A disturbance online estimation and double-duty ratio cooperative optimization control method for modular multilevel direct current power taking converter

Abstract

This invention discloses a method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter, comprising: sampling several parameters of the DC-DC converter; establishing a discrete prediction model in a dual duty cycle collaborative optimization control unit based on the output voltage and current state variables and the energy deviation state variables between the upper and lower bridge arms of the modular multilevel DC-DC converter; constructing an online disturbance estimation unit to estimate the disturbance in the dynamic equation of the filter inductor current in real time, and injecting the disturbance estimate into the discrete prediction model to form a disturbance correction discrete prediction model; constructing a multi-objective collaborative optimization objective function; performing rolling optimization on the multi-objective collaborative optimization objective function in the optimization control unit to solve for the optimal main duty cycle for adjusting the output voltage and the optimal phase shift duty cycle for energy balance control in the current control cycle, generating a bridge arm voltage control reference quantity, and applying it to the control process of the modular multilevel DC-DC converter.

Description

Technical Field

[0001] This invention relates to the field of auxiliary power supply and power electronic control technology for secondary equipment in DC distribution networks, and particularly to a method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC power converter. Background Technology

[0002] With the continuous advancement of new power system construction, DC distribution networks are gradually becoming an important development direction for power distribution systems due to their advantages such as high transmission efficiency, strong system adaptability, and ease of connecting distributed energy sources and DC loads. In DC distribution networks, secondary power equipment such as sensing and measurement devices, communication terminal devices, and relay protection devices are crucial components for ensuring the safe and stable operation of the system. These devices typically require a stable low-voltage DC power supply. Because DC distribution networks lack the electromagnetic induction energy extraction conditions found in traditional AC systems, directly powering secondary equipment from medium- and high-voltage DC buses usually faces challenges such as high input voltage, large voltage drop ratios, and stringent insulation requirements.

[0003] To address the aforementioned requirements, the self-powered system for secondary equipment in a DC distribution network can adopt a cascaded front-end and rear-end structure. Modular multilevel DC-DC converters, with their superior high-voltage expansion capabilities and modular voltage division characteristics, are suitable as the front-end power supply unit in self-powered DC distribution network secondary equipment scenarios. The rear-end uses an isolated DC / DC converter to further achieve isolation and voltage regulation. Since the front-end DC power supply unit is directly connected to the DC bus, its operating status not only affects the stability of the output voltage but also the energy distribution between bridge arms and the consistency of sub-module operation. Therefore, the design of the front-end control method is of great significance for the stable operation of the power supply unit.

[0004] Existing front-end control methods primarily focus on output voltage regulation and disturbance suppression. However, under conditions such as DC bus voltage fluctuations, model parameter deviations, and operational state coupling, mismatches easily occur between the ideal model and the actual object, thus affecting the accuracy of state prediction and dynamic regulation performance. Especially for modular multilevel DC-DC converters with multiple degrees of control freedom, constructing control laws solely around the output voltage makes it difficult to simultaneously address output voltage regulation and energy balance between bridge arms within the same control framework. Furthermore, if disturbances such as circuit parameter mismatches in the current path are not estimated online and the model is not corrected accordingly, the controller's disturbance rejection capability and optimized control effect under complex operating conditions will still be limited. Summary of the Invention

[0005] Technical Objective: To address the shortcomings of existing technologies, this invention discloses an online disturbance estimation and dual duty cycle collaborative optimization control method for a modular multilevel DC power supply converter. This method aims to improve the state prediction accuracy of the front-end DC power supply unit under conditions of input voltage disturbance and parameter deviation, while also taking into account power output voltage regulation and energy coordination control between bridge arms.

[0006] Technical solution: To achieve the above technical objectives, the present invention adopts the following technical solution.

