Control method and system of bidirectional dc-dc converter

Abstract

The application discloses a control method and system of a bidirectional DCDC converter, and relates to the technical field of bidirectional DCDC converter control. The method comprises the following steps: constructing a first parameter influence curve set and a second parameter influence curve set based on continuous time periods of historical first state data and historical second state data; constructing a transient influence relationship based on the first parameter influence curve set, the second parameter influence curve set and historical third state data according to time sequence correlation; calling a transient influence prediction value based on the matching result of the current state data of the bidirectional DCDC converter and the first parameter influence curve set and the second parameter influence curve set; and obtaining a bidirectional DCDC converter control strategy based on the transient influence prediction value, the transient influence relationship and the current state data according to minimum loss. The application has the beneficial effect that the influence of smooth transition and transient adjustment is balanced through minimum loss, and bidirectional balance of smooth transition and instantaneous switching is realized.

Description

Technical Field

[0001] This application relates to the field of bidirectional DC-DC converter control technology, and in particular to a control method and system for bidirectional DC-DC converters. Background Technology

[0002] A bidirectional DC-DC converter is a power electronic device that supports bidirectional energy flow and is widely used in power supply equipment, energy storage systems, electric vehicles, renewable energy, and other scenarios. In these applications, bidirectional DC-DC converters need to achieve flexible switching between charging and discharging modes to adapt to complex operating conditions, while also ensuring operational stability and reducing device losses throughout the entire operating cycle to extend their service life as much as possible.

[0003] However, in related technologies, it is difficult to balance operational stability and reduce device operating losses. For example, in PID control, PID parameters rely on empirical tuning, making it difficult to adapt to the dynamic operating conditions of bidirectional DC-DC converters. In predictive control technology, although dynamic changes in the control strategy are achieved through multi-objective optimization, the multi-objective optimization model is trained based on overall operating parameters. This leads to power device losses caused by transient switching being included in steady-state operating losses, causing prediction bias and ultimately resulting in poor adaptability of the control strategy.

[0004] The patent, "A Control Method, System, and Storage Medium for a Buck-Boost Bidirectional DC-DC Converter," publication number CN119582616A, published on March 7, 2025, specifically discloses the following methods: acquiring the DC bus voltage and battery voltage of the converter; acquiring the converter's switching frequency, dead time, and bus voltage, and calculating hysteresis parameters based on the switching frequency, dead time, and bus voltage; obtaining the correlation between the DC bus voltage and battery voltage based on the hysteresis parameters, and determining the required operating mode of the converter based on the correlation between the DC bus voltage and battery voltage. The operating modes include BUCK mode, BOOST mode, and upper transistor shoot-through mode; and adjusting the PWM waveform of the driving switching transistor to control the operation of the corresponding operating mode. This solution relies on the calculation of hysteresis parameters, and the control objective is the correctness of the control mode. It cannot reduce the impact of control parameter deviation and transient switching losses during actual control.

[0005] The patent "A Startup Control Method and System for a Bidirectional DC-DC Converter", publication number CN116760290A, publication date: September 15, 2023, specifically discloses the following steps for judging the startup command: After the bidirectional DC-DC converter startup control circuit outputs a DC-DC startup command to pre-start the DC-DC converter, the startup command is judged to obtain a first judgment result; the output voltage value comparison step: the startup voltage of the DC-DC converter is compared with the system startup voltage threshold, a DC-DC operation flag is set according to the first comparison result, and the startup program corresponding to the DC-DC operation flag is executed to drive the DC-DC converter; the pulse signal blocking control step: the output voltage value of the driven DC-DC converter is compared with the system startup voltage threshold to obtain a second comparison result, or the startup command is judged to be equal to 0 to obtain a second judgment result, and the pulse signal is blocked according to the second comparison result or the second judgment result. However, this scheme only aims at whether the startup is successful and whether the pulse is safe, and cannot reduce the impact of control parameter deviation and transient switching losses during actual control. Summary of the Invention

[0006] This application addresses the problem in existing technologies for controlling bidirectional DC-DC converters that cannot simultaneously ensure operational stability and reduce device operating losses. It provides a control method and system for bidirectional DC-DC converters. Based on different operating data from the power supply, charging, and switching states of the bidirectional DC-DC converter, a first parameter influence curve is constructed for the power supply state, and a second parameter influence curve is constructed for the charging state. Furthermore, transient impact calculations are performed under the switching state based on the temporal continuity of the states. This avoids ignoring transient moments, preventing instantaneous switching losses from being included in the stabilization adjustment process. The accuracy of calculating influencing factors under both steady-state and transient conditions is improved, allowing for early prediction of transient risks. By balancing the impact of smooth transition and transient adjustment through minimum loss, the losses caused by transient switching are reduced, while excessive smooth transition losses are avoided, achieving a bidirectional balance between smooth transition and instantaneous switching.

