DAB converter low circulating current regulation method and system facing power-rich area

Abstract

The present application relates to a kind of DAB converter low circulation regulation method and system for power enrichment area, belong to new energy direct current conversion and energy storage access technical field.The present application constructs multi-time scale collaborative regulation architecture.First, according to the nominal phase shift amount of power instruction calculation, and generate a limited number of candidate phase shift amount in its neighborhood;Second, based on the comprehensive cost function including output voltage error, circulating reactive power, inductance current stress and phase shift variation, the optimal fast phase shift adjustment amount is selected;Then, introduce fractional order integral-resonance corrector for compensation, superimposed final phase shift instruction is output;Finally, only when the working condition deviates or performance continues to degrade, trigger the patrol-group collaborative optimization mechanism to slowly self-tuning control parameters.The present application effectively reduces internal circulating current and calculation burden, improves the operating efficiency and stability of converter under wide voltage fluctuation scenario, applicable to photovoltaic, energy storage and electric vehicle charging and discharging scene.

Description

Technical Field

[0001] This invention belongs to the field of new energy DC-DC conversion and energy storage access technology, specifically relating to a low circulating current control method and system for DAB converters in power-rich areas. Background Technology

[0002] In areas rich in distributed power sources, rooftop solar PV, centralized solar PV, distributed wind power, user-side energy storage, and electric vehicle charging facilities are often connected to the distribution network simultaneously. Affected by sunlight, wind speed, load start-up and shutdown, and charging / discharging demands, local power output may fluctuate significantly within a short period. Traditional power distribution equipment primarily handles unidirectional, slowly changing power. However, these areas may experience reverse power supply at midday or power shortages in the evening or during peak charging times. Therefore, a DC-DC converter interface capable of bidirectional adjustment, electrical isolation, and modular expansion is required.

[0003] A dual active bridge (DAB) converter consists of a primary-side full-bridge, a secondary-side full-bridge, a high-frequency transformer, and its equivalent leakage inductance. The two full-bridges each generate a high-frequency square wave voltage. By adjusting the phase shift angle between the two square waves, the magnitude and direction of the transmitted power can be changed. This type of converter features bidirectional power transmission, electrical isolation, high power density, and high modularity, making it suitable for connecting distributed power source DC buses to energy storage batteries, charging loads, or DC buses of another voltage level.

[0004] However, the input voltage and power command in the distributed power source enrichment area change frequently, and the transformer leakage inductance, device voltage drop, bus capacitance, and load also change with temperature, aging, and operating point. If fixed parameters and fixed phase shift rules are still used, circulating currents can easily appear between the two full-bridge circuits, which only travel back and forth internally without doing work on the load. Circulating currents will increase the root mean square value of high-frequency inductor current, increase losses in switching devices, transformers, and conductors, and reduce the efficiency of the equipment under light loads and over a wide voltage range.

[0005] Existing dual active bridge converters typically employ single-phase-shift modulation. The controller compares the output voltage setpoint with the sampled value, and then obtains the phase shift command through a proportional-integral (PI) regulator. To improve the dynamic process, some schemes utilize the mathematical model of the dual active bridge to calculate the output voltage for the next control cycle based on the current voltage, current, and load conditions, and select the phase shift value with the smallest voltage error from several options. Other schemes employ fractional-order proportional, integral, or resonant regulators to improve adaptability to periodic disturbances and parameter variations.

[0006] For regulator parameters, common practices include offline tuning based on engineering experience, parameter sweeping, or a single particle swarm optimization algorithm. Some schemes directly incorporate population search algorithms into the real-time control loop, iterating repeatedly within each control cycle to simultaneously reduce voltage error and reactive power. Existing research has also attempted to combine fractional-order resonant correction, model-based phase shift selection, and mongoose search and particle swarm optimization to reduce reactive power within dual active bridges and improve conversion efficiency.

[0007] The existing technology has the following drawbacks:

[0008] First, many control schemes are selected based solely on output voltage error. Even if the output voltage tracks the given value, there may still be significant leakage current, circulating reactive power, and sudden phase shifts, thus failing to directly guarantee low-loss operation.

[0009] Second, fixed-parameter regulators typically perform well only around a rated operating point. When there are significant changes in the voltage on the distributed power source side, the voltage on the energy storage side, the load power, or the temperature, the original parameters may cause a slower response, increased overshoot, or an increase in high-frequency current.

[0010] Third, if complex population optimization is placed directly within the switching control cycle, the calculation time is difficult to determine, which may lead to control delay; if only offline tuning is performed once, it cannot adapt to the long-term multi-condition changes in the distributed power source rich area.

[0011] Fourth, in existing solutions, rapid phase shift adjustment, fractional-order error correction, and parameter optimization are often independent of each other, lacking clear data interfaces and time levels. When the controller experiences performance degradation, it is difficult to determine whether to change the phase shift, change the regulator parameters, or switch to another set of operating parameters.

[0012] Fifth, some papers lack consistent definitions of variables, phase shift angle normalization methods, and reactive power evaluation criteria, which hinders engineering implementation. This invention unifies primary-side calculation relationships, phase shift ranges, power directions, and performance indicators, enabling the control process to be directly deployed into digital controllers. Summary of the Invention

[0013] The purpose of this invention is to provide a low circulating current control method and system for DAB converters in power-rich regions, addressing the problems of large internal circulating reactive power, high high-frequency inductor current stress, and large computational burden for online tuning of composite control parameters in dual active bridge converters under conditions of frequent power fluctuations in photovoltaic, wind power, energy storage, and charging loads. This invention achieves synergistic suppression of circulating reactive power, high-frequency inductor current, and phase shift fluctuations while ensuring fast and accurate tracking of output voltage or transmitted power. Furthermore, it reduces the computational burden of real-time control by rationally allocating control tasks across different time scales.