[0007] A disturbance online estimation and dual duty cycle collaborative optimization control method for a modular multilevel DC power supply converter is applied to the front-stage DC power supply unit of a self-powered secondary equipment system in a DC distribution network. The front-stage DC power supply unit includes a modular multilevel DC power supply converter with several sub-modules (SM) connected in series. The modular multilevel DC power supply converter adopts a quasi-two-level stepped wave modulation strategy. Control methods include: S1, Input voltage of the DC power supply unit before sampling. Output voltage Output side filter inductor current Upper arm current Lower bridge arm current and the capacitor voltage of each submodule SM ; S2. In the dual duty cycle collaborative optimization control unit, a discrete prediction model is established based on the output side state variables of the modular multilevel DC power supply converter and the energy deviation state variables between the upper and lower bridge arms. The state variables of the discrete prediction model include the output side filter inductor current, output voltage, and energy deviation state variables between the upper and lower bridge arms. The control variables of the discrete prediction model include the main duty cycle and the phase shift duty cycle, which are used to realize dual duty cycle collaborative optimization control. S3. Construct an online disturbance estimation unit for the current channel, estimate the disturbance component in the dynamic equation of the filter inductor current in real time, and inject the disturbance estimate into the discrete prediction model to form a disturbance correction discrete prediction model. S4. For the discrete prediction model of disturbance correction, a multi-objective collaborative optimization objective function is constructed in the preset prediction time domain and control time domain. The multi-objective collaborative optimization objective function includes the output voltage tracking error term, the output side filter inductor current deviation term and the energy deviation term between bridge arms, and the control quantity changes are constrained. S5. In the dual duty cycle collaborative optimization control unit, the multi-objective collaborative optimization objective function is rolled to obtain the optimal main duty cycle of the output voltage and the optimal phase shift duty cycle of energy balance control in the current control cycle. S6. Based on the optimal main duty cycle and optimal phase shift duty cycle obtained from the dual duty cycle collaborative optimization control unit, a bridge arm voltage control reference quantity is generated and applied to the quasi-two-level stepped wave modulation process of the modular multilevel DC power supply converter. During the execution of the stepped switching of the bridge arm voltage, combined with the sub-module capacitor voltage sorting control matched with the stepped wave modulation, the current sub-module participating in the switching is determined to be engaged or disengaged, so that the modular multilevel DC power supply converter can achieve stable power supply and output voltage regulation under the DC distribution network bus voltage disturbance condition.

[0008] Beneficial effects: This invention improves the state prediction accuracy under circuit parameter deviations by constructing an online disturbance estimation unit to correct the prediction model of the dual duty cycle collaborative optimization unit; through the collaborative optimization control of the main duty cycle and the phase shift duty cycle, it realizes the coordinated control of power output voltage regulation and energy balance between bridge arms, thereby improving the power supply stability, output voltage dynamic adjustment performance and bridge arm energy coordination capability of the front-end DC power supply unit under DC distribution network bus disturbance conditions. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the self-powered system structure of the secondary equipment of the DC distribution network according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the key waveforms of quasi-two-level ladder wave modulation in an embodiment of the present invention; Figure 3 This is a flowchart of the capacitor sorting and voltage equalization control of the submodule based on stepped wave modulation in an embodiment of the present invention. Figure 4 This is a control block diagram of an online disturbance estimation and dual duty cycle collaborative optimization control method for a modular multilevel DC power supply converter according to an embodiment of the present invention. Detailed Implementation

[0010] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0011] Example: As attached Figure 1 As shown in this embodiment, a disturbance online estimation and dual duty cycle collaborative optimization control method for a modular multilevel DC power supply converter is applied to the front-stage DC power supply unit of a self-powered secondary equipment system in a DC distribution network. The front-stage DC power supply unit includes a modular multilevel DC power supply converter with several sub-modules SM connected in series. The modular multilevel DC power supply converter adopts a quasi-two-level stepped wave modulation strategy.

[0012] In this embodiment, the self-powered system for secondary equipment in the DC distribution network adopts a cascaded front-end and rear-end structure. The front-end refers to the front-end DC power extraction unit, which includes a modular multilevel DC-DC converter consisting of several sub-modules (SM) connected in series. The rear-end is an isolated DC / DC converter, which is connected to the secondary power equipment to provide stable power to sensing and measuring devices, communication terminal devices, relay protection devices, and other secondary equipment requiring low-voltage DC power. In this embodiment, the DC-DC converter in the front-end DC power extraction unit is used to directly draw power from the DC distribution network bus and complete the primary voltage reduction, outputting a voltage... The intermediate DC bus voltage is used by the subsequent isolated DC / DC converter to further achieve isolation, voltage regulation, and power supply to low-voltage equipment. Therefore, the front-end DC power extraction unit constitutes the energy harvesting stage of the entire self-powered system, and its operating status directly affects the input quality of the subsequent power supply unit and the stability of the entire self-powered system. Therefore, this invention mainly focuses on the control method of the front-end DC power extraction unit in a self-powered system.

[0013] As attached Figure 1 As shown, in this embodiment, the front-end DC power supply unit includes a modular multilevel DC power supply converter structure, specifically including an upper bridge arm inductor connected to the DC distribution network bus. , and the inductance of the upper bridge arm The upper bridge arm chain module and the lower bridge arm chain module connected to the upper bridge arm chain module; output side filter inductor. and filter capacitor After being connected in series, it is bridging both ends of the lower arm chain module. The upper arm chain module and the lower arm chain module are respectively composed of... N The system consists of several sub-modules (SMs) connected in series. Each SM is a half-bridge sub-module, and each half-bridge sub-module includes two power switching devices. , and sub-module capacitors The circuit structure of the half-bridge submodule is existing technology. Figure 1 The specific circuit structure of a half-bridge sub-module is given in the paper, and will not be elaborated here.