[0007] To achieve the above technical objectives, this application provides a technical solution: a control method for a bidirectional DC-DC converter, comprising the following steps: acquiring historical first-state data based on the power supply state of the bidirectional DC-DC converter, acquiring historical second-state data based on the charging state of the bidirectional DC-DC converter, and acquiring historical third-state data based on the switching state of the bidirectional DC-DC converter; constructing a first parameter influence curve set based on continuous time periods of the historical first-state data, and constructing a second parameter influence curve set based on continuous time periods of the historical second-state data; constructing a transient influence relationship based on time-series correlation using the first parameter influence curve set, the second parameter influence curve set, and the historical third-state data; retrieving a transient influence prediction value based on the matching result between the current state data of the bidirectional DC-DC converter and the first and second parameter influence curve sets; and acquiring a bidirectional DC-DC converter control strategy based on minimum loss using the transient influence prediction value, the transient influence relationship, and the current state data, and executing the control of the bidirectional DC-DC converter using the bidirectional DC-DC converter control strategy.

[0008] Furthermore, the construction of transient influence relationships based on time-series correlation using the first parameter influence curve set, the second parameter influence curve set, and historical third-state data includes: constructing a transient calculation set using four parameter influence curves of arbitrary continuous time series, each transient calculation set containing two first parameter influence curves, two second parameter influence curves, and three historical third-state data; constructing a first curve deviation using the two first parameter influence curves, constructing a second curve deviation using the two second parameter influence curves; and constructing a transient influence relationship using the first curve deviation, the second curve deviation, and the historical third-state data.

[0009] Furthermore, the construction of the first curve deviation using two first parameter influence curves and the construction of the second curve deviation using two second parameter influence curves includes: discretizing the first parameter influence curve and the second parameter curve based on a fixed step size, performing difference processing to obtain the first parameter grid points and the second parameter grid points, calculating the difference between the first parameter grid points and the second parameter grid points based on the point-by-point deviation, and performing feature extraction on the difference between the first parameter grid points and the second parameter grid points to obtain the first curve deviation and the second curve deviation.

[0010] Furthermore, the step of extracting features from the differences in the first parameter grid points and the differences in the second parameter grid points to obtain the first curve deviation and the second curve deviation includes: extracting features from the differences in the first parameter grid points and the differences in the second parameter grid points to obtain difference features; calculating the difference average and extracting extreme difference for the difference features to obtain the average deviation and extreme difference, using the average difference and extreme difference as the global deviation of the first curve deviation and the global deviation of the second curve deviation; performing cluster analysis on the difference features to obtain clusters with different degrees of difference, using the clusters as the local deviation of the first curve deviation and the local deviation of the second curve deviation.

[0011] Furthermore, the construction of transient influence relationship based on first curve deviation, second curve deviation and historical third state data includes: obtaining common historical third state data of first curve deviation and second curve deviation, and constructing transient influence relationship based on the similarity features of first curve deviation and second curve deviation and common historical third state data.

[0012] Furthermore, the construction of transient influence relationships based on the similarity features of the first curve deviation and the second curve deviation and the common historical third state data includes: obtaining similar clustering features of the first curve deviation and the second curve deviation based on clustering similarity and a preset clustering similarity threshold; obtaining similar factor features of the first curve deviation and the second curve deviation based on factor similarity and a preset factor similarity threshold; and constructing transient influence relationships by performing correlation analysis on the similar clustering features, similar factor features, and the common historical third state data.

[0013] Furthermore, the step of retrieving the transient impact prediction value based on the matching result of the current state data of the bidirectional DC-DC converter with the first parameter influence curve set and the second parameter influence curve set includes: if the current state of the bidirectional DC-DC converter is charging, then the first parameter influence curve in the first parameter influence curve set is retrieved according to the matching result of the second parameter influence curve set, and the transient impact prediction value is retrieved using the first parameter influence curve; if the current state of the bidirectional DC-DC converter is discharging, then the second parameter influence curve in the second parameter influence curve set is retrieved according to the matching result of the first parameter influence curve set, and the transient impact prediction value is retrieved using the second parameter influence curve.

[0014] Furthermore, based on the minimum smoothing loss, the initial bidirectional DC-DC converter control strategy is obtained from the transient impact prediction value and the current state data; the real-time difference between the real-time state data and the parameter impact curve is obtained, and the initial bidirectional DC-DC converter control strategy is corrected by the deviation value in the transient impact relationship that matches the real-time difference.

[0015] Another technical solution provided in this application is a control system for a bidirectional DC-DC converter, used to implement the method described above, comprising: a data acquisition unit for acquiring state data of the bidirectional DC-DC converter; a curve construction unit for constructing parameter influence curves based on historical state data; a transient influence analysis unit for constructing transient influence curves based on the parameter influence curves; and a control analysis unit for outputting a control strategy for the bidirectional DC-DC converter based on the current state data, parameter influence curves, and transient influence relationships.

[0016] Furthermore, it also includes: an auxiliary control circuit connected to the control analysis unit, used to assist in the magnetic reset of the bidirectional DC-DC converter. When the parameters of the bidirectional DC-DC converter control strategy do not meet the magnetic reset parameter threshold, the control analysis unit executes the auxiliary control circuit to start control.

