Singlechip-based wind-solar complementary power supply battery pack and power supply data tracing system

Abstract

The invention discloses a wind-solar complementary power supply battery pack and power supply data tracing system based on a single-chip microcomputer, and relates to the technical field of new energy power supply and data management, and the system comprises a single-chip microcomputer, and a data acquisition unit, a tracing storage unit, a feature tracing unit, a strategy optimization unit and an impedance diagnosis unit which are connected with the single-chip microcomputer. The traceability storage unit constructs an annular linked list structure through a ferroelectric memory, realizes erase-write balance through Hash operation, calculates a chained check value node by node to prevent tampering, and cooperates with breakpoint resume of a backup ferroelectric memory; the feature tracing unit extracts multi-scale fluctuation features and wind-solar coupling features through wavelet transform and compresses the features into feature fingerprints; the strategy optimization unit calculates the capacity attenuation rate of the battery pack and adjusts charging and discharging parameters, and cycle closed-loop verification is performed through verification; and the impedance diagnosis unit injects an excitation signal through the DC-DC converter to obtain an impedance spectrum and discriminate an attenuation mechanism. A closed loop is formed by taking the tracing storage unit as a hub, and storage life, data tamper-proofing and strategy adaptive optimization are considered.

Description

Technical Field

[0001] This invention relates to the field of new energy power supply and data management technology, specifically to a wind-solar hybrid power supply battery pack and power supply data traceability system based on a microcontroller. Background Technology

[0002] Wind-solar hybrid power supply systems are widely deployed in remote communication base stations, meteorological monitoring stations, and marine buoys—scenarios without mains power coverage. They maintain continuous power supply by using wind turbines and photovoltaic modules to complement each other and charge battery banks. During long-term operation, complete traceability of power supply data plays a crucial supporting role in fault diagnosis, performance evaluation, and operation and maintenance decisions.

[0003] Existing wind-solar hybrid power supply systems face the following technical bottlenecks in data traceability and energy management: Firstly, the write / erase cycle life of the storage medium is insufficient. Existing solutions mostly use EEPROM or Flash memory to record operational data, with write / erase cycles of approximately [missing information - likely related to lifespan]. Subsequent In high-frequency sampling scenarios, adopting a sequential write strategy will cause certain storage blocks to reach their erase / write limits first, thus limiting the system's maintenance-free cycle.

[0004] Secondly, data traceability and energy management are disconnected. The existing system's data recording module only performs a one-way process of acquisition, storage, and uploading, and the accumulated historical operating data is not fed back to the charge and discharge control stage. The battery management system relies on factory-preset fixed parameters and cannot adaptively adjust the control strategy according to the actual degradation of the battery pack.

[0005] Third, there is a lack of effective characterization of the time-varying characteristics of wind and solar power. Directly storing raw sampled data results in inefficient use of storage resources and makes it difficult to quickly extract characteristic information of typical operating conditions or failure modes from a large amount of historical data.

[0006] Fourth, there are limited means of online identification of battery degradation mechanisms on embedded platforms. Traditional electrochemical impedance spectroscopy testing relies on dedicated workstations, which are difficult to integrate into resource-constrained microcontroller platforms, thus restricting the refined management of different degradation mechanisms.

[0007] Fifth, data continuity is difficult to guarantee when communication links are unstable. Wireless communication links at remote sites are frequently interrupted due to weather and terrain. Existing systems lack offline caching and breakpoint resumption mechanisms, and operational data during communication interruptions are at risk of being lost, resulting in time gaps in traceability data and affecting subsequent attenuation rate analysis based on continuous historical data.

[0008] Therefore, there is an urgent need for a wind-solar complementary power supply data tracing system that can ensure long lifespan of storage media, bidirectional coupling of traceability data and energy management, efficient characterization of wind and solar features, online mechanism identification on embedded platforms, and data continuity assurance under communication interruption conditions. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a microcontroller-based wind-solar hybrid power supply battery pack and power supply data traceability system. The core concept lies in using a traceability storage unit as a data hub to drive the dynamic optimization of the charging and discharging control strategy using historical power supply data, forming a bidirectional coupled closed loop of traceability, analysis, optimization, and re-traceability.

[0010] The present invention adopts the following technical solution: A microcontroller-based wind-solar hybrid power supply battery pack and power supply data traceability system includes a microcontroller, and a data acquisition unit, a traceability storage unit, a feature traceability unit, a strategy optimization unit and an impedance diagnosis unit respectively connected to the microcontroller.

[0011] The data acquisition unit is used to collect wind and solar power data and battery pack operating parameters as power supply data, and then send the power supply data to the traceability storage unit.

[0012] The traceability storage unit includes a ferroelectric memory, which has a circular linked list structure for storing data in time sequence. The traceability storage unit determines the target storage address for each write operation through hash operation to achieve write-erase balance, and calculates a chain check value for each data node to prevent tampering.

[0013] The feature tracing unit is used to read wind and solar power data from the tracing storage unit, extract multi-scale fluctuation features and wind-solar coupling features through wavelet transform, generate fixed-length feature fingerprints through hash compression, and write them into the tracing storage unit.

[0014] The strategy optimization unit is used to read power supply data from the traceability storage unit, calculate the battery pack capacity decay rate and generate correction values ​​for charge and discharge parameters accordingly to adjust the charge and discharge parameters. The charge and discharge parameters before and after correction, as well as the battery pack operating parameters for each time period before and after correction, are written into the traceability storage unit. After the preset verification period ends, if the improvement in the decay rate does not reach the preset improvement threshold, the charge and discharge parameters before correction will be reverted.