[0014] To achieve the above objectives, the technical solution of the present invention is: a low circulating current self-tuning control method for DAB converters oriented towards power-rich regions, comprising:

[0015] Obtain the primary side DC voltage, secondary side DC voltage, high-frequency leakage inductance current, and secondary side load current of the dual active bridge converter module;

[0016] Calculate the nominal phase shift based on the current power command, primary DC voltage, and secondary DC voltage;

[0017] Based on the nominal phase shift, a local phase shift candidate set including a finite number of candidate values ​​is generated within a preset offset interval;

[0018] Based on each candidate phase shift in the local phase shift candidate set, the secondary side voltage, circulating current reactive power evaluation, root mean square value of leakage inductance current, and phase shift change are predicted for the next control cycle.

[0019] The comprehensive cost function of each candidate phase shift is calculated based on the secondary voltage, circulating current reactive power evaluation quantity, root mean square value of leakage inductance current and phase shift change quantity. The candidate phase shift quantity with the smallest comprehensive cost is selected as the fast phase shift adjustment quantity.

[0020] Obtain the phase shift correction value output by the fractional integral-resonant corrector, superimpose the rapid phase shift adjustment value and the phase shift correction value, and then perform amplitude limiting to generate the final phase shift command;

[0021] The final phase shift command is output to the phase shift modulation and drive module, which generates a drive pulse based on the final phase shift command and outputs it to the switching device of the dual active bridge converter module.

[0022] Monitor the current operating conditions. When the deviation of the operating conditions or the comprehensive performance index within the preset time window meets the triggering conditions, start the slow self-tuning unit to perform parameter optimization and write the optimization results into the operating condition parameter table.

[0023] Furthermore, the calculation of the nominal phase shift based on the current power command, primary-side DC voltage, and secondary-side DC voltage specifically includes:

[0024] In bidirectional power transfer mode, using the formula:

[0025]

[0026] The nominal phase shift was calculated. ,in, For switching frequency, To calculate the equivalent leakage inductance on the primary side, The transformer turns ratio This is the sampled value of the primary side DC voltage. This is the sampled value of the secondary side DC voltage. This is the current power command, and sgn(⋅) is the sign function.

[0027] Furthermore, the local phase shift candidate set is generated as follows:

[0028]

[0029] in, For nominal phase shift, For candidate intervals, denoted as the number of candidates on one side, and sat(⋅) as the phase shift limiting function. The local phase shift candidate set contains 2M+1 candidate values.

[0030] Furthermore, the formula for calculating the comprehensive cost function is as follows:

[0031]

[0032] in, The given value for the secondary voltage. The secondary side voltage of the next cycle is predicted based on the m-th candidate phase shift. This is the circulating reactive power evaluation quantity calculated based on the m-th candidate phase shift. This is the root mean square value of the leakage inductance current calculated based on the m-th candidate phase shift. For the m-th candidate phase shift, This is the final phase shift command of the previous control cycle. , , and These are the preset voltage tracking weight, circulating current reactive power weight, current stress weight, and phase shift change weight, respectively.

[0033] Furthermore, the transfer function of the fractional-order integral-resonant corrector is:

[0034]

[0035] in, For proportional gain, For fractional integral gain, For resonant gain, and For fractional order, The target perturbation angular frequency, Let be the resonant bandwidth, and s be the complex variable of the Laplace transform.

[0036] Furthermore, the superposition of the rapid phase shift adjustment amount and the phase shift correction amount, followed by amplitude limiting, specifically includes:

[0037] According to the formula Superposition and amplitude limiting are performed, among which, For the final phase shift command, For rapid phase shift adjustment, This is the phase shift correction amount. For the correction fusion coefficient, The maximum permissible normalized phase shift is given by sat(⋅), where sat(⋅) is the phase shift limiting function.

[0038] And a single-cycle rate of change constraint is applied to the superimposed phase shift command: ,in This is the preset maximum single-cycle phase shift change.

[0039] Furthermore, the startup of the slow self-tuning unit to perform parameter optimization specifically includes the following steps:

[0040] Initialize a parameter population containing N candidate individuals, with each individual corresponding to a set of control parameter vectors to be optimized. , For proportional gain, For fractional integral gain, For resonant gain, and For fractional order, , , and These are the preset voltage tracking weight, circulating current reactive power weight, current stress weight, and phase shift change weight, respectively.

[0041] In each iteration, based on the comparison between the sentinel random number and the search threshold, a roaming search or an avoidance search is selected to update the individual position and maintain population diversity.

[0042] The convergence calculation of individual positions is achieved by utilizing a particle swarm velocity and position co-update mechanism;

[0043] For failed individuals that repeatedly exceed the limit, experience performance degradation, or are less than a preset threshold distance, a rebirth operation is performed to generate a new individual near the global optimal position or within the full parameter space to replace the original failed individual.

[0044] The iteration stops when the maximum number of iterations is reached or the global optimal performance improvement is less than a preset threshold, and the optimal parameter set is output.

[0045] Furthermore, the monitoring of the current operating status specifically includes:

[0046] The configuration includes a condition vector comprising primary-side DC voltage, secondary-side DC voltage, power command, root mean square value of leakage inductance current, circulating current reactive power evaluation, and device temperature. ;

[0047] Calculate the current operating condition vector relative to the current operating condition center vector. Deviation of operating conditions , For the j-th component of the current working condition vector, Let j be the j-th component of the current working condition center vector. To prevent small positive numbers with a denominator of zero;

[0048] When the operating condition deviates Continuously exceeding the preset threshold Or, the overall performance index in the running data window continuously exceeds the preset threshold. When this occurs, the slow self-tuning unit is triggered.

