Pulse type ultra-low power consumption energy management system based on super capacitor

Abstract

The invention relates to the technical field of pulse load energy optimization control, in particular to a pulse type ultra-low power consumption energy management system based on a super capacitor, which comprises a pulse spectrum analysis unit, a capacitor bank dynamic tuning unit and a charge state compensation unit, a pulse spectrum analysis unit extracts a load pulse dominant frequency component through a window function weighting algorithm to generate a target frequency value, and a capacitor bank dynamic tuning unit inquires a mapping table according to the target frequency to adjust series-parallel connection combination of the super capacitors and injects a reverse reactance component to offset a resonant impedance difference value. And the charge state compensation unit predicts the residual charge quantity by using a Kalman filter, and triggers a charge redistribution strategy in a non-pulse period, thereby solving the problems of insufficient dynamic impedance matching and low charge balance efficiency in the traditional technology, realizing high-efficiency and low-consumption management of the energy of the super capacitor, improving the energy transmission efficiency, and reducing the energy consumption. And the endurance time of the equipment is prolonged.

Description

Technical Field

[0001] This invention relates to the field of pulsed load energy optimization control technology, and more specifically, to a pulsed ultra-low power energy management system based on supercapacitors. Background Technology

[0002] Pulse load energy optimization control is an important technology. In scenarios such as IoT sensors and portable electronic devices, supercapacitor energy management technology stores and releases pulse energy, which is the core means to ensure stable operation of the device in low power mode.

[0003] With the increasing miniaturization and low-power requirements of smart devices, building an energy management system that dynamically matches the characteristics of pulse loads is crucial for reducing system power consumption and extending device battery life. Traditional fixed-parameter energy management models are no longer suitable for the demands of variable pulse load frequencies and high-efficiency energy conversion. However, traditional supercapacitor energy management technology suffers from core problems such as insufficient dynamic impedance matching and low charge balancing efficiency. Existing solutions use fixed capacitor combinations and do not establish a real-time matching mechanism between pulse frequency and capacitor parameters. When the load pulse frequency jumps from 100Hz to 1kHz, the equivalent impedance matching deviation of the system exceeds the expected range, resulting in a decrease in energy transmission efficiency. The charge management method that relies solely on passive balancing cannot complete the charge redistribution of the supercapacitor group during the pulse interval. In addition, traditional systems lack cross-unit coordination mechanisms and cannot synchronously adjust capacitor configuration and charge strategy when the pulse spectrum changes, leading to an increased probability of system downtime under sudden pulse loads. This lack of dynamic matching capability and insufficient collaborative control ultimately results in low energy conversion efficiency and poor device battery life stability under pulse load scenarios, making it difficult to meet the long-term operation requirements of ultra-low power devices. To solve this technical problem, we provide a pulsed ultra-low power energy management system based on supercapacitors. Summary of the Invention

[0004] The purpose of this invention is to provide a pulsed ultra-low power energy management system based on supercapacitors to solve the problems mentioned in the background art.

[0005] 1. Due to insufficient dynamic impedance matching in traditional technology, the impedance matching deviation is large when the load pulse frequency changes, resulting in low energy transmission efficiency. Therefore, this case uses a capacitor bank dynamic tuning unit to perform adaptive capacitor matching and resonant impedance cancellation, which can dynamically adjust the capacitor combination and compensate impedance to improve energy transmission efficiency.

[0006] 2. Since traditional methods of charge balancing are inefficient and charge redistribution is difficult to complete during pulse intervals, this case uses a charge state compensation unit and a Kalman filter to predict the charge level and trigger a charge redistribution strategy, which can achieve charge balancing in non-pulse cycles and extend the device's battery life.

[0007] To achieve the above objectives, a pulsed ultra-low power energy management system based on supercapacitors is provided, comprising the following units:

[0008] The pulse spectrum analysis unit acquires the load-side current waveform in real time, extracts the amplitude-frequency spectrum of the pulse main frequency component through a window function weighting algorithm, generates the target frequency value based on the frequency point with the maximum amplitude in the amplitude-frequency spectrum, and calculates the pulse interval time based on the rising edge interval of adjacent pulses.