[0015] The impedance diagnostic unit is used to inject an excitation signal into the battery pack through a DC-DC converter to obtain the impedance spectrum, determine the attenuation mechanism based on the charge transfer impedance and diffusion impedance slope, and send the diagnostic results to the strategy optimization unit.

[0016] In the above technical solution, the traceability storage unit serves as the data hub of the entire system. All data reading and writing of the data acquisition unit, feature traceability unit, strategy optimization unit, and impedance diagnosis unit are completed through this unit, ensuring that the five functional units form an organic whole around the same traceability data flow.

[0017] Compared with the prior art, the present invention has the following beneficial effects: First, by replacing the traditional sequential write scheme of EEPROM with ferroelectric memory combined with a ring linked list structure and a hash-based write-erase balancing strategy, the write-erase lifespan of the storage medium is extended from... Sub-weight class upgraded to Sub-scale, meeting the technical requirements for storage reliability for long-term maintenance-free operation at remote sites.

[0018] Second, wavelet multi-scale decomposition and wavelet coherence analysis are used to extract the multi-scale fluctuation features and wind-solar coupling features of wind and solar power data and compress them into a fixed-length feature fingerprint, which reduces the storage resource consumption while retaining the core information required for fault mode matching.

[0019] Third, the battery pack capacity decay rate is deduced from the power supply data and the charging and discharging parameters are dynamically adjusted through the saturation correction function to form a closed-loop verification mechanism within the preset verification period. Compared with the open-loop control scheme with fixed parameters at the factory, this provides a technical means to adaptively correct the battery pack capacity based on the actual decay rate.

[0020] Fourth, by injecting an excitation signal into the battery pack through a DC-DC converter and extracting the impedance spectrum using digital lock-in amplification, the attenuation mechanism can be identified online on a microcontroller platform, which reduces system complexity and hardware cost compared to solutions that rely on dedicated electrochemical workstations.

[0021] Fifth, by backing up the ferroelectric memory to record breakpoint markers and retransmission status information, offline buffering is maintained during communication link interruption and breakpoint resume transmission is performed after communication is restored. The retransmitted power supply data is included in the traceability data chain after the integrity of the cyclic redundancy check value is verified. Compared with the scheme of directly discarding the running data when communication is interrupted, the time continuity of traceability data is guaranteed. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the architecture of a wind-solar hybrid power supply battery pack and power supply data traceability system based on a microcontroller, provided for an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0024] Reference manual attached Figure 1 This diagram illustrates a structural block diagram of a wind-solar hybrid power supply battery pack and power supply data traceability system based on a microcontroller, according to an embodiment of the present invention. This embodiment uses an STM32H743 microcontroller as the main controller, with a main frequency of 480MHz, and a built-in floating-point arithmetic unit and DSP instruction set, which are connected to the data acquisition unit, traceability storage unit, feature traceability unit, strategy optimization unit, and impedance diagnostic unit, respectively. It should be noted that those skilled in the art can select other microcontroller models based on hardware resource constraints; this embodiment of the present invention does not specifically limit this. In alternative solutions with limited hardware resources, the STM32F4 series can be used; in this case, the wavelet operation in the feature traceability unit needs to employ downsampling and fixed-point processing.

[0025] The data acquisition unit collects wind and solar power data and battery pack operating parameters through an analog-to-digital converter. The wind and solar power data includes wind power output. and photovoltaic power The battery pack operating parameters include battery voltage, battery current, state of charge, and temperature. The data acquisition unit uses wind and solar power data and battery pack operating parameters as power supply data and sends the power supply data to the traceability storage unit according to a preset sampling period. Optionally, the sampling period is 60 seconds. It is understood that those skilled in the art can adjust the value of the sampling period according to the application scenario, and this embodiment of the invention does not specifically limit this. In this embodiment of the invention, the data acquisition unit provides the original data source for all subsequent traceability analysis and control optimization. Its sampling accuracy and clock synchronization quality directly determine the effectiveness of downstream feature extraction and strategy optimization. Compared with the conventional BMS solution that only collects battery-side parameters, this unit simultaneously collects wind source-side power, providing a data foundation for subsequent wind-solar coupling feature analysis.

[0026] The traceability memory unit primarily uses ferroelectric memory as its storage medium. Ferroelectric memory achieves data storage based on the polarization reversal principle of ferroelectric thin films, with an erase / write lifetime of up to [missing information]. Second-order magnitude. Optionally, the ferroelectric memory is selected as the FM25V10 model, with a capacity of 128KB, and connected to the microcontroller via an SPI interface. It is understood that those skilled in the art can select other models of ferroelectric memory according to storage capacity requirements, and this embodiment of the invention does not specifically limit this. The ferroelectric memory has a circular linked list structure for storing data in time sequence. The memory is divided into... Each data node consists of an equal-capacity storage block, containing 8 bytes of timestamp, 16 bytes of wind and solar power data and battery pack operating parameters, 4 bytes of chained checksum, and 4 bytes of forward pointer, totaling 32 bytes. It should be noted that the extremely long write / erase life of ferroelectric memory refers to the storage medium itself being able to operate for over ten years without write / erase failure; the circular linked list uses an overwrite strategy, and the actual data retention time window is determined by both the storage capacity and the sampling period. Taking a 128KB capacity, 32 bytes per entry, and sampling once per minute as an example, the sliding window is approximately 2.8 days.

