A seamless switching method for operation modes of a multi-level high-voltage frequency converter
By predicting and canceling phase impacts in real time during the switching process of multi-level high-voltage frequency converters, the problems of phase slippage and current pulse at the moment of transfer are solved, achieving seamless switching and stable equipment operation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING HUICHUAN TECHNOLOGY CO LTD
- Filing Date
- 2025-12-22
- Publication Date
- 2026-05-29
AI Technical Summary
Existing synchronous transfer schemes cannot effectively avoid phase slippage and torque fluctuations and current pulses on the motor side during the switching process of multi-level high-voltage frequency converters, resulting in unstable equipment operation.
By observing and predicting the phase impact at candidate transfer times in real time, the transfer is carried out at the time with the smallest impact. The expected impact is actively offset by phase pre-shaping, and the mapping coefficient is subsequently corrected to optimize the switching process.
It effectively reduces phase slippage during the transfer, avoids torque fluctuations and current pulses on the motor side, improves switching accuracy and equipment operation stability, and reduces equipment wear and failure risks.
Smart Images
Figure CN121643586B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of high-voltage frequency converter control technology, and more specifically, to a method for seamless switching of operating modes of a multi-level high-voltage frequency converter. Background Technology
[0002] Medium and high voltage drive systems widely use multi-level high voltage frequency converters, including topologies such as cascaded units, three-level diodes, clamped flying capacitors, etc., to drive high-power induction motors or synchronous motors. In order to achieve switching between maintenance, operation, or energy-saving operation in the field, a synchronous transfer scheme is often configured. It is necessary to align the amplitude, frequency, and phase of the frequency converter output with the grid voltage within a short transfer window, and then complete the closed transition through a solid-state transfer switch or bypass switch. The engineering side has strict requirements for switching, such as controlling torque fluctuation overvoltage, shaft voltage, and electromagnetic compatibility limits to ensure seamless transfer.
[0003] The output of the multilevel topology is synthesized from discrete levels according to the modulation duty cycle, and is not a continuous value. Before and after entering the synchronous transfer, the carrier relationship of the control strategy, zero-sequence injection, and the power units involved in the operation may change. At the same time, the DC-side capacitor voltage of each power unit is affected by the load history and sampling time, and is difficult to be completely consistent. The superposition of these two types of changes at the transfer point will cause the equivalent phase voltage to have a small amplitude but concentrated time position step. Although this step does not significantly change the steady-state amplitude-frequency alignment result, it will form an instantaneous net slip in the spatial phase, which will lead to a short-time torque and current pulse response on the motor side.
[0004] Existing synchronous transfer schemes mostly rely on steady-state error for closed transition, such as checking the proximity of amplitude and frequency within a few fundamental frequency cycles, and issuing a transfer action after a predetermined threshold is met. This approach ignores the discrete level reconstruction effect within one to two switching cycles during the transfer instant. Even if the steady-state alignment is good, the equivalent voltage step at the transfer point will still cause a significant phase slip, ultimately resulting in the switching process failing to meet the stringent requirements of the engineering side for seamless transfer, and causing impact phenomena such as torque fluctuations and current pulses that affect the stable operation of the equipment. Summary of the Invention
[0005] This invention provides a method for seamless switching of operating modes of a multi-level high-voltage frequency converter, solving the technical problems mentioned in the background.
[0006] This invention provides a method for seamless switching of operating modes of a multi-level high-voltage frequency converter, comprising the following steps:
[0007] Step S101: During several switching cycles before the planned transfer, the three-phase voltages of the inverter and the grid are collected, and the instantaneous phase and instantaneous angular velocity of the inverter and the grid are calculated accordingly. A calculation rule is established to calculate the level reconstruction phase impact within a short time window with the transfer time as the center.
[0008] Step S102: Based on the level reconstruction information to be performed, the DC voltage of the power unit and the change, determine the equivalent output voltage step, and calculate the predicted level reconstruction phase impulse at the candidate time.
[0009] Step S103: From the set of candidate transfer times formed by the switching cycle, select the time that minimizes the absolute value of the predicted level reconstruction phase impact as the actual transfer time.
[0010] Step S104: In the switching cycle before the actual transfer time, apply a phase pre-shaping amount equal to the inverse of the predicted level reconstructed phase impulse to the modulation reference phase, while keeping the modulation reference amplitude unchanged.
[0011] Step S105: Perform a synchronous transfer action at the actual transfer time, and calculate the measured level reconstruction phase impact amount within a short time window according to the calculation rules;
[0012] Step S106: Based on the difference between the measured level reconstructed phase impulse and the predicted level reconstructed phase impulse, the switch state change and DC voltage change are mapped to an equivalent output voltage step, and the mapping coefficient is corrected accordingly.