[0014] The control objective of the front-end DC power extraction unit is to control the DC distribution network bus voltage. Even under conditions of fluctuations and deviations in control parameters, it can still maintain the basic stability of the intermediate DC bus voltage, while taking into account the average energy coordination between the upper and lower bridge arms and the consistency of the capacitor voltage of each submodule.

[0015] upper arm inductor The bridge arm current variation is limited, the commutation impact is suppressed, and the dynamic operation process is improved; the half-bridge submodule series structure is used to share the high voltage stress on the input side, and to realize power extraction from the high voltage DC bus and primary voltage reduction through modular superposition; the power extraction output side filter inductor and filter capacitor are used to smooth the power extraction output, reduce the output ripple and establish a more stable intermediate DC bus voltage. Figure 1 The diagram illustrates the structural relationship between the front-end DC-DC converter, the rear-end isolated DC / DC converter, and the secondary equipment load. It also schematically shows the main electrical quantities such as bridge arm voltage, bridge arm current, and system output voltage.

[0016] Control methods include: S1, Input voltage of the DC power supply unit before sampling. Output voltage Output side filter inductor current Upper arm current Lower bridge arm current and the capacitor voltage of each submodule SM ; S2. In the dual duty cycle collaborative optimization control unit, a discrete prediction model is established based on the output side state variables of the modular multilevel DC power supply converter and the energy deviation state variables between the upper and lower bridge arms. The state variables of the discrete prediction model include the output side filter inductor current, output voltage, and energy deviation state variables between the upper and lower bridge arms. The control variables of the discrete prediction model include the main duty cycle and the phase shift duty cycle, which are used to realize dual duty cycle collaborative optimization control. S3. Construct an online disturbance estimation unit for the current channel, estimate the disturbance components in the current dynamic equation in real time, and inject the disturbance estimate into the discrete prediction model to form a disturbance correction discrete prediction model. S4. For the discrete prediction model of disturbance correction, a multi-objective collaborative optimization objective function is constructed. The multi-objective collaborative optimization objective function includes the output voltage tracking error term, the output side filter inductor current deviation term, and the energy deviation term between bridge arms, and the control quantity changes are constrained. S5. In the dual duty cycle collaborative optimization control unit, the multi-objective collaborative optimization objective function is rolled to obtain the optimal main duty cycle of the output voltage and the optimal phase shift duty cycle of energy balance control in the current control cycle. S6. Based on the optimal main duty cycle and optimal phase shift duty cycle obtained from the dual duty cycle collaborative optimization control unit, a bridge arm voltage control reference quantity is generated and applied to the quasi-two-level stepped wave modulation process of the modular multilevel DC power supply converter. During the execution of the stepped switching of the bridge arm voltage, combined with the sub-module capacitor voltage sorting control matched with the stepped wave modulation, the current sub-module participating in the switching is determined to be engaged or disengaged, so that the modular multilevel DC power supply converter can achieve stable power supply and output voltage regulation under the DC distribution network bus voltage disturbance condition.

[0017] In this embodiment, the modular multilevel DC-DC converter adopts a quasi-two-level stepped-wave modulation strategy as the underlying driving method, and its key waveforms are as follows: Figure 2 As shown in the diagram, under this modulation strategy, the bridge arm voltage maintains a two-level output pattern on a macroscopic time scale to meet the volt-second balance requirements of the output-side filter inductor and the basic power transmission needs. During microscopic level switching, the bridge arm voltage does not undergo a one-time abrupt change, but rather a multi-stage stepped transition is achieved by sequentially connecting or disconnecting sub-modules. Through this method, the rate of change of the bridge arm voltage and the transient impact of switching can be effectively reduced while maintaining the overall modulation pattern. Figure 2 The relationship between bridge arm terminal voltage, output voltage, and bridge arm current during the modulation process is shown.

[0018] Each bridge arm comprises N series-connected half-bridge submodules. During the transient process of voltage switching at the bridge arm terminals, each transient update cycle preferably changes only the enabled or disabled state of one submodule, thus limiting the amplitude of a single voltage change at the bridge arm terminals to the range of the capacitor voltage of a single submodule. This disperses the originally large voltage step into multiple smaller-amplitude stepped transition processes, thereby reducing the voltage drop at the bridge arm terminals. This reduces stress and minimizes transient impact on devices during commutation. For high-voltage power supply scenarios in DC distribution networks, this approach helps improve the operational stability of modular multilevel structures and the reliability of device use.

[0019] The front-end DC power supply unit further incorporates submodule capacitor voltage equalization control to maintain long-term stable operation of the converter. Specifically, the submodule capacitor voltage equalization control process based on a quasi-two-level stepped-wave modulation strategy is as follows: Figure 3 As shown. In each control cycle, the controller samples the submodule capacitor voltage and bridge arm current of each bridge arm, and determines whether the bridge arm is in a stepped transient state (including rising transient and falling transient) based on the current operating stage of each bridge arm. Then, based on the bridge arm current direction and the relationship between the voltage of each submodule capacitor, the controller determines the order of submodule activation or deactivation.