[0009] The capacitor bank dynamic tuning unit performs adaptive capacitor matching and resonant impedance cancellation based on the target frequency value. The adaptive capacitor matching involves querying a preset frequency-capacitance mapping table based on the target frequency value to determine the target equivalent capacitance value. The series and parallel combination of supercapacitors is dynamically adjusted by switching the matrix to make the total capacitance value of the system match the target equivalent capacitance value. The resonant impedance cancellation involves calculating the difference between the equivalent inductive reactance and capacitive reactance of the system at the target frequency value, injecting a reverse reactance component with the same amplitude as the difference into the resonant compensation circuit, and generating an impedance matching completion flag signal.

[0010] The charge state compensation unit uses a Kalman filter to predict the remaining charge of the supercapacitor bank. When the impedance matching completion flag signal is equal to 1 and the pulse interval is greater than the minimum charge migration time threshold, a charge redistribution strategy is triggered during the non-pulse period.

[0011] As a further improvement to this technical solution, the execution of the window function weighting algorithm includes the following steps:

[0012] A Hanning window function is applied to the real-time acquired load-side current waveform to suppress spectral leakage. The windowed load-side current waveform is then converted into an initial spectrum using a fast Fourier transform. Frequency points with amplitudes exceeding a preset threshold are located in the initial spectrum, and the pulse dominant frequency component is selected. The frequency value of the pulse dominant frequency component is used as the basis for generating the target frequency value.

[0013] As a further improvement to this technical solution, the method for constructing the frequency-capacitance mapping table includes the following steps:

[0014] During the system initialization phase, a frequency sweep test signal is injected into the load end through a signal generator to obtain the optimal equivalent capacitance value of the supercapacitor group at different frequencies, and the optimal equivalent capacitance value of the supercapacitor group at different frequencies is recorded. The optimal equivalent capacitance value is determined by satisfying the minimum internal resistance of the system.

[0015] Establish a one-to-one mapping table between frequency values ​​and optimal equivalent capacitance values, and store it in non-volatile memory.

[0016] As a further improvement to this technical solution, the steps for dynamically adjusting the series-parallel combination of supercapacitors using a switching matrix include:

[0017] The required number of series capacitors and parallel capacitors are calculated based on the target equivalent capacitance value. The relay array is then controlled to reassemble the supercapacitor into a topology with series capacitors in series and parallel capacitors in parallel. The actual capacitance value after reassembly is detected by a bridge circuit. If the error exceeds the preset capacitance threshold, a secondary reconstruction is triggered.

[0018] As a further improvement to this technical solution, the operation of resonant impedance cancellation includes the following steps:

[0019] Based on the target frequency value, the system equivalent inductive reactance and capacitive reactance are obtained by querying the preset inductive reactance-capacitive reactance lookup table. The impedance difference is calculated based on the system equivalent inductive reactance and capacitive reactance. The adjustable inductor is controlled to generate a reverse reactance component with an amplitude equal to the impedance difference and injected into the compensation circuit. When the system internal resistance drops to the set threshold, the impedance matching completion flag signal is set to 1.

[0020] As a further improvement to this technical solution, the dynamic calibration of the inductive-capacitive reactance comparison table includes the following steps:

[0021] During each pulse interval, a white noise test signal is injected into the capacitor bank to measure the resonant frequency offset. The parameters of the lookup table are updated according to the formula: the impedance difference to be compensated = calibration coefficient * resonant frequency offset. The white noise test signal is used to excite the full-frequency response of the capacitor bank to detect the actual resonant characteristics. The resonant frequency offset is used to reflect the change of system impedance parameters. The two work together to achieve dynamic updating of the inductive reactance-capacitive reactance lookup table.

[0022] As a further improvement to this technical solution, the steps of the Kalman filter in predicting the residual charge include:

[0023] The system reads the terminal voltage and temperature of the supercapacitor bank, predicts the remaining charge through iterative state equations, corrects the prediction error using the terminal voltage measurement, and outputs the optimized remaining charge.

[0024] As a further improvement to this technical solution, the execution of the charge redistribution strategy includes the following steps:

[0025] When the impedance matching completion flag signal is equal to 1 and the pulse interval time is greater than the minimum charge migration time threshold, charge migration is initiated, and the single-cell capacitor with excessive charge is identified and used as a charge source. The charge source transfers energy to the single-cell capacitor with insufficient charge through a bidirectional DC-DC circuit. After the migration is completed, the charge status flag of all single cells is updated.