[0013] The beneficial effects of this invention are as follows: This invention first observes and predicts the phase impact at the candidate transfer moment in real time, selects the moment with the smallest impact to trigger the transfer, and then actively cancels the expected impact through single-cycle phase pre-shaping. Subsequently, the mapping coefficient is corrected based on the measured impact to optimize the subsequent prediction. This process effectively minimizes the phase slip within the extremely short window at the moment of transfer, avoids short-term torque fluctuations and current pulses on the motor side, and truly meets the stringent requirements of seamless transfer on the engineering side. At the same time, it is compatible with various multi-level topologies such as cascaded unit three-level diode clamping. In long-term use, it can continuously improve the switching accuracy and equipment operation stability, and reduce equipment losses and failure risks caused by switching impacts. Attached Figure Description
[0014] Figure 1 This is a flowchart of a method for seamless switching of operating modes of a multi-level high-voltage frequency converter according to the present invention;
[0015] Figure 2 This is a flowchart of step S102 of the present invention. Detailed Implementation
[0016] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.
[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" indicate that the element or object preceding the term encompasses the elements or objects listed following the term and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0018] like Figures 1-2 As shown, a method for seamless switching of operating modes of a multi-level high-voltage frequency converter includes the following steps:
[0019] Step S101: During several switching cycles before the planned transfer, the three-phase voltages of the inverter and the grid are collected, and the instantaneous phase and instantaneous angular velocity of the inverter and the grid are calculated accordingly. A calculation rule is established to calculate the level reconstruction phase impact within a short time window with the transfer time as the center.
[0020] Step S102: Based on the level reconstruction information to be performed, the DC voltage of the power unit and the change, determine the equivalent output voltage step, and calculate the predicted level reconstruction phase impulse at the candidate time.
[0021] Step S103: From the set of candidate transfer times formed by the switching cycle, select the time that minimizes the absolute value of the predicted level reconstruction phase impact as the actual transfer time.
[0022] Step S104: In the switching cycle before the actual transfer time, apply a phase pre-shaping amount equal to the inverse of the predicted level reconstructed phase impulse to the modulation reference phase, while keeping the modulation reference amplitude unchanged.
[0023] Step S105: Perform a synchronous transfer action at the actual transfer time, and calculate the measured level reconstruction phase impact amount within a short time window according to the calculation rules;
[0024] Step S106: Based on the difference between the measured level reconstructed phase impulse and the predicted level reconstructed phase impulse, the switch state change and DC voltage change are mapped to an equivalent output voltage step, and the mapping coefficient is corrected accordingly.
[0025] In one embodiment of the present invention, at the discrete sampling time of each switching cycle, the three-phase voltage of the inverter and the three-phase voltage of the grid are synchronously collected, and the three-phase voltage of the inverter and the three-phase voltage of the grid are multiplied by the linear transformation matrix from the three-phase stationary coordinate system to the two-phase stationary coordinate system respectively to obtain the two-phase orthogonal voltage components of the inverter and the two-phase orthogonal voltage components of the grid.
[0026] The inverter's two-phase quadrature voltage components and the grid's two-phase quadrature voltage components are respectively taken as independent variables and substituted into the arctangent function with quadrant discrimination to calculate the instantaneous phase of the inverter and the grid.
[0027] Calculate the difference between the instantaneous phase at the current sampling time and the instantaneous phase at the previous sampling time, and divide the difference by the switching period to obtain the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid.
[0028] It should be noted that the switching cycle represents the update cycle of the PWM (Pulse Width Modulation) signal in the multi-level high-voltage inverter, which is the basic time unit for discrete sampling and control. This cycle remains fixed during the same switching process. The discrete sampling time represents the specific sampling time point determined based on the switching cycle. The discrete sampling time is equal to the sampling index value multiplied by the switching cycle, where the sampling index value is a positive integer starting from 0. For example, the first sampling time is 1 multiplied by the switching cycle, the second sampling time is 2 multiplied by the switching cycle, and so on. The inverter three-phase voltage represents the A-phase, B-phase, and C-phase voltage signals output from the inverter side of the multi-level high-voltage inverter, which are the core electrical quantities reflecting the inverter's output status. The grid three-phase voltage represents the A-phase, B-phase, and C-phase voltage signals from the AC grid side connected to the inverter. It is the target synchronization voltage during inverter switching and needs to be collected synchronously with the inverter three-phase voltage to eliminate timing deviations.
[0029] It should be noted that synchronous acquisition can be achieved through hardware timing synchronization. Specifically, a unified sampling trigger clock (clock frequency is an integer multiple of the switching cycle) is set on the inverter control board. This clock simultaneously triggers the voltage sensor sampling circuits on both the inverter side and the grid side, ensuring that the sampling time deviation of the two voltage signals is less than 1 microsecond. If discrete sensors are used, the sampling signals must be transmitted through coaxial cables to avoid timing delays caused by differences in cable length. The acquired data must be stored in the same data buffer, and stored according to the sampling time index.
