A multi-machine parallel network type inverter communication-free fast power scheduling method

Abstract

The application provides a multi-machine parallel network type inverter communication-free fast power scheduling method. The method independently deploys a virtual oscillator controller with Andronov-Hopf oscillator as the core in each parallel inverter locally, and each inverter is interconnected without communication link; in the steady state, each machine relies on the nonlinear limit cycle dynamics of the virtual oscillator controller to realize power communication-free sharing through the common coupling point; when the inverter receives a scheduling instruction, the target steady-state phase angle difference is inversely solved based on the local impedance estimation, and an auxiliary reference state is generated, the phase of the main oscillator is continuously evolved to the target working condition through one-way phase traction, and a short-time compensation voltage is applied to compress the residual error of the physical circuit when the optimal switching trigger condition is met on the alpha-beta two axes; after the event is over, the local impedance estimation value is iteratively modified based on the active power residual error, and the scheduling unit is automatically exited according to the synchronization criterion.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter control technology, specifically a fast power dispatching method for multi-machine parallel grid-connected inverters without communication. Background Technology

[0002] With the rapid development of high-proportion renewable energy grid connection, distributed energy storage, and integrated power generation, grid-load, and energy storage devices, grid-connected inverters, as power electronic interfaces capable of autonomously constructing voltage and frequency, have become key equipment in new power systems. Compared to grid-connected control, grid-connected control can operate stably in weak grids or even isolated grids without relying on phase-locked loops (PLLs), and can provide inertia support and fast power response for the system. It is suitable for typical engineering scenarios such as multi-unit parallel operation, energy storage clusters, and microgrids containing various distributed power sources. Meanwhile, with the gradual establishment of fast frequency response ancillary service mechanisms in new power systems, renewable energy power plants are required to complete active power reconfiguration within milliseconds under external dispatch commands to achieve the goal of "dispatch instead of switching" for rapid power support. This places higher demands on the response speed and stability of grid-connected inverters under power dispatch transients. In multi-unit parallel operation scenarios, inter-unit communication links are often difficult to deploy or maintain reliably, thus requiring a fast dispatch mechanism capable of multi-unit collaboration under conditions of no communication.

[0003] Existing grid-based control methods still have significant limitations when facing the aforementioned requirements for rapid scheduling of multiple parallel units. First, traditional methods, such as droop control and virtual synchronous machine control, are constrained by the bandwidth of the low-pass filter in the power measurement stage. Their power scheduling response time is typically in the tens to hundreds of milliseconds, which is insufficient to meet the rapid support requirements within a hundred milliseconds. Increasing the filter bandwidth can speed up the response, but it also amplifies measurement noise and weakens the stability margin. Second, virtual oscillator control is based on nonlinear limit cycle dynamics and has a relatively fast response speed. However, existing related improvement methods are mostly based on single-unit grid-connected scenarios. Directly applying them to multi-unit parallel operation will cause the controller state reconfiguration caused by the scheduling unit to generate instantaneous voltage mismatch across the filter inductor. The resulting transient inrush current will spread to other parallel units through the common coupling point, which is particularly significant under weak grid conditions and can easily cause power oscillations and overcurrent risks in the parallel system. Furthermore, existing fast dispatch methods typically rely on pre-defined, precisely given filter inductance and line impedance parameters. However, in multi-unit parallel systems, the equivalent impedance experienced by each unit drifts with changes in the number of parallel units connected, switching status, and operating conditions, making it difficult to guarantee dispatch accuracy under fixed parameters. In addition, some existing improvement methods introduce external grid phase-locking or inter-unit synchronization signals during fast dispatch, disrupting the original phase-locked loop-free characteristics and distributed, non-communication-based operation characteristics of grid-based control, thus reducing the applicability of the solution in multi-unit parallel engineering scenarios. Therefore, how to enable grid-based inverters to achieve rapid dynamic response when power dispatch commands change, maintain robustness against parallel equivalent impedance drift, and prevent significant transient disturbances from spreading to other parallel units under the premise of multi-unit parallel, non-communication-based operation is a critical problem that urgently needs to be solved in the field of grid-based control. Summary of the Invention

[0004] The purpose of this invention is to provide a fast power dispatching method for multi-machine parallel grid-connected inverters without communication, so as to overcome the shortcomings of existing technologies, such as slow dispatching response, transient disturbances spreading to non-dispatchable units, dispatching accuracy being sensitive to parallel equivalent impedance drift, and lack of communication coordination mechanism.

[0005] The technical solution to achieve the purpose of this invention is: a fast power dispatching method for multi-machine parallel grid-connected inverters without communication, comprising the following steps:

[0006] Step 1: Deploy a virtual oscillator controller with an Andronov-Hopf oscillator as its core locally on each parallel inverter. Each inverter is connected to the grid in parallel through a common coupling point, and there is no communication link between the inverters.

[0007] The inverter's grid-connected three-phase current is collected in real time and the αβ axis current is obtained after abc / αβ coordinate transformation. The local αβ axis current command value is calculated based on the current active power command and reactive power command. The error between the αβ axis current and the local αβ axis current command value is sent to the virtual oscillator controller. Under steady-state operation, each inverter maintains its own voltage source external characteristics by relying on the nonlinear limit cycle dynamics of the virtual oscillator controller, and realizes the non-communication sharing of active and reactive power of each machine through the natural electrical coupling of the common coupling point.