[0027] Each time new data is written, the trace storage unit determines the target storage address for each write through hash operation to achieve write-erase balance. Specifically, the target block number is calculated according to formula (1). The formula for calculating the target block number is shown in formula (1):

[0028] in, The target block number is specified, and its value range is defined as follows: non-negative integers, It is a lightweight hash function. The current write timestamp is expressed in Unix time and in seconds. It is a global write counter, a 32-bit unsigned integer, that automatically wraps back to zero after overflow. This is the bitwise XOR operator. The total number of storage blocks is a positive integer and , The modulo operation is performed. The trace storage unit performs an XOR operation on the current write timestamp and the global write counter, performs a hash operation on the result, and then takes the modulo of the total number of storage blocks to obtain the target block number to determine the target storage address. Belongs to the time domain. Belonging to the counting field, the XOR operation between the two followed by hash mapping ensures that the write address is in all... Approximately uniform distribution across the blocks. Optionally, The CRC-32 algorithm is used. It is understood that those skilled in the art can choose other hash functions based on computing resources and uniformity requirements; this embodiment of the invention does not specifically limit this choice.

[0029] Furthermore, when the target block When the cumulative number of erase / write cycles exceeds the preset equalization threshold, a linear probing method is used to sequentially check... , The process continues until a usable block that has not exceeded the threshold is found, at which point it jumps to the next usable block. The trace storage unit maintains a global wear count table to record the cumulative number of erases and writes for each block.

[0030] In this embodiment of the invention, formula (1) distributes the write operation to different blocks through hash mapping, which is different from the traditional sequential write scheme that writes sequentially from a fixed starting address. The latter causes the number of erase / write operations of the head block to be much higher than that of the tail block when the sampling period is constant, resulting in the head block reaching the erase / write limit first and the whole block becoming invalid. The hash mapping after XORing the timestamp and the counter in formula (1) introduces pseudo-randomness, which can still maintain the balance of wear between blocks in scenarios with non-uniform sampling periods.

[0031] The chain check value is calculated for each data node in the traceable storage unit to prevent tampering. The chain check value for each data node is calculated according to formula (2). The formula for calculating the chain check value is shown in formula (2):

[0032] in, For the first The chained checksum of each data node is a dimensionless 32-bit integer. For the first Data content fields of each data node For the first The chained checksum of each data node is the checksum output of the predecessor data node. The data concatenation operator joins the beginning and end of two byte streams together. For the same hash function as in formula (1), Indexing data nodes and The chained checksum of the initial data node. A preset seed value is used. The chain checksum is calculated by concatenating the data content of the current data node with the chain checksum of the previous data node and then performing a hash operation. The chain checksum of each new data node is determined by its own data and the chain checksum of its predecessor, forming a unidirectional dependency chain. When verifying data integrity, the chain checksum is recalculated sequentially starting from any data node and compared with the stored value. If they are inconsistent, the data node or its predecessor data has been changed.

[0033] Furthermore, to improve the fault tolerance of the check chain, an independent checkpoint is set at fixed intervals between data nodes to store the independent check value of that data node. When an inconsistency is detected in a segment of the check chain, the check chain can be rebuilt from the nearest checkpoint, preventing the failure of a single data node from causing all subsequent checks to fail.

[0034] In this embodiment of the invention, formula (2) makes the chained checksum of each data node dependent on the data content of all its predecessor data nodes. Any single-point tampering will cause the entire checksum chain from that point to mismatch, which is different from the method of calculating the checksum independently for each data node. The latter can only detect whether the data of a single data node has been modified, and cannot detect sequential tampering or insertion / deletion operations between data nodes. The chained structure additionally provides the ability to verify the sequential consistency between data nodes.

[0035] The traceability storage unit also includes a backup ferroelectric memory, which is independently configured from the main ferroelectric memory. The backup ferroelectric memory stores breakpoint markers and retransmission status information and does not participate in the writing of regular power supply data. When the signal quality of the communication module remains below a preset threshold, the traceability storage unit detects a communication link interruption and switches to offline caching mode. In offline caching mode, the main ferroelectric memory continuously writes power supply data and records breakpoint markers in the backup ferroelectric memory at preset intervals. The breakpoint marker includes an interruption timestamp, data offset address, and cyclic redundancy check (CRC) value. After the communication link is restored, the range of unuploaded power supply data is determined based on the breakpoint markers in the backup ferroelectric memory, triggering retransmission. After retransmission, the retransmitted power supply data undergoes integrity verification based on the CRC value. The retransmitted power supply data is synchronously incorporated into the existing chained verification chain; specifically, the backup ferroelectric memory retains the chained verification value of the last successfully written data node before the communication link interruption. The starting data node is retransmitted and the chain check value of formula (2) is recalculated with the chain check value as the predecessor, so as to maintain the continuity of the check chain.

[0036] In this embodiment of the invention, the mechanism ensures that power supply data is not lost during communication link interruptions and that the retransmitted power supply data is seamlessly connected to the normally written power supply data on the verification chain, unlike the method of discarding data during offline periods. The latter would cause time gaps in traceability data at remote sites with unstable communication, affecting the accuracy of subsequent strategy optimization units in analyzing attenuation rates based on continuous historical data.