[0030] It should be noted that the linear transformation matrix from a three-phase stationary coordinate system to a two-phase stationary coordinate system is used to convert three-dimensional three-phase voltage signals (phase A, phase B, and phase C) into two-dimensional orthogonal voltage signals (α-axis and β-axis), simplifying phase and vector calculations. The matrix elements are fixed coefficients. The first row of this linear transformation matrix consists of 1, -1 / 2, and -1 / 2, respectively, while the second row consists of 0, √3 / 2, and √3 / 2, respectively. During calculation, the matrix is first multiplied by two-thirds, and then multiplied by the inverter's three-phase voltage and the grid's three-phase voltage according to matrix multiplication rules to obtain the corresponding two-phase orthogonal voltage components. The inverter's two-phase orthogonal voltage components represent the α-axis and β-axis voltage components obtained after the above linear transformation of the inverter's three-phase voltage. These components are perpendicular (orthogonal) to each other and together constitute the projection of the inverter's output voltage space vector onto the stationary coordinate system. The inverter's α-axis voltage component is equal to two-thirds multiplied by (inverter A-phase voltage minus half of inverter B-phase voltage minus half of inverter C-phase voltage); the inverter's β-axis voltage component is equal to two-thirds multiplied by (√3 / 2 inverter B-phase voltage minus √3 / 2 inverter C-phase voltage); together, they constitute the inverter's two-phase quadrature voltage components.
[0031] It should be noted that the two-phase quadrature voltage components of the power grid represent the α-axis and β-axis voltage components obtained after the three-phase voltages of the power grid undergo the above linear transformation. Their function is to correspond with the two-phase quadrature voltage components of the inverter and to be used for subsequent calculations of phase difference and angular velocity difference. The α-axis voltage component of the power grid is equal to two-thirds multiplied by (the voltage of phase A of the power grid minus half of the voltage of phase B of the power grid minus half of the voltage of phase C of the power grid); the β-axis voltage component of the power grid is equal to two-thirds multiplied by (the voltage of phase B of the power grid minus the voltage of phase C of the power grid). Together, they constitute the two-phase quadrature voltage components of the power grid.
[0032] It should be noted that the arctangent function with quadrant discrimination, when calculating the inverter's instantaneous phase, uses the inverter's β-axis voltage component as the numerator and the inverter's α-axis voltage component as the denominator; when calculating the grid's instantaneous phase, it uses the grid's β-axis voltage component as the numerator and the grid's α-axis voltage component as the denominator. The function output is the corresponding instantaneous phase (in radians or degrees), and it can automatically determine the specific quadrant within the phase range of 0 to 360 degrees. The inverter's instantaneous phase reflects the spatial vector direction of the inverter's output voltage. The grid's instantaneous phase reflects the spatial vector direction of the grid voltage. The inverter's instantaneous angular velocity reflects the rate of change of the inverter's instantaneous phase; the inverter's instantaneous angular velocity is equal to the inverter's instantaneous phase at the current sampling moment minus the inverter's instantaneous phase at the previous sampling moment, and the difference is then divided by the switching period. The instantaneous angular velocity of the power grid is used to reflect the rate of change of the instantaneous phase of the power grid. The instantaneous angular velocity of the power grid is equal to the instantaneous phase of the power grid at the current sampling time minus the instantaneous phase of the power grid at the previous sampling time, and the difference is then divided by the switching period.
[0033] In one embodiment of the present invention, the discrete sampling time index corresponding to the planned transfer time is determined, and the discrete summation interval covered by the short time window is determined with the discrete sampling time index as the center and the number of sampling points contained in a preset multiple of the switching cycle as the half width.
[0034] Within the discrete summation interval, the difference between the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid is calculated, and the difference is multiplied by the switching cycle to obtain the single-point slip. All single-point slips within the discrete summation interval are accumulated and summed to obtain the level reconstruction phase impulse.
[0035] It should be noted that the discrete sampling time index corresponding to the planned transition time represents the index value (positive integer) for converting the planned transition time (continuous time) into a discrete sampling time. The discrete sampling time index corresponding to the planned transition time is equal to the planned transition time divided by the switching period, and the result is rounded down (i.e., discarding the decimal part and keeping the integer part); for example, if the planned transition time is 10.8 times the switching period, the index value is 10. The preset multiplier represents the coefficient used to determine the half-width of the short time window, which is a positive integer. It needs to be determined based on the switching period and the response time of level reconstruction to ensure that the short time window only covers the instantaneous process of level reconstruction. It is usually taken as 1 or 2, that is, the half-width is 1 or 2 times the switching period.
