A control system for a multi-voltage bus hybrid microgrid

Abstract

The embodiment of the application discloses a kind of control systems of multi-voltage bus hybrid microgrid, including control layer and hardware layer;Wherein, medium voltage DC bus is used to connect the distributed energy module of power higher than first set power value;Low-voltage DC bus is used to connect equipment components, equipment components at least include power less than the DC load of set value and data center load, which are lower than second set power value;AC bus is used to compatible traditional grid and AC load;Main control layer is used to realize the closed-loop regulation of voltage, current;Auxiliary control layer is based on state estimation, bus impedance identification and virtual inertia control, realizes the consistency of multi-bus voltage and frequency;Comprehensive power management layer is based on model predictive control and optimization scheduling algorithm realizes cross bus energy balance and power sharing.The application solves the technical problems that the control layer and hardware layer of multi-voltage bus hybrid microgrid in the prior art are independent of each other, lack of cooperative relationship, and dynamic self-adaptive adjustment is poor.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a control system for a multi-voltage bus hybrid microgrid. Background Technology

[0002] In recent years, with the large-scale integration of renewable energy sources such as photovoltaics and wind power, distributed energy systems have exhibited characteristics of multi-type, high volatility, and strong randomness. Traditional single-voltage-level microgrid architectures, such as pure AC or pure DC systems, are struggling to meet the system stability and flexibility requirements when dealing with multi-source interconnection and the dynamic characteristics of energy flow. To achieve coordinated energy supply at different voltage levels and in different forms, hybrid AC / DC microgrids with multi-Voltage Buses (HM-MG) are gradually becoming an important direction for the development of smart grids.

[0003] However, the existing multi-voltage bus hybrid microgrid control system has the following main problems: (1) The control layer and the hardware layer are independent of each other and lack structural coupling design, resulting in inconsistent response when voltage, power and frequency change dynamically, and low system stability margin. (2) Most current algorithms adopt fixed parameter control or centralized optimization, which is difficult to cope with the high frequency fluctuation of renewable energy output and cannot achieve energy distribution adjustment in milliseconds. Summary of the Invention

[0004] This invention provides a control system for a multi-voltage bus hybrid microgrid, which solves the technical problems in the prior art where the control layer and hardware layer of the multi-voltage bus hybrid microgrid are independent of each other, lack coordination, and have poor dynamic adaptive adjustment.

[0005] This invention provides a control system for a multi-voltage bus hybrid microgrid, including a control layer and a hardware layer; the control layer and the hardware layer achieve millisecond-level state synchronization through a time synchronization communication bus;

[0006] The hardware layer includes a medium-voltage DC bus, a low-voltage DC bus, and an AC bus; the control layer includes a main control layer, an auxiliary control layer, and a comprehensive power management layer.

[0007] The medium-voltage DC bus is used to connect distributed energy modules with power higher than a first set power value, wherein the distributed energy module includes at least a photovoltaic array, a wind turbine, and an energy storage system;

[0008] The low-voltage DC bus is used to connect equipment components with power lower than a second set power value, and the equipment components include at least DC loads with power lower than the set value and data center loads.

[0009] The AC busbar is designed to be compatible with traditional power grids and AC loads;

[0010] The main control layer is used to realize closed-loop regulation of voltage and current;

[0011] The auxiliary control layer achieves multi-bus voltage and frequency consistency based on state estimation, bus impedance identification, and virtual inertia control.

[0012] The integrated power management layer achieves cross-bus energy balance and power sharing based on model predictive control and optimized scheduling algorithms.

[0013] Furthermore, the main control layer utilizes a short-time model predictive control algorithm to perform millisecond-level prediction and adjustment based on bus voltage deviation and frequency change rate, thereby suppressing transient disturbances.

[0014] Furthermore, the optimization objective function J of the short-time model predictive control algorithm short for:

[0015] ;

[0016] Among them, V bus (k) represents the bus voltage at predicted time k; V ref This is the voltage reference value; P gen (k) represents the system's power generation; P load (k) represents the system load power; Q and R are the voltage deviation weighting matrix and power deviation weighting matrix, respectively; N p For the number of prediction steps; || Q This represents the weighted L2 norm, used to penalize bias.

[0017] The control constraints of the optimization objective function include:

[0018] ;

[0019] ;

[0020] Among them, I conv This refers to the inverter output current. and These are the upper and lower safety limits for bus voltage, respectively.

[0021] Furthermore, the auxiliary control layer includes a state estimation module; the state estimation module estimates the system frequency change rate, bus voltage dynamics, and line impedance based on the extended Kalman filter algorithm.

[0022] Furthermore, the integrated power management layer utilizes a long-term model predictive control algorithm to optimize energy scheduling within a minute-level timescale, comprehensively considering energy routing costs, energy storage degradation costs, and line losses, thereby achieving cross-bus energy balance and power sharing.

