An inverter

By combining nominal model predictive control with the theory of bounded invariant sets of disturbances to solve the problems of slow dynamic response and insufficient robustness of existing inverters under nonlinear loads, efficient and fast voltage stability control is achieved.

CN122908568BActive Publication Date: 2026-11-17JIANGSU XINKE NEW ENERGY CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611381449.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-09-08
Publication Date
2026-11-17
Estimated Expiration
2046-09-08

AI Technical Summary

Technical Problem

Existing inverter control strategies struggle to achieve efficient and rapid dynamic response and steady-state control when faced with nonlinear, time-varying, and frequently changing loads. In particular, they suffer from heavy computational burden and insufficient robustness under model mismatch and external disturbance conditions.

Method used

An inverter control method is adopted, which combines nominal model predictive control and the theory of bounded invariant set of disturbances. By calculating the state and control increment constraints offline, the online calculation burden is reduced. Furthermore, by organically superimposing the feedforward duty cycle, nominal MPC control quantity and feedback correction quantity, a highly robust voltage stability control is achieved.

Benefits of technology

It can maintain an extremely low output voltage distortion rate and a short dynamic recovery time under severe load fluctuations and filter parameter changes, with overall performance far exceeding existing PI control, deadbeat control and standard MPC solutions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122908568B_ABST
    Figure CN122908568B_ABST
Patent Text Reader

Abstract

This invention relates to the field of inverter control and discloses an inverter for achieving highly robust voltage stability control in inverter applications with limited digital control resources. The inverter includes acquiring actual operating parameters, reference parameters, and filter parameters of the inverter to calculate the feedforward control quantity; calculating error parameters, using a steady-state Kalman filter observer to acquire the error change state and combining it with the integral state to construct an augmented state vector; determining the state deviation based on this vector and the nominal state vector to obtain the feedback correction quantity, and solving the model predictive control problem under a preset constraint set to obtain the nominal control quantity; combining the feedforward control quantity, the feedback correction quantity, and the nominal control quantity to determine the total control quantity, obtaining a control signal to drive the inverter. This invention integrates a feedforward, state observation, and MPC composite control architecture, effectively suppressing nonlinear disturbances and significantly improving the control accuracy and system robustness of the inverter.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of inverter control, and more particularly to an inverter. Background Technology

[0002] With the rapid development of new energy power generation and microgrid technologies, inverters, as the core interface devices connecting the DC side and AC loads / grids, directly determine the performance and reliability of the entire system through their output power quality. In applications such as high-end industrial drives, uninterruptible power supplies, and grid-connected power generation, loads often exhibit nonlinear, time-varying, and frequent abrupt changes, posing significant challenges to inverter control algorithms.

[0003] Currently, the mainstream control strategies for inverters mainly include: Proportional-integral control combined with repetitive control: While it can effectively suppress periodic disturbances, its dynamic response is slow under aperiodic load changes, and it is difficult to achieve optimal control under multiple constraints. Deadbeat control: Relies on precise circuit parameters, is extremely sensitive to model mismatch, and is prone to causing system oscillations or even instability in actual operation. Standard model predictive control: Although it has the ability to handle multivariable constraints and excellent dynamic performance, it suffers from heavy computational burden and insufficient robustness to external disturbances and model uncertainties.

[0004] Existing composite control schemes still face the following technical bottlenecks that urgently need to be addressed: Traditional feedforward control based on differential flatness often requires obtaining the derivative information of the load current. However, in actual systems, the derivative of the load current is difficult to measure directly and contains a lot of high-frequency noise, which leads to distortion of the feedforward quantity and deteriorates the system performance.

[0005] In voltage closed-loop control, due to sampling delay and non-ideal factors, it is difficult to accurately obtain the rate of change of voltage error, which leads to the estimation deviation of the state observer and thus affects the accuracy of feedback control.

[0006] Standard MPC suffers from significant deviations between the nominal predicted trajectory and the actual system trajectory when model mismatch exists, leading to a decline in control performance. Solving the optimization problem online incurs enormous computational costs, making it difficult to meet the real-time requirements of high-frequency inverters; while explicit MPC often lacks an effective compensation mechanism for state deviations.

[0007] Therefore, we propose an inverter to solve the above problems. Summary of the Invention

[0008] This invention provides an inverter for achieving highly robust voltage stability control in inverter applications where digital control resources are limited.

[0009] The first aspect of the present invention provides an inverter, the inverter comprising: a parameter acquisition module, configured to acquire actual operating parameters of the inverter, corresponding reference parameters and their derivatives, and acquire filter parameters; a control quantity calculation module, configured to calculate a feedforward control quantity based on the actual operating parameters, the reference parameters and their derivatives, and the filter parameters; and a state observation augmentation module, configured to calculate an error parameter based on the actual operating parameters and the reference parameters, acquire the changing state of the error parameter using an observer, integrate the error parameter to obtain an integral state, and combine the error parameter with the reference parameters. The changing state and the integral state are merged to construct an augmented state vector; the model predictive control module is used to determine the state deviation based on the augmented state vector and the nominal state vector of the previous control cycle to obtain the feedback correction amount; obtain a preset constraint set, solve the model predictive control problem based on the nominal state vector of the previous control cycle and under the preset constraint set to obtain the nominal control amount, and recursively update the nominal state vector of the next control cycle; the total control amount module is used to combine the feedforward control amount, the feedback correction amount and the nominal control amount to determine the total control amount, so as to obtain the control signal to drive the inverter.

[0010] Optionally, in a first implementation of the first aspect of the present invention, the actual operating parameters include output voltage, inductor current, and load current; the reference parameters and their derivatives include reference voltage, the first derivative of the reference voltage, and the second derivative of the reference voltage. The parameter acquisition module acquires the actual operating parameters of the inverter, the corresponding reference parameters, and their derivatives, specifically for: The output voltage, the inductor current, and the load current are synchronously sampled and digitally averaged to obtain the output voltage sample value, the inductor current sample value, and the load current sample value. Based on the preset frequency and phase, the reference voltage, the first derivative of the reference voltage, and the second derivative of the reference voltage are obtained from the reference table.

[0011] Optionally, in a second implementation of the first aspect of the present invention, the filter parameters include the DC bus voltage value, the filter inductance value, the equivalent series resistance value of the filter inductance, and the filter capacitor value; the feedforward control quantity is a feedforward duty cycle signal. The control quantity calculation module is specifically used for: The first feedforward component is obtained by multiplying and adding the second derivative of the reference voltage with the value of the filter inductor and the value of the filter capacitor. The second feedforward component is obtained by multiplying and adding the first derivative of the reference voltage with the equivalent series resistance of the filter inductor and the value of the filter capacitor. The reference voltage is used as the third feedforward component; The fourth feedforward component is obtained by multiplying the sampled load current value by the equivalent series resistance value of the filter inductor. The feedforward duty cycle signal is obtained by adding the first feedforward component, the second feedforward component, the third feedforward component, and the fourth feedforward component and then dividing by the DC bus voltage value.

[0012] Optionally, in a third implementation of the first aspect of the present invention, the error parameter is a voltage error; the change state is an estimated value of the voltage error change rate; and the integral state is an error integral state. The state observation augmentation module is specifically used for: The voltage error is calculated based on the output voltage sample value and the reference voltage. The voltage error and the total control quantity output from the previous control cycle are input to a pre-established steady-state Kalman filter, and the estimated value of the voltage error change rate is obtained through recursive calculation. The error integral state of the current control cycle is obtained by adding the product of the voltage error and the control cycle constant to the error integral state of the previous control cycle. The voltage error, the estimated rate of change of the voltage error, and the error integral state of the current control cycle are combined into an augmented state vector.