[0036] It should be noted that the planned transfer time is not randomly determined. It needs to be initially estimated based on the difference between the grid frequency and the inverter output frequency to ensure that the calculation of step S101 is started when the frequency deviation is less than 0.5Hz. Specifically, the average value of the instantaneous angular velocity of the grid (the average value of 10 consecutive switching cycles) and the average value of the instantaneous angular velocity of the inverter are calculated in real time. If the frequency deviation (angular velocity difference divided by 2π) corresponding to the difference between the two is less than 0.5Hz, then the time of the 10th switching cycle after the current time is set as the preliminary value of the planned transfer time. Then, the discrete sampling time index is determined based on this preliminary value, and the subsequent short time window calculation is started.
[0037] It should be noted that the short time window represents a discrete sampling interval symmetrically positioned to the left and right of the discrete sampling time index corresponding to the planned transfer time. This interval is used to accurately capture phase impacts caused by level reconstruction. The starting index of the short time window is equal to the discrete sampling time index corresponding to the planned transfer time minus a preset multiple, and the ending index is equal to the discrete sampling time index plus the preset multiple. For example, when the preset multiple is 1 and the index is 10, the short time window covers sampling times from indices 9 to 11. The discrete summation interval represents the range of discrete sampling indices that completely correspond to the short time window; that is, the starting index is the planned transfer time index minus a preset multiple, and the ending index is the planned transfer time index plus the preset multiple. The single-point slip represents the instantaneous phase slip of the inverter relative to the grid at a single sampling time within the discrete summation interval. The single-point slip is equal to the difference between the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid at that sampling time, multiplied by the switching period. The phase impact of level reconstruction represents the sum of the slip amounts of all single points within a short time window. It is the core indicator for measuring the magnitude of phase disturbance during level reconstruction. The smaller the value, the smaller the phase impact during switching. The phase impact of level reconstruction is equal to the sum of the slip amounts of each single point at each sampling moment within the discrete summation interval. For example, when the interval contains 3 sampling points, the total impact is the sum of the slip amounts of these 3 points.
[0038] In one embodiment of the present invention, a set of candidate transition times aligned with the switching period is determined, and the following calculation steps are performed for each candidate transition time in the set:
[0039] Step S201: Obtain the level reconstruction information at the candidate transfer time, quantify it into the switching state change of each power unit, and at the same time obtain the current DC voltage value, the switching state before reconstruction, and the DC voltage change of each power unit.
[0040] Step S202: Calculate the product of the switch state change and the DC voltage value, calculate the product of the switch state and the DC voltage change before reconstruction, add the two products to obtain the unit-level step, and accumulate the unit-level step of all power units in each phase to obtain the equivalent output voltage step.
[0041] Step S203: Obtain the instantaneous phase of the inverter, calculate the cosine and sine values of the instantaneous phase to form a unit direction vector, multiply the equivalent output voltage step by the unit direction vector to obtain the voltage vector step in the two-phase stationary coordinate system;
[0042] Step S204: Construct the current voltage space vector using the two-phase orthogonal voltage components of the inverter, and swap the positions of the two-phase orthogonal voltage components and invert one of them to construct a direction vector orthogonal to the current voltage space vector;
[0043] Step S205: Calculate the dot product of the direction vector and the voltage vector step, calculate the square of the magnitude of the current voltage space vector, divide the dot product by the square of the magnitude, and obtain the predicted level reconstruction phase impulse at the candidate transition time.
[0044] It should be noted that the candidate transition time set represents a collection of multiple times aligned with the switching cycle, each time being an integer multiple of the switching cycle. This set is used to select the optimal transition time, and its size typically covers 1 to 2 fundamental cycles. It is generated based on a switching cycle timer; for example, one candidate time is selected for each switching cycle, and 20 consecutive times form the set. The candidate transition time refers to a single time within the candidate transition time set. Level reconstruction information indicates the specific changes in the inverter's modulation method or the power units involved in the operation at the candidate transition time, read from the switching strategy pre-stored in the inverter control module. The switch state change represents the difference in the power unit's switch state before and after level reconstruction, used to indicate the change in the unit's output level. Its value range is -1, 0, and 1 (-1 indicates the level changes from 1 to 0 or 0 to -1, 0 indicates the level remains unchanged, and 1 indicates the level changes from -1 to 0 or 0 to 1). The switch state change equals the power unit's switch state at the candidate transition time minus the power unit's switch state before reconstruction, where the switch state before reconstruction is the unit state of the previous switching cycle before the candidate time. The switching state before reconstruction indicates the switching state of the power unit before the level reconstruction occurs (i.e., one switching cycle before the candidate transfer time), and the value is -1, 0, 1 (-1 indicates that the lower bridge arm is on and the upper bridge arm is off, 1 indicates that the upper bridge arm is on and the lower bridge arm is off, and 0 indicates that both the upper and lower bridge arms are off).