[0023] Furthermore, the objective function J of the long-term model predictive control algorithm long for:

[0024] ;

[0025] Among them, C grid P is the unit price of electricity purchased from the power grid. grid For grid-connected power; C deg λ is the energy storage degradation cost coefficient; ΔSOC is the change in energy storage state of charge; λ is the bus energy loss weighting factor; ||P route || is the L2 norm of the path power vector for power transmission, used to characterize line power loss; N T To predict the total number of steps.

[0026] Furthermore, the control layer also uses a preset event triggering mechanism to detect the disturbance type of the hardware layer in real time, and triggers the control mode switching based on the detected disturbance type. The disturbance type includes at least: load change, photovoltaic output fluctuation or bus voltage deviation exceeding the threshold. The control mode includes long-time model predictive control and short-time model predictive control.

[0027] Furthermore, the multi-voltage bus hybrid microgrid constructed by the hardware layer includes multiple inverter units;

[0028] Each inverter unit includes a distributed virtual inertia and damping adaptive module;

[0029] The distributed virtual inertia and damping adaptive module dynamically adjusts the inertia coefficient and damping coefficient based on the inverter unit's available power, state of charge, and temperature to achieve distributed frequency support and local oscillation suppression.

[0030] Furthermore, the hardware layer also includes an energy router;

[0031] The medium-voltage DC bus and the low-voltage DC bus are connected by an energy router;

[0032] The energy router optimizes power paths based on a minimum cost flow model.

[0033] Furthermore, the objective function of the minimum cost flow model is: ;

[0034] Among them, C ij For the equivalent cost of the channel, R line,ij For the line resistance, η conv,ij For conversion efficiency, Priority ij k1, k2, and k3 are load priority coefficients, where k1, k2, and k3 are weighting factors.

[0035] The constraints of the minimum cost flow model are:

[0036] ;

[0037] ;

[0038] Among them, P ij P represents the power flow from bus node i to node j. ji The power flow is from bus node j to node i; Let be the power generation capacity of node i; Let be the load power of node i; , These are the lower and upper limits for line power transmission, used to constrain the power flow between buses from exceeding the rated capacity.

[0039] Furthermore, the hardware layer also includes a bidirectional power converter;

[0040] The AC bus and the medium-voltage DC bus are connected by a bidirectional power converter.

[0041] Furthermore, the main control layer is also equipped with a collaborative operation mechanism of short-time model predictive control and distributed virtual inertia;

[0042] The cooperative constraint condition between the short-time model predictive control and the distributed virtual inertia is:

[0043] ;

[0044] Where, ΔP MPC (t) represents the model-predicted change in the controller's output power; Inertia power compensation provided for the i-th inverter unit.

[0045] Furthermore, the control layer is also equipped with a fault-tolerance and delay compensation mechanism;

[0046] When the communication duration between the main control layer, the auxiliary control layer, and the integrated power management layer is continuously delayed by a preset duration a certain number of times, the control layer activates a preset prediction compensation model and adds a delay compensation item to the corresponding control command.

[0047] This invention discloses a control system for a multi-voltage bus hybrid microgrid, comprising a control layer and a hardware layer. The control layer and hardware layer achieve millisecond-level state synchronization via a time-synchronization communication bus. The hardware layer includes a medium-voltage DC bus, a low-voltage DC bus, and an AC bus. The control layer includes a main control layer, an auxiliary control layer, and a comprehensive power management layer. The medium-voltage DC bus connects distributed energy modules with power exceeding a first set power value, wherein the distributed energy modules include at least photovoltaic arrays, wind turbines, and energy storage systems. The low-voltage DC bus connects equipment components with power below a second set power value, including at least DC loads with power less than the set value and data center loads. The AC bus is compatible with traditional power grids and AC loads. The main control layer implements closed-loop regulation of voltage and current. The auxiliary control layer achieves multi-bus voltage and frequency consistency based on state estimation, bus impedance identification, and virtual inertia control. The comprehensive power management layer achieves cross-bus energy balance and power sharing based on model predictive control and optimized scheduling algorithms. The embodiments of the present invention adopt a hierarchical collaborative structure of "three-layer bus-three-layer control", which solves the technical problems of the control layer and hardware layer of multi-voltage bus hybrid microgrids being independent of each other and lacking collaborative relationship, and having poor dynamic adaptive adjustment in the prior art. It realizes intelligent distribution of energy flow among multiple buses and high-precision control of power quality, thereby improving the dynamic robustness and economic operation level of the system. Attached Figure Description

[0048] Figure 1 This is a structural diagram of a control system for a multi-voltage bus hybrid microgrid provided in an embodiment of the present invention. Detailed Implementation

[0049] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0050] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this invention are used to distinguish different objects, not to limit a specific order. The various embodiments of this invention described below can be performed individually or in combination with each other; the embodiments of this invention do not impose specific limitations in this regard.