[0013] Optionally, in a fourth implementation of the first aspect of the present invention, the preset constraint set includes a tightening state constraint set and a tightening control increment constraint set; The model prediction control module is specifically used for: The augmented state vector is subtracted from the nominal state vector of the previous control cycle to obtain the state deviation. Obtain the pre-designed feedback gain matrix, and multiply the feedback gain matrix by the state deviation to obtain the feedback correction amount; Using the nominal state vector of the previous control cycle as the initial state, the model predictive control problem is solved under the tightened state constraint set and the tightened control increment constraint set to obtain the nominal control quantity sequence; The first element in the nominal control quantity sequence is extracted as the nominal control quantity, and the nominal state vector of the next control cycle is recursively obtained based on the nominal control quantity and the initial state.

[0014] Optionally, in a fifth implementation of the first aspect of the present invention, the total control quantity is the total duty cycle; and the control signal is a pulse width modulation signal. The total control quantity module is specifically used for: The initial total duty cycle is obtained by adding the feedforward control quantity, the feedback correction quantity, and the nominal control quantity. The initial total duty cycle is subjected to a limiting process to obtain the total duty cycle; The total duty cycle is converted into the pulse width modulation signal, and the total duty cycle is stored as the historical total control quantity for the next control cycle.

[0015] Optionally, in a sixth implementation of the first aspect of the present invention, a constraint set calculation module is also included: Based on a pre-defined disturbance bound, a bounded invariant set of state deviations is calculated; wherein, the disturbance bound is determined by the inverter's maximum load current, the maximum relative deviation of the filter inductor, the maximum relative deviation of the filter capacitor, and the minimum duty cycle resolution of the pulse width modulation. Obtain the pre-defined augmented state constraint set and nominal control quantity constraint set; The augmented state constraint set is subtracted from the bounded invariant set of state deviations to obtain the tightened state constraint set. The set of nominal control constraints and the set obtained by applying the feedback gain matrix to the bounded invariant set of state deviations are subtracted to obtain the set of tightened control increment constraints.

[0016] Optionally, in a seventh implementation of the first aspect of the present invention, the control quantity calculation module is further configured to: Before multiplying the load current sample value with the equivalent series resistance value of the filter inductor to obtain the fourth feedforward component, the load current sample value is subjected to a first-order low-pass filter. Before multiplying and adding the second derivative of the reference voltage with the filter inductor value and the filter capacitor value to obtain the first feedforward component, the second derivative of the reference voltage is subjected to amplitude limiting.

[0017] Optionally, in an eighth implementation of the first aspect of the present invention, the state observation augmentation module obtains the estimated voltage error rate of change through recursive calculation, specifically for: Obtain the voltage error state prediction value obtained in the previous control cycle and the gain matrix of the steady-state Kalman filter; Based on the predicted voltage error state value, the voltage error, and the gain matrix, a recursive calculation is performed to obtain the estimated voltage error state value for the current control cycle. The voltage error rate of change estimate is extracted from the voltage error state estimate of the current control cycle; The state observation augmentation module is further configured to, after obtaining the error integral state of the current control cycle, compare the error integral state of the current control cycle with the integral threshold to perform a limiting update on the error integral state of the current control cycle.

[0018] Optionally, in a ninth implementation of the first aspect of the present invention, the steady-state Kalman filter in the state observation augmentation module calculates the estimated voltage error rate of change using the following recursive formula with a duty cycle attenuation factor: ; In the formula, This is an estimate of the rate of change of voltage error; This is the predicted value of the voltage error change rate in the previous control cycle; These are the Kalman gain coefficients corresponding to the rate of change dimension in the gain matrix; For voltage error; This is the predicted value for voltage error status; This is the attenuation tuning factor; This represents the total control output of the previous control cycle; the subscript k represents the current control cycle; k-1 represents the previous control cycle.

[0019] The mechanism of this invention is as follows: It combines nominal model predictive control with the theory of bounded invariant sets of disturbances, and reduces online optimization to lookup and recursive calculation based on offline partitioned explicit control laws by tightening state and control increment constraints offline. This reduces the online computational burden while ensuring that the system always meets hard constraints under disturbances and model uncertainties. It forms a composite control strategy that combines the accuracy of flat control, the disturbance rejection of state observation and the ability to handle predictive control constraints. This solves the problems of response lag and parameter sensitivity of traditional cascade control and overcomes the engineering bottleneck of large online computation of conventional MPC. Beneficial effects: It counterintuitively abandons the derivative information of the load current and only uses the load current itself combined with the derivative of the reference voltage to calculate the feedforward duty cycle. It introduces a steady-state Kalman filter and uses the total duty cycle of the previous cycle and the current voltage error to recursively estimate the voltage error change rate. This eliminates the risk of high-frequency noise being amplified through the differentiating element and greatly improves the purity and reliability of the feedforward signal. The Kalman filter provides the optimal state estimation, enabling the system to obtain a fast dynamic response and maintain extremely high steady-state accuracy when facing load changes.

[0020] The control architecture is split into two parallel dimensions: offline nominal MPC and online deviation feedback correction. Based on the disturbance bound, the bounded invariant set of state deviations is calculated offline. Then, through inter-set subtraction operations, the tightened state constraint set and the tightened control increment constraint set are derived, pre-emptively addressing the model uncertainty and the impact of external disturbances at the constraint level. This ensures that even with parameter drift or strong disturbances during online control, the actual operating state will never exceed physical limits. Combined with explicit control law lookup tables, the complex online optimization is simplified to a single matrix multiplication and addition, enabling robust optimization calculations to be completed in microseconds. This allows high-end MPC algorithms to be practically applied in low-cost embedded controllers.

[0021] The feedforward duty cycle, nominal MPC control quantity, and feedback correction quantity are organically superimposed. The three are not simply connected in parallel. The feedforward processes the known reference trajectory and load disturbance, the nominal MPC optimizes the future dynamics and meets the tightening constraints, and the feedback correction quantity compensates for model mismatch and residual bias in real time. The system has the speed of feedforward, the global optimization of MPC, and the strong robustness of feedback. Even with drastic load fluctuations and changes in filter parameters with temperature, the output voltage distortion rate can still be controlled at an extremely low level, and the dynamic recovery time is extremely short. The overall performance far exceeds that of existing PI control, deadbeat control and standard MPC solutions. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of one embodiment of the inverter in this invention; Figure 2 This is a schematic diagram of another embodiment of the inverter in this invention; Figure 3 This is a schematic diagram of one embodiment of the control device for the inverter in this invention; Figure 4 This diagram illustrates how the output voltage, inductor current, and load current are sampled by an ADC, then digitally averaged to obtain the reference voltage and its first and second derivatives. Detailed Implementation

[0023] This invention provides an inverter for achieving highly robust voltage stability control in inverter applications with limited digital control resources. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the inverter in this invention includes: 101. Parameter acquisition module, used to collect the inverter's output voltage, inductor current and load current, and determine the reference voltage, the first derivative of the reference voltage and the second derivative of the reference voltage.