[0017] The beneficial effects of this application are as follows: 1. By constructing parameter influence curves under different states, historical similar curves can be directly retrieved based on the current state data. Thus, the initial control strategy can be built based on the historical similar curves, improving the initial response efficiency and reducing the device operating losses caused by transient switching. At the same time, based on the construction of transient influence relationships, the common influencing factors and their degree of influence under all modes are reflected. Thus, when the real-time state deviates from the historical state, the initial control strategy can be corrected based on the common influencing factors and their degree of influence to ensure operational stability as much as possible.

[0018] 2. By performing rapid correction of control parameters through transient influence relationships, the control parameters are limited to the initial bidirectional DC-DC converter control parameters, avoiding over-adjustment. Furthermore, the correction process does not require overly complex calculations, which can improve the efficiency of real-time adjustment and thus reduce response latency. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the control method of the bidirectional DC-DC converter in this application.

[0020] Figure 2 This is a schematic diagram of the auxiliary control circuit of the control system of the bidirectional DC-DC converter of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] like Figure 1 As shown in the first embodiment of this application, the control method for a bidirectional DC-DC converter includes the following steps:

[0023] Historical first state data is obtained based on the power supply state of the bidirectional DC-DC converter, historical second state data is obtained based on the charging state of the bidirectional DC-DC converter, and historical third state data is obtained based on the switching state of the bidirectional DC-DC converter.

[0024] A set of influence curves for the first parameter is constructed based on continuous time periods of historical first state data, and a set of influence curves for the second parameter is constructed based on continuous time periods of historical second state data.

[0025] Based on temporal correlation, a transient influence relationship is constructed using the set of influence curves for the first parameter, the set of influence curves for the second parameter, and historical third-state data.

[0026] The transient impact prediction value is retrieved based on the matching results of the current state data of the bidirectional DC-DC converter with the first parameter influence curve set and the second parameter influence curve set;

[0027] Based on the minimum loss, the bidirectional DC-DC converter control strategy is obtained from the transient impact prediction value, transient impact relationship and current state data, and the bidirectional DC-DC converter control is executed using the bidirectional DC-DC converter control strategy.

[0028] In this embodiment, different operating data of the bidirectional DC-DC converter in power supply, charging, and switching states are acquired, and a first parameter influence curve is constructed in the power supply state and a second parameter influence curve is constructed in the charging state. The transient influence in the switching state is calculated based on the temporal continuity of the states, so as to avoid the transient moment being ignored and the loss of instantaneous switching is included in the stabilization adjustment process. This improves the accuracy of the calculation of influencing factors in steady state and transient state, so as to predict transient risks in advance. By balancing the impact of smooth transition and transient adjustment with minimum loss, the loss caused by transient switching is reduced, and the transition loss caused by excessive smooth transition is avoided, thus achieving a two-way balance between smooth transition and instantaneous switching.

[0029] Specifically, the status data includes voltage data, current data, and feedback compensation parameters. These parameters can be current loop parameters or voltage loop parameters, such as phase shift angle and duty cycle. During the operation of the bidirectional DC-DC converter, to ensure output stability, current loop feedback control or voltage loop feedback control is used to adjust the feedback compensation parameters. Different bidirectional DC-DC converter topologies correspond to different feedback compensation parameters, which can be selected based on the actual bidirectional DC-DC converter topology. In this case, the first parameter influence curve and the second parameter influence curve are constructed based on the changes in the feedback compensation parameters with the voltage and current data.

[0030] For example, in a historical time series, the bidirectional DC-DC converter is in a charging state in the first time segment, a discharging state in the second time segment, a charging state in the third time segment, and a discharging state in the fourth time segment. At this time, a first parameter influence curve is constructed based on the historical first state data of the first time segment, a second parameter influence curve is constructed based on the historical first state data of the third time segment, a third parameter influence curve is constructed based on the historical second state data of the second time segment, and a fourth parameter influence curve is constructed based on the historical second state data of the fourth time segment. The set of first parameter influence curves includes the first parameter influence curves corresponding to the first and third time segments, and the set of second parameter influence curves includes the second parameter influence curves corresponding to the second and fourth time segments.

[0031] There is a first transient timing corresponding to the switching state between the first and second time segments, a second transient timing corresponding to the switching state between the second and third time segments, and a third transient timing corresponding to the switching state between the third and fourth time segments. The historical third state data of the first, second, and third transient timings are obtained.

[0032] Based on time-series correlation, transient influence relationships are constructed using the set of influence curves for the first parameter, the set of influence curves for the second parameter, and historical third-state data, including:

[0033] A transient calculation set is constructed using four parameter influence curves of arbitrary continuous time series. Each transient calculation set contains two first parameter influence curves, two second parameter influence curves, and three historical third state data.

[0034] The first curve deviation is constructed using two curves influencing the first parameter, and the second curve deviation is constructed using two curves influencing the second parameter.

