet coefficient as a weight to perform a weighted average of the real parts of the instantaneous impedance at these high-frequency points, thereby obtaining the ohmic internal resistance. Subsequently, under the condition of fixed ohmic internal resistance, the system selects the mid-frequency effective frequency points located between the preset low-frequency boundary and the preset high-frequency boundary, and uses a weighted nonlinear least squares fitting method to identify the charge transfer resistance, constant-phase element parameters, and fractional-order exponent, and determines the characteristic relaxation time of the polarization branch accordingly. The above parameters, along with the direction identifier, cell identifier, and pulse event association identifier, are encapsulated into an equivalent circuit parameter data package.

[0039] This step further compresses the impedance spectrum output by S20 into a set of parameters with electrochemical meaning. The ohmic internal resistance mainly comes from the high-frequency response and is less affected by the polarization establishment process; the charge transfer resistance, constant phase element parameters, and fractional exponent mainly come from the mid-frequency response and can reflect the degree of interfacial reaction and diffusion dispersion. By fixing the ohmic internal resistance first and then fitting the polarization branch, the number of parameters to be identified is reduced, and the fitting is easier to converge; the retention of direction identifiers and event identifiers allows S40 to statistically analyze similar parameters according to the charging direction or the discharging direction.

[0040] Traditional equivalent circuit parameter identification, by fitting charging and discharging pulses together, may average the interface response differences between the energy recovery and accelerated discharge phases of the battery cell. This can lead to residual polarization affecting the ohmic resistance or cause the charge transfer resistor to bear the error that should be explained by directional differences. Step S30 uses direction as the identification boundary and processes high-frequency ohmic information and mid-frequency polarization information in layers. This ensures that parameter extraction under random pulse conditions retains the true directional differences of the battery cell without causing overall fitting deviation due to a few low-quality frequency points.

[0041] In a set of discharge pulse data from 10 battery cells, the frequency domain impedance data packet of a certain cell retained multiple high-frequency effective points near 500 Hz and multiple mid-frequency effective points between 20 Hz and 120 Hz. The system first uses the real part of the impedance at the high-frequency points and combines it with the weighted current wavelet coefficients to determine the cell's ohmic internal resistance as approximately 0.52 mΩ. Then, under the condition of a fixed ohmic internal resistance, it uses the mid-frequency points to fit the charge transfer resistance, constant-phase element parameters, and fractional exponent, obtaining the characteristic relaxation time of the polarization branch in that direction as approximately 0.24 s. The parameter results, along with the cell identifier, discharge direction identifier, and same pulse event identifier, are written into the equivalent circuit parameter data packet.

[0042] Step S40: Based on the equivalent circuit parameter data package, a robust statistical mechanism with median absolute deviation is used to perform the cell consistency assessment task, and the assessment results of abnormal cells are output. The robust statistical mechanism based on absolute median deviation refers to a statistical processing method that, under the same directional identifier, uses the median of the ohmic internal resistance sequence and the charge transfer resistance sequence as the center of the group, and forms a robust scale based on the absolute deviation of each parameter from the median. Abnormal cells are then identified according to a preset multiple interval. The abnormal cell evaluation result includes not only a cell number, but also data such as the abnormality category, directional identifier, pulse event association identifier, abnormal cell parameters, group median, and polarization branch characteristic relaxation time for subsequent thermal runaway quantification.

[0043] In this embodiment, 1.4826 is a preset robust scaling normalization coefficient used when the median absolute deviation is used to approximate the standard deviation. It is not re-identified with individual battery pack samples. This coefficient is derived from the fact that the median absolute deviation under the standard normal distribution is approximately 0.6745 of the standard deviation, and its reciprocal is approximately 1.4826. The system fixes this coefficient for the ohmic internal resistance sequence and charge transfer resistance sequence under the same direction identifier, so that the normalized median absolute deviation is on the same scale as the traditional standard deviation, thereby enabling the two-fold and three-fold thresholds to correspond to the determination of aging concern cells and faulty cells in this scheme.