[0025] It is understood that the executing entity of this invention can be the control device of an inverter, or it can be a terminal or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0026] It should be noted that, at the hardware level, the system is configured as a single-phase AC inverter with a target rated output voltage RMS value of 220V and a rated output frequency of 50Hz. The sampling frequency of the control system is set to 10000Hz, i.e., each control cycle is 100μs. The system is equipped with high-precision Hall voltage and Hall current sensors at the output, and the microprocessor integrates a 12-bit resolution analog-to-digital converter.

[0027] Physical quantity acquisition: At a sampling moment during a certain control cycle of the inverter operation (assuming it's 1 ms after the start of the inverter's fundamental cycle), the microprocessor sends a synchronous sampling command to the analog-to-digital converter. Through the voltage sensor and conditioning circuit, the actual output voltage of the inverter at this moment is read as 95.5V. Simultaneously, through current sensors connected in series in the filter inductor circuit and the load output circuit respectively, the microprocessor reads that the actual inductor current is 4.2A and the actual load current is 4.0A. The microprocessor converts these three analog signals into corresponding digital quantities and stores them in a register.

[0028] Determining the reference data: The microprocessor needs to synchronously determine the control reference target at the current moment. Based on the AC output target of 50Hz and 220V RMS, the microprocessor internally knows that the peak voltage of the target sine wave is approximately 311V and the angular frequency is approximately 314rad / s.

[0029] Since the current sampling time is 1 ms after the start of the fundamental period, the electrical phase of the corresponding sinusoidal signal is 18°. The microprocessor calculates the following reference data using its internal sine wave generator or a lookup table: Calculate the sine value at this phase and multiply it by the voltage peak value to obtain the current instantaneous reference voltage value of approximately 96.1V.

[0030] The first derivative of the reference signal is represented as a cosine wave of the same frequency, with its amplitude being the product of the peak value of the original voltage and the angular frequency. The microprocessor calculates that the first derivative of the current reference voltage is approximately 92870 V / s.

[0031] The second derivative of the reference voltage is a sine wave out of phase with the original sine wave, and its amplitude is the product of the peak value of the original voltage and the square of the angular frequency. The microprocessor calculates that the second derivative of the reference voltage at this point is approximately -9,470,000 V / s. 2 These six states and reference data are stored in the cache as known conditions for use in calculating the feedforward duty cycle.

[0032] 102. Control quantity calculation module, used to calculate the feedforward duty cycle with the output voltage as the flat output based on the reference voltage, the first derivative of the reference voltage, the second derivative of the reference voltage, the load current, the value of the filter inductor, the equivalent series resistance value of the filter inductor, and the value of the filter capacitor, and obtain the feedforward duty cycle signal.

[0033] It should be noted that the microprocessor uses differential flat control logic to directly derive and calculate the required reference drive duty cycle based on the inherent hardware parameters and the state data collected in the previous step.

[0034] Assume the DC bus supply voltage of this single-phase inverter is 400V. The fixed parameters of the system hardware are: filter inductor value of 2mH (i.e., 0.002H), filter inductor equivalent series resistance of 0.1Ω, and filter capacitor value of 10μF (i.e., 0.00001F). The microprocessor retrieves the current data: reference voltage of 96.1V, first derivative of reference voltage of 92870V / s, and second derivative of reference voltage of -9470000V / s. 2 The load current is 4.0A.

[0035] The microprocessor calculates the feedforward value of the capacitor current required to maintain the output voltage as expected. Multiplying the first derivative of the reference voltage by the filter capacitor value, the theoretical current flowing through the filter capacitor is approximately 0.93A.

[0036] Adding the capacitor current to the actual load current, we find that the theoretical target current that the inductor circuit needs to provide is 4.93A.

[0037] Calculate the voltage bias across the inductor and its equivalent resistance. The voltage drop across the resistance is obtained by multiplying the target inductor current by the equivalent series resistance, resulting in 0.49V. The dynamic induced voltage drop across the inductor itself is obtained by multiplying the second derivative of the reference voltage by the values ​​of the filter capacitor and the filter inductor, yielding a dynamic voltage drop of -0.19V.

[0038] Then, the reference voltage, resistor voltage drop, and inductor dynamic voltage drop are added together (96.1V plus 0.49V minus 0.19V) to obtain the total feedforward target voltage that the inverter bridge arm needs to output in the current cycle, which is 96.4V. The total feedforward target voltage is divided by the DC bus voltage (96.4V divided by 400V) to calculate the feedforward duty cycle at the current moment, which is 24.1% (value 0.241), and stored as the feedforward duty cycle signal in the designated register.

[0039] 103. State observation augmentation module, used to calculate voltage error based on output voltage and reference voltage, estimate voltage error change rate using state observer, integrate voltage error to obtain error integral state, and combine voltage error, voltage error change rate and error integral state into augmented state vector.

[0040] It should be noted that the microprocessor converts the scattered signals acquired earlier into comprehensive state information for closed-loop feedback. The control cycle is 100μs (i.e., 0.0001s).

[0041] The microprocessor reads the current reference voltage as 96.1V, while the actual output voltage is 95.5V. Subtracting the actual output voltage from the reference voltage yields a current voltage error of 0.6V, indicating that the current actual voltage is slightly lower than the target requirement.

[0042] To avoid directly amplifying high-frequency noise via differential amplification, the microprocessor runs an offline-designed state observer algorithm. The current 0.6V voltage error is input to the state observer, and the algorithm is derived by combining the internal state variables from the previous control cycle with the observation gain matrix. After filtering and correction, the observer outputs an estimated voltage error rate of change of 980V / s.

[0043] Using discretized digital integration, the microprocessor multiplies the current voltage error of 0.6V by the control period of 0.0001s, resulting in an integral increment of 0.00006V·s for the current period. It then retrieves the historical cumulative error integral value from the previous period (assumed to be 0.01500V·s) recorded in memory and adds the new value to the historical value, resulting in an updated error integral state of 0.01506V·s.

[0044] The microprocessor combines these three variables into a one-dimensional data set to obtain the augmented state vector at the current moment: the first element is the voltage error 0.6, the second element is the voltage error change rate 980, and the third element is the error integral state 0.01506. This vector is stored in a specified memory address to await feedback correction calls.

[0045] 104. The model predictive control module is used to subtract the augmented state vector from the nominal state vector obtained by recursion in the previous control cycle to obtain the state deviation. The feedback correction is obtained by multiplying the feedback gain designed offline by the state deviation. The nominal state vector obtained by recursion in the previous control cycle is used as the initial state. The nominal model predictive control problem is solved under the tightened state constraint set and tightened control increment constraint set calculated offline to obtain the nominal control quantity. The nominal state vector of the next control cycle is then recursively obtained based on the nominal control quantity.

[0046] It should be noted that the microprocessor performs feedback compensation and online optimization calculations based on pipe model predictive control in this step.

[0047] The microprocessor extracts the nominal state vector recursively derived at the end of the previous control cycle. Assume its predicted current voltage error is 0.58, the error change rate is 960, and the error integral state is 0.01490. The microprocessor subtracts the corresponding elements of the actual augmented state vector from the predicted nominal state vector to obtain the state deviation, and then calls the pre-calculated feedback gain matrix coefficients offline. The feedback gain is multiplied by the corresponding state deviations and summed to obtain a feedback correction of 0.0054, the sum of the three products.