[0020] Specifically, during the transient process of voltage rise at the bridge arm, when the bridge arm current is positive, the lower-voltage submodule is prioritized for activation to increase the charging opportunity of the lower-voltage submodule and allow its capacitor voltage to return to the average value. When the bridge arm current is negative, the higher-voltage submodule is prioritized for activation to utilize the instantaneous current direction to promote a faster voltage drop. During the transient process of voltage fall at the bridge arm, when the bridge arm current is positive, the higher-voltage submodule is prioritized for deactivation; when the bridge arm current is negative, the lower-voltage submodule is prioritized for deactivation. Through this sorting control method, submodules with different voltage levels can obtain differentiated activation or deactivation opportunities under different current directions and different transient stages, thereby improving the consistency of capacitor voltages of each submodule and reducing the voltage dispersion between submodules.

[0021] In S2, the process of constructing the discrete prediction model includes: Establish the state variables and control variables of the discrete prediction model; State variables of the discrete prediction model The calculation formulas include: , in, This refers to the output-side filter inductor current. This refers to the output voltage, i.e., the regulated state of the intermediate DC bus. The energy deviation between the upper and lower bridge arms is used to characterize whether there is a shift in the average energy distribution between the upper and lower bridge arms. By incorporating the above state variables into the discrete prediction model of the dual duty cycle collaborative optimization control unit, the control method can simultaneously consider the objectives of power output voltage regulation, current dynamic adjustment, and energy coordination control between bridge arms within the same optimization framework.

[0022] When a modular multilevel DC-DC converter employs a quasi-two-level stepped-wave modulation strategy, it has a main duty cycle d and a phase-shift duty cycle under quasi-two-level phase-shift modulation. Two control degrees of freedom. The main duty cycle d is primarily used to adjust the intermediate DC bus voltage. Phase shift duty cycle It is mainly used to regulate the energy exchange process between the upper and lower bridge arms. That is, the main duty cycle d corresponds to the main regulation channel at the power output port, while the phase shift duty cycle... An auxiliary adjustment channel for coordinating the energy of the corresponding bridge arms. By simultaneously optimizing the two control variables mentioned above, the average energy deviation of the bridge arms can be suppressed while ensuring output voltage stability.

[0023] To simultaneously describe the output port state and the inter-arm energy deviation state of the modular multilevel DC-DC converter, the control input of the discrete predictive model is selected. Control quantity The calculation formula is: , in, d For the main duty cycle of the modular multilevel DC-DC power converter, d s The phase shift duty cycle of the modular multilevel DC-DC power converter; Construct the average state equation and the state equation for the energy deviation between the upper and lower bridge arms on the output side of the front-stage DC power supply unit; the average state equation on the output side of the front-stage DC power supply unit is: , in, This refers to the output-side filter inductor in the front-end DC power supply unit. Let be the DC distribution network bus voltage at time t; Let t be the output-side filter inductor current. Let be the output voltage at time t; For the output-side filter capacitor in the front-end DC power supply unit; The main duty cycle at time t; The equivalent resistance of the output port of the front-end DC power supply unit.

[0024] The state equations for the energy deviation between the upper and lower bridge arms include: , in, The rate of change of energy deviation between bridge arms; This is the equivalent adjustment coefficient of the phase shift duty cycle to the energy state between bridge arms; This refers to the phase shift duty cycle.

[0025] A continuous state-space prediction model is constructed by combining the average state equation of the output side of the front-end DC power extraction unit and the state equation of the energy deviation between the upper and lower bridge arms. The calculation formulas of the continuous state-space prediction model include: , in, This is the vector of the rate of change of state variables; The system matrix of the continuous state-space prediction model represents the dynamic coupling relationship between state variables; For a continuous state-space prediction model, the control matrix represents the control effect of the control quantity on the changes of each state quantity. The forward Euler method is used to discretize the average state equation and the state equation for the energy deviation between the upper and lower bridge arms in the continuous state-space prediction model, resulting in a discrete prediction model. The calculation formulas for the discrete prediction model include: , in, To control the sampling period; For the first kDiscrete state vectors at each control sampling time; For the first k +1 predicted state vectors at control sampling times; For the first k The control quantity at each control sampling time; This is the discrete state transition matrix, representing the influence of the current state quantity on the state quantity at the next sampling time. This is the discrete control input matrix, representing the influence of the current control quantity on the state quantity at the next sampling time.