[0049] Furthermore, after writing the optimization results into the operating condition parameter table, a parameter smoothing migration step is also included:

[0050] When the controller calls the parameter groups for different operating condition zones based on the operating condition identification results, it uses the formula... Perform parameter transition, where, This is the currently used control parameter group. For the old parameter set, For the newly called parameter group, This is the transition coefficient, which increases linearly from 0 to 1 within a preset transition time.

[0051] The present invention also provides a low circulating current self-tuning control system for DAB converters oriented towards power-rich regions, which performs the method described above and includes: a sampling module, a multi-time-scale low circulating current controller, a phase-shift modulation and driving module, and a dual active bridge conversion module.

[0052] The sampling module is used to collect the primary side DC voltage, secondary side DC voltage, high-frequency leakage inductance current and load current, and send the sampling signal to the multi-time-scale low circulating current controller.

[0053] The multi-timescale low-circulation current controller includes an operating condition identification unit, a fast phase shift adjustment unit, a fractional-order resonance correction unit, a slow self-tuning unit, a command fusion and constraint unit, and an operating condition parameter table.

[0054] The operating condition identification unit is used to identify the current operating condition based on the sampled signal and call the parameter group in the operating condition parameter table;

[0055] The fast phase shift adjustment unit is used to perform the steps of constructing a local phase shift candidate set, calculating the comprehensive cost, and selecting the fast phase shift adjustment amount.

[0056] The fractional-order resonance correction unit is used to generate phase shift correction amount;

[0057] The instruction fusion and constraint unit is used to superimpose the rapid phase shift adjustment amount and the phase shift correction amount, and output the final phase shift instruction after performing amplitude limiting and rate of change constraints.

[0058] The slow self-tuning unit is used to perform the parameter optimization steps of sentinel search and group cooperative iteration;

[0059] The phase-shift modulation and driving module is used to generate driving pulses according to the final phase-shift command to control the switching devices in the dual active bridge converter module.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1. This invention incorporates voltage tracking, circulating reactive power, current stress, and phase shift smoothness into the same candidate evaluation index. Compared to schemes that only compare output voltage error, this avoids maintaining a large internal round-trip current even when the voltage already meets the requirements.

[0062] 2. The fast layer evaluates only a finite number of candidate values ​​near the nominal phase shift, reducing the computational load from full-range search to a fixed small-scale computation; the slow parameter optimization is triggered on demand, avoiding population iteration in every switching cycle, which makes it easy to implement on digital signal processors or microcontrollers.

[0063] 3. The fractional-order integral-resonant corrector simultaneously handles slow-varying deviations and periodic disturbances, adapting to voltage fluctuations on the distributed power source side, voltage changes at the energy storage end, and the start-up and shutdown of charging loads. Through fusion coefficients and phase shift rate constraints, abrupt changes in control variables can be reduced.

[0064] 4. Sentinel patrol, avoidance, and regeneration of failed individuals are used to maintain the diversity of the parameter population, while particle velocity and position updates are used to accelerate local convergence. The combination of the two can reduce the risk of premature convergence of a single particle swarm.

[0065] 5. Operating parameter tables, hysteresis switching, and smooth parameter transitions enable the controller to cover wide input voltage, bidirectional power, and multiple load conditions, making it suitable for long-term multi-condition operation in areas with abundant distributed power sources. The multi-module power distribution method can also reduce the overload risk of individual devices based on temperature and available capacity. Attached Figure Description

[0066] Figure 1 This is a system application architecture diagram of the present invention in a distributed power source rich area.

[0067] Figure 2 This is a schematic diagram of the main circuit of the dual active bridge converter module used in this invention.

[0068] Figure 3 This is a structural block diagram of the multi-timescale low-circulation controller of the present invention.

[0069] Figure 4 This is the overall flowchart of the low-circulation self-tuning control method of the present invention.

[0070] Figure 5 This is a flowchart of the sentinel-group cooperative parameter optimization process of the present invention. Detailed Implementation

[0071] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0072] This invention provides a low circulating current self-tuning control method and system for DAB converters in areas with abundant power sources. It is applicable to distributed power source areas, such as scenarios where rooftop photovoltaics, centralized photovoltaics, distributed wind power, user-side energy storage, and electric vehicle charging facilities are simultaneously connected to the distribution network.

[0073] like Figure 1 The diagram shown illustrates the system application architecture of this invention in a distributed power source rich area. The system includes a distributed power source unit 110, a local DC bus 120, a dual active bridge converter module 130, an energy storage or DC load 140, a sampling module 150, a multi-timescale low circulating current regulator 160, a phase-shifting modulation and drive module 170, and an optional area energy management interface 180.

[0074] The distributed power supply unit 110 may include a photovoltaic array, a wind power rectifier unit, a fuel cell, user-side energy storage, or any combination of the above devices. A local DC bus 120 is used to collect the distributed power supply output and provide primary-side DC voltage to the dual active bridge converter module 130.

[0075] like Figure 2 As shown, the dual active bridge converter module 130 includes a primary-side full bridge 131, a high-frequency transformer and equivalent leakage inductance unit 132, a secondary-side full bridge 133, a primary-side bus capacitor, and a secondary-side bus capacitor. The primary-side full bridge 131 is connected to the local DC bus 120; the secondary-side full bridge 133 is connected to energy storage, electric vehicle charging / discharging ports, DC loads, or a DC bus of another voltage level. Both the primary-side full bridge 131 and the secondary-side full bridge 133 consist of four controllable switching devices, with the upper and lower switches of each bridge arm complementarily conducting and having a dead time set.

[0076] The sampling module 150 collects at least the primary side DC voltage, secondary side DC voltage, high-frequency leakage inductance current, and secondary side load current. Depending on protection and optimization needs, it can also collect switching device temperature, transformer temperature, primary side current, secondary side current, insulation status, and energy storage state of charge. The sampled data, after isolation, conditioning, and analog-to-digital conversion, are sent to the multi-timescale low-circulating current controller 160.