[0044] The system groups the equivalent circuit parameter data packets corresponding to each cell according to the direction identifier, forming ohmic resistance sequences and charge transfer resistance sequences in the same direction. For each parameter sequence, the system first calculates the absolute deviation of each sample relative to the median, then multiplies the median of this absolute deviation by 1.4826 to obtain the normalized median absolute deviation. Then, for any cell, the system calculates the deviation of the cell's ohmic resistance from the median ohmic resistance in the same direction, and the deviation of the cell's charge transfer resistance from the median charge transfer resistance in the same direction. When any deviation exceeds three times the corresponding median absolute deviation, the cell is identified as a faulty cell; when any deviation exceeds two times but does not exceed three times the corresponding median absolute deviation, the cell is identified as an aging-prone cell.

[0045] The resulting abnormal cell assessment provides a complete thermal calculation entry point for the S50. The abnormality location information clarifies which cell needs tracking, the abnormality category indicates whether it's a fault or an aging concern, the direction identifier and pulse event association marker point to the original pulse event, and the ohmic internal resistance, charge transfer resistance, polarization branch characteristic relaxation time, and their corresponding median provide the parameter basis for subsequent calculations of additional heat generation from ohmic and polarization losses. This step is not simply an alarm triggering process, but rather a process of organizing consistency deviations into a data structure that allows for further quantification of thermal risk.

[0046] If the traditional rule of adding or subtracting three times the standard deviation from the mean is used, a severely degraded cell will simultaneously increase both the mean and the standard deviation, causing other cells that have deviated from the normal distribution but to a lesser extent to be masked by the threshold. The median and median absolute deviation are less sensitive to outliers, maintaining stability in the center and scale of the battery pack even when abnormal samples already exist. For series-connected battery packs, this robust statistical approach is more suitable for scenarios in online monitoring where the sample size is limited, the proportion of abnormal cells is unknown, and their operating directions differ significantly.

[0047] Taking the parameter evaluation of 10 cells in the discharge direction as an example, the system obtained a median ohmic internal resistance of 0.51 mΩ and an absolute deviation of 0.012 mΩ. Cell No. 6 has an ohmic internal resistance of 0.56 mΩ, which deviates from the median absolute deviation by more than three times, and is therefore identified as a faulty cell. The deviation of the charge transfer resistance of cell No. 2 is between two and three times the median absolute deviation, and is therefore marked as a cell of concern for aging. The system encapsulates the cell identification, fault category, discharge direction identification, pulse event association identification, ohmic internal resistance, charge transfer resistance, polarization branch characteristic relaxation time, and corresponding median of cell No. 6 into an abnormal cell evaluation result.

[0048] Step S50: Based on the abnormal cell evaluation results, the thermal loss quantification task is performed using a pulse event correlation and integral heat generation mechanism, and the thermal loss quantification results are output.

[0049] The pulse event correlation and integral heat generation mechanism refers to a processing mechanism that, based on the direction identifier and pulse event correlation identifier in the abnormal cell evaluation results, retrieves the corresponding bus current waveform segment from the pulse event data packet generated in S10, and combines the ohmic internal resistance increment and charge transfer resistance increment of the abnormal cell relative to the median within the group to integrally quantify the additional heat generation during the pulse duration. The thermal loss quantification results include the additional heat generation of a single pulse and the thermal management control information generated when the preset heat generation warning threshold is exceeded.

[0050] The system first reads the direction identifier and pulse event association identifier from the abnormal cell evaluation results, and then finds the corresponding bus current waveform segment in the pulse event data packet to obtain the pulse current. Next, the system compares the ohmic internal resistance of the abnormal cell with the median ohmic internal resistance. Only when the ohmic internal resistance of the abnormal cell is higher than the median is the excess taken as the ohmic internal resistance increment; the same positive increment rule is applied to the charge transfer resistance. Then, during the pulse current duration, the system performs integral calculations based on the square of the pulse current, the ohmic internal resistance increment, the charge transfer resistance increment, and the characteristic relaxation time of the polarization branch. The polarization loss is then time-decayed and weighted according to the characteristic relaxation time of the polarization branch to obtain the additional heat generated by ohmic loss and the additional heat generated by polarization loss, respectively. These two are added together to form the additional heat generated by a single pulse.