[0048] The relevant numerical calculation relationships are shown in Table 1 below: Table 1 Classification of state physical quantities Augmented state vector (actual value) Nominal state vector (predicted value) State deviation (actual minus forecast) Offline feedback gain Voltage error 0.60000 0.58000 0.02000 0.0500 Error change rate 980 960 20 0.0001 Error integral state 0.01506 0.01490 0.00016 15.000 The microprocessor uses the nominal state vector recursively derived from the previous cycle as the initial observation state and initiates the quadratic programming algorithm. Considering that the total duty cycle operating range of the physical devices is limited to -0.95 to 0.95, in order to reserve sufficient safety adjustment space for the feedforward duty cycle (currently 0.241) and feedback correction, the system significantly tightens the control quantity boundaries allocated to nominal optimization. Assume that the tightened nominal control quantity constraint boundaries are set to -0.45 to 0.45. The microprocessor solves for the minimum cost function within these tightened constraint boundaries, obtaining the optimal nominal control quantity for the current cycle as 0.0120.

[0049] The microprocessor substitutes the current nominal control quantity 0.0120 and the current nominal state vector into the discretized nominal model, and predicts that the nominal voltage error of the next control cycle will converge to 0.45, the error change rate will decrease to 850, and the error integral state evolution will be 0.01495. The new value constitutes the nominal state vector of the next cycle.

[0050] 105. Total control quantity module, used to add and limit the feedforward duty cycle signal, nominal control quantity and feedback correction quantity to obtain the total duty cycle, which is used to obtain the pulse width modulation signal.

[0051] It should be noted that the microprocessor summarizes the calculation results from the aforementioned different dimensions at the physical layer and converts them into electrical signals that can be executed by the underlying driver hardware.

[0052] The system directly adds the three components together: 0.241 plus 0.0120 plus 0.0054, to obtain the theoretical total duty cycle of the current cycle as 0.2584.

[0053] To prevent damage to inverter power devices caused by extreme duty cycles, the system internally presets a safe physical boundary for the duty cycle (i.e., -0.95 to 0.95). The microprocessor compares the theoretical total duty cycle with the safe boundary. Since 0.2584 falls within this range, the limiting module determines it is safe and allows the circuit to proceed as originally set.

[0054] The logic for combining duty cycles and limiting is shown in Table 2 below: Table 2 Duty cycle component / parameter category numerical size Signal Source and Function Description Feedforward duty cycle signal 0.2410 Step 102 calculates the base duty cycle that provides the sustaining reference voltage. Nominal control quantity 0.0120 Step 104 calculates and guides the system to evolve along the ideal nominal trajectory. Feedback correction amount 0.0054 Step 104 calculates and compensates for the state observation bias in actual operation. Theoretical total duty cycle 0.2584 The sum of the above three components represents the ideal demand. Amplitude limiting safety boundary ±0.9500 Preset hardware physical security protection thresholds (upper and lower limits) Actual output total duty cycle 0.2584 The amplitude limit boundary was not triggered, and the output was based on the original theoretical total duty cycle value. The current inverter system has a switching control cycle of 100μs, and the microprocessor's internal hardware timer continuously counts at a fixed high-frequency clock. The microprocessor converts the total duty cycle of 25.84% into a time parameter: within this 100μs physical cycle, the equivalent high-level drive conduction time is 25.84μs.

[0055] The microprocessor writes the digital clock count value corresponding to 25.84μs into the comparator register of the internal pulse width modulation generator. When the hardware timer count value increments and matches the value in the comparator register, the external output pin flips, resulting in a digital pulse square wave with a high level width precisely corresponding to 25.84μs. This square wave is then transmitted to the gate drive circuit to control the operation of the power switching devices.

[0056] Please see Figure 2 and Figure 4 Another embodiment of the inverter in this invention includes: 201. Parameter acquisition module, used to collect the inverter's output voltage, inductor current and load current, and determine the reference voltage, the first derivative of the reference voltage and the second derivative of the reference voltage.

[0057] Specifically, at the start of the control cycle, sampling of the output voltage, inductor current, and load current is synchronously triggered, and the sampled data is digitally averaged to obtain the output voltage sample value, inductor current sample value, and load current sample value. Based on the frequency and phase of the reference voltage, the reference voltage is read from a pre-stored sine reference table, and the first derivative and second derivative of the reference voltage are read from a pre-stored derivative reference table to obtain the reference voltage, the first derivative of the reference voltage, and the second derivative of the reference voltage.

[0058] It should be noted that the target rated output voltage RMS value of this single-phase AC inverter is set to 220V, and the rated output frequency is set to 50Hz. The microprocessor's control and switching frequency is set to 10000Hz, meaning that the duration of each control cycle is 100μs.

[0059] At the 2ms mark within the fundamental cycle of the inverter operation, a new control cycle begins. At this time, the timer inside the microprocessor generates a synchronization trigger signal at the start of the cycle, directly activating the analog-to-digital converter to synchronously sample the output voltage, inductor current, and load current channels.

[0060] To eliminate the high-frequency switching noise picked up by the sensor, the system adopts an oversampling strategy, that is, to continuously perform four conversions and acquisitions on each physical quantity within a very short time (within 2μs).

[0061] For the output voltage channel, the four raw digital values ​​continuously acquired by the analog-to-digital converter correspond to actual voltages of 181.2V, 181.6V, 181.8V, and 181.4V, respectively. The microprocessor adds these four values ​​together and divides them by 4 to perform digital averaging, resulting in a smoothed output voltage sample value of 181.5V.

[0062] For the inductor current channel, the four raw current values ​​collected consecutively were 8.4A, 8.6A, 8.5A and 8.5A. After digital averaging, the inductor current sample value was 8.5A.

[0063] For the load current channel, the four raw current values ​​collected consecutively were 7.9A, 8.1A, 8.0A and 8.0A. After digital averaging, the resulting load current sample value was 8.0A.

[0064] To save on the overhead of trigonometric function floating-point operations in real-time control by the microprocessor, the system pre-stores three data tables in non-volatile memory: a sine reference table, a first-order derivative reference table, and a second-order derivative reference table.

[0065] Given that the inverter output frequency is 50Hz, corresponding to a complete sine wave fundamental cycle of 20ms. At a control frequency of 10000Hz, one fundamental wave cycle contains 200 control cycles. Therefore, the three tables pre-obtained by the system each contain 200 discrete data points, corresponding to index values ​​from 0 to 199.

[0066] The current time is 2ms after the start of the fundamental frequency period. The microprocessor calculates the current phase index value as 20 based on the runtime. The microprocessor uses this index value as an address pointer to synchronously access the pre-stored data table. The microprocessor reads the current reference voltage, which is 182.9V, from the 20th address of the sine reference table.

[0067] The microprocessor reads the first derivative of the reference voltage at that moment from address 20 of the first derivative reference table; the value is 79076 V / s. This value reflects the slope characteristic of the current reference voltage being in the rising phase.

[0068] The microprocessor reads the second derivative of the reference voltage at that moment from address 20 of the second derivative reference table; the value is -18055431 V / s². This value reflects the change in the curvature of the current reference waveform.

[0069] 202. Control quantity calculation module, used to calculate the feedforward duty cycle with the output voltage as the flat output based on the reference voltage, the first derivative of the reference voltage, the second derivative of the reference voltage, the load current, the value of the filter inductor, the equivalent series resistance value of the filter inductor, and the value of the filter capacitor, and obtain the feedforward duty cycle signal.