[0008] Step 2: Each inverter independently detects power dispatch commands locally; when an inverter receives a power dispatch command, proceed to Step 3; if no command is received, the inverter maintains the independent operation of the virtual oscillator controller from Step 1.

[0009] Step 3: The event-triggered scheduling unit solves the target steady-state phase angle difference and the current steady-state phase angle difference based on the local impedance estimation and the grid-connected power flow equation, and generates a local auxiliary reference state; the phase of the main oscillator is continuously evolved from the current operating condition to the target operating condition through unidirectional phase traction;

[0010] Step 4: The event-triggered scheduling unit independently monitors the optimal switching trigger conditions on the α and β coordinate axes respectively; when either axis first meets the trigger condition, a short-time compensation voltage is applied to the reference voltage corresponding to that axis to compress the residual error of the physical circuit of that axis; the other axis independently performs the same short-time compensation voltage application process when its corresponding trigger condition is met.

[0011] Step 5: In the instantaneous neighborhood after each short-time compensation voltage removal, the inverter samples the active power change rate as the residual evaluation index. Based on the sign and magnitude of the residual evaluation index, the deviation direction and degree of the local impedance estimate relative to the true value are inferred, and the local impedance estimate is corrected according to the discrete iterative update law.

[0012] Step 6: The controller continuously monitors the phase difference between the main oscillator phase and the auxiliary reference state; when the phase difference meets the synchronization criterion, the auxiliary reference state and unidirectional phase traction simultaneously exit, and the inverter returns to step 1.

[0013] Compared with the prior art, the significant advantages of this invention are:

[0014] (1) Based on the nonlinear limit cycle dynamics of the virtual oscillator controller, the present invention enables each parallel inverter to achieve power sharing through natural electrical coupling at a common coupling point without any inter-machine communication in steady state. It has distributed, highly reliable and easily expandable parallel operation capabilities. Compared with the existing parallel coordination schemes that require master-slave communication or synchronization signals, it significantly improves the engineering feasibility and system robustness.

[0015] (2) The present invention constrains the fast dispatching process to be completed locally by the dispatching unit. The dispatching unit is guided by the local auxiliary reference state and short-term compensation voltage, so that it can complete the power reconstruction within tens of milliseconds. The non-dispatching unit naturally absorbs the common coupling point disturbance by relying on the inherent convergence characteristics of the virtual oscillator controller, without the need for additional control actions or to obtain dispatching information from other units. This effectively suppresses the spread of dispatching transients within the parallel system, and significantly shortens the dispatching response time compared with the traditional virtual oscillator control in the case of multiple units in parallel operation in a weak power grid.

[0016] (3) After each short-time compensation voltage is removed, the present invention performs discrete iterative correction on the local impedance estimate based on the active power change rate, so that the scheduling accuracy is robust to the drift of the parallel equivalent impedance. This self-correction process relies entirely on local sampling of the inverter, without the need for additional identification of signal injection or inter-machine communication, which solves the engineering pain point that the existing fast scheduling method relies on pre-calibrated parameters and the parameters are difficult to accurately give in the scenario of multiple machines in parallel.

[0017] (4) The event-triggered scheduling unit of the present invention is activated only during the scheduling event and automatically exits according to the synchronization criterion after the new steady state is established, without changing the original steady-state power sharing accuracy and control structure of the parallel system; the method has no special dependence on the inverter main circuit topology and filter form, and is applicable to typical grid-type inverter structures such as three-phase full bridge and LCL filter, and has strong engineering versatility.

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the topology of a multi-machine parallel grid-connected inverter system according to an embodiment of the present invention.

[0020] Figure 2 This is a logical block diagram of the local control architecture proposed in this invention.

[0021] Figure 3 This is a schematic diagram of the energy landscape and limit cycle convergence characteristics of the virtual oscillator controller described in this invention.

[0022] Figure 4 This is a schematic diagram of the working timing of the event triggering scheduling unit described in this invention.

[0023] Figure 5 This is a flowchart of the iterative correction process for local impedance parameters based on power residuals as described in this invention.

[0024] Figure 6 This is a waveform comparison of the power current at the common coupling point between the scheduling method proposed in this invention and the traditional virtual oscillator control under the power step condition of two machines in parallel in a weak power grid.

[0025] Figure 7 This is a waveform comparison of the active power of the two inverters under the power step condition of two-machine parallel operation in a weak power grid, using the scheduling method proposed in this invention and the traditional virtual oscillator control.

[0026] Figure 8 The diagram shows a comparison of the three-phase current waveforms of the two inverters under the power step condition of a dual-machine parallel operation in a weak power grid, using the scheduling method proposed in this invention and the traditional virtual oscillator control. Detailed Implementation

[0027] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0028] This invention proposes a fast power dispatching method for multi-machine parallel grid-connected inverters without communication, wherein...

[0029] Multi-machine parallel grid-connected inverter system topology as follows Figure 1 The example shown illustrates a scenario where two grid-connected inverters are connected in parallel to a three-phase power grid; those skilled in the art can easily extend this to scenarios where three or more grid-connected inverters are connected in parallel. The system includes a first grid-connected inverter, a second grid-connected inverter, a common coupling point, and a three-phase power grid.