[0037] The feature tracing unit reads wind and solar power data from the tracing storage unit, specifically the wind power data sequence within a preset time window. and photovoltaic power data sequence Multi-scale fluctuation features and wind-solar coupling features are extracted using wavelet transform. Optionally, the time window is set to 24 hours. It is understood that those skilled in the art can adjust the value of the time window according to the daily variation cycle of wind and solar resources, and the embodiments of the present invention do not specifically limit this.

[0038] The feature tracing unit performs multi-scale wavelet decomposition on wind and solar power data separately for wind power and photovoltaic power, and performs the following on the wind power power data sequence: Layer discrete wavelet decomposition yields A total of one detail layer and one approximation layer The wavelet coefficients are grouped, and the information entropy of the energy distribution of the wavelet coefficients at each scale is used to characterize the multi-scale fluctuation characteristics. The formula for calculating the multi-scale wavelet energy entropy is shown in formula (3):

[0039] in, The energy entropy is a multi-scale wavelet energy, expressed in nat. For the first The energy of the layer wavelet coefficients is defined as the sum of the squares of all discrete wavelet coefficients in that layer, and the unit is . , The wavelet decomposition level is a positive integer. For layer index and and to Corresponding to each detail layer and Corresponding approximation layer, To find the dummy index and traverse the range Same and Independent of , It is the natural logarithm function. The denominator is... The sum of all layer energies is given, ensuring that the sum of the energy percentages of each layer is 1. This is true when all energy is concentrated on a single scale. Approaching zero, when energy is uniformly distributed across all scales Reaching the maximum value In the implementation of the project, The term is set to its limit value so that its contribution is zero. Wavelet energy entropy is calculated on the photovoltaic power data sequence using the same method to obtain the multi-scale fluctuation characteristics of wind power and photovoltaic power respectively.

[0040] Optionally, The value is set to 3, and the wavelet basis function is Daubechies-4. It is understood that those skilled in the art can adjust the wavelet basis function according to the signal's frequency range and time resolution requirements. The selection of wavelet basis functions may be adjusted, but the embodiments of the present invention do not impose specific limitations on this.

[0041] Furthermore, wavelet decomposition was implemented on the STM32H743 using optimized FIR filtering functions from the CMSIS-DSP library. The computation time for a 3-level decomposition of 1024 points of data was approximately 2ms. On the STM32F4 series, the data can be downsampled to 256 points and computation can be performed using a fixed-point Q15 format to meet real-time requirements.

[0042] In this embodiment of the invention, formula (3) compresses the multi-scale energy distribution into a single scalar index. A low value indicates that power fluctuations are dominated by a specific frequency band. A high value indicates that the wave energy is dispersed across multiple frequency bands, unlike the method of directly storing wavelet coefficients at each layer. The latter starts with hundreds of dimensions, which significantly reduces the traceable time depth under the condition of limited ferroelectric memory capacity, while energy entropy reduces the data dimension to one dimension while retaining the ability to distinguish wave patterns.

[0043] To characterize the coupling degree between wind power and photovoltaic power at various scales, the wavelet coherence coefficient is used to characterize the wind-solar coupling characteristics. The formula for calculating the wavelet coherence coefficient is shown in formula (4):

[0044] in, The wavelet coherence coefficients are dimensionless and their values ​​range from 1 to 2. , These are wavelet scaling parameters, corresponding to different time resolutions, and are dimensionless positive real numbers. The time shift parameter is in seconds. The cross-wavelet spectrum of wind power series and photovoltaic power series, with units of , The wavelet spectrum of the wind power sequence is given in units of 1. , The small wavelet spectrum of the photovoltaic power sequence is given in units of 1. , The time-frequency domain smoothing operator performs weighted averaging on the cross spectrum and the self spectrum in the time and scale neighborhoods, respectively. This is for modulo operation. The subscript *w* represents wind power, the subscript *s* represents photovoltaic power, and the subscript *ws* represents the cross-spectrum of the two. It should be noted that smoothing is a necessary step to obtain a meaningful coherent estimate; the ratio of the unsmoothed original cross-spectrum to the self-spectrum is always equal to 1, and it lacks the ability to distinguish between strong and weak coupling. After smoothing, and It is always positive in the region where there is non-zero energy in the power sequence. Approaching 1 indicates that the power of the two paths is on the scale. ,time Highly coupled at the point, Approaching 0 indicates that the two are independent at that point.

[0045] Furthermore, for all scales and time translation On The average coherence coefficient curves at each scale are obtained by taking the mean value over the time dimension, and stored as a fixed-length feature fingerprint as a feature of wind-solar coupling.

[0046] In this embodiment of the invention, formula (4) has time-frequency joint resolution capability, which can distinguish the coupling differences of wind and solar power at different time scales, unlike the method that only calculates the global Pearson correlation coefficient. The latter compresses the correlation of the entire time period into a scalar, and cannot distinguish the scale-dependent characteristics of wind and solar power being negatively correlated at short time scales and positively correlated at long time scales. In wind-solar complementary systems, the coupling differences at different scales have different technical guiding significance for the formulation of charging and discharging strategies.

[0047] Multi-scale fluctuation characteristics and wind-solar coupling characteristics are combined using hash compression to generate a fixed-length feature fingerprint. Specifically, the energy entropy at each scale is... The average coherence coefficients at each scale and the start and end timestamps of the corresponding time windows are compressed using CRC-32 hashing to generate fixed-length feature fingerprints. Each feature fingerprint has a fixed data size of 32 bytes and is written to the traceability storage unit. Based on a 24-hour window, sampling once per minute, and 32 bytes per original record, the original data is approximately 46KB. The fixed-length feature fingerprints reduce storage size by about three orders of magnitude compared to the original data, and can be used to perform similarity matching with historical feature fingerprints to assist in fault mode retrieval.