[0070] Specifically, the pre-stored DC bus voltage value, filter inductance value, equivalent series resistance value of filter inductance, and filter capacitor value are read; the second derivative of the reference voltage is multiplied and added with the filter inductance value and the filter capacitor value to obtain the first feedforward component; the first derivative of the reference voltage is multiplied and added with the equivalent series resistance value of filter inductance and the filter capacitor value to obtain the second feedforward component; and the reference voltage is used as the third feedforward component. The fourth feedforward component is obtained by multiplying the load current by the equivalent series resistance of the filter inductor; the feedforward duty cycle signal is obtained by adding the first, second, third and fourth feedforward components and dividing by the DC bus voltage; the derivative information of the load current is not used in the calculation of the feedforward duty cycle signal.

[0071] Furthermore, before multiplying the load current by the equivalent series resistance of the filter inductor to obtain the fourth feedforward component, the acquired load current is subjected to a first-order low-pass filter to obtain the filtered load current signal, and the filtered load current signal is used in the calculation of the fourth feedforward component. Before multiplying and adding the second derivative of the reference voltage with the values ​​of the filter inductor and filter capacitor to obtain the first feedforward component, the second derivative of the reference voltage is subjected to amplitude limiting to obtain the amplitude-limited second derivative signal, which is then used in the calculation of the first feedforward component. The first-order low-pass filtering and amplitude limiting are used to suppress the interference of high-frequency noise on the feedforward duty cycle signal.

[0072] It should be noted that the microprocessor uses pre-set hardware parameters, combined with the acquired reference and sampled signals, to independently calculate and synthesize multiple feedforward components using differential flatness theory.

[0073] The control cycle is still at the 2ms mark of the fundamental cycle. The microprocessor reads the core physical parameters of the single-phase inverter hardware platform: DC bus voltage is 400V, filter inductance is 0.002H, equivalent series resistance of filter inductance is 0.1Ω, and filter capacitor is 0.00001F. The microprocessor retrieves the current status data: reference voltage is 182.9V, first derivative of reference voltage is 79076V / s, and second derivative of reference voltage is -18055431V / s. 2 The load current sampling value is 8.0A.

[0074] To suppress high-frequency noise interference, the microprocessor preprocesses some sensitive signals.

[0075] The microprocessor inputs the sampled load current value into a first-order low-pass digital filter. Assuming the filtered load current buffered in the previous control cycle is 7.95A, combined with the current sampled value of 8.0A and the preset filter coefficients, the current filtered load current signal obtained after first-order low-pass filtering and smoothing is 7.96A.

[0076] The system's pre-set safety limit threshold for the second derivative of the reference voltage is ±25,000,000 V / s. 2 The microprocessor will convert the current second derivative to -18055431 V / s. 2 Compared to this threshold, since it is within the safe range, the limiter does not truncate it, and the original value of the second derivative signal after limiting remains -18055431V / s. 2 .

[0077] The microprocessor calculates the four feedforward voltage components using four sets of arithmetic channels: 1. First feedforward component: Reflects the dynamic voltage drop generated by the capacitor current on the inductor. Multiplying the second derivative signal after limiting (-18055431V / s²) with the filter inductor value (0.002H) and the filter capacitor value (0.00001F) yields the first feedforward component as -0.36V.

[0078] 2. Second feedforward component: Reflects the voltage drop across the equivalent resistance of the inductor caused by the capacitor current. Multiplying the first derivative of the reference voltage (79076V / s) by the equivalent series resistance of the filter inductor (0.1Ω) and the value of the filter capacitor (0.00001F), the second feedforward component is calculated to be 0.08V.

[0079] 3. Third feedforward component: Directly reflects the basic requirements of the target output voltage. The reference voltage is directly assigned to this component, which is 182.9V.

[0080] 4. Fourth feedforward component: Reflects the static voltage drop across the resistor caused by the load current. Multiplying the filtered load current signal (7.96A) by the equivalent series resistance of the filter inductor (0.1Ω), the fourth feedforward component is calculated to be 0.80V.

[0081] The derivative of the load current was not used in this process, which effectively avoided numerical divergence caused by sudden load changes.

[0082] The microprocessor adds the first feedforward component (-0.36V), the second feedforward component (0.08V), the third feedforward component (182.9V), and the fourth feedforward component (0.80V) to obtain the target inverter bridge arm output voltage of 183.42V.

[0083] Divide it by the DC bus voltage (400V) to get a value of 0.45855. After rounding, the feedforward duty cycle signal is determined to be 0.4586 and stored in the register.

[0084] 203. State observation augmentation module, used to calculate voltage error based on output voltage and reference voltage, estimate voltage error change rate using state observer, integrate voltage error to obtain error integral state, and combine voltage error, voltage error change rate and error integral state into augmented state vector.

[0085] Specifically, the voltage error signal is calculated based on the output voltage and the reference voltage; the voltage error signal and the total duty cycle of the output from the previous control cycle are input to a pre-designed steady-state Kalman filter, and the estimated value of the voltage error rate of change is obtained through recursive calculation; the error integral state obtained from the previous control cycle is added to the product of the voltage error signal and the control cycle constant to obtain the error integral state of the current control cycle; the voltage error signal, the estimated value of the voltage error rate of change, and the error integral state of the current control cycle are combined into an augmented state vector; the augmented state vector is used for subsequent calculation of the state deviation.

[0086] Furthermore, it also includes: reading the steady-state Kalman filter gain matrix pre-stored in the controller's non-volatile memory; performing recursive calculations based on the voltage error state prediction value, voltage error signal, and steady-state Kalman filter gain matrix obtained in the previous control cycle to obtain the voltage error state estimate for the current control cycle; extracting the voltage error rate of change estimate from the voltage error state estimate for the current control cycle; after calculating the error integral state for the current control cycle, comparing the error integral state for the current control cycle with the pre-set upper and lower limits of integration to obtain the limited error integral state, and replacing the error integral state for the current control cycle with the limited error integral state; merging the limited error integral state with the voltage error signal and the voltage error rate of change estimate into an augmented state vector.

[0087] It should be noted that the microprocessor primarily constructs the comprehensive error state for closed-loop control through discrete numerical calculations and an adaptive filtering algorithm incorporating total duty cycle compensation. (Control period constant) s.

[0088] The current reference voltage is read as 182.9V, and the actual output voltage is 181.5V. Subtracting the output voltage from the reference voltage, the calculated voltage error signal for the current control cycle is 1.4V, indicating a certain lag in the current output.

[0089] The microprocessor reads the steady-state Kalman filter gain matrix and retrieves the data from the previous control cycle. Assuming the total duty cycle of the output in the previous cycle is 0.46, the system recursively predicts the current voltage error state to be 1.35V, and the predicted voltage error rate of change is 8100.00V / s.

[0090] To suppress nonlinear disturbances in the dead zone of the bridge arm, this embodiment uses the following recursive formula for the state observer with a duty cycle attenuation factor: ; The formula parameters are as follows: This is the estimated rate of change of the current voltage error (V / s). The predicted value of the voltage error change rate in the previous cycle (8100.00V / s). The Kalman gain coefficient in the rate of change dimension (preset to 2500s⁻¹). The current voltage error signal is 1.4V. The voltage error state prediction value for the previous cycle is 1.35V. This is the attenuation tuning factor (set to 0.1). The total duty cycle output in the previous control cycle is 0.46; the subscript k represents the current control cycle; k-1 represents the previous control cycle.