[0035] Transient initial influence relationships are constructed using the first curve deviation, the second curve deviation, and historical third state data.

[0036] The first parameter influence curve is derived from the power supply state of the bidirectional DC-DC converter, and the second parameter influence curve is derived from the charging state of the bidirectional DC-DC converter. The deviation between the first parameter influence curves reflects the impact of transient switching on the power supply state, and the deviation between the second parameter influence curves reflects the impact of transient switching on the charging state, thus reflecting the transient effects under the same state. Furthermore, the common effects caused by transient switching are further identified based on the deviations of the first and second curves.

[0037] For example, taking a dual active bridge converter topology, the first state data includes input voltage, output voltage, inductor current, and phase shift angle; the second state data includes input voltage, output voltage, inductor current, and duty cycle; and the third state data includes the switching time, voltage surge values ​​before and after the switching, current surge values, and switching duration. It can be understood that the third state data is actually calculated based on the voltage and current data before and after the switching, as well as the feedback compensation parameters.

[0038] It should be noted that, to ensure the accuracy of the transient impact relationship construction, the state data is first preprocessed based on the operating condition stability threshold to filter out continuous stable periods in the power supply and charging states, excluding fluctuations caused by switching moments or faults. In this embodiment, the operating condition stability threshold is 2% for output voltage ripple and 5% for inductor current fluctuation. When the output voltage ripple is less than or equal to 2% and the inductor current fluctuation is less than or equal to 5% within a preset time, it is considered to be in a stable operating period, and unstable operating periods are filtered out. In other embodiments, the operating condition stability threshold can also be set according to actual stability requirements.

[0039] Based on the first state data, a first parameter influence curve is constructed based on the changes in phase shift angle with input voltage, output voltage, and inductor current. Based on the second state data, a second parameter influence curve is constructed based on the changes in duty cycle with input voltage, battery voltage, and charging current. Piecewise polynomial fitting algorithms, random forest regression, and other methods can be used to construct the curves, thereby obtaining the nonlinear relationship of multivariate synergistic influence. This is to adapt to the situation where the phase shift angle adjustment of the bidirectional DC-DC converter responds to input / output voltage fluctuations and inductor current changes, reflecting the stability of bidirectional energy flow as various factors change.

[0040] Among them, constructing the first curve deviation using two first parameter influence curves and constructing the second curve deviation using two second parameter influence curves include:

[0041] Based on the fixed step size discretization of the first parameter influence curve, interpolation processing is performed to obtain the first parameter grid points. The difference of the first parameter grid points is calculated based on the point-by-point deviation. The first curve deviation is obtained by feature extraction of the difference of the first parameter grid points.

[0042] Interpolation is performed on the second parameter influence curve based on a fixed step size to obtain the second parameter grid points. The difference between the second parameter grid points is calculated based on the point-by-point deviation, and feature extraction is performed on the difference between the second parameter grid points to obtain the second curve deviation.

[0043] In this embodiment, based on the fixed-step discretization of the first and second parameter influence curves, three-dimensional grid points corresponding to the first and second parameter influence curves are obtained. Interpolation is performed using methods such as three-dimensional linear interpolation or Kriging interpolation to generate parameter grid points. For each parameter grid point, point-by-point deviation calculations are performed to calculate the phase shift angle difference between the first and third time segments under the same input voltage, output voltage, and inductor current. The duty cycle difference between the second and fourth time segments under the same input voltage, battery capacity, and charging current is also calculated. Based on the phase shift angle and duty cycle differences, difference features are extracted to construct the first curve deviation and the second curve feature deviation. These difference features reflect the factors significantly affected by transient switching after a transient switch occurs. In this embodiment, the difference features are physical quantities related to the phase shift angle and duty cycle, such as current and voltage.

[0044] Specifically, the feature extraction of the first curve deviation from the differences in the first parameter grid points includes:

[0045] Feature extraction is performed on the differences between the grid points of the first parameter to obtain the difference features;

[0046] The difference features are averaged and extreme differences are extracted to obtain the average deviation and extreme difference. The average difference and extreme difference are used as the global deviation of the first curve deviation.

[0047] Cluster analysis is performed on the differential features to obtain clusters with different degrees of difference, and the clusters are used as local deviations of the first curve deviation.

[0048] Correspondingly, feature extraction of the differences in the second parameter grid points to obtain the second curve deviation includes:

[0049] Feature extraction is performed on the differences between the grid points in the second parameter to obtain the difference features;

[0050] The difference features are averaged and extreme differences are extracted to obtain the average deviation and extreme difference. The average difference and extreme difference are used as the global deviation of the second curve deviation.

[0051] Cluster analysis is performed on the differential characteristics to obtain clusters with different degrees of difference, and the clusters are used as local deviations of the second curve deviation.