[0048] The strategy optimization unit reads power supply data from the traceability storage unit, specifically historical charge-discharge cycle data within a preset time window, calculates the battery pack capacity decay rate, and generates correction values ​​for the charge-discharge parameters accordingly to adjust the parameters. Optionally, the time window is set to 7 days. It is understood that those skilled in the art can adjust the time window value according to the decay cycle of the battery chemistry system, and this embodiment of the invention does not specifically limit this.

[0049] The normalized decay rate of the maximum available capacity within the preset time window is calculated according to formula (5). The formula for calculating the normalized decay rate is shown in formula (5):

[0050] in, This is the normalized decay rate, representing the ratio of the maximum usable capacity to the initial capacity of the window within each charge-discharge cycle, and is dimensionless. The maximum available capacity at the start of the time window, in Ah. , The maximum available capacity at the end of the time window, in Ah. This represents the number of charge / discharge cycles within the time window and is a positive integer. The value is zero when there is no charge / discharge cycle within the time window, and the decay rate calculation is not triggered at this time. It is calculated by reverse engineering from the charge and discharge traceability data recorded in the ferroelectric memory, rather than relying on the factory nominal value.

[0051] In this embodiment of the invention, formula (5) uses traceable data rather than the factory nominal value as the calculation basis, so that the degradation rate reflects the actual operating state of the battery pack, which is different from the method of obtaining the health status offline based on constant current discharge test. The latter requires interrupting normal power supply to perform a dedicated test process, which is difficult to implement in unattended remote sites.

[0052] Normalized decay rate Exceeding the preset attenuation threshold At that time, the strategy optimization unit generates a correction value for the charging cutoff voltage through the saturation correction function, specifically calculated according to formula (6). The formula for calculating the correction value of the charging cutoff voltage is shown in formula (6):

[0053] in, This is a correction value for the charging cutoff voltage, expressed in volts (V), and is always negative or zero, meaning the cutoff voltage is reduced. To correct the gain coefficient, which is a dimensionless positive real number, The maximum permissible correction voltage amplitude is expressed in volts (V). The preset attenuation threshold is dimensionless and , It is the hyperbolic tangent function. Fraction Since both the numerator and denominator are dimensionless, the fraction is also a dimensionless real number. Just over When the hyperbolic tangent function output is close to the linear region, the correction value is approximately proportional to the overshoot; when Far exceeding At this point, the output of the hyperbolic tangent function tends to saturate, and the correction value is limited to [value missing]. Within. The corrected charging cutoff voltage is the original cutoff voltage plus... .

[0054] Optionally, The value is 0.8. The value is 0.2V. Values It is understood that those skilled in the art can refer to the technical manual of the selected battery chemistry system for further information. , and The value of is adjusted, but the embodiments of the present invention do not impose specific limitations on this.

[0055] In this embodiment of the invention, the hyperbolic tangent saturation characteristic of formula (6) has a self-limiting effect in the large deviation region, which is different from the linear correction strategy. The latter will increase the correction value proportionally when the decay rate changes abruptly, which may lead to the cutoff voltage being excessively reduced, thus causing undercharging problems. Formula (6) through The parameters provide a deterministic upper bound on the correction magnitude, balancing correction response speed and system stability.

[0056] Furthermore, the strategy optimization unit also monitors the variance of the state of charge (SOC) fluctuation within a preset statistical window. When the variance exceeds a preset variance threshold, it determines that the depth of discharge is too large and accordingly increases the discharge termination voltage to reduce the SOC fluctuation range. Optionally, the statistical window is set to 24 hours, the variance threshold is set to 0.01, and the discharge termination voltage is increased in increments of 0.05V each time with a cumulative increase not exceeding 0.2V. It is understood that those skilled in the art can adjust the values ​​of the above parameters according to the sensitivity of the battery chemistry system to the depth of discharge, and the embodiments of the present invention do not specifically limit this.

[0057] The strategy optimization unit writes the charging and discharging parameters before and after the correction, as well as the battery pack operating parameters for each time period before and after the correction, into the traceability storage unit and marks the version number. After the preset verification period ends, the strategy optimization unit reads the power supply data within that period again and calculates the relative change in the normalized decay rate before and after the correction as the improvement magnitude according to formula (7). The formula for calculating the improvement magnitude is shown in formula (7):

[0058] in, To improve the magnitude and make it dimensionless, To correct the normalized decay rate within the preset verification period, and to ensure it is dimensionless. , The normalized decay rate within the preset verification period after correction is dimensionless. This indicates that the decay has slowed down after the correction. This indicates accelerated decay.

[0059] Furthermore, when Reaching the preset improvement threshold The time-marking strategy correction is effective and maintains the current charge / discharge parameters; Below However, if the value is positive, maintain the current charge and discharge parameters and extend the observation period by a preset verification cycle; If the value is negative, the improvement in the degradation rate does not reach the preset improvement threshold. The system immediately reverts to the pre-correction charge / discharge parameters and limits the charge / discharge current to a preset percentage of the rated value. Simultaneously, an alarm flag is set, and an alarm message is sent to maintenance personnel via the communication link, indicating that the battery pack may have experienced a sudden hardware failure or irreversible degradation requiring manual investigation. Optionally, The value is set to 0.2, and the preset verification period is 7 days. It is understood that those skilled in the art can adjust the system security margin according to the system's requirements. The value of the preset verification period is adjusted, but the embodiments of the present invention do not impose specific limitations on this.