[0051] This step converts the abnormal electrical parameters of the S40 into physical quantities usable for thermal management. Only positive increments above the median are counted to avoid misidentifying low-resistance cells as additional heat sources. Using the current waveform from the original pulse event ensures that the generated heat corresponds to the actual current amplitude, duration, and event direction. Introducing the characteristic relaxation time of the polarization branch prevents polarization loss from being simply equated to constant resistance heating, instead reflecting the different thermal contributions of polarization establishment at the beginning and end of the pulse. The final output thermal loss quantification result includes both numerical risk and the basis for subsequent actions such as active cooling control commands, power derating control commands, or abnormal warning information.

[0052] Existing consistency assessment schemes often stop at indicating that the internal resistance of a particular cell is too high. The thermal management system can only react based on the average temperature or maximum temperature difference of the entire pack, which is prone to control lag. Step S50 re-correlates the abnormal parameters with specific pulse events, which can explain how much more heat the abnormality will generate in a single braking recovery or start-up discharge pulse, allowing the diagnostic results to be directly incorporated into the thermal safety control chain. For robot battery packs that frequently start and stop, this event-based quantification method is more effective than simple resistance alarms in reflecting the localized thermal risks caused by short-term high-current pulses.

[0053] Continuing with the example of the aforementioned abnormal cell No. 6, during a 0.4 s-long initial acceleration discharge pulse with a peak value of approximately 40 A, the system reads the bus current waveform from the corresponding pulse event data packet. Cell No. 6 has an ohmic internal resistance approximately 0.05 mΩ higher than the group median, a charge transfer resistance approximately 0.20 mΩ higher than the group median, and a polarization branch characteristic relaxation time of approximately 0.24 s. The system quantifies the ohmic loss and polarization loss according to the pulse current change process, obtaining an additional heat generation of approximately 0.50 J for this single pulse. If the preset heat generation warning threshold is 0.40 J, the thermal loss quantification result simultaneously includes a power derating control command and an anomaly warning information to limit subsequent high-rate discharge pulses and prompt maintenance personnel to re-inspect the cell.

[0054] Example 2: Furthermore, the present invention provides an intelligent robot battery pack cell consistency evaluation system, which employs an intelligent robot battery pack cell consistency evaluation method described in the above embodiments, and can solve the technical problem of intelligent robot battery pack cell consistency evaluation. The beneficial effects of the intelligent robot battery pack cell consistency evaluation system provided by the present invention are the same as those of the intelligent robot battery pack cell consistency evaluation method described in the above embodiments, and other technical features of the intelligent robot battery pack cell consistency evaluation system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0055] Example 3: This invention provides an intelligent robot battery pack cell consistency evaluation device. The device includes at least one processor and a memory communicatively connected to the processor. The memory stores instructions executable by the processor, which are then executed to enable the processor to perform the intelligent robot battery pack cell consistency evaluation method described in Example 1. The intelligent robot battery pack cell consistency evaluation device in this invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This intelligent robot battery pack cell consistency evaluation device is merely an example and should not limit the functionality or scope of the invention. The device may also include a processing unit (e.g., a central processing unit, a graphics processing unit), which can perform various appropriate actions and processes based on a program stored in a read-only memory or a program loaded from a storage device into a random access memory. The random access memory also stores various programs and data required for the operation of an intelligent robot battery pack cell consistency evaluation device. The processing unit, read-only memory, and random access memory are interconnected via a bus. The I / O interface is also connected to the bus. Typically, the following systems can be connected to the I / O interface: input devices including touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices including liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices including magnetic tapes, hard disks, etc.; and communication devices. The communication device allows the intelligent robot battery pack cell consistency evaluation device to communicate wirelessly or wiredly with other devices to exchange data. While an intelligent robot battery pack cell consistency evaluation device with various systems has been described, it should be understood that implementing or possessing all the described systems is not required. Alternatively, more or fewer systems can be implemented.

[0056] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for evaluating the consistency of battery cells in an intelligent robot battery pack. The computer program product provided by this invention can solve the technical problem of evaluating the consistency of battery cells in an intelligent robot battery pack. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the method for evaluating the consistency of battery cells in an intelligent robot battery pack provided in the above embodiments, and will not be repeated here.

[0057] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a read-only memory. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0058] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0059] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the present invention and its equivalents, the present invention also intends to include these modifications and variations.