[0030] Each grid-connected inverter consists of a DC power supply, a three-phase full-bridge circuit, and an LCL filter connected sequentially. The three-phase full-bridge circuit comprises six switching transistors forming three bridge arms. Each transistor is connected in parallel with an anti-parallel diode and a parasitic capacitor. The upper and lower transistors of each bridge arm are connected in series to form their respective arms. The midpoint of each bridge arm is designated as its output terminal. The LCL filter includes an inverter-side inductor L1 and a parallel filter capacitor C. f The inverter-side inductor L1 is connected to the output terminals of each phase of the three-phase full-bridge circuit at one end, and to the common coupling point via the grid-side inductor L2 at the other end. A parallel filter capacitor Cf is connected between the inverter-side inductor L1 and the grid-side inductor L2. The LCL filter output terminals of the two grid-connected inverters are connected in parallel to the same common coupling point, which is then connected via the grid-side line inductor L... g With grid-side line resistance R g The grid-side impedance is connected to the three-phase power grid.

[0031] There is no communication link between the two grid-connected inverters; each inverter is equipped with its own independent sampling module, controller and drive circuit. The local controller of each inverter runs the same control logic and independently determines the occurrence of scheduling events and executes corresponding control actions based solely on local sampling information.

[0032] In some embodiments, a multi-machine parallel grid-connected inverter system is constructed based on the following specific parameters: grid-side line voltage RMS value 380V, DC-side voltage 800V, grid frequency 50Hz; LCL filter parameters are inverter-side inductance L1 = 0.4mH, and parallel filter capacitor C f =25μF, grid-side inductance L2=0.12mH; grid-side line parameters are grid-side line inductance L g =1.2mH, grid-side line resistance R g =1.8Ω, the equivalent short-circuit ratio of the dual-machine system under these parameters is approximately 2.6, corresponding to typical weak grid operating conditions. The virtual oscillator controller parameter is the voltage scaling factor k. v =230V / A, current scaling factor k i =0.066A / A, velocity constant ξ=30, virtual capacitance C=0.55F, virtual inductance L=18.422μH, oscillator rated amplitude X n =1V, rotation angle φ R =π / 2. The event-triggered scheduling unit parameters are phase traction gain K=50, phase angle-power ratio factor K. P =3.3113×10⁻ 6 rad / W, short-time compensation voltage duration reference value Δt=220μs, short-time compensation voltage amplitude U inj =20V.

[0033] A fast power dispatch method for multi-machine parallel grid-connected inverters without communication, whose local control architecture is as follows: Figure 2 As shown, each inverter is locally and independently deployed with a two-layer control structure consisting of a bottom-level virtual oscillator controller and an upper-level event-triggered scheduling unit. Under steady-state operation, the bottom-level virtual oscillator controller independently maintains the inverter's voltage source characteristics and shares power with other parallel units via a common coupling point. The upper-level event-triggered scheduling unit is temporarily activated when the inverter receives a power scheduling command, guiding the inverter to quickly complete the operating condition switch, and automatically exits based on synchronization criteria after a new steady state is established. This invention first analyzes the mechanism of multi-machine parallel operation without communication coordination from the perspective of the nonlinear dynamics of the virtual oscillator controller. The virtual oscillator controller described in this invention is constructed with a nonlinear LC oscillation circuit as its core, and its state equation includes a nonlinear damping term acting on the output voltage amplitude. Under the dominant effect of this nonlinear damping term, the controller's state trajectory exhibits a dynamic characteristic of natural convergence to the steady-state limit cycle, and the output voltage amplitude is continuously clamped to the rated value V. n Nearby. Specifically, the voltage amplitude deviation Ṽ is defined as V−V n The phase difference deviation from the grid voltage is δ̃=δ−δ * The following scalar exists along the controller trajectory:

[0034]

[0035] This scalar is in the target steady state (V=V n , δ=δ * The minimum value is taken at () and decreases monotonically along the continuous dynamics of the controller, thus ensuring that the controller state trajectory asymptotically converges to the target steady-state limit cycle. This dynamic characteristic has a clear engineering meaning: any inverter using the virtual oscillator controller possesses inherent self-recovery capability after being disturbed at the common coupling point, and can reconverge to its own steady-state limit cycle without external control actions or inter-machine synchronization signals. It is this inherent characteristic that enables multiple grid-type inverters connected in parallel at the common coupling point to achieve power sharing without communication in steady state through natural electrical coupling at the common coupling point, relying on the equivalent voltage source characteristics of the virtual oscillator; in scheduling transients, after local disturbances of the scheduling unit are transmitted to the other parallel units through the common coupling point, the other parallel units passively absorb the disturbances based on the convergence characteristics of their own controllers, and can recover naturally without anticipating the scheduling event. The above mechanism provides a theoretical basis for the present invention to achieve rapid power coordinated scheduling under multi-machine parallel conditions without communication.

[0036] The following combination Figures 1 to 5 Each implementation step is explained in detail below:

[0037] Step 1: Deployment and steady-state operation of the local virtual oscillator controller

[0038] Each inverter has a locally deployed virtual oscillator controller with an Andronov-Hopf oscillator at its core.