[0060] In this embodiment of the invention, formula (7) ensures that each strategy correction has a quantifiable effect evaluation index, and the fallback logic ensures that invalid corrections will not continue to affect the battery pack, unlike unverified open-loop correction schemes. The latter executes for a long time once a correction value is generated, making it impossible to detect whether the correction produces the expected effect or whether it has a negative impact. The closed-loop structure incorporates the consequences of control decisions into the traceability data chain, providing experience accumulation for subsequent corrections.

[0061] The impedance diagnostic unit injects an excitation signal into the battery pack via a DC-DC converter to obtain the impedance spectrum online. Specifically, the impedance diagnostic unit superimposes low-amplitude excitation signals from a preset frequency set onto the DC-DC converter drive signal, and superimposes a sinusoidal component with an amplitude of a preset percentage of the rated duty cycle onto the DC-DC converter duty cycle control signal, sequentially superimposing low-amplitude excitation signals from each frequency in the preset frequency set onto the DC-DC converter drive signal. Provide incentives, For frequency index and , This represents the total number of frequency points in the frequency set and is a positive integer. The preset frequency set is selected at logarithmic intervals between the preset lower frequency limit and the preset upper frequency limit. Several frequency points are selected, and the frequencies are evenly distributed on the logarithmic frequency axis to cover the characteristic frequency bands of the electrochemical impedance spectroscopy. Optionally, the preset percentage is 5%, the lower frequency limit is 0.05Hz, and the upper frequency limit is 100Hz. The value is 128. It is understood that those skilled in the art can adjust the value of the frequency set according to the impedance spectrum frequency range of the target battery and storage capacity constraints; this embodiment of the invention does not specifically limit this. It should be noted that low-frequency excitation requires a relatively long phase-locked integration time, and the sweep frequency measurement of all frequency points must be completed within a preset diagnostic cycle. Optionally, the diagnostic cycle is set to 24 hours, and the sweep frequency measurement is scheduled during periods of low system load and when the battery is idle or undergoing low-current float charging, in order to reduce the interference of operating current fluctuations on impedance measurement.

[0062] The microcontroller's ADC synchronously acquires the voltage and current responses of the battery pack, and extracts the complex impedance at each frequency through digital lock-in amplification to construct an impedance spectrum. Specifically, the voltage response acquired by the ADC is multiplied point-by-point with a reference sine and cosine signal of the same frequency, and then low-pass filtered to extract the in-phase component of the voltage. and orthogonal components Performing the same operation on the current response yields the in-phase component. and orthogonal components The complex impedance at each frequency is calculated according to formula (8). The formula for calculating the complex impedance is shown in formula (8):

[0063] in, For frequency The complex impedance below and the unit is , For the first Each excitation frequency is expressed in Hz. For frequency index and , The total number of frequency points is a positive integer. For frequency The in-phase component of the voltage response, in units of V. For frequency The orthogonal components of the voltage response, with units of V. For frequency The in-phase component of the current response, in units of Ampere (A). For frequency The orthogonal components of the current response, with units of A. The imaginary unit. The modulus of the denominator. Corresponding to the current amplitude, under normal excitation conditions, the battery has finite impedance, therefore the current response is not zero. The impedance amplitude can be obtained from the real and imaginary parts. and phase angle ,all The complex impedance at each frequency point constitutes the impedance spectrum.

[0064] In this embodiment of the invention, the digital phase-locked amplification in formula (8) excludes noise at non-excitation frequencies and DC-DC switching harmonics from the extraction results through narrowband filtering, which differs from the direct FFT spectrum analysis method. The latter suffers from severe spectrum leakage under low signal-to-noise ratio conditions, and for actual operating conditions with high DC-DC switching harmonic content, FFT is difficult to effectively separate the excitation frequency component from the switching harmonic component.

[0065] The impedance diagnostic unit uses compressed sensing to compress the impedance spectrum and store it in the traceability storage unit. (The rest of the text appears to be incomplete and requires further context.) The complex impedance values ​​at each frequency point are arranged as a column vector. , dimension and The battery impedance spectrum is controlled by a few equivalent circuit parameters, exhibiting a smooth curve in the frequency dimension, and its energy in the orthogonal transform domain is concentrated in a few coefficients. Based on this, a sparsed basis matrix is ​​pre-constructed. , dimension . It is built offline during system initialization and stored in the microcontroller firmware. Optionally, A discrete cosine transform basis is used. It is understood that those skilled in the art can select other orthogonal bases based on the frequency domain characteristics of the impedance spectrum; this embodiment of the invention does not specifically limit this selection. Impedance spectrum vector It can be represented as , where the coefficient vector Only a few non-zero or significantly non-zero components correspond to the basis vectors carrying the main energy under the orthogonal basis. It should be noted that the battery impedance spectrum consists of three smooth curves: ohmic internal resistance, charge transfer arc, and Warburg diffusion tail. Its frequency domain shape is constrained by a few physical parameters, and after orthogonal transformation, the energy is highly concentrated, satisfying the sparsity premise. Construct the compressed observation equation. The compressed sensing observation equation is shown in equation (9):

[0066] in, The compressed observation vector has a dimension of , The observation matrix has a dimension of . And it is deterministically generated from a preset random seed. The original impedance spectrum vector has a dimension of . , It is a sparse basis matrix with dimension . It is an orthogonal matrix and stored in the firmware. Let be a sparse coefficient vector in the transform domain with dimension . , for Norm is the number of non-zero elements in a vector. The sparsity is a positive integer. For impedance spectrum dimension and , To compress the observation dimension and it is a positive integer and . No more than 6. The impedance diagnostic unit will only perform one diagnostic at a time. 3D observation vector Write to the traceability storage unit. During traceability, it is based on the data pre-stored in the firmware. and Construct a combined perception matrix Through sparse reconstruction using the orthogonal matching pursuit algorithm, in Solving for sparsity coefficients Then by Restore the complete impedance spectrum.