[0052] In this embodiment, the overall degree of difference between two time periods is obtained based on the average deviation of the parameter grid point differences. The maximum and minimum difference factors between the two time periods are obtained based on the maximum and minimum deviations of the parameter grid differences. The overall degree of difference, maximum difference value, maximum difference factor, minimum difference value, and minimum difference factor are used as global deviations. Simultaneously, cluster analysis is used to cluster the parameter grid point differences to obtain the difference factors and difference values ​​in low-bias clusters and high-bias clusters, which are used as local deviations. The global deviation reflects the overall difference between the two time periods and the boundary conditions of transient influences, while the local deviation determines the sensitivity range of transient influence factors, improving the accuracy and robustness of the transient influence relationship construction.

[0053] In other embodiments, local deviations can also be constructed by using the rate of change of the first parameter grid point difference with different factors to reflect the impact of local feature changes on the deviation.

[0054] The transient influence relationships constructed using the first curve deviation, the second curve deviation, and historical third-state data include:

[0055] Obtain the common historical third state data of the first curve deviation and the second curve deviation, and construct the transient influence relationship based on the similar characteristics of the first curve deviation and the second curve deviation and the common historical third state data.

[0056] During the switching process from discharge state to charging state to discharge state to charging state, the switching state from charging state to discharge state simultaneously affects the processes of discharge state to charging state to discharge state and charging state to discharge state to charging state. At this point, the historical third-state data corresponding to the switching state from charging state to discharge state is the common historical third-state data. Furthermore, based on cluster similarity and factor similarity, similarity features of the first curve deviation and the second curve deviation are obtained. These similarity features, along with the common historical third-state data, are used to construct transient influence relationships. This reflects transient features that lead to simultaneous deviations in the charging and power supply states, improving the generalization ability of transient influence relationships. Moreover, the hardware losses caused during transient switching persist and simultaneously affect both the charging and power supply states, thus also demonstrating the hardware losses caused by transient switching.

[0057] Specifically, the transient influence relationship constructed based on the similar characteristics of the first curve deviation and the second curve deviation, and the common historical third-state data, includes:

[0058] Similarity clustering features between the first curve deviation and the second curve deviation are obtained based on cluster similarity and a preset cluster similarity threshold;

[0059] Based on factor similarity and a preset factor similarity threshold, similar factor features of the first curve deviation and the second curve deviation are obtained;

[0060] Transient influence relationships are constructed by performing correlation analysis on similar cluster features, similar factor features, and common historical third-state data.

[0061] In this embodiment, the preset cluster similarity threshold is 0.8, and the preset factor similarity threshold is 0.6. Cluster similarity is calculated based on the similarity between clusters with the first curve deviation and clusters with the second curve deviation, i.e., local deviation similarity is obtained. Factor similarity is calculated based on the maximum and minimum deviation factors of the first and second curve deviations, thereby reflecting the global and local feature similarity. It is understood that the similarity threshold can be set according to the actual deviation situation.

[0062] Based on linear regression and nonlinear analysis, correlation analysis is performed on pseudo-clustering features, similarity factor features, and common historical third-state data to extract transient features with the greatest impact on similar features and the degree of their influence. Multiple linear regression can be used for linear regression analysis, and gradient boosting trees can be used for nonlinear analysis. By identifying the correlation between transient features and similar features, transient influence relationships that cause bidirectional effects are obtained, reflecting transient switching features that may cause global deviations. Therefore, under new operating conditions, transient influence relationships can be used to compensate for some of the general effects, minimizing control deviations under new operating conditions.

[0063] Furthermore, based on the matching results between the current state data of the bidirectional DC-DC converter and the sets of influence curves for the first and second parameters, the transient influence prediction values ​​are retrieved, including:

[0064] If the bidirectional DC-DC converter is currently in the charging state, the first parameter influence curve in the first parameter influence curve set is retrieved based on the matching result of the second parameter influence curve set, and the transient influence prediction value is retrieved using the first parameter influence curve.

[0065] If the bidirectional DC-DC converter is currently in a discharge state, then the second parameter influence curve in the second parameter influence curve set is retrieved based on the matching result of the first parameter influence curve set, and the transient influence prediction value is retrieved using the second parameter influence curve.

[0066] When the bidirectional DC-DC converter is currently in a charging state, the current state data is matched with each of the second parameter influence curves in the second parameter influence curve set to obtain the corresponding second parameter influence curve. When the bidirectional DC-DC converter is currently in a discharging state, the current state data is matched with each of the first parameter influence curves in the first parameter influence curve set to obtain the corresponding first parameter influence curve. It can be understood that the first parameter influence curve corresponds to the historical power supply state of the bidirectional DC-DC converter, and the second parameter influence curve corresponds to the historical charging state of the bidirectional DC-DC converter. Therefore, matching is performed based on the current charging state and the historical charging state, and matching is performed based on the current discharging state and the historical discharging state. In this embodiment, the current state data is the bidirectional DC-DC converter operating data that includes at least a stable power supply state or charging state. Using a parameter influence curve with the same trend as the current state as a benchmark, a parameter influence curve with a time sequence related to that parameter influence curve is retrieved. For example, if the time sequence of the first parameter influence curve matched based on the current charging state is t1, then the second parameter influence curve with a time sequence of t2 is retrieved, and the initial parameter in the second parameter influence curve is used as the transient influence prediction value.