[0039] Real-time acquisition of inverter grid-connected three-phase current i abc After the abc / αβ coordinate transformation, the αβ axis current i is obtained. α with i β According to the current active power command P * With reactive power command Q * Calculate the local current command value i α * with i β * The error between the sampled current from the actual circuit and the commanded current is fed into the virtual oscillator controller. The virtual oscillator controller uses a nonlinear LC oscillator circuit as its core, and its state equation includes a nonlinear damping term acting on the output voltage amplitude. Under the action of this nonlinear damping term, the controller's state trajectory naturally converges to the steady-state limit cycle corresponding to the current power command, and the output voltage amplitude is continuously clamped to the rated value V. n Nearby. The output voltage of the virtual oscillator controller is amplified to a fixed amplitude and used as the inverter voltage reference signal v. αv β The inverter's three-phase full-bridge switching transistors are then driven by PWM modulation.

[0040] In a further embodiment, the state equation of the virtual oscillator controller in the αβ coordinate system is:

[0041]

[0042] Among them, v α and v β The output voltage state of the controller along the α and β axes, ξ is the convergence rate constant, and k is the output voltage state along the α and β axes. v k is the voltage scaling factor. i V is the current scaling factor. n ω is the effective value of the inverter's rated voltage. nom i is the inherent resonant frequency of the oscillator. α * and i β * i is the reference value for the αβ axis current. α and i β This represents the actual output current of the inverter.

[0043] In a further embodiment, according to the current active power command P * With reactive power command Q * Calculate the local current command value i α * with i β * The specific formula is:

[0044]

[0045] Where, θ g The phase angle is the grid voltage. The actual output current i of the physical inverter. α i β After sampling, compared with the reference value i α * i β * The difference, after rotational transformation, serves as the input signal for the oscillator, used to establish a power interaction channel between the controller and the power grid.

[0046] The output state v of the virtual oscillator controller α and v β After voltage scaling, the signal serves as the inverter's voltage reference signal. This signal is then modulated by PWM to drive the switching transistors, ultimately establishing an AC voltage at the inverter's output port. The output current is sampled and fed back to the controller input, forming a closed loop. In this closed loop, the state [v]... α , v β On the αβ plane with amplitude Vn It makes circular motion, forming the main oscillator.

[0047] The master oscillator is the orthogonal state variable [v] in the above process. α , v β Its phase angle is defined as:

[0048]

[0049] This phase angle characterizes the instantaneous phase of the controller's output voltage. The steady-state phase difference φ of the master oscillator... * With active power command P * The correspondence is as follows:

[0050]

[0051] Where ω is the grid angular frequency, L is the total series inductance, and V g This is the effective value of the grid voltage.

[0052] Under steady-state conditions with multiple inverters operating in parallel, each inverter's virtual oscillator controller independently operates the aforementioned dynamic process. Relying on its droopy-like active-frequency and reactive-voltage mapping relationship, active and reactive power are shared through natural electrical coupling at a common coupling point. This steady-state power sharing process is completely independent of any inter-inverter communication link or external synchronization signal. Figure 3 The energy landscape and limit loop convergence characteristics of the virtual oscillator controller are further illustrated: after a disturbance occurs, the system state naturally converges to the steady-state limit loop along the energy decrease direction; therefore, any inverter can recover to the steady state by relying on the convergence characteristics of its own controller after being disturbed by the common coupling point, without the need for additional control actions.

[0053] Step 2: Activation of the local scheduling instruction detection and event-triggered scheduling unit

[0054] Each inverter independently detects power dispatch commands locally. When an inverter does not receive a dispatch command, the controller operates independently as described in step 1 using the virtual oscillator controller. The event-triggered dispatch unit is in a sleep state and does not participate in the inverter control process; at this time, the inverter exhibits standard voltage source characteristics. When the inverter receives a new active power command P... * When the local event triggers the scheduling unit, it is activated and enters the fast scheduling process described in steps 3 to 6. The other parallel units that have not received the scheduling instructions still maintain the independent operation state described in step 1. They do not perceive or participate in the inverter's scheduling process and only passively respond to the disturbances caused by the inverter scheduling through the natural electrical coupling of the common coupling point.

[0055] Step 3: Local inverse solution of the phase angle difference of the target operating condition and generation of auxiliary reference states

[0056] After receiving a new power command, the inverter's event-triggered scheduling unit uses the local grid impedance estimate and the grid-connected power flow equation under the inductive network to solve for the target steady-state phase angle difference and the current steady-state phase angle difference:

[0057] For the inverter output channel under inductive circuit conditions, the active power command P * The phase angle difference δ with the target steady state * The following conditions must be met:

[0058]

[0059] Among them, V n V is the nominal value of the inverter output voltage amplitude. g X = ωL, where X is the voltage amplitude of the power grid. eq L is the equivalent impedance of the inverter. eq The inverter filter inductor L f Compared with the local impedance estimate L̂ g The sum is given. The target steady-state phase angle difference δ can be obtained by inverse solution of the above equation. * .

[0060] Similarly, by the current power command Similarly, the current steady-state phase angle difference δ0 can be obtained by inverse solution.

[0061]

[0062] The inverter generates a local auxiliary reference state based on this: the phase of the main oscillator at the moment of triggering is denoted as θ1(t). on ⁻ Then, the initial phase θ2 of the auxiliary reference state is generated locally by the following formula:

[0063]

[0064] The inverter initializes the auxiliary reference state to be δ ahead or behind the grid phase. * The position is used as a reference target for subsequent traction by the main oscillator. The auxiliary reference state, after being generated, operates at a rated angular frequency ω. nom Free evolution.