[0077] The multi-timescale low-circulating current controller 160 includes an operating condition identification unit 161, a fast phase shift adjustment unit 162, a fractional-order resonance correction unit 163, a slow self-tuning unit 164, an instruction fusion and constraint unit 165, an operating data window unit 166, and an operating condition parameter table 167. The controller 160 can be implemented by a digital signal processor, a microcontroller, a field-programmable gate array, a real-time controller, or a combination of the above devices.

[0078] The phase-shift modulation and drive module 170 receives the final phase-shift command, generates the complementary drive pulses required for the primary-side full bridge 131 and the secondary-side full bridge 133, and performs dead-time insertion, pulse blocking, overcurrent protection, and fault reset. The area energy management interface 180 is used to receive charging power, discharging power, bus voltage setpoints, or power direction commands given by the upper layer, and can also upload available capacity, temperature, and fault status to the upper layer.

[0079] The DC output terminal of the distributed power unit 110 is connected to the local DC bus 120, which is connected to the primary DC terminal of the dual active bridge converter module 130. The secondary DC terminal of the dual active bridge converter module 130 is connected to the energy storage or DC load 140. The input terminal of the sampling module 150 is connected across or in series with the aforementioned voltage and current detection positions, and its output terminal is connected to the data input terminal of the controller 160.

[0080] The input terminal of the operating condition identification unit 161 is connected to the sampling module 150 and the area energy management interface 180; the output terminal is connected to the fast phase shift adjustment unit 162, the fractional-order resonance correction unit 163, and the operating condition parameter table 167, respectively. The output terminals of the fast phase shift adjustment unit 162 and the fractional-order resonance correction unit 163 are both connected to the command fusion and constraint unit 165. The command fusion and constraint unit 165 is connected to the phase shift modulation and drive module 170, and the phase shift modulation and drive module 170 is connected to the drive terminals of each switching device of the dual active bridge converter module 130.

[0081] The running data window unit 166 records data such as voltage error, reactive power, inductor current, phase shift, efficiency, and temperature within a predetermined time window. When the operating condition identification result or window performance index meets the triggering conditions, the running data window unit 166 starts the slow self-tuning unit 164. The slow self-tuning unit 164 generates a new optimal parameter set and writes it into the operating condition parameter table 167. To prevent abrupt parameter changes, a ramp transition or weighted interpolation is used between the old and new parameters.

[0082] The system works as follows:

[0083] Two full-bridge circuits generate square wave voltages with a duty cycle of approximately 50% at the same switching frequency. A phase shift angle exists between the primary-side square wave and the secondary-side square wave referred to the primary side. When the phase shift angle is positive, energy flows from the primary side to the secondary side; when the phase shift angle is negative, energy flows from the secondary side to the primary side. As the absolute value of the phase shift angle increases, the transmitted power initially increases, but circulating current and current stress may also increase accordingly.

[0084] The controller 160 first calculates the nominal phase shift that meets the current power demand based on the power analysis relationship, and then generates several candidate values ​​only near the nominal phase shift. For each candidate value, the controller calculates the root mean square values ​​of the output voltage, circulating reactive power, and inductor current for the next control cycle based on the sampled voltage and current, and calculates the overall cost. The candidate value with the lowest cost is used as the fast phase shift adjustment.

[0085] The fractional-order resonant correction unit 163 compensates for long-term deviations in output voltage and periodic fluctuations caused by distributed power sources. This compensation amount is added to the rapid phase shift adjustment amount, and after phase shift range limitation, single-cycle change rate limitation, and power direction check, it is converted into a drive pulse by the phase shift modulation and drive module 170.

[0086] The slow self-tuning unit 164 does not participate in the repetitive calculations for each switching cycle. Parameter optimization is only initiated when the input voltage, output voltage, or power command crosses the preset operating condition boundaries, or when the voltage error, circulating reactive power, inductor current, and efficiency exceed thresholds within a continuous time window. This allows it to follow long-term operating condition changes in areas rich in distributed power sources without causing the real-time control cycle to time out due to iterative calculations.

[0087] like Figure 4 The diagram shown is the overall flowchart of the low circulating current self-tuning control method of the present invention. First, the variables are defined and the power relationship of the dual active bridge is analyzed:

[0088] Let the primary DC voltage be The secondary side DC voltage is The turns ratio of the primary side to the secondary side of the high-frequency transformer is The equivalent leakage inductance converted to the primary side is: The switching frequency is angular frequency is The phase shift angle is .in:

[0089]

[0090] Define normalized phase shift for:

[0091]

[0092] Under single-phase-shift modulation, neglecting device voltage drop and excitation current, the average transmission power of the dual active bridge can be written as:

[0093]

[0094] In the formula, This indicates that energy flows from the primary side to the secondary side. This indicates that energy flows from the secondary side to the primary side. If a different phase shift normalization method is used, only proportional modification is needed. The definition remains the same, and the control flow remains unchanged.

[0095] Let the secondary bus capacitance be The load current is The control cycle is Based on the capacitor charge balance, the discrete update relationship of the secondary voltage is as follows:

[0096]

[0097] This relationship uses actual sampling. , and It can be updated in real time according to changes in input voltage and load. If a leakage current sensor is configured, it can also utilize... The calculation results are corrected online.

[0098] This invention discloses a low circulating current self-tuning control method for DAB converters in power-rich regions, comprising the following steps:

[0099] Step S1: Device initialization and data acquisition

[0100] The controller reads the transformer turns ratio, equivalent leakage inductance, bus capacitance, switching frequency, allowable phase shift range, allowable current, and temperature threshold, and loads the initial parameter set. Then, it operates in control cycles... collection , , , and upper-level power setpoint or voltage setpoint .