[0051] Figure 1 This is a structural diagram of a control system for a multi-voltage bus hybrid microgrid provided in an embodiment of the present invention.

[0052] like Figure 1As shown, the control system of the multi-voltage bus hybrid microgrid is characterized by including a control layer 10 and a hardware layer 20; the control layer 10 and the hardware layer 20 achieve millisecond-level state synchronization through a time synchronization communication bus.

[0053] The hardware layer 20 includes a medium-voltage DC bus (MVDC Bus), a low-voltage DC bus (LVDC Bus), and an AC bus (ACBus); the control layer 10 includes a main control layer 11, an auxiliary control layer 12, and a comprehensive power management layer 13.

[0054] The medium-voltage DC bus (MVDC Bus) is used to connect distributed energy modules with power exceeding a first set power value. These distributed energy modules include at least photovoltaic arrays, wind turbines, and energy storage systems. The low-voltage DC bus (LVDC Bus) is used to connect equipment components with power below a second set power value. These equipment components include at least DC loads with power less than the set value and data center loads. The AC bus is used to be compatible with traditional power grids and AC loads.

[0055] The main control layer 11 is used to realize closed-loop regulation of voltage and current; the auxiliary control layer 12 realizes the consistency of voltage and frequency of multiple buses based on state estimation, bus impedance identification and virtual inertia control; the integrated power management layer 13 realizes cross-bus energy balance and power sharing based on model predictive control and optimized scheduling algorithms.

[0056] Specifically, attend Figure 1 The AC bus is typically 380V / 220V 50Hz AC power, mainly connected to the main power grid (i.e., the traditional power grid) at the grid connection point. It is also compatible with traditional AC loads such as motors, lighting, and home appliances. The MVDC bus is typically ±750V DC power, used to aggregate renewable energy units such as photovoltaic and wind power, as well as energy storage systems, including batteries and supercapacitors. The LVDC bus is typically 48V / 24V DC power, used to connect low-power DC loads such as servers, controllers, communication equipment, and data center loads.

[0057] The medium-voltage DC bus (MVDC) and AC bus (AC) are connected via a bidirectional power converter (BPC). The BPC is equipped with voltage / current sensors and a PWM (Pulse Width Modulation) control unit to achieve bidirectional energy flow and smooth regulation. The system constructs a cross-bus energy network through an energy router (ER) to achieve power sharing and optimal energy flow control among different bus levels.

[0058] The Primary Control Layer 11 is primarily responsible for rapid dynamic response and steady-state voltage and current closed-loop control; its control cycle is 1 millisecond, and it mainly runs on DSP (Digital Signal Processing) chips or FPGA (Field Programmable Gate Array) hardware. The Secondary Control Layer 12 achieves multi-bus voltage and frequency consistency through state estimation, bus impedance identification, and virtual inertia control; its control cycle is 10 to 50 milliseconds. The Tertiary Power Management Layer 13 employs Model Predictive Control (MPC) and energy routing optimization algorithms to perform global energy optimization on timescales ranging from 100 milliseconds to several minutes. The three control layers communicate and synchronize clocks through a Time Sensitive Network (TSN) to ensure consistency of control commands across layers. The primary control layer has a data sampling frequency of 10 kHz and a communication delay of no more than 50 microseconds.

[0059] The control system features a communication and time synchronization mechanism, employing a dual-link communication architecture: the primary link uses Time-Sensitive Networking (TSN) for synchronizing control commands, while the backup link uses an EtherCAT bus for data transmission. All control nodes share a unified clock source, with time drift controlled within 50 microseconds. In the event of a link failure, the system automatically switches to the redundant path and corrects the control output using a delay compensation algorithm, ensuring continuous and stable system operation.

[0060] The embodiments of the present invention adopt a hierarchical collaborative structure of "three-layer bus-three-layer control", which solves the technical problems of the control layer and hardware layer of multi-voltage bus hybrid microgrids being independent of each other and lacking collaborative relationship, and having poor dynamic adaptive adjustment in the prior art. It realizes intelligent distribution of energy flow among multiple buses and high-precision control of power quality, thereby improving the dynamic robustness and economic operation level of the system.

[0061] Optionally, the main control layer 11 uses a short-time model predictive control algorithm to perform millisecond-level prediction and adjustment based on the bus voltage deviation and frequency change rate, thereby suppressing transient disturbances.

[0062] Specifically, Short-Term Model Predictive Control (STP) can rapidly predict and adjust based on bus voltage deviation and frequency change rate within a millisecond to second timescale, thereby suppressing transient disturbances. STP focuses on voltage stability and frequency balance; to achieve precise control, a state-space predictive model is introduced to correct the control input in real time.

[0063] Specifically, state-space modeling is performed based on the system's electrical characteristics. First, the system is assumed to consist of n Distributed Power Generation Units (DGUs), each connected to the bus via an inverter unit. The system's state vector is defined as follows: ; where i di i qi Vd and Vq are the d / q axis components of the output current of the i-th inverter unit, respectively, in amperes (A), and the bus voltage components are the bus voltage components, respectively, in volts (V).