[0065] Since the actual operating frequency and amplitude of the auxiliary oscillator are the same as those of the main oscillator, after obtaining the auxiliary reference state phase θ2, the corresponding orthogonal state quantity [v] of the auxiliary oscillator... α2 , v β2 ] can be generated accordingly.

[0066] The generation process of the auxiliary reference state is completed entirely within the inverter controller, without relying on any external grid phase measurement or introducing a phase-locked loop (PLL) circuit, thus strictly maintaining the PLL-free characteristics of grid-type control.

[0067] After the auxiliary oscillator is generated, a unidirectional coupling relationship is established with the main oscillator. The amplitude and phase angle state equations of the coupled main oscillator controller are as follows:

[0068]

[0069] Where K is the coupling gain. P and Q are the actual active and reactive power outputs of the inverter, respectively. * and Q * This is a reference value for power scheduling.

[0070] At this point, the inverter introduces a unidirectional phase traction term K·sin(θ2−θ1) into the dynamic equation of the amplitude and phase of the main oscillator, continuously pulling the phase of the main oscillator from the vicinity of the current operating condition to the vicinity of the target operating condition. The selection of K makes the characteristic time of the coupling traction action significantly smaller than the phase convergence time of the virtual oscillator controller under natural dynamics, thereby accelerating the migration of the internal phase of the inverter to the target operating condition while maintaining the continuity of the controller state evolution.

[0071] Step 4: Separate detection of optimal switching trigger conditions for the αβ coordinate axes and application of short-time compensation voltage.

[0072] While the unidirectional phase traction mechanism described in step 3 enables the main oscillator phase to evolve rapidly toward the target operating condition, the inductor current in the inverter's physical circuit cannot change instantaneously. Therefore, residual errors in the physical state inevitably exist between the old and new operating conditions. To reduce these residual errors, the event-triggered scheduling unit independently monitors the optimal switching triggering conditions on both the α and β coordinate axes and applies a short-time compensation voltage to the reference voltage of the corresponding axis at the triggering time.

[0073] Specifically, for the incremental equivalent circuit on the αβ coordinate axis of the inverter, time-domain analysis of the first-order RL response shows that the following conditions are satisfied at the instant of operating condition switching:

[0074]

[0075] At this point, the initial amplitude of the transient component in the on-axis current response, which consists of an exponentially decaying term, is zero. Where Δv k Let L be the difference between the current output voltage of the master oscillator on the k-axis and the target steady-state voltage, where ω is the power grid angular frequency, and L is the voltage at the k-axis. f Here, ΔI is the inverter filter inductor, and ΔI is the incremental steady-state current amplitude pre-calculated from the steady-state parameters of the old and new operating conditions. Its expression is:

[0076]

[0077] Where, |ΔV m |=2V n ·sin(|δ * −δ0| / 2) is the magnitude of the difference vector between the two voltage vectors before and after the operating condition switch, R f For filtering resistors. The inverter continuously monitors the satisfaction of the above equation on the α-axis and β-axis respectively; since the switching conditions of the α-axis and β-axis are determined by the instantaneous phase angle, the triggering times of the two axes are naturally about a quarter of the fundamental frequency period apart in timing. When the conditions are satisfied independently, the two axes respectively activate the short-time compensation voltage application process of the corresponding axis.

[0078] When a certain axis, such as the α axis, first meets the triggering condition, a short-time compensation voltage U is applied to the reference voltage corresponding to the α axis. inj The polarity of the short-time compensation voltage is opposite to the direction of the instantaneous deviation value Δi(t0) of the current at the trigger moment of the modulation wave α-axis, in order to actively compress the residual error of the physical circuit. Its magnitude can generally be set as a certain percentage of the rated voltage, with a typical value of about 30V in a 220V AC power grid scenario. The duration Δt of the short-time compensation voltage is calculated analytically based on the inverter filter parameters and the instantaneous current value at the trigger moment using the following formula:

[0079]

[0080] In the formula, L̂ f This is the estimated value of the local filter inductance. The physical meaning of the above formula is that by selecting a compensation voltage amplitude and duration that matches the residual error current, the inverter physical circuit can accurately fall near the steady-state current trajectory corresponding to the new operating condition at the moment the compensation voltage is removed. Under the typical parameters of this embodiment, Δt is usually on the order of hundreds of microseconds, much smaller than the fundamental period of 20ms.

[0081] When the β-axis subsequently meets its corresponding triggering condition, the same compensation voltage application process is independently performed on the β-axis, and the duration is also calculated analytically according to the above formula. The short-time compensation voltage application processes of the α-axis and β-axis are independent of each other and their timing is staggered by about a quarter of the fundamental frequency period, jointly completing the compression of the physical circuit residual error of the output current on the αβ coordinate axes. Figure 4 The complete working sequence of the event-triggered scheduling unit is further illustrated, including the sequential triggering of the first and second phases after the scheduling event occurs, the application process of the short-time compensation voltage of the corresponding axis, and the complete sequence of the event-triggered scheduling unit from activation to self-exit.

[0082] Step 5: Iterative correction of local impedance parameters based on power residuals

[0083] The analytical calculation of the short-time compensation voltage duration Δt in step 4 depends on the local filter inductance estimate L̂.f In practical engineering, both the filter inductance and grid-connected impedance can shift with changes in temperature, magnetic circuit saturation, and the switching status of parallel units. Under multi-unit parallel operation conditions, the equivalent impedance actually experienced by each inverter will also drift with the number and operating status of other parallel units, and the equivalent impedance experienced by different units will be different. Therefore, the deviation between the estimated local impedance and the true value is inevitable, and may lead to overcompensation or undercompensation in the compensation process in step 4. To address the above parameter uncertainties, the inverter iteratively corrects the estimated local impedance based on the power residual after each short-time compensation voltage removal.