[0101] The sampled values ​​are first subjected to amplitude limiting, outlier removal, and low-delay filtering. Voltage, current, and power can be normalized to rated values, allowing devices of different power levels to share the same control process.

[0102] Step S2: Operating condition identification and parameter group retrieval

[0103] Construct the working condition vector:

[0104]

[0105] In the formula, This is the root mean square value of the leakage inductance current. To reflect the circulating reactive power evaluation quantity that reflects the internal round-trip energy, This refers to the temperature of the device or transformer. According to... The corresponding parameter group is retrieved from the operating condition parameter table 167, which specifies the input voltage range, output voltage range, and power range.

[0106] To avoid repeated switching near the operating condition boundaries, entry and exit thresholds are set for each operating condition zone, forming a hysteresis band between them. Parameter group switching is performed using:

[0107]

[0108] in, The value will gradually increase from zero to one within the predetermined transition period.

[0109] Step S3: Calculate the nominal phase shift

[0110] In positive power transmission and In the implementation method, the nominal phase shift is obtained from the average power relationship:

[0111]

[0112] In bidirectional power transmission, it can be uniformly written as:

[0113]

[0114] When the value within the square root is less than zero or the nominal phase shift exceeds the allowable range, the power command and phase shift are saturated, and the current transmittable power limit is returned to the area energy management interface 180.

[0115] Step S4: Construct a local phase shift candidate set

[0116] nominal phase shift Construct a finite candidate set for the center:

[0117]

[0118] in, For candidate intervals, The number of candidates on one side. This represents the phase shift limiting function. Typically, only three, five, or seven candidate values ​​need to be set. The value can be temporarily increased when the input voltage or power changes rapidly. Decrease in steady state To balance search range and phase shift resolution.

[0119] Step S5: Calculate candidate states and circulation indicators

[0120] For the candidate set of the th Phase shift Calculate the secondary voltage for the next control cycle:

[0121]

[0122] Based on the segmented slope of the bridge arm voltage and leakage inductance current corresponding to the candidate phase shift, the root mean square value of the candidate leakage inductance current can be obtained. Alternatively, a lookup table can be established using the measured leakage inductance current from the most recent switching cycles, and interpolation can be performed based on the phase shift and voltage ratio.

[0123] When evaluating circulating reactive power based on the difference between single-side-view power and active power, the following method can be used:

[0124]

[0125] in, This represents the active power under candidate phase shift. Alternatively, the integral of the reverse energy of the product of leakage inductance current and the voltages of both bridges within one switching cycle can be used as a circulating current index. The purpose of both methods is to evaluate the internal round-trip energy not output to the load.

[0126] Step S6: Calculate the overall cost and select the fast phase shift amount

[0127] Calculate the overall cost for each candidate phase shift:

[0128]

[0129] in, , , and These are the weights for voltage tracking, circulating current reactive power, current stress, and phase shift change, respectively. Selection:

[0130]

[0131] As a rapid phase shift adjustment, the controller will not choose a phase shift that significantly increases circulating current or current stress in order to reduce a small voltage error.

[0132] Step S7: Fractional order resonance correction

[0133] Define the output voltage error:

[0134]

[0135] One expression for a fractional integral-resonant corrector is:

[0136]

[0137] in, , and These represent proportional, fractional integral, and resonant gain, respectively. and For fractional order, The target perturbation angular frequency, This is the resonant bandwidth. Preferably, and The phase shift correction amount is obtained after discretization of the corrector. .

[0138] The final phase shift command is:

[0139]

[0140] Simultaneously satisfying the single-period rate of change constraint:

[0141]

[0142] in, This is the correction fusion coefficient. This coefficient can be appropriately reduced during large disturbances, allowing the fast layer to handle the main power regulation; and appropriately increased during steady-state conditions, enabling the fractional-order resonant corrector to eliminate residual bias and periodic fluctuations.

[0143] Step S8: Generate drive pulse and perform protection.

[0144] Phase-shifting modulation and driving module 170 according to Complementary pulses are generated on both sides of the full bridge. Upon detection of overcurrent, overvoltage, undervoltage, overtemperature, or drive fault, the phase shift is immediately frozen or reduced; the drive is blocked when a severe fault threshold is reached. After the fault is recovered, the phase shift recovers from zero or a safe value via a ramp, preventing bias magnetization and current surges in the high-frequency transformer.

[0145] Step S9: Determine if slow self-tuning is triggered.

[0146] Based on the current working condition center Based on this, define the deviation from the operating condition:

[0147]

[0148] in, To prevent small positive numbers with a denominator of zero. When exceeding the threshold continuously Or the overall performance index in the running data window continuously exceeds the threshold. When the slow self-tuning unit 164 is activated, it will be started. If it is not triggered, the next control cycle will proceed directly.

[0149] In the above method, the vector of parameters to be optimized in slow self-tuning can be written as:

[0150]

[0151] Depending on the actual controller resources, optimization can also be performed on only a subset of parameters. (The last part, "with length as...", appears to be an unrelated fragment and is left untranslated.) The data window evaluation parameter group, and the comprehensive performance function are:

[0152]

[0153] in:

[0154]

[0155]

[0156] This represents the root mean square value of the circulating reactive power within the window. This represents the root mean square value of the leakage inductance current within the window. For conversion efficiency, , , , and This is for normalized weights. If the equipment does not have the capability for online efficiency measurement, the last term can be replaced with an estimate based on conduction loss and switching loss.

[0157] like Figure 5 The diagram shown is a flowchart of the sentinel-swarm cooperative parameter optimization process of the present invention. The specific optimization steps are as follows:

[0158] First, determine the upper and lower boundaries of each parameter. and ,generate One initial candidate individual:

[0159]

[0160] in, A random vector between 0 and 1 This indicates that corresponding elements are multiplied.