Claims

1. A method for evaluating the consistency of battery cells in an intelligent robot battery pack, characterized in that, The methods include: Step S10: Obtain battery pack operating data, and based on the battery pack operating data, use a pulse event triggering and segment association caching mechanism to execute an effective pulse event capture task and output a pulse event data packet; Step S20: Based on the pulse event data packet, the instantaneous impedance spectrum construction task is performed using the Morlet wavelet transform mechanism with adaptive center frequency of charge state, and the frequency domain impedance data packet is output. Step S30: Based on the frequency domain impedance data packet, a fractional-order equivalent circuit model and a direction decoupling identification mechanism are used to perform the cell equivalent circuit parameter extraction task, and output the equivalent circuit parameter data packet; Step S30 specifically includes: Step S301: Establish a fractional-order equivalent circuit model of the battery cell. The fractional-order equivalent circuit model includes an ohmic internal resistance and a polarization branch composed of a charge transfer resistor and a constant-phase element connected in parallel. The ohmic internal resistance is connected in series with the polarization branch. The constant-phase element is characterized by the constant-phase element parameter and the fractional-order exponent. The fractional-order exponent is a dimensionless parameter that is greater than zero and less than or equal to one. Step S302: Divide the instantaneous impedance spectrum into a charging direction impedance spectrum and a discharging direction impedance spectrum according to the direction identifier; for each direction impedance spectrum, select high-frequency effective frequency points with frequencies greater than the preset high-frequency boundary from the effective frequency points; use the square of the amplitude of the current wavelet coefficient corresponding to each of the high-frequency effective frequency points as the weight to perform a weighted average of the real part of the instantaneous impedance at each of the high-frequency effective frequency points to obtain the ohmic internal resistance. Step S303: Under the condition of fixing the ohmic internal resistance, for each direction impedance spectrum, select the mid-frequency effective frequency point located between the preset low-frequency boundary and the preset high-frequency boundary from the effective frequency points; use a weighted nonlinear least squares fitting method to identify the charge transfer resistance, constant phase element parameters, and fractional exponent; determine the polarization branch characteristic relaxation time based on the identified charge transfer resistance, constant phase element parameters, and fractional exponent; and encapsulate the ohmic internal resistance, the charge transfer resistance, the constant phase element parameters, the fractional exponent, the polarization branch characteristic relaxation time, the direction identifier, the cell identifier, and the pulse event association identifier to obtain the equivalent circuit parameter data package; Step S40: Based on the equivalent circuit parameter data package, a robust statistical mechanism using the median absolute deviation is employed to perform a cell consistency assessment task, and the assessment results for abnormal cells are output; Step S40 specifically includes: Step S401: Group the equivalent circuit parameter data packets corresponding to each cell according to the direction identifier, and obtain the ohmic internal resistance sequence and charge transfer resistance sequence under the same direction identifier respectively; Step S402: For each of the ohmic internal resistance sequences and each of the charge transfer resistance sequences, calculate the median and the absolute deviation of the median of the corresponding parameter sequences, wherein the absolute deviation of the median satisfies: Where MAD(X) represents the absolute deviation of the median, X represents the sequence of ohmic internal resistances or charge transfer resistances under the same directional identifier, median(·) represents the median operation, |·| represents the absolute value operation, and 1.4826 represents the preset robust scale normalization coefficient. Step S403: For any given cell, calculate the deviation of the cell's internal resistance from the median internal resistance under the same direction identifier, and the deviation of the cell's charge transfer resistance from the median charge transfer resistance under the same direction identifier; when the deviation of the internal resistance or the deviation of the charge transfer resistance is greater than three times the absolute deviation of the median of the corresponding parameter sequence, the cell is determined to be a faulty cell; when the deviation of the internal resistance or the deviation of the charge transfer resistance is greater than twice the absolute deviation of the median of the corresponding parameter sequence and less than or equal to three times the absolute deviation of the median of the corresponding parameter sequence, the cell is determined to be an aging-concealed cell; the cell identifier of the faulty cell or the aging-concealed cell is used as abnormal cell location information, and the abnormal category, the direction identifier, the pulse event association identifier, the internal resistance, the charge transfer resistance, the polarization branch characteristic relaxation time, the median internal resistance, and the median charge transfer resistance are associated and encapsulated to obtain the abnormal cell evaluation result; Step S50: Based on the abnormal cell evaluation results, the thermal loss quantification task is performed using a pulse event correlation and integral heat generation mechanism, and the thermal loss quantification results are output.