[0026] The construction process of the perturbation-corrected discrete prediction model in S3 includes: The total disturbance includes input voltage fluctuations in the front-end DC power supply unit and deviations in circuit model parameters. In this invention, an online disturbance estimation unit for the current channel is constructed, introducing the total disturbance as an extended disturbance variable into the state equation of the filter inductor current. Since the main duty cycle channel is sensitive to input voltage fluctuations and model parameter deviations, the filter inductor current equation is rewritten to improve model prediction accuracy as follows: , in, for t The total unknown disturbance term at any given time includes input voltage fluctuations of the preceding DC power supply unit and deviations in circuit model parameters. By uniformly incorporating these uncertainties into the total disturbance term, a foundation can be provided for subsequent online disturbance estimation and dynamic model correction.

[0027] Define the state variables for perturbation estimation Perturbation estimation of state variables The calculation formulas include: , in, The first disturbance estimation state variable represents the output-side filter inductor current. ; The second disturbance is used to estimate the state variable, representing the unknown term F. ; State variables estimated based on disturbance A continuous estimated state model is established, and the calculation formulas include: , in, The equivalent control gain of the main duty cycle channel; This represents the rate of change of the total disturbance term.

[0028] State variables estimated based on disturbance An online disturbance estimation unit for the current path is constructed, and the calculation formula includes: , in, This is the estimated value of the filter inductor current; This is the estimated total disturbance. The rate of change of the estimated value of the filter inductor current; The rate of change of the total disturbance estimate; and To estimate the gain coefficient.

[0029] After discretizing the online estimation unit for the disturbance in the filter inductor current channel, the discrete estimation equation can be obtained, and the calculation formula includes: , in, and The first k The estimated values ​​of the filtered inductor current and the total disturbance at each control sampling time; For the first k The main duty cycle at each control sampling time; For the first k The output voltage sample value at each control sampling moment; For the first k The sampled value of the filter inductor current at each control sampling moment; and According to the first k The sample size and estimate at the nth control sampling time are recursively updated to obtain the nth control sampling time. k +1 estimated values ​​of the filtered inductor current channel and total disturbance at control sampling time.

[0030] The aforementioned discrete estimation equations are used to update the current state estimate and the total disturbance estimate in each control sampling period, where the total disturbance estimate serves as the input for subsequent correction of the current state prediction model and construction of the disturbance correction discrete prediction model.

[0031] Furthermore, based on the discrete estimation equation, a modified current state prediction model is obtained, and the calculation formula includes: , in, The first value obtained after correction of the total disturbance estimate k +1 predicted value of the filter inductor current at each control sampling time; For the first k The total disturbance estimate obtained by the online disturbance estimation unit at each control sampling time.

[0032] Specifically, the aforementioned discrete prediction model is a nominal prediction model without a disturbance compensation term. The discrete estimation equation is used to obtain the estimated value of the total disturbance in the current channel in real time. The corrected current state prediction model is used to compensate for this disturbance estimate in the filter inductor current prediction process. Based on this, the disturbance compensation term of the output inductor current channel is written into the state-space form, thus obtaining the calculation formula of the disturbance correction discrete prediction model as follows: , in, This is the effect matrix of the disturbance estimate on the discrete state prediction.

[0033] Compared to the uncorrected nominal prediction model, i.e. the discrete prediction model, the corrected prediction model, i.e. the disturbance-corrected discrete prediction model, introduces the total disturbance estimate into the state prediction process, enabling the model to compensate for the impact of input fluctuations and parameter deviations on state prediction in real time within the control cycle, thereby improving the converter's adaptability under complex operating conditions.

[0034] In S4, a multi-objective collaborative optimization objective function is constructed within the preset prediction and control time domains. The construction process includes: To express the online total disturbance estimation and dual duty cycle collaborative optimization control method in a standard optimization solution form, it is necessary to incorporate the future time domain, i.e., the future... The state of each step is expressed as a linear function of the current state and the control increment sequence. Therefore, this embodiment uses an incremental control expression: , in, For the first k -1 control sampling moment has already applied the control vector to the front-end DC power supply unit; For the first k The control increment vector to be optimized at each control sampling time; For the first k The main duty cycle increment at each control sampling moment; For the first k The phase shift duty cycle increment at each control sampling time.

[0035] Furthermore, in the control time domain The inner minimization yields the multi-objective collaborative optimization objective function: , in, To optimize the objective function for multiple objectives in a collaborative manner; and These are the reference voltage and the reference current, respectively. , and These are output voltage error, filter inductor current deviation, and inter-bridge arm energy deviation, respectively. Weighting coefficients; The number of steps to predict the future; , and Based on the first k The state information prediction at the sampling time is obtained for the first time. k + p The output voltage, filter inductor current, and energy deviation between the upper and lower bridge arms of each prediction step; and These are penalty terms for changes in the main duty cycle and the phase shift duty cycle, respectively. This represents the number of control increment steps that need to be solved during the current rolling optimization process; and The first k + p The main duty cycle increment and phase shift duty cycle increment of each control step.