[0161] Second, calculate the performance function for each candidate individual. Based on the comparison between the sentinel random number and the search threshold, choose between a roving search or an avoidance search. A roving search expands the population's coverage within the parameter space; an avoidance search moves some individuals away from the current dense region to reduce the probability of premature convergence to a local optimum.

[0162] Third, after completing the patrol or avoidance update, the particle swarm velocity and position relationship are used for cooperative convergence:

[0163]

[0164]

[0165] in, For inertial weights, and The learning coefficient, Let i be the velocity of the i-th individual in the t-th iteration. For location, For the first The historical best position of an individual To be the globally optimal position This indicates that out-of-range parameters will be projected back into the allowed set. .

[0166] Fourth, for individuals that repeatedly exceed the limit, experience performance degradation, or are too close to each other, the failed individuals are regenerated. A number of new individuals are generated in the vicinity of the global optimum and within the full parameter range, so as to maintain both local fine search and global exploration capabilities.

[0167] Fifth, recalculate and update the performance function. and The process stops when the maximum number of iterations is reached, or when the global optimal performance improvement is less than a threshold in several consecutive iterations, and the optimal parameter set is output. .

[0168] Slow self-tuning can be performed offline on a digital simulation or hardware-in-the-loop platform before equipment commissioning, or it can be performed at low frequency using a shadow model, backup processing core, or area controller during equipment operation. When executed online, new parameters are first verified for stability range, phase shift range, and current upper limit before being written into operating parameter table 167. Unverified candidate parameters are not directly used to drive power devices.

[0169] When a local DC system contains multiple parallel dual active bridge modules, the regional energy management interface 180 first allocates power commands based on the available capacity, temperature margin, and energy storage state of charge of each module. The power input for each module can be:

[0170]

[0171] in, This is the total system power command. For the first The available power of each module, To reflect the allocation coefficients of temperature, health status, and energy storage margin, The number of modules is specified. Each module still uses the local fast phase shift adjustment and slow self-tuning of this invention, so that even if communication is temporarily interrupted, it can maintain local stable operation under the most recent power setting.

[0172] When photovoltaic output increases rapidly and energy storage allows charging, the total power command is positive, and multiple dual active bridges send excess energy to the energy storage or charging end. When the local DC bus power is insufficient and energy storage allows discharging, the total power command reverses, and the dual active bridges send energy back to the distributed power source side or local bus through negative phase shift. This bidirectional operation mode can alleviate reverse power flow and short-term power fluctuations in areas with abundant distributed power sources.

[0173] Implementation Examples

[0174] This embodiment uses a distributed photovoltaic (PV) DC bus to transmit power to energy storage batteries or electric vehicle charging loads via a dual active bridge converter as an example. This scenario can be equivalently represented by an isolated bidirectional DC-DC converter interface under conditions of PV output fluctuations, energy storage voltage changes, and frequent switching of charging loads in a distributed power-rich area.

[0175] In this embodiment, the photovoltaic side consists of multiple photovoltaic modules connected in series and parallel. The open-circuit voltage of a single photovoltaic module is approximately... The short-circuit current is approximately The maximum power point voltage is approximately The maximum power point current is approximately Under high light intensity, the output voltage of the photovoltaic array is approximately... The output current is approximately The corresponding output power is approximately Therefore, this embodiment corresponds to approximately Distributed photovoltaic power conversion scenarios at various levels.

[0176] The primary side arm voltage of the dual active bridge converter is approximately The secondary side reduced voltage is approximately The voltage transformation relationship between the primary and secondary sides of a high-frequency transformer is approximately... The converter uses inter-bridge phase shifting to adjust the direction and magnitude of power transmission. The high-frequency leakage inductance current is approximately [value missing] during stable operation. to They change periodically.

[0177] The load side uses an energy storage battery or an electric vehicle power battery model, with a nominal battery voltage of approximately Rated capacity is approximately The initial state of charge is The battery model response time is set to The parameters described above are only for illustrating the embodiments of the present invention and do not constitute a limitation on the power level, voltage level, or load type of the converter.

[0178] After the controller is powered on, it reads the primary side voltage, secondary side voltage, leakage inductance parameters, current protection threshold, phase shift constraint, and the initial parameter set corresponding to each operating condition. The initial search range for the proportional coefficient, integral coefficient, resonant coefficient, and auxiliary gain in the fractional-order composite correction branch is set as follows:

[0179]

[0180] The search range for the fractional order is set as follows:

[0181]

[0182] The controller collects the primary side voltage according to the set sampling period. Secondary voltage High-frequency inductor current Load current and power command It also limits the amplitude of the sampled values ​​and judges abnormal states.

[0183] Within each control cycle, the fast phase shift adjustment layer first calculates the nominal phase shift based on the primary side voltage, secondary side voltage, leakage inductance parameters, and power command. In this embodiment, five candidate phase shifts are set around the nominal phase shift, and the candidate set is represented as follows:

[0184]

[0185] in, The interval for candidate phase shifts. All candidate phase shifts are limited to the phase shift range allowed by the converter.

[0186] For each candidate phase shift, the controller calculates the corresponding output voltage, transmitted power, circulating reactive power, root mean square value of high-frequency inductor current, and phase shift change, and evaluates them using a comprehensive cost function. The comprehensive cost function simultaneously considers output tracking error, circulating reactive power, current stress, and phase shift jitter. The candidate phase shift with the lowest cost is selected as the fast adjustment quantity for the current control cycle.

[0187] Subsequently, the rapid adjustment is superimposed on the output of the fractional-order composite correction branch. The fractional-order integrator is used to eliminate steady-state and slowly varying deviations in bus voltage or transmitted power, while the resonant circuit is used to suppress periodic fluctuations. The superimposed phase shift command, after amplitude and rate-of-change limiting, is sent to the phase-shift modulation and drive module.