[0026] As a further improvement to this technical solution, the setting of the minimum charge migration time threshold includes:

[0027] The minimum charge migration time threshold is calculated by adding the system response delay time and the safety margin time to the base time required for a single charge migration based on the target frequency value. The response delay time is the fixed time required for system hardware switching. The base time required is determined by the target frequency value, which is the theoretical shortest cycle of charge migration. The safety margin time is the buffer time reserved to cope with sudden load fluctuations.

[0028] As a further improvement to this technical solution, a cross-unit collaboration step is also included:

[0029] After completing the charge redistribution, the charge state compensation unit sends a spectrum resampling command to the pulse spectrum analysis unit. The pulse spectrum analysis unit starts a new round of current waveform acquisition according to the command. The capacitor bank dynamic tuning unit receives the new target frequency value and refreshes the frequency-capacitance mapping table priority.

[0030] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0031] In the pulsed ultra-low power energy management system based on supercapacitors, the pulse spectrum analysis unit extracts the load pulse main frequency component in real time, providing a precise target frequency value for the dynamic tuning of the capacitor bank. This enables the total system capacitance and resonant impedance to dynamically match the load characteristics, effectively reducing impedance matching deviation and improving energy transmission efficiency. The capacitor bank dynamic tuning unit adjusts the series and parallel combination of supercapacitors through a switching matrix and injects a reverse reactance component to offset the resonant impedance difference, thereby optimizing the system's internal resistance and ensuring efficient energy transmission. The charge state compensation unit uses a Kalman filter to predict the remaining charge of the supercapacitor bank and triggers a charge redistribution strategy during non-pulse periods, reducing the charge deviation of individual capacitors, delaying the overall capacity decay of the capacitor bank, and extending the equipment's endurance. Cross-unit collaborative steps ensure that the frequency-capacitor mapping table priority is refreshed in a timely manner after charge redistribution, ensuring that the system synchronously adjusts the capacitor configuration and charge strategy when the pulse spectrum changes, reducing the probability of system downtime under sudden loads, and achieving efficient and low-power management of supercapacitor energy. Attached Figure Description

[0032] Figure 1 This is an overall block diagram of the present invention.

[0033] The meanings of the labels in the diagram are as follows:

[0034] 1. Pulse spectrum analysis unit; 2. Capacitor bank dynamic tuning unit; 3. Charge state compensation unit. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] This invention provides a pulsed ultra-low power energy management system based on supercapacitors. Please refer to [link / reference]. Figure 1 As shown, it includes the following units:

[0037] The pulse spectrum analysis unit 1 acquires the load-side current waveform in real time, extracts the amplitude-frequency spectrum of the pulse main frequency component through a window function weighting algorithm, generates the target frequency value based on the frequency point with the maximum amplitude in the amplitude-frequency spectrum, and calculates the pulse interval time based on the rising edge interval of adjacent pulses.

[0038] The capacitor bank dynamic tuning unit 2 performs adaptive capacitor matching and resonant impedance cancellation based on the target frequency value. The adaptive capacitor matching involves querying a preset frequency-capacitance mapping table according to the target frequency value to determine the target equivalent capacitance value. The series and parallel combination of supercapacitors is dynamically adjusted by switching matrix to make the total capacitance value of the system match the target equivalent capacitance value. The resonant impedance cancellation involves calculating the difference between the equivalent inductive reactance and capacitive reactance of the system at the target frequency value, injecting a reverse reactance component with the same amplitude as the difference into the resonant compensation circuit, and generating an impedance matching completion flag signal.

[0039] The charge state compensation unit 3 uses a Kalman filter to predict the remaining charge of the supercapacitor bank. When the impedance matching completion flag signal is equal to 1 and the pulse interval is greater than the minimum charge migration time threshold, the charge redistribution strategy is triggered during the non-pulse period.