[0064] The control input vector is: The output vector is: Where f represents the bus frequency. The system dynamic model can be expressed as: Where A is the system state matrix, reflecting bus impedance and electrical coupling; B is the control input matrix, reflecting the controller's regulation of system variables; E is the disturbance matrix, reflecting the impact of load changes and renewable energy output fluctuations; and d is the disturbance vector, including the load current change ΔI. load With photovoltaic output fluctuation ΔP PV .

[0065] The system dynamic model obtains a linear approximation equation through small-signal linearization: This provides a state equation basis for short-time model predictive control.

[0066] Optionally, the optimization objective function Jshort of the short-time model predictive control algorithm is:

[0067] ;

[0068] Where Vbus(k) is the bus voltage at prediction time k; Vref is the voltage reference value; Pgen(k) is the system generating power; Pload(k) is the system load power; Q and R are the voltage deviation weighting matrix and the power deviation weighting matrix, respectively; Np is the prediction step number; ||Q represents the weighted L2 norm, used to penalize deviations;

[0069] The control constraints for optimizing the objective function include:

[0070] ;

[0071] ;

[0072] Where Iconv is the inverter output current; and These are the upper and lower safety limits for bus voltage, respectively.

[0073] Specifically, the short-time model predictive control algorithm has a prediction step size of 1 millisecond and a prediction time window of 0 to 10 seconds. When the load power change rate (|ΔP / P total When |>20%) or the bus voltage deviation |△V|>6%, the short-time MPC is triggered, the optimal control input vector is calculated in real time, and the output is sent to the PWM unit of the main control layer 11 for execution.

[0074] Optionally, the auxiliary control layer 12 includes a state estimation module; the state estimation module estimates the system frequency change rate, bus voltage dynamics, and line impedance based on the extended Kalman filter algorithm.

[0075] Specifically, to improve the accuracy of the system model, the auxiliary control layer 12 includes a state estimation module based on the Extended Kalman Filter (EKF). This module is used to estimate the system frequency change rate, bus voltage dynamics, and line impedance.

[0076] Specifically, the equation for the extended Kalman filter is:

[0077] ;

[0078] ;

[0079] Where, x k = [f, df / dt, V] bus ] T , represents the system state vector, df / dt is the rate of change of frequency, used to reflect the system's inertial response speed; y k = [f meas V meas ] T , is a measurable output, f meas V meas These represent the actual measured frequency and voltage, respectively; v k n k Let F represent the process noise and G represent the measurement noise, and let F, G, and H represent the system state matrix, input matrix, and output matrix, respectively. The extended Kalman filter algorithm can achieve state updates within a 10-millisecond period, with a frequency estimation error of less than ±0.05 Hz. The estimation results are used to correct the MPC model and calculate the virtual inertia, thereby improving control accuracy.

[0080] Optionally, the integrated power management layer 13 utilizes a long-term model predictive control algorithm (Long-Term MPC) to optimize energy dispatch by comprehensively considering energy routing costs, energy storage degradation costs, and line losses within a minute-level time scale, thereby achieving cross-bus energy balance and power sharing.

[0081] Specifically, long-term MPC operates on a minute-level timescale with the goal of minimizing system energy costs.

[0082] Optionally, the objective function J of the long-term model predictive control algorithm long for:

[0083] ;

[0084] Among them, C grid P is the unit price of electricity purchased from the power grid. grid For grid-connected power; C deg λ is the energy storage degradation cost coefficient; ΔSOC is the change in energy storage state of charge; λ is the bus energy loss weighting factor; ||P route || is the L2 norm of the path power vector for power transmission, used to characterize line power loss; N T To predict the total number of steps.

[0085] Specifically, the long-term MPC main output power distribution reference trajectory P ref (t) is provided to the short-term MPC for real-time tracking, thereby achieving cross-timescale control coordination. Its optimization constraints include the upper and lower limits of energy storage SOC, line capacity limits, and bus voltage constraints. The long-term control cycle is 10 to 3600 seconds, and the objective function and constraint parameters are updated in real time through a rolling horizontal optimization mechanism.

[0086] Optionally, the control layer 10 also uses a preset event triggering mechanism to detect the disturbance type of the hardware layer 20 in real time, and triggers the control mode switching based on the detected disturbance type. The disturbance type includes at least: load change, photovoltaic output fluctuation or bus voltage deviation exceeding the threshold, and the control mode includes long-term model predictive control and short-term model predictive control.

[0087] Specifically, the control layer 10 is also equipped with an event-triggered switching mechanism. When a sudden load change, photovoltaic output fluctuation, or bus voltage deviation exceeding the threshold is detected, the system automatically switches from long-term MPC to short-term MPC to ensure that the controller still has a fast response in the event of an emergency.