[0067] Optionally, The value is 16. It is understandable that those skilled in the art can adjust the value based on the trade-off between storage capacity and reconstruction accuracy. The value of is adjusted, but the embodiments of the present invention do not impose specific limitations on this.

[0068] In this embodiment of the invention, formula (9) reduces the storage amount for each diagnosis from... Dimensions dropped to This approach, within the constraints of limited ferroelectric memory capacity, increases the traceability time depth of the impedance spectrum, unlike methods that directly store the complete impedance spectrum. The latter... Furthermore, when each complex component occupies 8 bytes, each diagnosis requires 1024 bytes of storage space. Compressed sensing reduces this storage requirement to 128 bytes.

[0069] The impedance diagnostic unit extracts the charge transfer impedance from the recovered complete impedance spectrum based on an equivalent circuit model. and diffusion impedance slope Each value is compared with its initial calibration reference value, and the mechanism discrimination coefficient is calculated according to formula (10). The formula for calculating the mechanism discrimination coefficient is shown in formula (10):

[0070] in, The mechanistic discrimination coefficient is a dimensionless positive real number. The current charge transfer impedance is expressed in units of 1 / 2π. and , The charge transfer impedance reference value recorded during the initial calibration of the system during its first run is the initial calibration reference value of the charge transfer impedance, and the unit is... and , The current diffusion impedance slope in units of , The diffusion impedance slope reference value recorded during initial calibration is the initial calibration reference value of the diffusion impedance slope, and the unit is... Same and , This is a dimensionless positive real number used as a weighting factor for the diffusion term and is used to adjust the relative weights of the charge transfer impedance term and the diffusion impedance term. This is for absolute value operations. The first term of formula (10) The ratio of charge transfer impedance to its initial calibration reference value reflects the increase in charge transfer impedance relative to the initial calibration reference value; the second term It is the weighted deviation of the diffusion impedance slope relative to its initial calibration reference value, reflecting the weighted relative amount of the diffusion characteristics deviating from the initial calibration reference value.

[0071] It should be noted that, and Initial calibration is performed during the first system run, and different battery chemistry systems and battery packs of different capacities have different initial calibration reference values. Optionally, The value is set to 1.5. It is understood that those skilled in the art can determine the relative sensitivity of the two decay mechanisms—charge transfer and diffusion—in the target battery chemistry system. The value of is adjusted, but the embodiments of the present invention do not impose specific limitations on this.

[0072] Furthermore, the charge transfer deviation ratio is defined. and diffusion deviation ratio The attenuation mechanism is determined based on the relationship between the two factors and their respective preset thresholds: when Exceeding the preset charge transfer threshold and The diffusion threshold was not exceeded. When the decay mechanism is determined to be the thickening of the solid electrolyte interface film; when Exceed and Exceed When the degradation mechanism is determined to be lithium plating; when Not exceeding and Exceed At that time, the attenuation mechanism was determined to be electrolyte attenuation. Optionally, The value is 1.5. The value is set to 0.3. It is understood that those skilled in the art can determine the appropriate value based on the impedance degradation characteristics of the target battery chemistry system. and The value of is adjusted, but this embodiment of the invention does not specifically limit this. The impedance diagnosis unit sends the diagnosis results to the strategy optimization unit, which adjusts the charge and discharge parameters differently according to different attenuation mechanisms: when the solid electrolyte interface film thickens, the charging current is reduced to slow down the film growth rate; during lithium plating, charging is paused and forced to stand; and when the electrolyte decays, a maintenance warning is marked.

[0073] In this embodiment of the invention, formula (10) uses two parameters, charge transfer impedance and diffusion impedance slope, to jointly determine the attenuation mechanism, which differs from the method of setting a fixed threshold using only a single impedance parameter. The latter cannot distinguish between impedance changes caused by the obstruction of the charge transfer process and the obstruction of the diffusion process, two different physical mechanisms, and lacks quantitative basis for scenarios requiring differentiated maintenance strategies. Formula (10) uses weighting factors... Adjusting the relative weights of the two mechanisms allows the mechanism discrimination coefficient to be adapted to the degradation characteristics of different battery chemistry systems.