[0067] It should be noted that in this embodiment, the first parameter influence curve set, the second parameter influence curve set, and the transient influence relationship are pre-built. The bidirectional DC-DC converter has two states switching during operation, so the first parameter influence curve and the second parameter influence curve can be pre-built in an interleaved manner. When obtaining the transient influence prediction value, it is only necessary to call the pre-built first parameter influence curve set and the second parameter influence curve set.

[0068] It is understandable that, since it is rare for the current state data to have the exact same trend as the historical state data, the first parameter influence curve and the second parameter influence curve are matched based on the maximum similarity. For example, the second parameter influence curve is retrieved by matching the current state data with the first parameter influence curve that is most similar to the first parameter influence curve in the set of first parameter influence curves.

[0069] The control strategy for the bidirectional DC-DC converter, based on minimum loss, is obtained from the predicted transient impact value, transient impact relationship, and current state data.

[0070] Based on the minimum smoothing loss, the initial bidirectional DC-DC converter control strategy is obtained from the transient impact prediction value and the current state data.

[0071] The real-time difference between the real-time state data and the parameter influence curve is obtained, and the initial bidirectional DC-DC converter control strategy is corrected by the deviation value that matches the real-time difference in the transient influence relationship.

[0072] The initial control strategy is calculated by analyzing historical switching scenarios, allowing for advance adjustment of feedback compensation parameters to reduce transient switching losses and avoid unnecessary losses caused by excessive stability. Furthermore, by analyzing the real-time differences between real-time state data and the matched parameter influence curves, the global and local deviations matching these real-time differences in the transient influence relationship are obtained. This involves matching the differences with transient characteristics within the transient influence relationship to retrieve the corresponding deviation values. For example, in constructing the transient influence relationship, the phase shift angle corresponding to the same voltage data before and after the switching state may differ due to current surges. If the real-time difference matches this current surge characteristic during operation, the phase shift angle deviation value is retrieved based on the corresponding difference value. This corrects the initial bidirectional DC-DC converter control strategy, enabling control parameter adjustments in response to real-time disturbances, improving stability while reducing power device losses caused by transient switching.

[0073] Specifically, smoothing losses include transient transition losses and steady-state operating losses, and a loss model is pre-built:

[0074] ;

[0075] ;

[0076] ;

[0077] Where P represents the smoothing loss. This represents the steady-state operating loss. Indicates transient loss. Indicates the steady-state loss of MOS. This represents the steady-state loss of the inductor core. This represents the steady-state loss due to parasitic resistance. This indicates voltage overshoot loss. Indicates current oscillation loss. Indicates reverse recovery loss. Indicates the duration of the transient transition.

[0078] The steady-state losses of MOS include steady-state conduction losses and steady-state switching losses. Steady-state conduction losses are obtained based on on-resistance and current thermal effects; steady-state switching losses are obtained based on switching frequency and energy loss per switching action; steady-state inductor core losses are obtained based on a core loss model; parasitic resistance losses are obtained based on inductor and capacitor parasitic resistances; voltage overshoot losses are obtained based on overshoot voltage and current; current oscillation losses are obtained based on oscillating current and parasitic resistance; and reverse recovery losses are obtained based on diode single reverse recovery energy loss and the number of reverse recovery cycles during transient processes. Loss models can be constructed based on the topology of the bidirectional DC-DC converter and manufacturer data, or trained using historical loss data.

[0079] The initial bidirectional DC-DC converter control parameters are obtained based on the minimum smoothing loss, transient impact prediction, and current state data. Simultaneously, as real-time differences change, the transient impact relationship is invoked to quickly correct the initial bidirectional DC-DC converter control parameters. For example, in the transient impact relationship, for every 5A current deviation, the phase shift angle increases by 0.03°. It can quickly correct the phase angle based on real-time differences. By performing rapid correction of control parameters through transient influence relationships, it is limited to the initial bidirectional DC-DC converter control parameters, avoiding over-adjustment. Moreover, the correction process does not require overly complex calculations, which can improve the efficiency of real-time adjustment and thus reduce response latency.

[0080] Understandably, in other cases, the status data also includes temperature data, incorporating the effect of temperature on component operation into the transient impact relationship, to avoid control deviations caused by component operation deviations under high temperature conditions.

[0081] As a second embodiment of this application, the control system of the bidirectional DC-DC converter includes:

[0082] The data acquisition unit is used to acquire the status data of the bidirectional DC-DC converter;

[0083] Curve construction unit, used to construct parameter influence curves based on historical state data;

[0084] The transient impact analysis unit is used to construct transient impact curves based on parameter impact curves.

[0085] The control analysis unit is used to output the bidirectional DC-DC converter control strategy based on the current state data, parameter influence curves, and transient influence relationships.