[0088] Optionally, the multi-voltage bus hybrid microgrid formed by hardware layer 20 includes multiple inverter units; each inverter unit includes a distributed virtual inertia and damping adaptive module; the distributed virtual inertia and damping adaptive module is based on the inverter unit's own available power P. avail,i State of charge (SOC) i and temperature T i Dynamically adjust the inertia coefficient M i With damping coefficient D i This enables distributed frequency support and local oscillation suppression.

[0089] Specifically, to enhance the dynamic stability of multi-inverter parallel systems in high-penetration renewable energy scenarios, the inverter units in hardware layer 20 are also equipped with a Distributed Virtual Inertia (DVI) and adaptive damping adjustment module. In this module, each inverter unit can dynamically allocate virtual inertia and damping coefficients according to its own power capacity, energy storage state, and thermal characteristics to simulate the inertia characteristics of a synchronous generator and suppress system power oscillations.

[0090] (1) Power dynamic equation.

[0091] The dynamic support behavior of the output power of the i-th inverter unit is defined as follows: ;

[0092] Wherein, ΔPi is the inertia support power provided by the inverter unit, reflecting its response capability to changes in system frequency; Mi is the virtual inertia coefficient, in units of kW·s / Hz, used to describe the power compensation amplitude caused by a unit frequency change; Di is the damping coefficient, in units of kW / Hz, used to suppress steady-state frequency deviation; f is the actual system frequency; and fref is the system reference frequency, usually 50Hz or 60Hz.

[0093] When the system frequency decreases (df / dt < 0), the inverter unit automatically increases its output power; when the frequency increases (df / dt > 0), the inverter unit decreases its output power, thus achieving inertia support for frequency changes. By calculating inertia and damping parameters in real time, distributed frequency support and local oscillation suppression are achieved. The virtual inertia module and the MPC control layer work together to ensure that the system can maintain frequency stability under power fluctuations.

[0094] (2) Adaptive adjustment of inertia and damping parameters.

[0095] The inertia coefficient and damping coefficient are defined as follows:

[0096] ;

[0097] ;

[0098] Where: P avail,i The available output power of the inverter unit; SOC i The state of charge (SBC) is a dimensionless value typically between 0 and 1; T i Here, α is the equipment temperature; α is the inertia scaling factor, typically taken as 0.1~0.5; and β is the damping scaling factor, typically taken as 0.05~0.2. The exponential term in the formula ensures that the inertia coefficient automatically decreases when the equipment temperature rises or the state of charge (SOC) deviates from the median, to prevent overload or over-discharge. In engineering implementation, the inertia and damping parameters are updated every 100 milliseconds, and measurement noise is eliminated through a low-pass filter with a filtering time constant of 20 milliseconds.

[0099] (3) State estimation and inertia scheduling process.

[0100] The system estimates the rate of frequency change (df / dt) in real time using an extended Kalman filter (EKF). Specifically, the estimation equation is:

[0101] ;

[0102] Wherein, the state vector x k = [f, df / dt, V] bus ] T The observation vector is y k = [f meas V meas ] T The rate of change of frequency after state estimation is used as the inertia scheduling input. The system dynamically adjusts M based on the estimation results. i and D i The updated parameters are sent to the local controller of each inverter unit for execution via the communication bus.

[0103] (4) Control constraints and execution logic.

[0104] To prevent overload caused by inertial response, a power limiting constraint is defined: ;in, = 0.2 P nom,i P nom,i This refers to the rated power of the inverter.

[0105] The inertia control process is as follows:

[0106] Step 1: Monitor system frequency and rate of change;

[0107] Step 2: EKF estimates df / dt;

[0108] Step 3: Adjust M according to SOC and temperature i Di ;

[0109] Step 4: Calculate ΔP i And issue a power command;

[0110] Step 5: When the power output exceeds the limit, enter the safety limiting mode.

[0111] (5) Algorithm running effect.

[0112] Under a 50% photovoltaic penetration rate, the system frequency deviation decreased from ±1.2 Hz to ±0.35 Hz; the frequency oscillation decay time was shortened from 1.2 seconds to 0.4 seconds. Experiments show that this distributed inertia control method can achieve multi-inverter coordination in milliseconds, significantly improving the dynamic stability of the system.

[0113] Optionally, hardware layer 20 also includes an energy router (ER); the medium-voltage DC bus MVDCBus and the low-voltage DC bus LVDC Bus are connected through the energy router; the energy router optimizes the power path based on the minimum cost flow model.

[0114] Specifically, in multi-bus hybrid microgrids, energy routing and fault isolation are key to maintaining the safe operation of the system. In order to improve the flexibility of energy interaction between multiple buses, an energy router is set up between the medium-voltage DC bus MVDC Bus and the low-voltage DC bus LVDC Bus. The energy router has an energy routing optimization algorithm based on the minimum cost flow model and a hierarchical soft isolation strategy.