[0091] Calculate the predicted residual: 1.4V - 1.35V = 0.05V.

[0092] Calculate the exponential decay term: .

[0093] State correction term: 2500s - 1 × 0.05V × 0.955 = 119.375V / s, rounded to two decimal places, is 119.38V / s.

[0094] The current estimated rate of change of voltage error is: 8100.00V / s + 119.38V / s = 8219.38V / s.

[0095] The error integral state from the previous cycle is retrieved as 0.02500 V·s. Multiplying the voltage error signal (1.4V) by the control cycle constant (0.0001s) yields the integral increment for the current cycle as 0.00014 V·s. After summing, the initial value of the current integral state is 0.02514 V·s.

[0096] The system's preset integration upper and lower limits are ±0.05000 V·s. Since 0.02514 V·s is within the safe range, this value is directly taken as the error integration state after limiting.

[0097] The microprocessor reorganizes the variables to obtain the current augmented state vector: voltage error signal 1.4, rate of change estimate 8219.38, error integral state 0.02514.

[0098] 204. The Model Predictive Control module is used to subtract the augmented state vector from the nominal state vector obtained by recursion in the previous control cycle to obtain the state deviation. The feedback correction is obtained by multiplying the feedback gain designed offline by the state deviation. The nominal state vector obtained by recursion in the previous control cycle is used as the initial state. The nominal model predictive control problem is solved under the tightened state constraint set and tightened control increment constraint set calculated offline to obtain the nominal control quantity. The nominal state vector of the next control cycle is then recursively obtained based on the nominal control quantity.

[0099] Specifically, the augmented state vector is subtracted from the nominal state vector obtained by recursion in the previous control cycle to obtain the state deviation; the feedback gain matrix stored offline is read, and the feedback gain matrix is ​​multiplied by the state deviation to obtain the feedback correction amount; the tightened state constraint set and tightened control increment constraint set obtained by subtraction between sets from the bounded invariant set of state deviation stored offline are read. Using the nominal state vector obtained from the previous control cycle as the initial state, the nominal model predictive control problem is solved under the constraints of the tightened state constraint set and the tightened control increment constraint set to obtain the nominal control quantity sequence. The first element is extracted from the nominal control quantity sequence as the nominal control quantity. The nominal state vector of the next control cycle is obtained by recursion based on the nominal control quantity and the initial state, and the nominal state vector is stored.

[0100] Further, solving the nominal model predictive control problem includes: reading the explicit control law partition information stored offline in the controller's non-volatile memory, which contains multiple state partitions and the control gain vector and bias constant value corresponding to each state partition; using the nominal state vector obtained recursively from the previous control cycle as the initial state vector, comparing it sequentially with the boundary conditions of each state partition to determine the target state partition to which the initial state vector belongs; and reading the control gain vector and bias constant value corresponding to the target state partition. The nominal control quantity is obtained by multiplying and adding the control gain vector and the initial state vector, and then adding the bias constant value. Based on the initial state vector and the nominal control quantity, a one-step recursive operation is performed using the nominal system state matrix and the nominal system input matrix stored offline to obtain and store the nominal state vector of the next control cycle. The nominal control quantity is used to add to the feedforward duty cycle signal and the feedback correction quantity in the subsequent operation, and the nominal state vector of the next control cycle is used to calculate the state deviation in the next control cycle.

[0101] It should be noted that the microprocessor performs online evaluation and state-space propagation based on an explicit model predictive control strategy. Retrieve the current augmented state vector (1.4, 8219.38, 0.02514) and the nominal state vector recursively derived from the previous cycle (1.35, 8100.00, 0.02500).

[0102] Subtracting the corresponding values ​​from the two sets: 1.4 minus 1.35 gives a deviation of 0.05; 8219.38 minus 8100.00 gives a deviation of 119.38; 0.02514 minus 0.02500 gives a deviation of 0.00014.

[0103] The coefficients of the offline feedback gain matrix were read as 0.04, 0.00005, and 10.0, respectively.

[0104] Calculate the feedback correction amount: 0.04 × 0.05 = 0.0020; 0.00005 × 119.38 ≈ 0.0060; 10.0 × 0.00014 = 0.0014. Adding them together, the feedback correction amount is 0.0094.

[0105] The microprocessor reads the explicit control law partition information from memory. The nominal state vector (1.35, 8100.00, 0.02500) from the previous cycle is used as the initial state vector for polyhedral boundary comparison.

[0106] Assume it belongs to the fifth zone. The control gain vector coefficients corresponding to the fifth zone are read as 0.010, 0.000005, and 0.4, and the bias constant value is 0.0050.

[0107] Perform multiplication and addition operations: 0.010 × 1.35 = 0.0135; 0.000005 × 8100.00 = 0.0405; 0.4 × 0.02500 = 0.0100. The sum is 0.0640. Adding the bias constant 0.0050, the nominal control value for the current control cycle is 0.0690. This value is safely limited within the preset ±0.120 tightening control range.

[0108] The microprocessor uses the nominal system state matrix and input matrix, combined with the initial state vector and the obtained nominal control quantity 0.0690, to perform discrete recursion.

[0109] The predicted nominal system state for the next control cycle is as follows: the nominal voltage error decreases to 1.20, the error change rate decreases to 7800.00, and the error integral state increases slightly to 0.02515.

[0110] 205. Total control quantity module, used to add and limit the feedforward duty cycle signal, nominal control quantity and feedback correction quantity to obtain the total duty cycle, which is used to obtain the pulse width modulation signal.

[0111] Specifically, the feedforward duty cycle signal, nominal control quantity, and feedback correction quantity are added to obtain the initial total duty cycle. The initial total duty cycle is compared with the preset upper and lower duty cycle values. If the initial total duty cycle is greater than the upper duty cycle value, the upper duty cycle value is taken as the total duty cycle after limiting. If the initial total duty cycle is less than the lower duty cycle value, the lower duty cycle value is taken as the total duty cycle after limiting. Otherwise, the initial total duty cycle is taken as the total duty cycle after limiting. The total duty cycle after limiting is written into the pulse width modulation comparison register to obtain the pulse width modulation signal. The total duty cycle after limiting is stored as the total duty cycle output in the previous control cycle for use by the state observer in the next control cycle.

[0112] It should be noted that the microprocessor performs physical layer synthesis of the duty cycle components of all the aforementioned control dimensions, converting them into digital pulse signals.

[0113] The microprocessor reads three core components: the feedforward duty cycle signal 0.4586, the nominal control quantity 0.0690, and the feedback correction quantity 0.0094.

[0114] Adding these three values ​​together: 0.4586 + 0.0690 + 0.0094, we get the initial total duty cycle as 0.5370.

[0115] The inverter's physical safety boundary is set as an upper limit of 0.95 and a lower limit of -0.95 for the duty cycle. The initial total duty cycle of 0.5370 is within the safe range, so the limiting module is allowed to proceed as originally set. After limiting is determined, the total duty cycle is 0.5370.