2. The method for evaluating the consistency of battery cells in an intelligent robot battery pack as described in claim 1, characterized in that, Step S10, which involves acquiring battery pack operating data, executing a valid pulse event capture task based on the battery pack operating data using a pulse event triggering and segment association caching mechanism, and outputting a pulse event data packet, specifically includes: Step S101: Synchronously collect the terminal voltage of each cell, the state of charge of each cell, and the bus current of the battery pack at a preset sampling frequency to form the battery pack operating data; subtract the bus current at the current sampling moment from the bus current at the adjacent previous sampling moment to obtain the bus current difference; calculate the sampling time interval between the current sampling moment and the adjacent previous sampling moment; calculate the absolute value of the bus current change rate based on the bus current difference and the sampling time interval; Step S102: Calculate the root mean square value for the absolute value of the bus current change rate within a preset time window before the current sampling time; multiply the root mean square value by a preset dynamic threshold coefficient to obtain a dynamic trigger determination value; when the absolute value of the bus current change rate is greater than the dynamic trigger determination value, and the sign of the bus current remains unchanged within a preset duration, determine that a valid pulse event has occurred, and determine the direction identifier based on the sign of the bus current relative to the preset current reference direction; Step S103: Write the terminal voltages of each cell within a preset time period before and after the triggering time of the valid pulse event into the associated buffer to obtain a voltage waveform segment; write the bus current within the corresponding time period into the associated buffer to obtain a bus current waveform segment; calculate the average state of charge of each cell within a preset time window before the triggering time to obtain the average state of charge; assign a cell identifier to each cell and generate a pulse event associated identifier for the valid pulse event; encapsulate the voltage waveform segment, the bus current waveform segment, the direction identifier, the average state of charge, the cell identifier, and the pulse event associated identifier to obtain the pulse event data packet.

3. The method for evaluating the consistency of battery cells in an intelligent robot battery pack as described in claim 2, characterized in that, Step S20, which involves using the Morlet wavelet transform mechanism with adaptive center frequency of charge state to perform instantaneous impedance spectrum construction based on the pulse event data packet and outputting frequency domain impedance data packet, specifically includes: Step S201: Detrend the voltage waveform segment and bus current waveform segment in the pulse event data packet; determine the adaptive center frequency based on the average state of charge, the preset reference center frequency, and the preset center frequency adjustment coefficient; perform Morlet wavelet transform on the detrendened voltage waveform segment and bus current waveform segment based on the adaptive center frequency to obtain the Morlet wavelet transform result; Step S202: Based on the Morlet wavelet transform results, obtain the voltage wavelet coefficients and the current wavelet coefficients, and map the voltage wavelet coefficients and the current wavelet coefficients to the corresponding frequency points; Step S203: Select effective frequency points from the frequency points that simultaneously meet the preset spectral energy ratio determination condition, the preset signal-to-noise ratio determination condition, and the preset effective frequency range determination condition that matches the average state of charge; calculate the instantaneous impedance value based on the ratio of the voltage wavelet coefficient to the current wavelet coefficient at each effective frequency point; construct an instantaneous impedance spectrum based on the instantaneous impedance value and the effective frequency points; and encapsulate the instantaneous impedance spectrum, the effective frequency points, the voltage wavelet coefficient, the current wavelet coefficient, the direction identifier, the cell identifier, and the pulse event association identifier to obtain the frequency domain impedance data packet.

4. The method for evaluating the consistency of battery cells in an intelligent robot battery pack as described in claim 3, characterized in that, Step S201, the step of determining the adaptive center frequency based on the average state of charge, the preset reference center frequency, and the preset center frequency adjustment coefficient, includes: Add 1 to the adjustment increment and calculate the square root of the sum to obtain the frequency adjustment term; multiply the preset reference center frequency by the frequency adjustment term to obtain the adaptive center frequency; The adjustment increment is obtained by multiplying the preset center frequency adjustment coefficient by the square of the deviation of the average state of charge from the preset normalized state of charge reference value; The preset normalized state of charge reference value is 0.5, and the preset center frequency adjustment coefficient is a preset dimensionless coefficient ranging from 0.1 to 0.