[0045] It should be noted that if the fundamental frequency of the power grid is 50Hz (period 20ms) and the switching period is 100μs (10kHz switching frequency), then two fundamental cycles contain 20ms ÷ 100μs = 200 switching cycles; therefore, the candidate transition time set contains 200 time points (one time point is taken for each switching cycle); this ensures coverage of all phase points within the fundamental cycle and avoids missing the minimum impact time due to an excessively small set size; if the fundamental frequency is 60Hz, the set size is calculated similarly based on two fundamental cycles (approximately 16.67ms) to ensure compatibility with different power grid frequencies, which will not be elaborated here.
[0046] It should be noted that the DC voltage change represents the difference between the current DC voltage value of the power unit and the DC voltage value of the previous sampling period, reflecting the dynamic change of the unit's DC voltage and used to correct the deviation in static voltage calculation. The unit-level step represents the contribution of a single power unit to the output voltage step caused by level reconstruction. The unit-level step is equal to the product of the switch state change and the current DC voltage value of the power unit, plus the product of the switch state before reconstruction and the DC voltage change. Adding these two products together yields the unit-level step of a single power unit. The equivalent output voltage step represents the sum of the unit-level steps of all power units in a certain phase of the inverter, reflecting the total voltage step caused by level reconstruction in that phase. The equivalent output voltage step is equal to the sum of the unit-level steps of all power units in that phase according to algebraic rules. If a phase has N power units, the unit-level step of each of the N units is added sequentially to obtain the equivalent output voltage step of that phase.
[0047] Specifically, candidate time A certain phase below (by The equivalent output voltage step (identified by phases A, B, and C). The calculation formula is as follows: ,
[0048] Where N represents the total number of power units contained in each phase of the multilevel frequency converter. Indicates candidate time The switching state change of the k-th power unit. Indicates candidate time The DC-side voltage of the k-th power unit. Indicates candidate time The switching state of the previous k-th power unit, Indicates candidate time The change in the DC voltage of the k-th power unit relative to the previous sampling time.
[0049] It should be noted that the unit direction vector represents a two-dimensional vector composed of the cosine and sine values of the inverter's instantaneous phase. Its magnitude is 1, and its direction is consistent with the direction of the inverter's current output voltage space vector. It is used to map the equivalent output voltage step of a single phase onto a two-phase stationary coordinate system. The first component of the unit direction vector is the cosine value of the inverter's instantaneous phase, and the second component is the sine value of the inverter's instantaneous phase. These two components together constitute the two-dimensional unit direction vector. The voltage vector step represents the projection of the equivalent output voltage step onto the two-phase stationary coordinate system. It is a vector with direction and magnitude, reflecting the disturbance of level reconstruction on the voltage space vector. The first component of the voltage vector step (α-axis component) is equal to the equivalent output voltage step multiplied by the first component of the unit direction vector (cosine value), and the second component (β-axis component) is equal to the equivalent output voltage step multiplied by the second component of the unit direction vector (sine value). These two components together constitute the voltage vector step on the two-phase stationary coordinate system.
[0050] It should be noted that the current voltage space vector represents a two-dimensional vector composed of the two-phase orthogonal voltage components of the inverter, reflecting the spatial state of the inverter output voltage at the candidate transition moment. The first component (α-axis component) of the current voltage space vector is the α-axis component of the two-phase orthogonal voltage components of the inverter, and the second component (β-axis component) is the β-axis component of the two-phase orthogonal voltage components of the inverter. The two components together constitute the current voltage space vector. The orthogonal direction vector represents a two-dimensional vector perpendicular to the current voltage space vector, used to separate the component in the voltage vector step that is most sensitive to phase changes. The first component of the orthogonal direction vector is equal to the negative value of the β-axis component of the two-phase orthogonal voltage components of the inverter, and the second component is equal to the α-axis component of the two-phase orthogonal voltage components of the inverter. The two components together constitute a direction vector orthogonal to the current voltage space vector. The dot product is equal to the first component of the orthogonal direction vector multiplied by the first component of the voltage vector step, plus the second component of the orthogonal direction vector multiplied by the second component of the voltage vector step. Adding the two products together gives the value of the dot product. The square of the magnitude of the current voltage space vector is equal to the square of the first component (α-axis component) of the current voltage space vector plus the square of the second component (β-axis component). Adding these two squares gives the value of the square of the magnitude. The predicted level reconstruction phase impact represents the instantaneous phase slip expected due to level reconstruction at the candidate transition moment. It is a core indicator for selecting the optimal transition moment; the smaller the absolute value, the smaller the phase impact. The predicted level reconstruction phase impact is equal to the dot product of the orthogonal direction vector and the voltage vector step, divided by the square of the magnitude of the current voltage space vector. The quotient is the predicted level reconstruction phase impact.