[0115] Alternatively, the objective function of the minimum cost flow model is: ;

[0116] Among them, C ij For the equivalent cost of the channel, R line,ij For the line resistance, η conv,ij For conversion efficiency, Priority ij k1, k2, and k3 are load priority coefficients, where k1, k2, and k3 are weighting factors.

[0117] The constraints of the minimum cost flow model are:

[0118] ;

[0119] ;

[0120] Among them, P ij P represents the power flow from bus node i to node j. ji The power flow is from bus node j to node i; Let be the power generation capacity of node i; Let be the load power of node i; , These are the lower and upper limits for line power transmission, used to constrain the power flow between buses from exceeding the rated capacity.

[0121] Specifically, suppose the system bus network is abstracted as a directed graph G(V, E), where V i E represents the bus node. ij This represents the energy flow channel between buses, with a power flow of P. ij The objective function is: , where C ij The equivalent cost of the channel is defined as: In the formula: R line,ij η is the line resistance; conv,ij For conversion efficiency, it is typically between 0 and 1; Priority ij k1 represents the load priority coefficient, with 0 for critical loads and 1 for non-critical loads; k1, k2, and k3 are weighting factors, typically set to 1:0.5:0.2.

[0122] Constraints: ; The optimization algorithm solves the problem using linear programming with a period of 100 milliseconds and outputs the optimal power path.

[0123] The energy router also features a soft isolation control mechanism. When a bus overvoltage V > 1.1 V is detected... ref When the short-circuit current or current rise rate dI / dt exceeds the set current threshold, the system enters soft isolation mode.

[0124] Specifically, in the event of a fault, the energy router implements soft isolation control, ensuring uninterrupted operation of critical loads through tiered load reduction and redundant channel switching. The soft isolation process consists of three phases: (1) Load reduction phase (0~200 ms): The system gradually reduces power by 20% every 200 milliseconds; (2) Isolation phase (200~1000 ms): If the fault persists, the corresponding channel is disconnected; (3) Recovery phase (after the fault is cleared): Power is gradually restored at the same rate. During the soft isolation process, the energy router redistributes power paths, prioritizing the continuous power supply to critical loads. Simulation verification shows that this strategy reduces the peak current of the system by 90% during fault switching, and the continuity of power supply to critical loads reaches 99.99%.

[0125] Optionally, hardware layer 20 also includes a bidirectional power converter; the AC bus and the medium-voltage DC bus MVDCBus are connected via a bidirectional power converter (BPC).

[0126] Optionally, the main control layer 11 is also equipped with a collaborative operation mechanism for short-time model predictive control and distributed virtual inertia; the collaborative constraints between short-time model predictive control and distributed virtual inertia are as follows:

[0127] ;

[0128] Where, ΔP MPC (t) represents the model-predicted change in the controller's output power; Inertia power compensation provided for the i-th inverter unit.

[0129] Specifically, to achieve dynamic consistency, the main control layer 11 also establishes cooperative constraints between short-time MPC and distributed virtual inertia. Short-time MPC is responsible for system-level power balancing; the virtual inertia module is responsible for frequency transient support. When the frequency fluctuation rate |df / dt| > 1.5 Hz / s, the MPC triggers the inertia module to increase M... i When the system stabilizes, |f - f ref With a frequency <0.1Hz, the inertia gradually recovers to its nominal value. This dual-layer constraint mechanism avoids mutual interference between MPC and inertia control, achieving an organic integration of predictive optimization and inertial response. System measurements show that under a 50% load abrupt change, the power oscillation decay time is reduced from 0.8 seconds to 0.25 seconds.

[0130] Optionally, the control layer 10 is also equipped with a fault-tolerance-delay compensation mechanism; when the communication duration between the main control layer 11, the auxiliary control layer 12 and the integrated power management layer 13 is delayed by a preset duration for a number of consecutive set times, the control layer starts the preset prediction compensation model and adds a delay compensation item to the corresponding control command.

[0131] Specifically, to address the control command latency issue and better handle anomalies such as communication delays and node disconnections, control layer 10 also incorporates a three-stage fault-tolerance mechanism: delay detection, prediction compensation, and degradation maintenance. When a communication delay exceeds 100 milliseconds for three consecutive times, the system activates the prediction compensation model. , where τ represents the number of detected time delay steps. This model incorporates a delay term compensation into the control commands, keeping the state prediction error within 5%.

[0132] When the master control node loses connection, each inverter unit enters Hold Mode, maintaining the most recent power setpoint Pi^last, with frequency support continuing to be provided by the virtual inertia module. Once communication is restored, state resynchronization is performed via timestamp comparison. Through hysteresis compensation for input signals, the system maintains stability even with communication delay τ. The system can maintain stable operation without voltage drops or frequency overshoots during a maximum 2-second master control interruption.