[0116] The data synthesis relationship and limiting logic are shown in Table 3 below: Table 3 Duty cycle component / parameter category numerical size Signal Source and Function Description Feedforward duty cycle signal 0.4586 Step 202 calculates the fundamental driving force that sustains the reference voltage operating trajectory. Nominal control quantity 0.0690 Step 204 calculates and guides the system state to evolve along the optimal constraint trajectory. Feedback correction amount 0.0094 Step 204 calculates and compensates for parameter perturbations and external deviations during actual operation. Initial total duty cycle 0.5370 The sum of feedforward, nominal, and feedback terms represents the total regulation demand. Duty cycle upper and lower limits ±0.95 Pre-defined hardware safety protection boundaries prevent direct short circuits in bridge arms. Total duty cycle after limiting 0.5370 If the value is within the upper and lower limits and truncation is not triggered, output the original value. With a physical cycle of 100μs, the microprocessor converts the total duty cycle of 53.70% into specific time parameters: the equivalent high-level drive conduction time within this cycle should be 53.70μs.

[0117] The digital clock count value corresponding to 53.70μs is written into the comparator register of the internal pulse width modulation generator. When the timer matches, the pin level is triggered to flip, resulting in a precise high-level square wave to drive the power device.

[0118] 206. The constraint set calculation module is used to solve the state feedback gain matrix offline based on the filter inductance value, the equivalent series resistance value of the filter inductance, the filter capacitor value, and the pre-set state weight matrix and control weight matrix; determine the bounded invariant set of state deviation based on the disturbance bound; perform inter-set subtraction operation between the augmented state constraint set and the bounded invariant set of state deviation to obtain the tightened state constraint set; perform inter-set subtraction operation between the nominal control quantity constraint set and the set obtained by applying the state feedback gain matrix to the bounded invariant set of state deviation to obtain the tightened control increment constraint set; write the state feedback gain matrix, the tightened state constraint set, and the tightened control increment constraint set into the controller's non-volatile memory; and subsequently read the state feedback gain matrix, the tightened state constraint set, and the tightened control increment constraint set from the controller's non-volatile memory for use.

[0119] Furthermore, the offline solution of the state feedback gain matrix, determination of the bounded invariant set of state deviations, calculation of the tightened state constraint set and tightened control increment constraint set includes: determining the disturbance bound based on the maximum allowable load current of the inverter, the maximum relative deviation of the filter inductor value, the maximum relative deviation of the filter capacitor value, and the minimum duty cycle resolution of the pulse width modulation; using the disturbance bound as input, calculating the bounded invariant set of state deviations using the iterative polyhedral set operation method, and stopping the iteration when the bounded invariant set of state deviations no longer expands, thus obtaining the bounded invariant set of state deviations; The augmented state constraint set and the bounded invariant state deviation set are subtracted to obtain the tightened state constraint set. The nominal control constraint set and the set obtained by applying the state feedback gain matrix to the bounded invariant state deviation set are subtracted to obtain the tightened control increment constraint set. The bounded invariant state deviation set, the tightened state constraint set, and the tightened control increment constraint set are all written into the controller's non-volatile memory for subsequent retrieval.

[0120] It should be noted that a discrete state-space model of the system is established based on the filter inductance value (0.002H), equivalent series resistance (0.1Ω), and capacitance (0.00001F). The state weights and control weight matrices are defined, and the optimal state feedback gain matrix is ​​obtained by solving the algebraic matrix equations. The gain coefficients for voltage error, error rate of change, and error integral are determined to be 0.04, 0.00005, and 10.0, respectively.

[0121] Based on the inverter's maximum allowable load current of 15A, the maximum relative deviation of inductor and capacitor values ​​(±10%), and a pulse width modulation resolution of 0.1%, a polyhedral disturbance boundary is established. An iterative polyhedral set operation method is employed, using a feedback gain matrix for closed-loop recursion. Iteration continues until the polyhedral boundary no longer expands, yielding a bounded, invariant set of state deviations, representing the maximum possible state shift of the system under the most extreme disturbance.

[0122] To allow sufficient safety margin for dynamic operation, the system performs subtraction operations between sets: State constraint processing: Original voltage error limit ±10.0V, rate of change limit ±20000V / s, integral limit ±0.050V·s. After subtracting the redundant boundary of the bounded invariant set of state deviations, the tightened state constraint set is obtained.

[0123] Control constraint handling: Considering that the feedforward duty cycle already consumes a significant amount of resources, the original safe adjustment range (nominal control constraint set) allocated to the nominal model predictive control is ±0.150. The equivalent duty cycle resources required for limit disturbance feedback are defined as ±0.030. Subtracting the two, the tightened control increment constraint set boundary is obtained as ±0.120.

[0124] The parameters for shrinking the constraint set are shown in Table 4 below: Table 4 Parameter Dimension Categories Original constraint set physical boundary State deviation bounded invariant set mapping extrema Tightened constraint set boundary after set subtraction Voltage error ±10.0V ±0.5V ±9.5V Voltage error change rate ±20000V / s ±1000V / s ±19000V / s Error integral state ±0.050V·s ±0.001V·s ±0.049V·s Duty cycle allocated to predictive control ±0.150 ±0.030 (Feedback Consumption Limit) ±0.120 The microprocessor encapsulates the state feedback gain matrix, the tightened state constraint set obtained from Table 4, and the tightened control increment constraint set data. This data is then uniformly programmed and written into non-volatile memory. During subsequent operation, the system only needs to quickly read and call this fixed parameter matrix set, reducing the computational load on real-time control calculations.

[0125] Figure 3 This is a schematic diagram of the structure of a control device for an inverter according to an embodiment of the present invention. This device 300 can vary considerably due to differences in configuration or performance. Device 300 includes a transmitter 301, a receiver 302, and a processor 303. The processor 303 can also be a controller. Figure 3 The device is designated as "controller / processor 303". Optionally, the device 300 may also include a modem processor 305, which may include an encoder 306, a modulator 307, a decoder 308, and a demodulator 309.

[0126] In one example, transmitter 301 modulates (e.g., analog-to-analog conversion, filtering, amplification, and up-conversion, etc.) the output sample to obtain an uplink signal, which is transmitted via an antenna to an access network device. On the downlink, the antenna receives the downlink signal transmitted by the access network device. Receiver 302 modulates (e.g., filtering, amplification, down-conversion, and digitization, etc.) the signal received from the antenna and provides an input sample. In modem processor 305, encoder 306 receives service data and signaling messages to be transmitted on the uplink and processes (e.g., formatting, encoding, and interleaving) the service data and signaling messages. Modulator 307 further processes (e.g., symbol mapping and modulation) the encoded service data and signaling messages and provides an output sample. Demodulator 309 processes (e.g., demodulates) the input sample and provides a symbol estimate. Decoder 308 processes (e.g., deinterleaving and decoding) the symbol estimate and provides decoded data and signaling messages to device 300. Encoder 306, modulator 307, demodulator 309, and decoder 308 can be implemented by a combined modem processor 305. These units perform processing according to the radio access technology adopted by the radio access network (e.g., LTE and other evolved systems access technologies). It should be noted that when device 300 does not include modem processor 305, the above-mentioned functions of modem processor 305 can also be performed by processor 303.

[0127] The processor 303 controls and manages the operation of the device 300, and is used to execute the processing procedures performed by the device 300 in the above embodiments of this disclosure. For example, the processor 303 is also used to execute various steps of the transmitting or receiving device in the above method embodiments, and / or other steps of the technical solutions described in the embodiments of this disclosure.

[0128] Furthermore, the device 300 may also include a memory 304 for storing program code and data for the device 300.

[0129] Understandable, Figure 3 Only a simplified design of device 300 is shown. In practical applications, device 300 can include any number of transmitters, receivers, processors, modem processors, memory, etc., and all devices that can implement the embodiments of this disclosure are within the protection scope of the embodiments of this disclosure.