[0051] In one embodiment of the present invention, a set of candidate transition times is obtained, wherein each candidate transition time in the set is an integer multiple of the switching period, and each candidate transition time corresponds one-to-one with a predicted level reconstruction phase impulse.
[0052] For each candidate transition time in the set, the absolute value operation is performed on the corresponding predicted level reconstructed phase impulse to obtain the non-negative absolute value metric of the candidate transition time.
[0053] Compare the non-negative absolute values of all candidate transition times in the set, select the candidate transition time with the smallest value, and determine that time as the actual transition time.
[0054] It should be noted that the non-negative absolute value metric represents a non-negative scalar obtained by taking the absolute value of the predicted level reconstruction phase impact, used to eliminate the positive and negative directional influence of the impact. The non-negative absolute value metric is equal to the absolute value of the predicted level reconstruction phase impact at the corresponding candidate transition time. If the predicted impact is positive, it is directly retained; if it is negative, its opposite positive value is taken, and the result is always greater than or equal to 0. The actual transition time represents the time with the smallest non-negative absolute value metric selected from the candidate transition time set. It is the time reference for the final execution of level reconstruction and mode switching, ensuring that the phase impact is minimized during switching.
[0055] It should be noted that a safety threshold for the preset non-negative absolute value metric is set (usually 0.005 radians, determined based on the inverter's load capacity). If the metric value of all candidate moments is greater than this threshold, it indicates that the current operating condition (such as excessive grid frequency fluctuation) is not suitable for switching. In this case, an operating condition waiting mechanism is triggered, which regenerates the candidate transfer moment set and calculates the metric value every 10 switching cycles until a candidate moment with a metric value less than or equal to the safety threshold is found, at which point the filtering is performed. This avoids forced switching under high impact risk, which could lead to equipment failure. The default time interval for candidate moments is one switching cycle, which is suitable for most operating conditions. If the grid frequency stability is poor (e.g., fluctuations exceeding 0.2Hz), the time interval can be adjusted to two switching cycles to reduce the number of candidate moments and thus reduce the computational load, while still covering two fundamental cycles. The adjustment rules need to be pre-stored in the inverter control program. Users can select the normal mode (one switching cycle interval) or the stable mode (two switching cycle intervals) according to the grid operating conditions to ensure efficient filtering under different operating conditions.
[0056] In one embodiment of the present invention, a pre-shaping time interval is determined with the actual transfer time as the endpoint and a duration of one switching cycle.
[0057] Obtain the predicted level reconstructed phase impulse corresponding to the actual transfer time, calculate the inverse of the predicted level reconstructed phase impulse, and use the calculation result as the phase pre-shaping amount;
[0058] For each modulation update time within the pre-shaping time interval, obtain the original modulation reference phase and the original modulation reference amplitude at that time;
[0059] The original modulation reference phase is added to the phase pre-shaping amount to obtain the updated modulation reference phase, and the original modulation reference amplitude is directly used as the updated modulation reference amplitude.
[0060] Calculate the cosine and sine values of the updated modulation reference phase, and multiply the cosine and sine values by the updated modulation reference amplitude to obtain the two orthogonal voltage components of the modulation reference in the two-phase stationary coordinate system.
[0061] It should be noted that the pre-shaping time interval represents a continuous time interval with the actual transition time as the endpoint and a duration equal to one switching cycle. The start time of the pre-shaping time interval is equal to the actual transition time minus one switching cycle, and the end time is the actual transition time. The phase pre-shaping amount represents a small angle quantity used to correct the modulation reference phase. Its core function is to offset the phase impact expected to be generated by the level reconstruction at the actual transition time. Its value has an inverse compensation relationship with the predicted impact amount. The phase pre-shaping amount is equal to the negative of the predicted level reconstruction phase impact amount at the corresponding actual transition time. If the predicted impact amount is positive, the phase pre-shaping amount takes an equally negative value; if the predicted impact amount is negative, the phase pre-shaping amount takes an equally positive value. The modulation update time indicates the specific time point within the pre-shaping time interval when the inverter updates the modulation reference signal (phase and amplitude), which is synchronized with the switching cycle of PWM control to ensure that the updated signal can drive the switching devices in time. The interval of the modulation update time is equal to one-half of the switching cycle (e.g., if the switching cycle is 100 microseconds, it is updated every 25 microseconds). All update times fall within the pre-shaping time interval, and the last update time is no more than 10 microseconds away from the actual transfer time.