[0086] In this embodiment, the data acquisition unit is connected to the curve construction unit and the control analysis unit, the curve construction unit is connected to the transient impact analysis unit and the control analysis unit, and the transient impact analysis unit is connected to the control analysis unit.

[0087] As a third embodiment of this application, the control system of the bidirectional DC-DC converter, applied in a dual active bridge converter topology, further includes:

[0088] The auxiliary control circuit, connected to the control analysis unit, is used to assist in the magnetic reset of the bidirectional DC-DC converter. When the parameters of the bidirectional DC-DC converter control strategy do not meet the magnetic reset parameter threshold, the control analysis unit executes the auxiliary control circuit to start control.

[0089] In this embodiment, the auxiliary control circuit includes an auxiliary winding connected to the high-voltage side resonant inductor, a rectifier diode connected to the auxiliary winding, a bleeder inductor connected to the auxiliary winding, and a bleeder switch connected to the bleeder inductor and the low-voltage side. By adding two auxiliary windings to the high-voltage side resonant inductor and forming a full-wave rectifier circuit through the rectifier diode, the bleeder switch is turned on when the high-voltage side resonant inductor resonates, and the remaining energy on the high-voltage side resonant inductor is slowly discharged to the low-voltage power supply terminal through the bleeder inductor, realizing energy recovery of the high-voltage side resonant inductor, completing the reset of the high-voltage side resonant inductor, and avoiding the overheating of the high-voltage side resonant inductor.

[0090] Specifically, such as Figure 2 As shown, in a dual active bridge converter topology, switches T1, T2, T3, T4, T5, T6, T7, and T8 form two complete H-bridge circuits. Lm is the low-voltage side energy storage boost inductor, Lr is the high-voltage side resonant inductor, Lm is the transducer transformer, rectifier diodes D1 and D2 are used for rectification output, Ci is the low-voltage side filter capacitor, Co is the high-voltage side filter capacitor, La is the bleeder inductor, and Ta is the bleeder switch. A full-wave rectifier circuit structure is constructed through two auxiliary windings, rectifier diodes D1 and D2, thereby reducing energy loss and heat generation of the high-voltage resonant inductor Lr during the boost process and recovering the energy of the high-voltage resonant inductor. Furthermore, during the buck process, Ta can be turned off, thus achieving soft switching of the high-voltage side switch.

[0091] Correspondingly, during the control process of the bidirectional DC-DC converter, when the phase shift angle in the control strategy of the bidirectional DC-DC converter does not meet the magnetic reset requirement, the auxiliary control circuit is activated to assist in the magnetic reset. This is to adapt to the extreme operating conditions where the adjustment value of the phase shift angle is not within the threshold range of the magnetic reset parameter, i.e., when the phase shift angle is not within 0-π / 2. It can still simultaneously achieve output voltage control based on the phase shift angle and achieve magnetic reset based on the auxiliary control circuit. The hardware settings ensure the stability and safety of regulation under extreme operating conditions.

[0092] In other cases, the component losses and energy recovery values ​​of the auxiliary control circuit are calculated as the minimum loss calculation items. When the phase shift angle of the control strategy does not meet the magnetic reset requirement, the loss caused by the decrease in transient switching smoothness stability due to the phase shift angle deviation and the loss caused by starting the auxiliary control circuit are used to determine whether it is necessary to adjust the phase shift angle and start the auxiliary control circuit according to the magnetic reset requirement, so as to perform the switching control of the bidirectional DC-DC converter with the minimum loss as much as possible.

[0093] The specific embodiments described above are preferred embodiments of the control method and system for the bidirectional DC-DC converter of this application, and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape and structure of this application are within the protection scope of this application.

Claims

1. A control method for a bidirectional DC-DC converter, characterized in that: Includes the following steps: Historical first state data is obtained based on the power supply state of the bidirectional DC-DC converter, historical second state data is obtained based on the charging state of the bidirectional DC-DC converter, and historical third state data is obtained based on the switching state of the bidirectional DC-DC converter. A set of influence curves for the first parameter is constructed based on continuous time periods of historical first state data, and a set of influence curves for the second parameter is constructed based on continuous time periods of historical second state data. Based on temporal correlation, a transient influence relationship is constructed using the set of influence curves for the first parameter, the set of influence curves for the second parameter, and historical third-state data. Based on the matching results of the initial time-series real-time state data collected under the current state of the bidirectional DC-DC converter and the set of influence curves of the first parameter and the second parameter, the transient influence prediction value is retrieved. Based on the minimum smoothing loss, the bidirectional DC-DC converter control strategy is obtained from the transient impact prediction value, transient impact relationship, real-time state data collected after the initial timing in the current state, and real-time state data collected after the initial timing in the current state. The bidirectional DC-DC converter control is then executed using the bidirectional DC-DC converter control strategy.