[0036] In the objective function of multi-objective collaborative optimization, constraints are imposed on changes, namely, penalty terms are added for changes in the main duty cycle and phase shift duty cycle of the control quantity to suppress excessively rapid changes in the control quantity; In the rolling optimization process of S5, the rolling optimization predicts the future state based on the discrete prediction model after disturbance correction in each control cycle, and solves for the optimal main duty cycle of the control output voltage and the optimal phase shift duty cycle of the control energy balance in the current control cycle.

[0037] In this invention, energy coordination between the upper and lower bridge arms is not implemented as an independent external balancing loop, but rather participates in unified optimization as a cooperative control objective in the disturbance correction discrete prediction model. Specifically, the energy deviation state quantity between the bridge arms is constructed based on the deviation between the average capacitor voltages of each submodule in the upper and lower bridge arms. Then, the phase shift duty cycle is solved by the collaborative optimization unit based on the energy deviation signal. The correction amount adjusts the energy exchange process between the upper and lower bridge arms, thereby suppressing the deviation in operating state between the bridge arms. It should be understood that the energy control between the bridge arms and the submodule capacitor sorting and voltage equalization control act at different levels: the former is used to coordinate the overall operating state between the upper and lower bridge arms, and the latter is used to improve the consistency of capacitor voltage between submodules within each bridge arm. The two work together to maintain the stable operation of the modular multilevel DC-DC converter.

[0038] It should be noted that the quasi-two-level stepped wave modulation and submodule capacitor voltage sequencing control belong to the bottom-level modulation execution stage of this invention. Its function is to generate bridge arm drive signals based on the main duty cycle and phase-shift duty cycle output by the dual-duty-cycle collaborative optimization unit, and to improve the consistency of submodule capacitor voltages within the same bridge arm during stepped switching. Energy coordination between upper and lower bridge arms belongs to the upper-level dual-duty-cycle collaborative optimization control stage. This stage introduces the energy deviation state quantity between bridge arms into the disturbance correction discrete prediction model and the multi-objective collaborative optimization objective function, allowing the phase-shift duty cycle to be solved synchronously during the rolling optimization process. Therefore, this invention adopts a hierarchical control approach combining bottom-level modulation and equalization control with upper-level online disturbance estimation and dual-duty-cycle collaborative optimization control to improve the feasibility and operational robustness of the entire front-end power supply unit under complex operating conditions.

[0039] Based on the above, the modular multilevel DC-DC converter control method of the present invention adopts a control method of online disturbance estimation of the current channel and dual duty cycle collaborative optimization, and its control block diagram is as follows. Figure 4 As shown. Figure 4 The controller includes an outer voltage loop controller, an online disturbance estimation unit, a dual duty cycle collaborative optimization unit, a sub-module sorting and equalization unit, and a pulse width modulation generation unit. The outer voltage loop controller, based on the error between the output voltage setpoint and the output voltage feedback value, generates a current reference value for the output-side filter inductor current via a PI controller. This reference value serves as the reference input for the dual duty cycle collaborative optimization unit. By coordinating the outer voltage loop PI controller with the inner loop's dual duty cycle collaborative optimization control, the output voltage regulation requirement can be transformed into a dynamic adjustment requirement for the output-side filter inductor current. This allows the controller to more directly regulate the system's energy transfer process while considering multi-variable coupling relationships.

[0040] In summary, in this embodiment, the specific execution process of the controller is as follows: First, feedback quantities such as input voltage, output voltage, output-side filter inductor current, bridge arm current, and capacitor voltage of each submodule are collected to establish a discrete prediction model; then, the online disturbance estimation unit is updated using the sampled state information to obtain the disturbance estimate at the current moment; then, the discrete prediction model is corrected based on the disturbance estimate to predict the system state within several sampling steps in the future; on this basis, a multi-objective collaborative optimization objective function is constructed, including output voltage tracking error, filter inductor current deviation, bridge arm energy deviation, and control quantity change constraints, and online rolling optimization is performed to obtain the optimal main duty cycle d for output voltage control and the optimal phase shift duty cycle for energy balance control at the current moment. Finally, the first optimal control variable corresponding to the current control cycle is applied, and sampling, estimation, prediction, and optimization are performed again in the next control cycle. Through the above-mentioned collaborative rolling optimization method, the controller can continuously correct control commands in real time according to the system operating status.

[0041] After obtaining the optimal main duty cycle d and the optimal phase shift duty cycle Subsequently, the pulse width modulator, combined with a quasi-two-level stepped wave modulation strategy, generates corresponding drive signals and controls the activation or deactivation of each sub-module, enabling the modular multilevel DC-DC converter to achieve stable power supply and output voltage regulation under DC bus voltage disturbance conditions.