[0074] In summary, in the system provided by this embodiment of the invention, the data acquisition unit collects wind and solar power data and battery pack operating parameters as power supply data; the traceability storage unit stores the power supply data in a time sequence using a circular linked list structure within the ferroelectric memory, and uses hash operations to achieve write-erase balance and chain-based checksums to prevent tampering; the backup ferroelectric memory ensures breakpoint continuation when the communication link is interrupted; the feature tracing unit extracts multi-scale fluctuation features and wind-solar coupling features from the wind and solar power data using wavelet transform, and generates a fixed-length feature fingerprint through hash compression, which is then written into the traceability storage unit; the strategy optimization unit calculates the battery pack capacity decay rate based on the power supply data and generates correction values ​​for charge and discharge parameters through a saturation correction function; after the preset verification period ends, the improvement magnitude is compared with a preset improvement threshold to determine whether to roll back; the impedance diagnosis unit injects an excitation signal into the battery pack through a DC-DC converter to obtain the impedance spectrum, which is then compressed and stored; based on the charge transfer impedance and diffusion impedance slope, the decay mechanism is determined, and the diagnosis results are fed back to the strategy optimization unit to achieve differentiated adjustment of charge and discharge parameters. The above units form a closed loop with the traceability storage unit as the data hub, enabling the storage of traceability data, feature extraction, strategy optimization, and impedance diagnosis to run collaboratively on the microcontroller platform.

[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A microcontroller-based wind-solar hybrid power supply battery pack and power supply data traceability system, characterized in that, It includes a microcontroller, and a data acquisition unit, a traceability storage unit, a feature traceability unit, a strategy optimization unit, and an impedance diagnosis unit, which are respectively connected to the microcontroller; The data acquisition unit is used to collect wind and solar power data and battery pack operating parameters as power supply data, and send the power supply data to the traceability storage unit; The traceability storage unit includes a ferroelectric memory, which has a circular linked list structure for storing data in time sequence. The traceability storage unit determines the target storage address for each write operation through hash operation to achieve write-erase balance, and calculates a chain check value for each data node to prevent tampering. The feature tracing unit is used to read the wind and solar power data from the tracing storage unit, extract multi-scale fluctuation features and wind-solar coupling features through wavelet transform, generate a fixed-length feature fingerprint through hash compression, and write it into the tracing storage unit. The strategy optimization unit is used to read the power supply data from the traceability storage unit, calculate the battery pack capacity decay rate and generate a correction value for the charging and discharging parameters to adjust the charging and discharging parameters. The charging and discharging parameters before and after the correction, as well as the battery pack operating parameters for each time period before and after the correction, are written into the traceability storage unit. After the preset verification period ends, if the improvement of the decay rate does not reach the preset improvement threshold, the charging and discharging parameters before the correction are returned. The impedance diagnostic unit is used to inject an excitation signal into the battery pack through a DC-DC converter to obtain an impedance spectrum, determine the attenuation mechanism based on the charge transfer impedance and diffusion impedance slope, and send the diagnostic results to the strategy optimization unit.

2. The system according to claim 1, characterized in that, The traceability storage unit performs an XOR operation on the current write timestamp and the global write counter, performs a hash operation on the result, and then takes the modulo of the total number of storage blocks to obtain the target block number to determine the target storage address; when the cumulative number of erases and writes of the target block exceeds the preset equalization threshold, it jumps to the next available block.

3. The system according to claim 1, characterized in that, The chain check value is calculated by concatenating the data content of the current data node with the chain check value of the previous data node and then performing a hash operation.

4. The system according to claim 1, characterized in that, The traceability storage unit also includes a backup ferroelectric memory for storing breakpoint markers and retransmission status information. When the traceability storage unit detects a communication link interruption, it switches to offline caching mode and records the breakpoint markers in the backup ferroelectric memory. After the communication link is restored, the range of unuploaded power supply data is determined based on the breakpoint markers, and retransmission is triggered.

5. The system according to claim 4, characterized in that, The breakpoint marker includes the interruption time timestamp, data offset address, and cyclic redundancy check value; after the retransmission is completed, the retransmitted power supply data is subjected to integrity verification based on the cyclic redundancy check value.

6. The system according to claim 1, characterized in that, The feature tracing unit performs multi-scale wavelet decomposition on the wind and solar power data separately for wind power and photovoltaic power. The information entropy of the energy distribution of wavelet coefficients at each scale is used to characterize the multi-scale fluctuation feature, and the wavelet coherence coefficient is used to characterize the wind-solar coupling feature. The multi-scale fluctuation feature and the wind-solar coupling feature are then hash-compressed to generate the fixed-length feature fingerprint.

7. The system according to claim 1, characterized in that, The strategy optimization unit calculates the normalized decay rate of the maximum available capacity within a preset time window. When the normalized decay rate exceeds a preset decay threshold, a correction value for the charging cutoff voltage is generated through a saturation correction function. After the preset verification period ends, the relative change in the normalized decay rate before and after the correction is calculated as the improvement magnitude.

8. The system according to claim 1, characterized in that, The impedance diagnostic unit superimposes low-amplitude excitation signals of each frequency in a preset frequency set into the DC-DC converter drive signal, extracts the complex impedance at each frequency through digital phase-locked amplification to form the impedance spectrum, compresses the impedance spectrum using a compressed sensing method and stores it in the traceability storage unit, and recovers the complete impedance spectrum through sparse reconstruction during traceability.

9. The system according to claim 8, characterized in that, The impedance diagnostic unit extracts the charge transfer impedance and the diffusion impedance slope from the recovered complete impedance spectrum, compares them with their respective initial calibration reference values, and calculates a theoretical discrimination coefficient based on the ratio of the charge transfer impedance to its initial calibration reference value and the weighted deviation of the diffusion impedance slope relative to its initial calibration reference value, so as to determine whether the attenuation mechanism is solid electrolyte interface film thickening, lithium plating, or electrolyte attenuation.

Images