[0188] During stable operation, the primary arm voltage of the dual active bridge remains at approximately [value missing]. The secondary side bridge arm voltage is maintained at approximately The primary and secondary voltages exhibit square waves with a certain phase difference, and the high-frequency inductor current is approximately... The periodic variation within the range indicates that the converter can continuously transmit power in a given direction.

[0189] To verify the dynamic adjustment capability of this invention under conditions of sudden power changes in distributed power sources, in When an input power or load disturbance is applied, the output voltage changes from approximately [a certain value]. Short-term drop to approximately The fast phase shift adjustment layer immediately recalculates the nominal phase shift and updates the local candidate set, while the fractional-order composite correction branch synchronously compensates for the remaining bias. The output voltage is approximately... Time to restore to Nearby, the recovery time after the disturbance is approximately .

[0190] During the battery load test, the initial state of charge of the battery was: The test ended at approximately The battery terminal voltage remains at approximately Within the specified range, this indicates that no significant battery terminal voltage oscillation occurred during load changes.

[0191] Under the same input voltage, output load, and power command conditions, the present invention is compared with conventional single-phase shift modulation methods and single-phase shift methods using proportional-integral regulators. The simulation results are shown in Table 1.

[0192] Table 1 Comparison of steady-state performance of different control methods

[0193]

[0194] Compared to conventional single-phase shift methods, the conversion efficiency of this invention is improved by approximately Percentage points; compared to the proportional-integral single-phase-shift method, the conversion efficiency is improved by approximately percentage points.

[0195] Comparing the absolute values ​​of circulating reactive power, this invention reduces approximately [amount missing] compared to conventional single-phase shift methods. Compared to the proportional-integral single-phase shift method, it reduces the cost by approximately The circulating reactive power of this invention is close to zero, indicating that the reactive component in the transmitted power that does not participate in effective energy exchange is significantly suppressed.

[0196] Under different load power conditions, the root mean square value of the high-frequency inductor current corresponding to this invention is lower than that of the other two methods. Simultaneously, the peak-to-peak value of the inductor current and the current ripple obtained by this invention are smaller, indicating that the constraints of the comprehensive cost function on the internal circulating current and current stress can reduce unnecessary high-frequency current.

[0197] When the input voltage, output voltage, or load power exceeds the preset operating condition boundary, or when the circulating reactive power, the root mean square value of the inductor current, and the output error exceed the threshold within a continuous time window, the controller activates the slow parameter self-tuning layer.

[0198] The slow parameter self-tuning layer employs a parameter optimization mechanism combining sentinel search and particle swarm optimization. Under the same objective function and termination conditions, the parameter optimization mechanism employed in this invention achieves optimal performance in the [missing information - likely a specific step or step]. The minimum fitness is reached around the [number]th iteration; among the comparative methods, the Green Anaconda optimization method reaches the minimum fitness around the [number]th iteration. To achieve the desired result in approximately [number] iterations, the particle swarm optimization method requires [number] iterations. In the next iteration, the gray wolf optimization method requires approximately The next iteration.

[0199] Therefore, compared with the three comparison methods mentioned above, the number of iterations required to achieve the target fitness in this invention is reduced by approximately [number missing]. , and After optimization, the new parameters are first verified for stability range, current upper limit and phase shift boundary, and then written into the operating parameter table through weighted interpolation or ramp transition method to avoid parameter abrupt changes affecting converter operation.

[0200] Under the experimental and loss statistics conditions described in this invention, the total power loss of the dual active bridge converter is approximately The switching loss is approximately The measured peak conversion efficiency is approximately The experiment showed that the phase shift relationship between the primary and secondary bridge arm voltages and the corresponding high-frequency inductor current waveform could be observed. When the load power changed abruptly, the controller adjusted the phase shift to change the transmission power accordingly, and the DC side voltage only fluctuated briefly before stabilizing.

[0201] It should be noted that the aforementioned The efficiency results are shown in the simulation comparison under the working conditions of this invention. The efficiency results are for the experimental and loss statistics conditions of this invention. The test conditions and loss statistics methods are different for the two, and they should not be directly converted as data under the same conditions.

[0202] In summary, this embodiment demonstrates that the present invention can maintain rapid tracking of output voltage and transmission power when distributed photovoltaic power output and energy storage load change, and significantly reduce circulating reactive power and high-frequency inductor current. By configuring fast phase shift selection, fractional-order composite correction, and slow parameter self-tuning on different time scales, it can also avoid the complex parameter optimization process occupying each switching control cycle.

[0203] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A low circulating current self-tuning control method for DAB converters oriented towards power-rich regions, characterized in that, include: Obtain the primary side DC voltage, secondary side DC voltage, high-frequency leakage inductance current, and secondary side load current of the dual active bridge converter module; Calculate the nominal phase shift based on the current power command, primary DC voltage, and secondary DC voltage; Based on the nominal phase shift, a local phase shift candidate set including a finite number of candidate values ​​is generated within a preset offset interval; Based on each candidate phase shift in the local phase shift candidate set, the secondary side voltage, circulating current reactive power evaluation, root mean square value of leakage inductance current, and phase shift change are predicted for the next control cycle. The comprehensive cost function of each candidate phase shift is calculated based on the secondary voltage, circulating current reactive power evaluation quantity, root mean square value of leakage inductance current and phase shift change quantity. The candidate phase shift quantity with the smallest comprehensive cost is selected as the fast phase shift adjustment quantity. Obtain the phase shift correction value output by the fractional integral-resonant corrector, superimpose the rapid phase shift adjustment value and the phase shift correction value, and then perform amplitude limiting to generate the final phase shift command; The final phase shift command is output to the phase shift modulation and drive module, which generates a drive pulse based on the final phase shift command and outputs it to the switching device of the dual active bridge converter module. Monitor the current operating conditions. When the deviation of the operating conditions or the comprehensive performance index within the preset time window meets the triggering conditions, start the slow self-tuning unit to perform parameter optimization and write the optimization results into the operating condition parameter table.

2. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The calculation of the nominal phase shift based on the current power command, primary-side DC voltage, and secondary-side DC voltage specifically includes: In bidirectional power transfer mode, using the formula: The nominal phase shift was calculated. ,in, For switching frequency, To calculate the equivalent leakage inductance on the primary side, The transformer turns ratio This is the sampled value of the primary side DC voltage. This is the sampled value of the secondary side DC voltage. This is the current power command, and sgn(⋅) is the sign function.

3. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The method for generating the local phase shift candidate set is as follows: in, For nominal phase shift, For candidate intervals, denoted as the number of candidates on one side, and sat(⋅) as the phase shift limiting function. The local phase shift candidate set contains 2M+1 candidate values.

4. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The formula for calculating the comprehensive cost function is as follows: in, The given value for the secondary voltage. The secondary side voltage of the next cycle is predicted based on the m-th candidate phase shift. This is the circulating reactive power evaluation quantity calculated based on the m-th candidate phase shift. This is the root mean square value of the leakage inductance current calculated based on the m-th candidate phase shift. For the m-th candidate phase shift, This is the final phase shift command of the previous control cycle. , , and These are the preset voltage tracking weight, circulating current reactive power weight, current stress weight, and phase shift change weight, respectively.

5. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The transfer function of the fractional-order integral-resonance compensator is: in, For proportional gain, For fractional integral gain, For resonant gain, and For fractional order, The target perturbation angular frequency, Let be the resonant bandwidth, and s be the complex variable of the Laplace transform.

6. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The process of superimposing the rapid phase shift adjustment amount and the phase shift correction amount and then performing amplitude limiting specifically includes: According to the formula Superposition and amplitude limiting are performed, among which, For the final phase shift command, For rapid phase shift adjustment, This is the phase shift correction amount. For the correction fusion coefficient, The maximum permissible normalized phase shift is given by sat(⋅), where sat(⋅) is the phase shift limiting function. And a single-cycle rate of change constraint is applied to the superimposed phase shift command: ,in This is the preset maximum single-cycle phase shift change.

7. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The process of starting the slow self-tuning unit to perform parameter optimization includes the following steps: Initialize a parameter population containing N candidate individuals, with each individual corresponding to a set of control parameter vectors to be optimized. , For proportional gain, For fractional integral gain, For resonant gain, For the correction fusion coefficient, and For fractional order, , , and These are the preset voltage tracking weight, circulating current reactive power weight, current stress weight, and phase shift change weight, respectively. In each iteration, based on the comparison between the sentinel random number and the search threshold, a roaming search or an avoidance search is selected to update the individual position and maintain population diversity. The convergence calculation of individual positions is achieved by utilizing a particle swarm velocity and position co-update mechanism; For failed individuals that repeatedly exceed the limit, experience performance degradation, or are less than a preset threshold distance, a rebirth operation is performed to generate a new individual near the global optimal position or within the full parameter space to replace the original failed individual. The iteration stops when the maximum number of iterations is reached or the global optimal performance improvement is less than a preset threshold, and the optimal parameter set is output.

8. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, The monitoring of the current operating status specifically includes: The configuration includes a condition vector comprising primary-side DC voltage, secondary-side DC voltage, power command, root mean square value of leakage inductance current, circulating current reactive power evaluation, and device temperature. ; Calculate the current operating condition vector relative to the current operating condition center vector. Deviation of operating conditions , For the j-th component of the current working condition vector, Let j be the j-th component of the current working condition center vector. To prevent small positive numbers with a denominator of zero; When the operating condition deviates Continuously exceeding the preset threshold Or the overall performance index in the running data window continuously exceeds the preset threshold. When this occurs, the slow self-tuning unit is triggered.

9. The low circulating current self-tuning control method for DAB converters oriented towards power-rich regions according to claim 1, characterized in that, After writing the optimization results into the operating condition parameter table, the parameter smoothing migration step is also included: When the controller calls the parameter groups for different operating condition zones based on the operating condition identification results, it uses the formula... Perform parameter transition, where, This is the currently used control parameter group. For the old parameter set, For the newly called parameter group, This is the transition coefficient, which increases linearly from 0 to 1 within a preset transition time.

10. A low circulating current self-tuning control system for a DAB converter oriented towards a power-rich region, executing the method as described in any one of claims 1-9, characterized in that, include: Sampling module, multi-timescale low-circulating current controller, phase-shifting modulation and driving module, and dual active bridge converter module; The sampling module is used to collect the primary side DC voltage, secondary side DC voltage, high-frequency leakage inductance current and load current, and send the sampling signal to the multi-time-scale low circulating current controller. The multi-timescale low-circulation current controller includes an operating condition identification unit, a fast phase shift adjustment unit, a fractional-order resonance correction unit, a slow self-tuning unit, a command fusion and constraint unit, and an operating condition parameter table. The operating condition identification unit is used to identify the current operating condition based on the sampled signal and call the parameter group in the operating condition parameter table; The fast phase shift adjustment unit is used to perform the steps of constructing a local phase shift candidate set, calculating the comprehensive cost, and selecting the fast phase shift adjustment amount. The fractional-order resonance correction unit is used to generate phase shift correction amount; The instruction fusion and constraint unit is used to superimpose the rapid phase shift adjustment amount and the phase shift correction amount, and output the final phase shift instruction after performing amplitude limiting and rate of change constraints. The slow self-tuning unit is used to perform the parameter optimization steps of sentinel search and group cooperative iteration; The phase-shift modulation and driving module is used to generate driving pulses according to the final phase-shift command to control the switching devices in the dual active bridge converter module.

Images