[0084] Specifically, the rate of change of active power sampled in the instantaneous neighborhood after the k-th short-time compensation voltage removal is used as the residual evaluation index:

[0085]

[0086] In the formula, t k off+ ε represents the instantaneous neighborhood after the k-th short-time compensation voltage is removed. The physical meaning of the residual evaluation index is: if the estimated local impedance matches the true value, the compensation process precisely brings the inverter's physical state to near the target steady-state trajectory, and after compensation removal, the inverter's active power no longer requires a significant natural adjustment process. k If the local impedance estimate is too small, the compensation duration will be too short and the compensation amount insufficient. After the compensation is removed, the inverter's active power will continue to rise towards the target value under the natural dynamics of the virtual oscillator controller. k If the local impedance estimate is too high, the compensation duration will be too long and the compensation amount will be excessive. After the compensation is removed, the inverter's active power will experience a short-term overshoot and then drop back. ε k It is negative.

[0087] Accordingly, the inverter corrects the local impedance estimate according to the following discrete iterative update law:

[0088]

[0089] In the formula, Let Lk be the estimated local filter inductance value used in the k-th scheduling event, η be the iteration gain, and sgn(ΔP) be the sign correction term for the current scheduling direction, which is positive under power up-adjustment conditions and negative under power down-adjustment conditions. The update process is executed once after each scheduling event and does not participate in the main control flow within a single scheduling event, thus not affecting the real-time control performance of the inverter in a single event. After several independent scheduling events, the estimated local filter inductance value Lk... f It will gradually converge to near the true value, corresponding to ε. kAs the value gradually approaches zero, the compensation accuracy of the inverter in subsequent scheduling events is continuously improved. Figure 5 The complete process of iterative correction of local impedance parameters based on power residuals is further illustrated.

[0090] The local impedance parameter iterative correction mechanism relies entirely on the inverter's local sampling information, requiring no additional identification signals to be injected into the system, nor any inter-machine communication or external parameter transmission. This makes it fully compatible with the multi-machine parallel operation without communication characteristic of this invention. In a multi-machine parallel system, each inverter independently runs the above correction process, and the local impedance estimate of each inverter converges to its respective perceived true equivalent impedance. This ensures that the scheduling accuracy of each inverter in the entire parallel system is robust to parallel equivalent impedance drift.

[0091] Step 6: Monitoring of synchronization criteria and automatic exit of the event-triggered scheduling unit

[0092] After completing the short-time compensation voltage application process for the α and β axes described in step 4 and the unidirectional phase traction described in step 3, the controller continuously monitors the phase difference between the main oscillator phase θ1 and the auxiliary reference state phase θ2. When this phase difference satisfies the synchronization criterion:

[0093]

[0094] At this point, it can be determined that the main oscillator state is fully consistent with the target operating condition, and the inverter executes the self-exit operation of the event-triggered scheduling unit: the auxiliary reference state stops generating, the unidirectional phase traction term is returned to zero, the event-triggered scheduling unit is completely released, and the inverter immediately returns to the independent operation state of the virtual oscillator controller described in step 1. Where The preset phase tolerance threshold is used, and its value reflects the allowable error between the phase difference of the virtual oscillator and the target phase difference.

[0095] The self-exit mechanism of the event-triggered scheduling unit has two important implications: First, the event-triggered scheduling unit is only activated during the scheduling event and automatically exits after a new steady state is established. Therefore, it does not change the original steady-state control structure and power sharing accuracy of the parallel system, ensuring that the fast scheduling mechanism of the present invention is fully compatible with the steady-state operation characteristics of the original virtual oscillator control. Second, the activation and exit of the event-triggered scheduling unit are determined autonomously based entirely on the local state and do not rely on any inter-machine communication or external control signals, which is consistent with the multi-machine parallel non-communication cooperative operation characteristics of the present invention.

[0096] While the single inverter performs the scheduling process, the remaining parallel units maintain the independent operation of the virtual oscillator controller described in step 1, neither sensing nor participating in the inverter's scheduling process. During the rapid power reconfiguration process, the scheduling unit transmits a certain degree of transient disturbance to the remaining parallel units via the common coupling point; the remaining parallel units naturally absorb this disturbance based on the self-recovering characteristics of their own virtual oscillator controllers under the nonlinear limit cycle dynamics described in step 1, and quickly return to steady-state operation after scheduling is completed. Thus, this invention achieves coordinated and rapid scheduling of a multi-unit parallel system without any communication requirements.

[0097] This invention utilizes the self-recovery characteristics of the virtual oscillator controller to enable the non-scheduler to passively absorb disturbances at the common coupling point, significantly shortening the overall scheduling response time of the parallel system under the premise of no communication, and improving the robustness of scheduling accuracy to parallel equivalent impedance drift.

[0098] To verify the effectiveness of this invention, tests were conducted on a dual-machine parallel experimental platform comprising two grid-connected inverters with a rated power of 15kW and weak grid conditions (equivalent short-circuit ratio of approximately 2.6). The operating conditions were set as follows: both inverters initially operated at an active power output of 10kW, with a total system active power output of 20kW; at a certain moment, a power step command was applied to the first inverter, increasing its active power command from 10kW to 20kW, while the active power command of the second inverter remained unchanged at 10kW.