[0062] It should be noted that the original modulation reference phase represents the original target phase of the inverter before pre-shaping at the modulation update time, generated by the current control algorithm (such as voltage closed-loop control), reflecting the normal phase trajectory before the switchover. The original modulation reference amplitude represents the original target voltage amplitude of the inverter before pre-shaping at the modulation update time, generated by the voltage setpoint or closed-loop control, reflecting the normal output voltage level before the switchover. The updated modulation reference phase is equal to the original modulation reference phase at the modulation update time plus the phase pre-shaping amount. If the calculated result exceeds the range of 0 to 360 degrees (or 0 to 2π radians), it needs to be normalized by adding or subtracting 360 degrees (or 2π radians). The updated modulation reference amplitude is equal to the original modulation reference amplitude at the modulation update time. The two orthogonal voltage components of the modulation reference in the two-phase stationary coordinate system represent the projection components of the pre-shaped modulation reference signal on the α-axis and β-axis. The first orthogonal voltage component (α-axis) is equal to the updated modulation reference amplitude multiplied by the cosine of the updated modulation reference phase; the second orthogonal voltage component (β-axis) is equal to the updated modulation reference amplitude multiplied by the sine of the updated modulation reference phase. Both components are instantaneous values that change with time.
[0063] It should be noted that the interval between modulation update times is equal to the switching period divided by 4 (i.e., one switching period contains 4 update times). If the switching period is 100 microseconds, the update interval is 25 microseconds, and the number of updates within the interval is 4 (e.g., 0 microseconds, 25 microseconds, 50 microseconds, 75 microseconds); if the switching period is 80 microseconds, the update interval is 20 microseconds, and the number of updates within the interval is 4 (0 microseconds, 20 microseconds, 40 microseconds, 60 microseconds), ensuring that the number of updates within each switching period is fixed at 4. If the updated modulation reference phase is greater than 360 degrees, subtract an integer multiple of 360 degrees from the phase until the result falls within the range of 0 to 360 degrees; if the phase is less than 0 degrees, add an integer multiple of 360 degrees to the phase until the result falls within the range of 0 to 360 degrees; for example, when the phase is 370 degrees, subtracting 360 degrees gives 10 degrees, and when the phase is -5 degrees, adding 360 degrees gives 355 degrees, ensuring that the phase is continuous without jumps.
[0064] It should be noted that the maximum allowable value of the preset phase pre-shaping amount is 0.01 radians (approximately 0.57 degrees). If the absolute value of the calculated pre-shaping amount exceeds this value, it indicates that the current predicted impact amount is too large and is not suitable for pre-shaping. In this case, a re-timing should be triggered, i.e., returning to step S103 to re-select the actual transfer time. The maximum allowable value is determined based on the response speed of the inverter's switching devices (exceeding 0.01 radians will cause the switching devices to fail to respond to phase changes in a timely manner), avoiding the introduction of new phase jump impacts by excessive pre-shaping. This will not be elaborated upon here.
[0065] In one embodiment of the present invention, the discrete sampling time index corresponding to the actual transfer time is determined, and the number of sampling points covered by the half-width of the short time window is determined according to the switching period of a preset multiple, thereby establishing a short time window sampling interval centered on the actual transfer time and covering the sampling points before and after; and the hardware is controlled to perform a synchronous transfer action at the actual transfer time.
[0066] For each discrete sampling moment within the short time window sampling interval, the inverter's three-phase voltage and the grid's three-phase voltage are collected. The inverter's three-phase voltage and the grid's three-phase voltage are then multiplied by the linear transformation matrix from the three-phase stationary coordinate system to the two-phase stationary coordinate system to obtain the inverter's two-phase orthogonal voltage components and the grid's two-phase orthogonal voltage components.
[0067] In one embodiment of the present invention, the two-phase quadrature voltage components of the inverter and the two-phase quadrature voltage components of the grid are respectively substituted into the arctangent function with quadrant discrimination to calculate the instantaneous phase of the inverter and the instantaneous phase of the grid.
[0068] Calculate the difference between the instantaneous phase at the current sampling time and the instantaneous phase at the previous sampling time, and divide the difference by the switching period to obtain the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid;
[0069] Within the short-time-window sampling interval, the difference between the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid is calculated. The difference is multiplied by the switching cycle to obtain the single-point slip. All single-point slips within the short-time-window sampling interval are accumulated and summed to obtain the measured level reconstruction phase impulse.