[0040] To accurately extract the dominant frequency component of the load pulse and provide a precise target frequency value for subsequent capacitor bank tuning, pulse spectrum analysis unit 1 needs to execute a window function weighting algorithm. The specific implementation method of the window function weighting algorithm is as follows:

[0041] Because the real-time acquired load-side current waveform is prone to spectral leakage due to truncation during spectral analysis, a Hanning window function needs to be applied to suppress spectral leakage and provide a clean time-domain signal for subsequent spectral conversion. The Hanning window function is a smoothing window function that reduces the abrupt changes at the edges of the time-domain signal, thereby reducing leakage in spectral analysis. Spectral leakage is the phenomenon of energy spreading in the frequency domain when a non-periodic signal is truncated; windowing can significantly suppress this leakage, making the spectral analysis results more accurately reflect the true frequency components of the signal. To convert the time-domain signal into frequency-domain characteristics, a Fast Fourier Transform (FFT) is used to process the windowed current waveform, converting it into the initial spectrum, i.e., the 1024-point time-domain sequence. The signal is converted into a 513-point frequency domain sequence, with each point corresponding to the amplitude and phase information of a frequency component. The generated initial spectrum can intuitively reflect the frequency characteristics of the load pulse, providing a data basis for the selection of the main frequency component. In order to extract effective frequency features from the initial spectrum, it is necessary to locate the frequency points in the initial spectrum whose amplitude exceeds a preset threshold. The preset threshold is dynamically set according to the system noise level. All frequency points are traversed in the spectrum, and the frequency points with amplitudes greater than the threshold are marked as effective frequency components. The frequency point with the largest amplitude is the pulse main frequency component. The pulse main frequency component is selected, and the frequency value of the pulse main frequency component is used as the basis for generating the target frequency value, ensuring that the target frequency value is highly consistent with the actual main frequency of the load pulse. This solves the impedance matching deviation problem caused by inaccurate frequency identification in traditional technology.

[0042] To establish a matching relationship between pulse frequency and the equivalent capacitance value of the supercapacitor bank, and to provide data support for subsequent adaptive capacitance matching, a frequency-capacitance mapping table needs to be constructed during the system initialization phase. The method for constructing the frequency-capacitance mapping table includes the following steps:

[0043] Since the optimal equivalent capacitance value of the system at different frequencies needs to be determined by actual measurement, a frequency sweep test signal is injected into the load end through a signal generator during the system initialization phase. The frequency sweep signal adopts a linear frequency sweep method, with the frequency gradually increasing from 100Hz to 10kHz, and the sweep step size is 100Hz. Each frequency point maintains a stable output for 100ms. The frequency sweep test signal is a signal whose frequency changes continuously within a certain range according to a certain pattern. It is used to test the response characteristics of the system at different frequencies. The linear frequency sweep method covers the target frequency range to ensure that comprehensive frequency characteristic data is obtained. At the same time as injecting the frequency sweep signal, the optimal equivalent capacitance value of the supercapacitor bank at different frequencies is measured by an LCR bridge, and the system internal resistance data is collected simultaneously to ensure that the obtained equivalent capacitance value can accurately match the actual load requirements. The LCR bridge is an instrument used to measure inductance, capacitance, and resistance. In order to enable the supercapacitor bank to achieve efficient energy transfer at different frequencies, To achieve optimal performance, the equivalent capacitance value that minimizes the system's internal resistance needs to be determined. For each swept frequency point, multiple equivalent capacitance values ​​and their corresponding system internal resistance are recorded. By traversing all swept frequency points, the optimal equivalent capacitance value at each frequency is selected, and the optimal equivalent capacitance value of the supercapacitor bank at different frequencies is recorded. The optimal equivalent capacitance value is determined based on minimizing the system's internal resistance. To quickly query the optimal equivalent capacitance value corresponding to a frequency during system operation, a one-to-one mapping table needs to be established between each frequency point and its corresponding optimal equivalent capacitance value. The mapping table is stored in key-value pairs, where the key is the frequency value and the value is the corresponding optimal equivalent capacitance value. The established mapping table is stored in non-volatile memory via the SPI interface to ensure that the data is not lost after power failure. The stored mapping relationship can be continuously optimized and updated through subsequent self-learning algorithms to continuously improve the accuracy of capacitor matching and provide reliable data support for the capacitor bank dynamic tuning unit 2.