[0042] The drive signal is generated by combining a quasi-two-level stepped wave modulation strategy. During each switching cycle, the bridge arm voltage maintains a two-level output on a macroscopic level. During the micro level switching process, the sub-modules are sequentially connected or disconnected to form a stepped transition in the bridge arm voltage, thereby reducing the voltage change rate and switching impact at the bridge arm.

[0043] The control method also includes submodule capacitor voltage sorting control in conjunction with quasi-two-level step wave modulation. This sorting control operates during the step switching process of the bridge arm voltage rising and falling transients. In each transient update cycle, based on the direction of the bridge arm current and the magnitude of the candidate submodule capacitor voltages currently participating in the step switching, the order of submodules being put into or removed during this step switching is determined. This aims to improve the consistency of submodule capacitor voltages while limiting single voltage jumps at the bridge arm. Specifically, during the bridge arm voltage rising transient, when the bridge arm current is positive, candidate submodules with lower capacitor voltages are preferentially put into operation; when the bridge arm current is negative, candidate submodules with higher capacitor voltages are preferentially put into operation. During the bridge arm voltage falling transient, when the bridge arm current is positive, candidate submodules with higher capacitor voltages are preferentially removed; when the bridge arm current is negative, candidate submodules with lower capacitor voltages are preferentially removed.

[0044] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter, characterized in that, A front-end DC power supply unit is applied to the self-powered system of secondary equipment in DC distribution network; the front-end DC power supply unit includes a modular multilevel DC power supply converter consisting of several sub-modules SM connected in series; the modular multilevel DC power supply converter adopts a quasi-two-level stepped wave modulation strategy. Control methods include: S1, Input voltage of the DC power supply unit before sampling. Output voltage Output side filter inductor current Upper arm current Lower bridge arm current and the capacitor voltage of each submodule SM ; S2. In the dual duty cycle collaborative optimization control unit, a discrete prediction model is established based on the output side state variables of the modular multilevel DC power supply converter and the energy deviation state variables between the upper and lower bridge arms. The state variables of the discrete prediction model include the output side filter inductor current, output voltage, and energy deviation state variables between the upper and lower bridge arms. The control variables of the discrete prediction model include the main duty cycle and the phase shift duty cycle, which are used to realize dual duty cycle collaborative optimization control. S3. Construct an online disturbance estimation unit for the current channel, estimate the disturbance component in the dynamic equation of the filter inductor current in real time, and inject the disturbance estimate into the discrete prediction model to form a disturbance correction discrete prediction model. S4. For the discrete prediction model of disturbance correction, a multi-objective collaborative optimization objective function is constructed in the preset prediction time domain and control time domain. The multi-objective collaborative optimization objective function includes the output voltage tracking error term, the output side filter inductor current deviation term and the energy deviation term between bridge arms, and the change of control quantity is constrained. S5. In the dual duty cycle collaborative optimization control unit, the multi-objective collaborative optimization objective function is rolled to obtain the optimal main duty cycle of the output voltage and the optimal phase shift duty cycle of energy balance control in the current control cycle. S6. Based on the optimal main duty cycle and optimal phase shift duty cycle obtained from the dual duty cycle collaborative optimization control unit, a bridge arm voltage control reference quantity is generated and applied to the quasi-two-level stepped wave modulation process of the modular multilevel DC power supply converter. During the execution of the stepped switching of the bridge arm voltage, combined with the sub-module capacitor voltage sorting control matched with the stepped wave modulation, the current sub-module participating in the switching is determined to be engaged or disengaged, so that the modular multilevel DC power supply converter can achieve stable power supply and output voltage regulation under the condition of DC distribution network bus voltage disturbance.

2. The disturbance online estimation and dual duty cycle collaborative optimization control method for a modular multilevel DC-DC power supply converter according to claim 1, characterized in that: In S2, the process of constructing the discrete prediction model includes: Establish the state variables and control variables of the discrete prediction model; Construct the average state equation and the state equation for the energy deviation between the upper and lower bridge arms on the output side of the front-end DC power supply unit. A continuous state-space prediction model is constructed by combining the average state equation of the output side of the front-end DC power supply unit and the state equation of the energy deviation between the upper and lower bridge arms. The forward Euler method is used to discretize the average state equation and the state equation of energy deviation between the upper and lower bridge arms in the continuous state-space prediction model, so as to obtain the discrete prediction model.

3. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 2, characterized in that: The average state equations on the output side of the front-end DC power supply unit include: ， in, This refers to the output-side filter inductor in the front-end DC power supply unit. Let be the DC distribution network bus voltage at time t; Let t be the output-side filter inductor current. Let be the output voltage at time t; For the output-side filter capacitor in the front-end DC power supply unit; The main duty cycle at time t; The equivalent resistance of the output port of the front-stage DC power supply unit; The state equations for the energy deviation between the upper and lower bridge arms include: ， in, The rate of change of energy deviation between bridge arms; This is the equivalent adjustment coefficient of the phase shift duty cycle to the energy state between bridge arms; This refers to the phase shift duty cycle.

4. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 1, characterized in that: The construction process of the perturbation-corrected discrete prediction model in S3 includes: An online disturbance estimation unit for the current path is constructed, and the total disturbance is introduced as an extended disturbance variable into the state equation of the filter inductor current. Define the disturbance estimation state variables; the disturbance estimation state variables X include the output-side filter inductor current and the unknown disturbance term; Establish a continuous estimated state model based on the disturbance estimated state variable X; Construct an online disturbance estimation unit based on the disturbance estimation state variable X; After discretizing the online estimation unit for the disturbance of the current channel, the discrete estimation equation can be obtained; The modified current state prediction model is obtained based on the discrete estimation equation; Construct a discrete prediction model with perturbation correction.

5. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 4, characterized in that: An online disturbance estimation unit is constructed based on the disturbance estimation state variable X. The calculation formula includes: ， in, This is the estimated value of the filter inductor current; This is the estimated total disturbance. The rate of change of the estimated value of the filter inductor current; The rate of change of the total disturbance estimate; The equivalent control gain of the main duty cycle channel; and To estimate the gain coefficient; This refers to the output-side filter inductor in the front-end DC power supply unit. Let be the output voltage at time t; This is the output-side filter inductor current.

6. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 4, characterized in that: The calculation formulas for the disturbance-corrected discrete prediction model include: ， in, For the first k Discrete state vectors at each control sampling time; For the first k +1 predicted state vectors at control sampling times; For the first k The control quantity at each control sampling time; This is the discrete state transition matrix, representing the influence of the current state quantity on the state quantity at the next sampling time. This is the discrete control input matrix, representing the influence of the current control quantity on the state quantity at the next sampling time. This is the matrix representing the effect of the disturbance estimate on the discrete state prediction. For the first k The total disturbance estimate obtained by the online disturbance estimation unit at each control sampling time.

7. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 1, characterized in that: In S4, a multi-objective collaborative optimization objective function is constructed within the preset prediction and control time domains. The calculation formula includes: ， in, To optimize the objective function for multiple objectives in a collaborative manner; and These are the reference voltage and the reference current, respectively. , and These are output voltage error, filter inductor current deviation, and inter-bridge arm energy deviation, respectively. Weighting coefficients; The number of steps to predict the future; , and Based on the first k The state information prediction at the sampling time is obtained for the first time. k + p The output voltage, filter inductor current, and energy deviation between the upper and lower bridge arms of each prediction step; and These are penalty terms for changes in the main duty cycle and the phase shift duty cycle, respectively. This represents the number of control increment steps that need to be solved in the current rolling optimization process; and The first k + p The main duty cycle increment and phase shift duty cycle increment of each control step.

8. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 1, characterized in that: In S6, the optimal main duty cycle is obtained. d and optimal phase shift duty cycle d s Subsequently, the pulse width modulator, combined with a quasi-two-level stepped wave modulation strategy, generates corresponding drive signals and controls the activation or deactivation of each sub-module, enabling the modular multilevel DC-DC converter to achieve stable power supply and output voltage regulation under DC bus voltage disturbance conditions.

9. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 8, characterized in that: The drive signal is generated by combining a quasi-two-level stepped wave modulation strategy. During each switching cycle, the bridge arm voltage maintains a two-level output on a macroscopic level. During the micro level switching process, the sub-modules are sequentially connected or disconnected to form a stepped transition in the bridge arm voltage, thereby reducing the voltage change rate and switching impact at the bridge arm.

10. A method for online disturbance estimation and dual duty cycle collaborative optimization control of a modular multilevel DC-DC converter according to claim 1, characterized in that: The control method also includes submodule capacitor voltage sorting control in conjunction with quasi-two-level step wave modulation. This sorting control operates during the step switching process of the bridge arm voltage rising and falling transients. In each transient update cycle, based on the direction of the bridge arm current and the magnitude of the candidate submodule capacitor voltages currently participating in the step switching, the order of submodules being put into or removed during this step switching is determined. This aims to improve the consistency of submodule capacitor voltages while limiting single voltage jumps at the bridge arm. Specifically, during the bridge arm voltage rising transient, when the bridge arm current is positive, candidate submodules with lower capacitor voltages are preferentially put into operation; when the bridge arm current is negative, candidate submodules with higher capacitor voltages are preferentially put into operation. During the bridge arm voltage falling transient, when the bridge arm current is positive, candidate submodules with higher capacitor voltages are preferentially removed; when the bridge arm current is negative, candidate submodules with lower capacitor voltages are preferentially removed.

Images