2. The control method for the bidirectional DC-DC converter as described in claim 1, characterized in that: The construction of transient influence relationships based on temporal correlation using the first parameter influence curve set, the second parameter influence curve set, and historical third-state data includes: A transient calculation set is constructed using four parameter influence curves of arbitrary continuous time series. Each transient calculation set contains two first parameter influence curves, two second parameter influence curves, and three historical third state data. The first curve deviation is constructed using two curves influencing the first parameter, and the second curve deviation is constructed using two curves influencing the second parameter. Transient influence relationships are constructed using the deviations of the first curve, the second curve, and historical third-state data.

3. The control method for the bidirectional DC-DC converter as described in claim 2, characterized in that: The construction of the first curve deviation using two first parameter influence curves and the construction of the second curve deviation using two second parameter influence curves include: Based on the fixed step size discretization of the first parameter influence curve and the second parameter influence curve, difference processing is performed to obtain the first parameter grid point and the second parameter grid point. The difference between the first parameter grid point and the second parameter grid point is calculated based on the point-by-point deviation. Feature extraction is performed on the difference between the first parameter grid point and the second parameter grid point to obtain the first curve deviation and the second curve deviation.

4. The control method for the bidirectional DC-DC converter as described in claim 3, characterized in that: The step of extracting features from the first parameter grid point difference and the second parameter grid point difference to obtain the first curve deviation and the second curve deviation includes: Feature extraction is performed on the differences between grid points of the first parameter and the differences between grid points of the second parameter to obtain difference features; The difference features are averaged and extreme differences are extracted to obtain the average deviation and extreme difference. The average difference and extreme difference are used as the global deviation of the first curve deviation and the global deviation of the second curve deviation. Cluster analysis is performed on the differential characteristics to obtain clusters with different degrees of difference. These clusters are used as local deviations of the first curve deviation and the second curve deviation.

5. The control method for the bidirectional DC-DC converter as described in claim 2, characterized in that: The construction of transient influence relationships using the first curve deviation, the second curve deviation, and historical third-state data includes: Obtain the common historical third state data of the first curve deviation and the second curve deviation, and construct the transient influence relationship based on the similar characteristics of the first curve deviation and the second curve deviation and the common historical third state data.

6. The control method for the bidirectional DC-DC converter as described in claim 5, characterized in that: The construction of transient influence relationships based on the similarity features of the first curve deviation and the second curve deviation, and common historical third-state data, includes: Similarity clustering features between the first curve deviation and the second curve deviation are obtained based on cluster similarity and a preset cluster similarity threshold; Based on factor similarity and a preset factor similarity threshold, similar factor features of the first curve deviation and the second curve deviation are obtained; Transient influence relationships are constructed by performing correlation analysis on similar cluster features, similar factor features, and common historical third-state data.

7. The control method for the bidirectional DC-DC converter as described in claim 1, characterized in that: The transient impact prediction value is retrieved by matching the initial time-series real-time acquired state data based on the current state of the bidirectional DC-DC converter with the first parameter influence curve set and the second parameter influence curve set. If the bidirectional DC-DC converter is currently in the charging state, the first parameter influence curve in the first parameter influence curve set is retrieved based on the matching result of the second parameter influence curve set, and the transient influence prediction value is retrieved using the first parameter influence curve. If the bidirectional DC-DC converter is currently in a discharge state, then the second parameter influence curve in the second parameter influence curve set is retrieved based on the matching result of the first parameter influence curve set, and the transient influence prediction value is retrieved using the second parameter influence curve.

8. The control method for the bidirectional DC-DC converter as described in claim 1, characterized in that: The bidirectional DC-DC converter control strategy, based on minimum smoothing loss and using transient impact predictions, transient impact relationships, real-time state data acquired during the initial timing sequence in the current state, and real-time state data acquired after the initial timing sequence in the current state, includes: Based on the minimum smoothing loss, the initial bidirectional DC-DC converter control strategy is obtained by predicting transient effects and acquiring real-time state data of the initial timing sequence under the current state. The real-time difference between the state data collected after the initial timing sequence in the current state and the parameter influence curve is obtained, and the initial bidirectional DC-DC converter control strategy is corrected by the deviation value that matches the real-time difference in the transient influence relationship.

9. A control system for a bidirectional DC-DC converter, used to implement the method as described in any one of claims 1 to 8, characterized in that: include: The data acquisition unit is used to acquire the status data of the bidirectional DC-DC converter; Curve construction unit, used to construct parameter influence curves based on historical state data; The transient impact analysis unit is used to construct transient impact relationships based on parameter impact curves. The control analysis unit is used to output a bidirectional DC-DC converter control strategy based on the state data collected in real time after the initial timing sequence in the current state, the state data collected in real time after the initial timing sequence in the current state, the parameter influence curve, and the transient influence relationship.

10. The control system for the bidirectional DC-DC converter as described in claim 9, characterized in that: Also includes: The auxiliary control circuit, connected to the control analysis unit, is used to assist in the magnetic reset of the bidirectional DC-DC converter. When the parameters of the bidirectional DC-DC converter control strategy do not meet the magnetic reset parameter threshold, the control analysis unit executes the auxiliary control circuit to start control.

Images