5. When the deviation of the average state of charge from the preset normalized state of charge reference value increases, the adaptive center frequency is increased to improve the impedance characteristic resolution capability in the transition region between the high-frequency and mid-frequency bands.

5. The method for evaluating the consistency of battery cells in an intelligent robot battery pack as described in claim 1, characterized in that, Step S50, which involves performing thermal runaway quantification based on the abnormal cell evaluation results using a pulse event correlation and integral heat generation mechanism, and outputting the thermal runaway quantification results, specifically includes: Step S501: Based on the direction identifier and the pulse event association identifier in the abnormal cell evaluation result, read the corresponding bus current waveform segment from the pulse event data packet to obtain the pulse current; Step S502: Compare the ohmic internal resistance of the abnormal cell with the median ohmic internal resistance. If the ohmic internal resistance of the abnormal cell is greater than the median ohmic internal resistance, the difference between the two is determined as the ohmic internal resistance increment; otherwise, the ohmic internal resistance increment is determined as zero. Compare the charge transfer resistance of the abnormal cell with the median charge transfer resistance. If the charge transfer resistance of the abnormal cell is greater than the median charge transfer resistance, the difference between the two is determined as the charge transfer resistance increment; otherwise, the charge transfer resistance increment is determined as zero. During the duration of the pulse current, perform integral calculation based on the square of the pulse current, the ohmic internal resistance increment, the charge transfer resistance increment, and the characteristic relaxation time of the polarization branch. Then, apply time-decay weighting to the polarization loss according to the characteristic relaxation time of the polarization branch to obtain the additional heat generated by ohmic loss and the additional heat generated by polarization loss, respectively. Add the additional heat generated by ohmic loss and the additional heat generated by polarization loss to obtain the additional heat generated by a single pulse. Step S503: Compare the single-pulse additional heat generation with a preset heat generation warning threshold; when the single-pulse additional heat generation is greater than the preset heat generation warning threshold, generate thermal management control information, the thermal management control information including at least one of active heat dissipation control command, power derating control command and abnormal warning information; associate and encapsulate the single-pulse additional heat generation and the thermal management control information to obtain the thermal loss quantification result.

6. A smart robot battery pack cell consistency evaluation system, applied to the smart robot battery pack cell consistency evaluation method according to any one of claims 1 to 5, characterized in that, The intelligent robot battery pack cell consistency evaluation system includes: The pulse event data construction module is used to acquire battery pack operating data, and based on the battery pack operating data, it uses a pulse event triggering and segment association caching mechanism to execute an effective pulse event capture task and output a pulse event data packet. The instantaneous impedance spectrum construction module is used to perform the instantaneous impedance spectrum construction task based on the pulse event data packet using the Morlet wavelet transform mechanism with adaptive center frequency of charge state, and output frequency domain impedance data packet; The equivalent circuit parameter identification module is used to perform the cell equivalent circuit parameter extraction task based on the frequency domain impedance data packet using a fractional-order equivalent circuit model and a direction decoupling identification mechanism, and output the equivalent circuit parameter data packet. The consistency evaluation module is used to perform a cell consistency evaluation task based on the equivalent circuit parameter data package using a robust statistical mechanism of median absolute deviation, and output the evaluation results of abnormal cells. The thermal loss quantification and early warning module is used to perform thermal loss quantification tasks based on the abnormal cell evaluation results using a pulse event correlation and integral heat generation mechanism, and output the thermal loss quantification results.

7. A smart robot battery pack cell consistency evaluation device, characterized in that, The intelligent robot battery pack cell consistency evaluation device includes: a memory, a processor, and an intelligent robot battery pack cell consistency evaluation program stored in the memory and executable on the processor. When the intelligent robot battery pack cell consistency evaluation program is executed by the processor, it implements an intelligent robot battery pack cell consistency evaluation method according to any one of claims 1 to 5.

8. A computer program product, characterized in that, The computer program product includes an intelligent robot battery pack cell consistency evaluation program, which, when executed by a processor, implements an intelligent robot battery pack cell consistency evaluation method according to any one of claims 1 to 5.

Images