[0133] In this embodiment of the invention, (1) Hardware implementation. The system controller adopts a collaborative architecture of TI C2000 DSP and NXP i.MX8 processor, with the main control layer having a cycle of 1 millisecond (11 cycles), the auxiliary control layer having a cycle of 10 milliseconds (12 cycles), and the integrated power management layer having a cycle of 50 to 100 milliseconds (13 cycles). The communication interface uses a TSN + EtherCAT dual bus, with a data sampling rate of 5 kHz and a delay of less than 50 microseconds. (2) Software implementation. The control algorithm runs on a real-time operating system (RTOS) and adopts a task priority scheduling mechanism:

[0134] Task 1: Master control (highest priority);

[0135] Task 2: State estimation and auxiliary control;

[0136] Task 3: MPC and Energy Optimization;

[0137] Task 4: Communication Synchronization and Fault Detection.

[0138] (3) Experimental verification. Real-time simulation tests were conducted using the RT-LAB platform: photovoltaic power fluctuation was ±20%, bus voltage fluctuation suppression rate was 78%; energy storage power regulation response time was reduced from 200 ms to 45 ms; system frequency deviation during fault switching did not exceed ±0.2 Hz. The experimental results show that the control system of the multi-voltage bus hybrid microgrid provided in this embodiment of the invention can significantly improve the power quality and reliability of the hybrid microgrid.

[0139] In summary, the control system for the multi-voltage bus hybrid microgrid provided in this embodiment of the invention has the following advantages:

[0140] (1) Layered collaborative control. For the first time, a three-layer collaborative control structure of "main control - auxiliary control - integrated power management" is proposed. The hardware and algorithm are deeply integrated to achieve dynamic and steady-state dual optimization across time scales (milliseconds to minutes). Through unified architecture design, time scale collaboration and information exchange between the main, auxiliary and management layers are realized, avoiding multi-layer control conflicts. The dual time domain MPC and distributed virtual inertia collaborative control reduce the bus frequency deviation by 70% and the voltage fluctuation is less than ±2%. The control algorithm fully considers the non-ideal factors of hardware (delay, temperature rise, line impedance) and achieves real-time correction through parameter adaptation, which improves the feasibility of engineering. (2) Virtual inertia and MPC fusion control. By coupling short-time MPC and distributed virtual inertia control, predictive regulation and inertia response are realized simultaneously under frequency disturbances, which significantly improves frequency stability and anti-disturbance capability. (3) Parallel execution of energy routing and soft isolation. Based on the minimum cost flow energy path optimization and soft isolation mechanism, the energy flow is automatically adjusted in the event of a fault or sudden change to keep the power supply to the critical load uninterrupted, and the system continuity rate reaches 99.99%. (4) Communication fault tolerance and delay compensation capability. Through the predictive compensation model and delay monitoring mechanism, the system can still maintain stable operation even within 2 seconds of main control interruption, and the voltage drop will not exceed 2%. (5) Hardware and software collaborative integrated architecture. The control level is deployed collaboratively on DSP, FPGA and embedded platforms to realize multi-functional real-time parallel processing such as voltage and current control, energy optimization and communication synchronization. (6) Comprehensive energy flow optimization. The energy routing strategy based on the minimum cost flow model takes into account line loss, energy storage life and load priority to achieve economic operation. (7) Flexible fault isolation. Through the soft isolation strategy, the system switches to a safe state within 0.2 seconds after fault detection to ensure the continuity of power supply to critical loads. (8) Good engineering feasibility and verification results. Through RT-LAB and actual demonstration area testing, it is proved that the architecture proposed in this embodiment of the invention operates stably in complex hybrid bus environment and significantly reduces system oscillation and power fluctuation. (9) Strong engineering promotion. The architecture has a high degree of modularity and can be flexibly extended to multiple scenarios such as new energy parks, electric vehicle swapping stations, and data center power supply.

[0141] In summary, by establishing a coupled short-time and long-time MPC model, dynamic steady-state coordinated control at different time scales is achieved. Extended Kalman filtering is used for state estimation, and distributed virtual inertia and damping adaptive algorithms are combined to achieve dynamic frequency support among multiple inverters. The system further introduces energy routing based on a minimum cost flow model and a hierarchical soft isolation strategy to ensure power supply continuity for critical loads under fault conditions. This method can maintain stable operation under communication delays and control node disconnection conditions, significantly improving the power quality, frequency stability, and operational reliability of the hybrid microgrid, and possesses good engineering feasibility and application value.