[0099] like Figure 6 As shown, the horizontal axis represents time, and the vertical axes represent the active power (kW), reactive power (kVar), and three-phase current (A) at the point of common coupling, respectively. The figure illustrates the overall dynamic response of the parallel system at the point of common coupling under the above operating conditions. Figure 6 (a) Figure 6 (c) represents the system response curve when using the traditional virtual oscillator control method. Figure 6 (b) Figure 6 (d) represents the system response curve when the method described in this invention is used.

[0100] The comparative results show that, when the same power step command is applied to the first inverter, the active power at the common coupling point (CCP) takes a relatively long transition period to establish a new equilibrium value using the traditional virtual oscillator control method, with a transient duration of approximately 180 ms. However, using the method described in this invention, the CCP quickly establishes a new equilibrium value, with the transient duration shortened to approximately 30 ms, and no significant overshoot or oscillation occurs during the transition. The response results of the three-phase current at the CCP show a consistent trend. This verifies that the method proposed in this invention has a significant advantage over traditional control methods in terms of overall system scheduling response speed under multi-machine parallel operation conditions.

[0101] like Figure 7 As shown, the horizontal axis represents time, and the vertical axis represents the active power (unit: kW) and reactive power (unit: kVar) of the first and second inverters, respectively. The figure illustrates the dynamic power response of the two inverters under the above operating conditions. Figure 7 (a) Figure 7 (c) represents the power response curves of the first inverter and the second inverter when using the traditional virtual oscillator control method; Figure 7 (b) Figure 7 (d) represents the power response curves of the two inverters when the method described in this invention is used.

[0102] Comparative results show that, when using the method described in this invention, the first inverter undertaking the scheduling task can complete power command tracking from 10kW to 20kW in a shorter time; the second inverter experiences smaller transient disturbance amplitudes and shorter durations during this process, subsequently recovering to its original output level of 10kW. In contrast, when using the traditional virtual oscillator control method, the power transition process of the first inverter is significantly longer, and the time required for the entire system to complete power redistribution is significantly increased. This verifies that the method proposed in this invention, while improving the response speed of the scheduling unit, can control the disturbance of scheduling transients to other parallel units within a small range, preventing the spread of oscillations within the parallel system.

[0103] like Figure 8 As shown, the horizontal axis represents time, and the vertical axis represents the three-phase output current (unit: A) of the first and second inverters, respectively. The figure further illustrates the transient response of the two inverters under the aforementioned operating conditions from a current perspective. Figure 8 (a) Figure 8 (c) represents the current response curve when using the traditional virtual oscillator control method. Figure 8 (b) Figure 8 (d) represents the current response curve when the method described in this invention is used.

[0104] The comparative results show that, using the method described in this invention, the output current amplitude of the first inverter quickly adjusts and matches the new power demand after a power command step, with a transition process of approximately 40ms. The second inverter experiences a small current fluctuation corresponding to power redistribution during the transient process, then quickly recovers to a steady-state level. In contrast, using the traditional virtual oscillator control method, the current transition process of the first inverter is approximately 150ms, and the overall dynamic response speed is significantly slower. These experimental phenomena further verify, from the current perspective, the rapid dispatch capability and effective suppression of disturbances from other parallel units under actual weak grid multi-unit parallel operation conditions proposed in this invention.

[0105] Finally, it should be noted that the above examples are merely some specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments and can have many variations. For example, the method described in this invention can be applied to the main circuit topology of grid-connected inverters other than three-phase full-bridge inverters, and the LCL filter structure can be replaced with other typical filter structures such as L or LC; the local impedance parameter iterative correction mechanism based on power residuals described in this invention can also be further extended by combining other online parameter identification algorithms; the multi-machine parallel operation scenario described in this invention can be naturally extended to the parallel operation of three or more grid-connected inverters. All variations that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should be considered within the scope of protection of this invention.

Claims

1. A fast power dispatching method without communication for multi-machine parallel grid-connected inverters, characterized in that, Includes the following steps: Step 1: Deploy a virtual oscillator controller with an Andronov-Hopf oscillator as its core locally on each parallel inverter. Each inverter is connected to the grid in parallel through a common coupling point, and there is no communication link between the inverters. The inverter's grid-connected three-phase current is collected in real time and the αβ axis current is obtained after abc / αβ coordinate transformation. The local αβ axis current command value is calculated based on the current active power command and reactive power command. The error between the αβ axis current and the local αβ axis current command value is sent to the virtual oscillator controller. Under steady-state operation, each inverter maintains its own voltage source external characteristics by relying on the nonlinear limit cycle dynamics of the virtual oscillator controller, and realizes the non-communication sharing of active and reactive power of each machine through the natural electrical coupling of the common coupling point. Step 2: Each inverter independently detects power dispatch commands locally; when an inverter receives a power dispatch command, proceed to Step 3; if no command is received, the inverter maintains the independent operation of the virtual oscillator controller from Step 1. Step 3: The event-triggered scheduling unit solves the target steady-state phase angle difference and the current steady-state phase angle difference based on the local impedance estimation and the grid-connected power flow equation, and generates a local auxiliary reference state; The phase of the master oscillator is continuously evolved from the current operating condition to the target operating condition through unidirectional phase traction; Step 4: The event-triggered scheduling unit independently monitors the optimal switching trigger conditions on the α and β coordinate axes respectively; when either axis first meets the trigger condition, a short-time compensation voltage is applied to the reference voltage corresponding to that axis to compress the residual error of the physical circuit of that axis; the other axis independently performs the same short-time compensation voltage application process when its corresponding trigger condition is met. Step 5: In the instantaneous neighborhood after each short-time compensation voltage removal, the inverter samples the active power change rate as the residual evaluation index. Based on the sign and magnitude of the residual evaluation index, the deviation direction and degree of the local impedance estimate relative to the true value are inferred, and the local impedance estimate is corrected according to the discrete iterative update law. Step 6: The controller continuously monitors the phase difference between the main oscillator phase and the auxiliary reference state; when the phase difference meets the synchronization criterion, the auxiliary reference state and unidirectional phase traction simultaneously exit, and the inverter returns to step 1.

2. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, The state equation of the virtual oscillator controller in the αβ coordinate system is: Among them, v α and v β The output voltage state of the controller along the α and β axes, ξ is the convergence rate constant, and k is the output voltage state along the α and β axes. v k is the voltage scaling factor. i V is the current scaling factor. n ω is the effective value of the inverter's rated voltage. nom i is the oscillator's inherent resonant frequency. α * and i β * i is the reference value for the αβ axis current. α and i β This represents the actual output current of the inverter.

3. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, According to the current active power command P * With reactive power command Q * Calculate the local αβ axis current command value i α * with i β * The specific formula is: Where, θ g V is the phase angle of the grid voltage. n This is the effective value of the inverter's rated voltage.

4. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, The formulas for the target steady-state phase angle difference and the current steady-state phase angle difference, which are derived by the event-triggered scheduling unit based on local impedance estimation and inverse solution of the grid-connected power flow equation, are as follows: In the formula, For the target steady-state phase angle difference, V n V is the nominal value of the inverter output voltage amplitude. g X represents the grid voltage amplitude, and X represents the inverter's equivalent impedance. The active power command is given by ω, where ω is the grid angular frequency and L is the total series inductance. Current power command; Generated local auxiliary reference state Specifically: In the formula, The phase of the master oscillator is triggered at the instant.

5. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, The specific method by which the event-triggered scheduling unit continuously evolves the phase of the master oscillator from the current operating condition to the target operating condition through unidirectional phase traction is as follows: An auxiliary oscillator is generated using an auxiliary reference state. This auxiliary oscillator is unidirectionally coupled to the main oscillator. The amplitude and phase angle state equations of the coupled main oscillator controller are as follows: Where K is the coupling gain, P and Q are the actual active and reactive power outputs of the inverter, respectively. * Q * These represent the current active power command and reactive power command, respectively, where ξ is the convergence rate constant, and k is the reactive power command. v k is the voltage scaling factor. i V is the current scaling factor. n ω is the effective value of the inverter's rated voltage. nom θ1 is the inherent resonant frequency of the oscillator, C is the virtual capacitance value inside the virtual oscillator, and θ1 and θ2 are the phase angles of the main oscillator and the auxiliary oscillator, respectively.

6. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, The optimal switching trigger condition is as follows: In the formula, , Let ω be the difference between the current output voltage of the master oscillator on the k-axis and the target steady-state voltage, where ω is the power grid angular frequency, and L is the voltage at which the mains oscillator operates. f ΔI is the inverter filter inductor, and ΔI is the incremental steady-state current amplitude calculated from the steady-state parameters of the old and new operating conditions.

7. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, When the α-axis first meets the triggering condition, a short-time compensation voltage U is applied to the reference voltage corresponding to the α-axis. inj The polarity of the short-time compensation voltage is opposite to the direction of the instantaneous deviation value Δi(t0) of the current at the moment of triggering the modulation wave α-axis. The duration Δt of the short-time compensation voltage is specifically as follows: In the formula, L̂ f This is the estimated value of the local filter inductance.

8. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, Within the instantaneous neighborhood after each short-time compensation voltage removal, the inverter samples the active power change rate as a residual evaluation index. Based on the sign and magnitude of the residual evaluation index, the direction and degree of deviation of the local impedance estimate relative to the true value are inferred. The specific method for correcting the local impedance estimate according to the discrete iterative update law is as follows: The rate of change of active power sampled in the instantaneous neighborhood after the k-th short-time compensation voltage removal is used as the residual evaluation index. : In the formula, t k off+ This represents the instantaneous neighborhood after the k-th short-time compensation voltage is removed. Active power; The inverter corrects the local impedance estimate according to the following discrete iterative update law: In the formula, This is the estimated value of the local filter inductance used in the k-th scheduling event. η is the estimated value of the local filter inductance used in the (k+1)th scheduling event, η is the iterative gain, sgn(ΔP) is the sign correction term for the current scheduling direction, which is positive under the power up-adjustment condition and negative under the power down-adjustment condition, and ΔP is the change in active power scheduling.

9. The method for fast power dispatching without communication for multi-machine parallel grid-connected inverters according to claim 1, characterized in that, The specific form of the synchronization criterion is as follows: Where θ1 is the phase of the master oscillator, θ2 is the phase of the local auxiliary reference state, and δ θ This is the preset phase tolerance threshold.

Images