[0044] To accurately match the equivalent capacitance of the supercapacitor bank to the target equivalent capacitance value, thereby achieving dynamic impedance matching between the system and the load pulse frequency, it is necessary to dynamically adjust the series and parallel combination of supercapacitors through a switching matrix. The specific implementation method is as follows:

[0045] Since the series and parallel combination of supercapacitors directly affects their equivalent capacitance, it is necessary to first calculate the required number of series and parallel capacitors based on the target equivalent capacitance. The nominal capacitance of a single supercapacitor is known. The formula for calculating the equivalent capacitance of a series capacitor is: The formula for calculating the equivalent capacitance of parallel capacitors is: Therefore, the overall equivalent capacitance value is By working backward from this formula, when the target equivalent capacitance value is... When this is done, the required number of series capacitors can be calculated. and the number of parallel capacitors This provides a precise quantitative basis for subsequent capacitor reconfiguration and controls the relay array to reconfigure the supercapacitors into a topology of series-connected capacitors and parallel-connected capacitors. To adjust the supercapacitor bank to the target series-parallel structure, the connection method of the supercapacitors needs to be controlled by the relay array. The relay array consists of multiple electromagnetic relays, each corresponding to a connection path for one supercapacitor. For example, when it is calculated that three supercapacitors need to be connected in parallel, the three corresponding paths are closed by controlling the relays, thus connecting the three supercapacitors in parallel. In this way, the supercapacitor bank can be reconfigured into... A series, The parallel topology meets the system's real-time requirements for dynamic impedance matching. A bridge circuit is used to detect the actual capacitance value after reassembly. To ensure that the deviation between the equivalent capacitance value of the reassembled supercapacitor bank and the target equivalent capacitance value is within the allowable range, the bridge circuit is used for precise detection of the actual capacitance value. This allows for accurate measurement of the actual capacitance value after reassembly. The measured actual capacitance value is compared with the target equivalent capacitance value. If the error exceeds a preset capacitance threshold, a secondary reassembly is triggered. During the secondary reassembly, the required number of series and parallel capacitors is recalculated, and the relay array is controlled again to adjust the topology until the error between the actual capacitance value and the target equivalent capacitance value is within the preset threshold. This effectively improves energy transmission efficiency and provides strong support for the efficient operation of the pulsed ultra-low power energy management system.

[0046] To achieve dynamic matching of the system's resonant impedance and improve energy transfer efficiency, a resonant impedance cancellation operation needs to be performed after obtaining the target frequency value. The resonant impedance cancellation operation includes the following steps:

[0047] Since the equivalent inductive reactance and capacitive reactance of the system at different frequencies need to be determined by a pre-stored lookup table, the preset inductive reactance-capacitive reactance lookup table is queried based on the target frequency value generated by the pulse spectrum analysis unit 1. This lookup table was established through frequency sweep testing during the system initialization phase and stores the equivalent capacitive reactance of the supercapacitor bank and the equivalent inductive reactance of the line parasitic inductance at different frequencies. The query process is implemented through hardware-accelerated hash table retrieval to ensure real-time acquisition of impedance parameters, providing an accurate data foundation for subsequent impedance difference calculation. Based on the target frequency value, the preset inductive reactance-capacitive reactance lookup table is queried to obtain the system's equivalent inductive reactance and capacitive reactance. To compensate for the system's resonant impedance difference, the impedance difference is calculated based on the system's equivalent inductive reactance and capacitive reactance. The difference is the impedance that needs to be compensated. By controlling the number of turns of the adjustable inductor, a reverse reactance component with an amplitude equal to the impedance difference is generated at the target frequency and injected into the compensation circuit to achieve accurate compensation of the resonant impedance. To verify the impedance matching effect, the change of the system internal resistance is monitored in real time. When the system internal resistance drops from the initial value to below the set threshold after the reverse reactance component is injected, it indicates that the resonant impedance has been effectively canceled. At this time, the impedance matching completion flag signal is set to 1 by the digital logic circuit. This flag signal is used to trigger the operation of the subsequent charge state compensation unit 3, which solves the impedance mismatch problem caused by the parasitic inductance of the line and provides a double guarantee for the efficient transmission of supercapacitor energy.

[0048] To ensure that the inductive-capacitive reactance lookup table is updated synchronously with the actual impedance parameters of the system, and to compensate for matching deviations caused by component aging or environmental changes, dynamic calibration of the lookup table must be performed within each pulse interval. The dynamic calibration of the inductive-capacitive reactance lookup table includes the following steps:

[0049] Since the white noise test signal contains wide-band energy and can fully excite the actual resonant characteristics of the capacitor bank, a white noise test signal is injected into the capacitor bank through a signal generator during each pulse interval. The pulse interval is the time period from the end of the load pulse to the start of the next pulse. To obtain the real-time changes in the system impedance parameters, a spectrum analyzer is used to analyze the response signal of the capacitor bank under white noise excitation, measuring the deviation of its actual resonant frequency from the theoretical resonant frequency in the reference table. The theoretical value of the resonant frequency can be obtained by formula... Calculation, where This is the theoretical value of the resonant frequency. For the system's equivalent inductance, As the equivalent capacitance, to convert the resonant frequency offset into the impedance difference that needs to be compensated, a preset calibration formula is used: impedance difference to be compensated = calibration coefficient * resonant frequency offset; the updated parameters are written to non-volatile memory via the I2C bus, overwriting the original lookup table data; white noise test signal is used to excite the full-band response of the capacitor bank to detect the actual resonant characteristics; resonant frequency offset is used to reflect the change of system impedance parameters; the two work together to achieve dynamic updating of the inductive reactance-capacitive reactance lookup table.

[0050] To accurately predict the remaining charge of a supercapacitor bank and provide a reliable basis for charge redistribution strategies, a Kalman filter is needed to perform the prediction operation. The steps of the Kalman filter to predict the remaining charge include:

[0051] Since the charge of a supercapacitor is strongly correlated with its terminal voltage and temperature, the terminal voltage of the supercapacitor bank is first read using a high-precision ADC, while the temperature of the bank is collected using a thermistor to provide accurate initial information for subsequent predictions. To predict the charge at the next moment based on the current state, a Kalman filter is used to iteratively predict the remaining charge using the state equation. The state equation of the supercapacitor establishes the charge... With terminal voltage ,temperature The dynamic relationship is expressed as follows: ,in, The state transition function is obtained by fitting the capacitor charge-discharge characteristic curve. This provides a basic predicted value for process noise and subsequent error correction. To improve prediction accuracy, the latest terminal voltage measurement value needs to be used to correct the prediction error, outputting the optimized residual charge. The measurement correction equation for the Kalman filter is as follows: ;in, To predict the amount of charge, The Kalman gain is calculated from the system noise covariance matrix. As a measurement function, the predicted charge is converted into a theoretical terminal voltage. The corrected remaining charge is output to the charge state compensation unit 3 through the UART interface, which can accurately reflect the actual charge of the supercapacitor bank. This provides precise data support for the determination of the start-up conditions of the charge redistribution strategy and the energy transfer process, ensuring the effectiveness and efficiency of charge migration.

[0052] To address the issue of uneven charge distribution among individual capacitors within a supercapacitor bank and extend the overall lifespan of the bank, a charge redistribution strategy needs to be implemented under specific conditions. The execution of this strategy includes the following steps:

[0053] Since charge redistribution must be performed when the system impedance matching is complete and there is a sufficient time window, the charge redistribution process is triggered and charge migration is started when the impedance matching completion flag signal output by the capacitor bank dynamic tuning unit 2 is equal to 1 and the pulse interval time is greater than the minimum charge migration time threshold. The response time of this startup mechanism is less than 1ms, avoiding migration when impedance mismatch or insufficient time is required. To accurately locate the individual capacitors that need energy transfer, the remaining charge of each individual capacitor is predicted by a Kalman filter, and a charge threshold is set. Individuals exceeding the threshold are identified as overcharged capacitors and serve as charge sources. The identification of charge sources is achieved through successive comparators to ensure that energy is transferred from overcharged cells to undercharged cells, avoiding energy waste caused by blind migration. To efficiently complete charge migration, the charge source is connected to the undercharged cell through a bidirectional DC-DC circuit. The conduction time of the switching transistor is controlled by pulse width modulation to transfer energy to the undercharged cell. After the migration is completed, the charge status flag of all cells is updated through the I2C bus. This solves the problem of overall performance degradation of the capacitor bank caused by charge imbalance and provides a guarantee for the long-term stable operation of pulsed ultra-low power devices.