[0130] The present invention also provides a control device for an inverter, the control device for the inverter including a memory and a processor, the memory storing computer-readable instructions, and when the computer-readable instructions are executed by the processor, causing the processor to perform the steps of the inverter in the above embodiments.

[0131] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the inverter.

[0132] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0133] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0134] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An inverter, characterized in that, include: The parameter acquisition module is used to acquire the actual operating parameters of the inverter, the corresponding reference parameters and their derivatives, and the filter parameters of the inverter; wherein, the actual operating parameters include output voltage, inductor current and load current; the reference parameters and their derivatives include reference voltage, the first derivative of the reference voltage and the second derivative of the reference voltage; the filter parameters include DC bus voltage value, filter inductance value, filter inductance equivalent series resistance value and filter capacitor value; The control quantity calculation module is used to calculate the feedforward control quantity based on the actual operating parameters, the reference parameters and their derivatives, and the filter parameters; wherein, the feedforward control quantity is a feedforward duty cycle signal; the control quantity calculation module is specifically used to: multiply and add the second derivative of the reference voltage with the filter inductor value and the filter capacitor value to obtain a first feedforward component; multiply and add the first derivative of the reference voltage with the equivalent series resistance value of the filter inductor and the filter capacitor value to obtain a second feedforward component; use the reference voltage as a third feedforward component; multiply the load current with the equivalent series resistance value of the filter inductor to obtain a fourth feedforward component; add the first feedforward component, the second feedforward component, the third feedforward component, and the fourth feedforward component and divide by the DC bus voltage value to obtain the feedforward duty cycle signal; The state observation augmentation module is used to calculate error parameters based on the actual operating parameters and the reference parameters, use an observer to obtain the changing state of the error parameters, integrate the error parameters to obtain the integral state, and merge the error parameters, the changing state and the integral state to construct an augmented state vector. The model predictive control module is used to determine the state deviation based on the augmented state vector and the nominal state vector of the previous control cycle to obtain the feedback correction amount; obtain a preset constraint set; solve the model predictive control problem based on the nominal state vector of the previous control cycle and under the preset constraint set to obtain the nominal control amount; and recursively update the nominal state vector of the next control cycle. The total control quantity module is used to combine the feedforward control quantity, the feedback correction quantity, and the nominal control quantity to determine the total control quantity, so as to obtain a control signal to drive the inverter.

2. The inverter according to claim 1, characterized in that, The parameter acquisition module acquires the actual operating parameters of the inverter, the corresponding reference parameters, and their derivatives, specifically for: The output voltage, the inductor current, and the load current are synchronously sampled and digitally averaged to obtain the output voltage sample value, the inductor current sample value, and the load current sample value. Based on the preset frequency and phase, the reference voltage, the first derivative of the reference voltage, and the second derivative of the reference voltage are obtained from the reference table.

3. An inverter according to claim 2, characterized in that, The error parameter is the voltage error; the change state is the estimated rate of change of the voltage error; the integral state is the error integral state. The state observation augmentation module is specifically used for: The voltage error is calculated based on the output voltage sample value and the reference voltage. The voltage error and the total control quantity output from the previous control cycle are input to a pre-established steady-state Kalman filter, and the estimated value of the voltage error change rate is obtained through recursive calculation. The error integral state of the current control cycle is obtained by adding the product of the voltage error and the control cycle constant to the error integral state of the previous control cycle. The voltage error, the estimated rate of change of the voltage error, and the error integral state of the current control cycle are combined into an augmented state vector.

4. An inverter according to claim 1, characterized in that, The preset constraint set includes a tightening state constraint set and a tightening control increment constraint set; The model prediction control module is specifically used for: The augmented state vector is subtracted from the nominal state vector of the previous control cycle to obtain the state deviation. Obtain the pre-designed feedback gain matrix, and multiply the feedback gain matrix by the state deviation to obtain the feedback correction amount; Using the nominal state vector of the previous control cycle as the initial state, the model predictive control problem is solved under the tightened state constraint set and the tightened control increment constraint set to obtain the nominal control quantity sequence; The first element in the nominal control quantity sequence is extracted as the nominal control quantity, and the nominal state vector of the next control cycle is recursively obtained based on the nominal control quantity and the initial state.

5. An inverter according to claim 1, characterized in that, The total control quantity is the total duty cycle; the control signal is a pulse width modulation signal. The total control quantity module is specifically used for: The initial total duty cycle is obtained by adding the feedforward control quantity, the feedback correction quantity, and the nominal control quantity. The initial total duty cycle is subjected to a limiting process to obtain the total duty cycle; The total duty cycle is converted into the pulse width modulation signal, and the total duty cycle is stored as the historical total control quantity for the next control cycle.

6. An inverter according to claim 4, characterized in that, It also includes a constraint set calculation module: Based on a pre-defined disturbance bound, a bounded invariant set of state deviations is calculated; wherein, the disturbance bound is determined by the inverter's maximum load current, the maximum relative deviation of the filter inductor, the maximum relative deviation of the filter capacitor, and the minimum duty cycle resolution of the pulse width modulation. Obtain the pre-defined augmented state constraint set and nominal control quantity constraint set; The augmented state constraint set is subtracted from the bounded invariant set of state deviations to obtain the tightened state constraint set. The set of nominal control constraints and the set obtained by applying the feedback gain matrix to the bounded invariant set of state deviations are subtracted to obtain the set of tightened control increment constraints.

7. An inverter according to claim 2, characterized in that, The control quantity calculation module is also used for: Before multiplying the load current sample value with the equivalent series resistance value of the filter inductor to obtain the fourth feedforward component, the load current sample value is subjected to a first-order low-pass filter. Before multiplying and adding the second derivative of the reference voltage with the filter inductor value and the filter capacitor value to obtain the first feedforward component, the second derivative of the reference voltage is subjected to amplitude limiting.

8. An inverter according to claim 3, characterized in that, The state observation augmentation module obtains the estimated voltage error rate of change through recursive calculation, specifically for: Obtain the voltage error state prediction value obtained in the previous control cycle and the gain matrix of the steady-state Kalman filter; Based on the predicted voltage error state value, the voltage error, and the gain matrix, a recursive calculation is performed to obtain the estimated voltage error state value for the current control cycle. The voltage error rate of change estimate is extracted from the voltage error state estimate of the current control cycle; The state observation augmentation module is further configured to, after obtaining the error integral state of the current control cycle, compare the error integral state of the current control cycle with the integral threshold to perform a limiting update on the error integral state of the current control cycle.

9. An inverter according to claim 8, characterized in that, The steady-state Kalman filter in the state observation augmentation module calculates the estimated voltage error rate of change using the following recursive formula with a duty cycle attenuation factor: ; In the formula, This is an estimate of the rate of change of voltage error; This is the predicted value of the voltage error change rate in the previous control cycle; These are the Kalman gain coefficients corresponding to the rate of change dimension in the gain matrix; For voltage error; This is the predicted value for voltage error status; This is the attenuation tuning factor; This represents the total control output of the previous control cycle; the subscript k represents the current control cycle; k-1 represents the previous control cycle.

Citation Information

Patent Citations

  • Inverter control system of sensor without back electromotive force and control method thereof

    CN106374765A

  • Kalman filtering active damping iterative learning model prediction control method of grid-connected inverter

    CN122178445A