[0142] In the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0143] Finally, it should be noted that the above are merely preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

Claims

1. A control system for a multi-voltage bus hybrid microgrid, characterized in that, It includes a control layer and a hardware layer; the control layer and the hardware layer achieve millisecond-level state synchronization through a time synchronization communication bus; The hardware layer includes a medium-voltage DC bus, a low-voltage DC bus, and an AC bus; the control layer includes a main control layer, an auxiliary control layer, and a comprehensive power management layer. The medium-voltage DC bus is used to connect distributed energy modules with power higher than a first set power value, wherein the distributed energy module includes at least a photovoltaic array, a wind turbine, and an energy storage system; The low-voltage DC bus is used to connect equipment components with power lower than a second set power value, and the equipment components include at least DC loads with power lower than the set value and data center loads. The AC busbar is designed to be compatible with traditional power grids and AC loads; The main control layer is used to realize closed-loop regulation of voltage and current; The auxiliary control layer achieves multi-bus voltage and frequency consistency based on state estimation, bus impedance identification, and virtual inertia control. The integrated power management layer achieves cross-bus energy balance and power sharing based on model predictive control and optimized scheduling algorithms.

2. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The main control layer uses a short-time model predictive control algorithm to perform millisecond-level prediction and adjustment based on bus voltage deviation and frequency change rate, thereby suppressing transient disturbances.

3. The control system for the multi-voltage bus hybrid microgrid according to claim 2, characterized in that, The optimization objective function J of the short-time model predictive control algorithm short for: ； Among them, V bus (k) represents the bus voltage at predicted time k; V ref This is the voltage reference value; P gen (k) represents the system's power generation; P load (k) represents the system load power; Q and R are the voltage deviation weighting matrix and power deviation weighting matrix, respectively; N p For the number of prediction steps; || Q This represents the weighted L2 norm, used to penalize bias. The control constraints of the optimization objective function include: ； ； Among them, I conv This refers to the inverter output current. and These are the upper and lower safety limits for bus voltage, respectively.

4. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The auxiliary control layer includes a state estimation module; the state estimation module estimates the system frequency change rate, bus voltage dynamics, and line impedance based on the extended Kalman filter algorithm.

5. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The integrated power management layer utilizes a long-term model predictive control algorithm to optimize energy scheduling within a minute-level timescale, taking into account energy routing costs, energy storage degradation costs, and line losses, thereby achieving cross-bus energy balance and power sharing.

6. The control system for the multi-voltage bus hybrid microgrid according to claim 5, characterized in that, The objective function J of the long-term model predictive control algorithm long for: ； Among them, C grid P is the unit price of electricity purchased from the power grid. grid For grid-connected power; C deg λ is the energy storage degradation cost coefficient; ΔSOC is the change in energy storage state of charge; λ is the bus energy loss weighting factor; ||P route || is the L2 norm of the path power vector for power transmission, used to characterize line power loss; N T To predict the total number of steps.

7. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The control layer also uses a preset event triggering mechanism to detect the disturbance type of the hardware layer in real time, and triggers the control mode switching based on the detected disturbance type. The disturbance type includes at least: load change, photovoltaic output fluctuation or bus voltage deviation exceeding the threshold. The control mode includes long-term model predictive control and short-term model predictive control.

8. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The multi-voltage bus hybrid microgrid formed by the hardware layer includes multiple inverter units. Each inverter unit includes a distributed virtual inertia and damping adaptive module; The distributed virtual inertia and damping adaptive module dynamically adjusts the inertia coefficient and damping coefficient based on the inverter unit's available power, state of charge, and temperature to achieve distributed frequency support and local oscillation suppression.

9. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The hardware layer also includes an energy router; The medium-voltage DC bus and the low-voltage DC bus are connected by an energy router; The energy router optimizes power paths based on a minimum cost flow model.

10. The control system for the multi-voltage bus hybrid microgrid according to claim 9, characterized in that, The objective function of the minimum cost flow model is: ; Among them, C ij For the equivalent cost of the channel, R line,ij For the line resistance, η conv,ij For conversion efficiency, Priority ij k1, k2, and k3 are load priority coefficients, where k1, k2, and k3 are weighting factors. The constraints of the minimum cost flow model are: ； ； Among them, P ij P represents the power flow from bus node i to node j. ji The power flow is from bus node j to node i; Let be the power generation capacity of node i; Let be the load power of node i; , These are the lower and upper limits for line power transmission, used to constrain the power flow between buses from exceeding the rated capacity.

11. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The hardware layer also includes a bidirectional power converter; The AC bus and the medium-voltage DC bus are connected by a bidirectional power converter.

12. The control system for the multi-voltage bus hybrid microgrid according to claim 2, characterized in that, The main control layer is also equipped with a collaborative operation mechanism of short-time model predictive control and distributed virtual inertia. The cooperative constraint condition between the short-time model predictive control and the distributed virtual inertia is: ； Where, ΔP MPC (t) represents the model-predicted change in the controller's output power; Inertia power compensation provided for the i-th inverter unit.

13. The control system for the multi-voltage bus hybrid microgrid according to claim 1, characterized in that, The control layer is also equipped with a fault-tolerance and delay compensation mechanism; When the communication duration between the main control layer, the auxiliary control layer, and the integrated power management layer is continuously delayed by a preset duration a certain number of times, the control layer activates a preset prediction compensation model and adds a delay compensation item to the corresponding control command.

Images