[0054] To ensure the charge redistribution strategy can be fully executed within the pulse interval and avoid interruption by subsequent pulse loads, a reasonable minimum charge migration time threshold needs to be set. The minimum charge migration time threshold setting includes:

[0055] Since the target frequency reflects the periodic characteristics of the pulsed load, the theoretical shortest period of charge transfer needs to be determined based on it. The base time required for a single charge transfer is then calculated using the target frequency, as shown in the formula. ;in This is an empirical coefficient. The target frequency value represents the shortest time required to complete a single charge migration under ideal conditions, ensuring the match between the theoretical period and the actual migration process. This provides a core benchmark for threshold setting. Considering the fixed delay in system hardware switching, the response delay time needs to be included in the threshold calculation, superimposed with the system response delay time and the safety margin time. The response delay time refers to the time from triggering the charge migration command to the actual start-up of the bidirectional DC-DC circuit, including the microcontroller instruction execution time and the relay action time. This delay time is measured using an oscilloscope and is deterministic. Superimposing it ensures that the threshold covers the time consumed in the hardware preparation phase, avoiding delays in migration startup. To prevent sudden pulse loads from interfering with the charge migration process, a safety margin time needs to be reserved. The safety margin time is set according to the load fluctuation characteristics of the application scenario, and is determined by analyzing the fluctuations of historical load data. The amplitude is determined to ensure a buffer space in case of sudden pulses at the end of the migration. The minimum charge migration time threshold is obtained by summing the three time components mentioned above. The minimum charge migration time threshold is equal to the required base time + system response delay time + safety margin time. After the threshold is set, it needs to be verified by simulated pulse load testing. When the pulse interval time is greater than 38ms, the charge migration can be completed completely; if it is less than 38ms, it will be automatically skipped to avoid the migration being interrupted by pulses before completion. The response delay time is the fixed time required for system hardware switching. The required base time is determined by the target frequency value, which is the theoretical shortest cycle of charge migration. The safety margin time is the buffer time reserved to cope with sudden load fluctuations. Through the above process of setting the minimum charge migration time threshold, the system realizes a complete threshold construction process from theoretical base time calculation to hardware delay compensation and dynamic margin configuration. This process is linked with the start conditions of the charge redistribution strategy. The threshold serves as a benchmark for judging whether the pulse interval time is sufficient, ensuring that the migration process is executed within the safe time window.

[0056] Including cross-unit collaboration steps:

[0057] After completing the charge redistribution, the charge state compensation unit 3 sends a spectrum resampling command to the pulse spectrum analysis unit 1. The pulse spectrum analysis unit 1 starts a new round of current waveform acquisition according to the command. The capacitor bank dynamic tuning unit 2 receives the new target frequency value and refreshes the frequency-capacitance mapping table priority.

[0058] In this invention, the pulse spectrum analysis unit 1 extracts the load pulse main frequency component through a window function weighting algorithm to generate the target frequency value. The capacitor bank dynamic tuning unit 2 adjusts the series and parallel combination of supercapacitors according to the target frequency by querying the mapping table and injects a reverse reactance component to offset the resonant impedance difference. The charge state compensation unit 3 uses a Kalman filter to predict the remaining charge and triggers a charge redistribution strategy during non-pulse periods. This solves the problems of insufficient dynamic impedance matching and low charge balance efficiency in traditional technologies, realizes efficient and low-consumption management of supercapacitor energy, improves energy transmission efficiency, and extends the equipment's battery life.

[0059] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A pulsed ultra-low power energy management system based on supercapacitors, characterized in that, Includes the following units: The pulse spectrum analysis unit (1) collects the current waveform at the load end in real time, extracts the amplitude-frequency spectrum of the pulse main frequency component through the window function weighting algorithm, generates the target frequency value based on the maximum amplitude frequency point in the amplitude-frequency spectrum, and calculates the pulse interval time based on the rising edge interval of adjacent pulses. The capacitor bank dynamic tuning unit (2) performs adaptive capacitor matching and resonant impedance cancellation based on the target frequency value. The adaptive capacitor matching is to query the preset frequency-capacitance mapping table according to the target frequency value, determine the target equivalent capacitance value, and dynamically adjust the series and parallel combination of supercapacitors through the switching matrix so that the total capacitance value of the system matches the target equivalent capacitance value. The resonant impedance cancellation is to calculate the difference between the equivalent inductive reactance and capacitive reactance of the system under the target frequency value, inject the reverse reactance component with the same amplitude and difference in the resonant compensation circuit, and generate an impedance matching completion flag signal. The charge state compensation unit (3) uses a Kalman filter to predict the remaining charge of the supercapacitor bank. When the impedance matching completion flag signal is equal to 1 and the pulse interval time is greater than the minimum charge migration time threshold, the charge redistribution strategy is triggered in the non-pulse period.

2. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: The execution of the window function weighting algorithm includes the following steps: A Hanning window function is applied to the real-time acquired load-side current waveform to suppress spectral leakage. The windowed load-side current waveform is then converted into an initial spectrum using a fast Fourier transform. Frequency points with amplitudes exceeding a preset threshold are located in the initial spectrum, and the pulse dominant frequency component is selected. The frequency value of the pulse dominant frequency component is used as the basis for generating the target frequency value.

3. The pulsed ultra-low power energy management system based on supercapacitors according to claim 2, characterized in that: The method for constructing the frequency-capacitance mapping table includes the following steps: During the system initialization phase, a frequency sweep test signal is injected into the load end through a signal generator to obtain the optimal equivalent capacitance value of the supercapacitor group at different frequencies, and the optimal equivalent capacitance value of the supercapacitor group at different frequencies is recorded. The optimal equivalent capacitance value is determined by satisfying the minimum internal resistance of the system. Establish a one-to-one mapping table between frequency values ​​and optimal equivalent capacitance values, and store it in non-volatile memory.

4. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: The steps for dynamically adjusting the series-parallel combination of supercapacitors using the switching matrix include: The required number of series capacitors and parallel capacitors are calculated based on the target equivalent capacitance value. The relay array is then controlled to reassemble the supercapacitor into a topology with series capacitors in series and parallel capacitors in parallel. The actual capacitance value after reassembly is detected by a bridge circuit. If the error exceeds the preset capacitance threshold, a secondary reconstruction is triggered.

5. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: The operation of canceling the resonant impedance includes the following steps: Based on the target frequency value, the system equivalent inductive reactance and capacitive reactance are obtained by querying the preset inductive reactance-capacitive reactance lookup table. The impedance difference is calculated based on the system equivalent inductive reactance and capacitive reactance. The adjustable inductor is controlled to generate a reverse reactance component with an amplitude equal to the impedance difference and injected into the compensation circuit. When the system internal resistance drops to the set threshold, the impedance matching completion flag signal is set to 1.

6. The pulsed ultra-low power energy management system based on supercapacitors according to claim 5, characterized in that: The dynamic calibration of the inductive-capacitive reactance comparison table includes the following steps: During each pulse interval, a white noise test signal is injected into the capacitor bank to measure the resonant frequency offset. The parameters of the lookup table are updated according to the formula: the impedance difference to be compensated = calibration coefficient * resonant frequency offset. The white noise test signal is used to excite the full-frequency response of the capacitor bank to detect the actual resonant characteristics. The resonant frequency offset is used to reflect the change of system impedance parameters. The two work together to achieve dynamic updating of the inductive reactance-capacitive reactance lookup table.

7. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: The steps of the Kalman filter predicting the remaining charge include: The system reads the terminal voltage and temperature of the supercapacitor bank, predicts the remaining charge through iterative state equations, corrects the prediction error using the terminal voltage measurement, and outputs the optimized remaining charge.

8. The pulsed ultra-low power energy management system based on supercapacitors according to claim 7, characterized in that: The execution of the charge redistribution strategy includes the following steps: When the impedance matching completion flag signal is equal to 1 and the pulse interval time is greater than the minimum charge migration time threshold, charge migration is initiated, and the single-cell capacitor with excessive charge is identified and used as a charge source. The charge source transfers energy to the single-cell capacitor with insufficient charge through a bidirectional DC-DC circuit. After the migration is completed, the charge status flag of all single cells is updated.

9. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: The setting of the minimum charge migration time threshold includes: The minimum charge migration time threshold is calculated by adding the system response delay time and the safety margin time to the base time required for a single charge migration based on the target frequency value. The response delay time is the fixed time required for system hardware switching. The base time required is determined by the target frequency value, which is the theoretical shortest cycle of charge migration. The safety margin time is the buffer time reserved to cope with sudden load fluctuations.

10. The pulsed ultra-low power energy management system based on supercapacitors according to claim 1, characterized in that: It also includes cross-unit collaboration steps: After completing the charge redistribution, the charge state compensation unit (3) sends a spectrum resampling command to the pulse spectrum analysis unit (1). The pulse spectrum analysis unit (1) starts a new round of current waveform acquisition according to the command. The capacitor bank dynamic tuning unit (2) receives the new target frequency value and refreshes the frequency-capacitance mapping table priority.

Images