[0070] It should be noted that the number of sampling points covered by the short time window half-width is equal to a preset multiple (1 or 2), because the short time window half-width is a preset multiple multiplied by the switching cycle, and each switching cycle corresponds to one discrete sampling point. Synchronous transfer action represents the sequence of hardware actions executed at the actual transfer moment; its core is to complete the inverter's switch from the current operating mode to the target mode. The measured level reconstruction phase impact represents the sum of all single-point slip amounts within the short time window sampling interval, reflecting the true magnitude of the phase impact during the actual switching process. The measured level reconstruction phase impact is equal to the sum of the single-point slip amounts at each discrete sampling moment within the short time window sampling interval, calculated algebraically. If the interval contains 3 sampling points, it is the sum of the 3 single-point slip amounts.
[0071] Specifically, the measured level reconstruction phase impulse. The calculation formula is as follows:
[0072] ,
[0073] Where K represents the number of sampling points contained in the half-width of the short time window. This represents the instantaneous angular velocity of the inverter at the k-th discrete moment. This represents the instantaneous angular velocity of the power grid at the k-th discrete moment. Indicates the switching cycle. This represents the discrete sampling index corresponding to the actual transition time.
[0074] It should be noted that the synchronous transfer action is triggered by the inverter's switching control module. This module shares the same clock (100MHz) with the sampling control module, ensuring synchronization between the trigger signal and the sampling signal. The trigger signal transmission path is through the switching control module, the optocoupler isolation circuit, and the driver board. Optocoupler isolation prevents high-voltage side interference from affecting the low-voltage control signal. After receiving the signal, the driver board drives the inverter's IGBTs, SSTS, and bypass contactor according to the pre-stored timing sequence, with a transmission delay of ≤2 microseconds, ensuring accurate timing. After the measured level reconstruction phase impulse is calculated, it is stored in an impulse buffer with a capacity of 100 entries (stores the measured values of the most recent 100 switches). Each data entry contains three fields: measured value, actual transfer time, and preset multiple, stored in chronological order, with newer data overwriting the oldest data. During subsequent calls, the corresponding measured value is read according to the actual transfer time index, ensuring accurate matching with the predicted value for that switch and avoiding the retrieval of incorrect data.
[0075] In one embodiment of the present invention, the measured level reconstruction phase impulse and the predicted level reconstruction phase impulse corresponding to the actual transfer time are obtained, and the difference between the measured level reconstruction phase impulse and the predicted level reconstruction phase impulse is calculated to obtain the phase impulse difference.
[0076] Acquire the switching state changes of each power unit at the actual transfer time, the current DC voltage value, the switching state before the actual transfer time, and the DC voltage changes;
[0077] Calculate the product of the switch state change and the DC voltage value, calculate the product of the switch state and the DC voltage change before the actual transfer moment, add these two products to obtain the unit-level step value, and accumulate the unit-level step values of all power units in each phase to obtain the equivalent output voltage step.
[0078] In one embodiment of the present invention, the current voltage space vector is constructed by the two-phase quadrature voltage components of the inverter obtained at the actual transfer moment, and the positions of the two-phase quadrature voltage components are swapped and one of them is inverted to construct a direction vector orthogonal to the current voltage space vector.
[0079] Calculate the square of the magnitude of the current voltage space vector, divide the phase impulse difference by the product of the equivalent output voltage step and the square of the magnitude to obtain the normalized error, and multiply the normalized error by the preset update step size to obtain the adjustment coefficient.
[0080] The adjustment coefficient is multiplied by the direction vector to obtain the correction vector, and then the correction vector is added to the mapping coefficient vector corresponding to the current time to complete the correction of the mapping coefficient.
[0081] Specifically, with respect to the actual transfer time Corresponding inverter instantaneous phase Related mapping coefficients The calculation formula is as follows: ,
[0082] in Indicates the update step size. Indicates the phase impact difference value. Indicates the actual transfer time The square of the corresponding inverter voltage space vector magnitude, Indicates the actual transfer time The equivalent output voltage step of a certain phase. It represents the orthogonal direction vector of the inverter space voltage vector corresponding to the actual transfer moment.
[0083] It should be noted that the preset update step size represents a very small positive value used to control the magnitude of the mapping coefficient correction. This avoids excessively large corrections that could cause coefficient oscillations (instability), or excessively small corrections that could lead to slow convergence. The value needs to balance stability and correction efficiency. The preset update step size is a constant within a fixed range (initial value is between 0.001 and 0.01, such as 0.005), and does not require real-time calculation. The adjustment coefficient is equal to the normalization error multiplied by the preset update step size. If the normalization error is positive, the adjustment coefficient is positive; if the normalization error is negative, the adjustment coefficient is negative, and the magnitude is controlled by the step size. The correction vector represents the product of the adjustment coefficient and the orthogonal direction vector. It is the incremental correction term of the mapping coefficient vector. The vector dimension is the same as the mapping coefficient vector to ensure direct superposition correction. Each component of the correction vector is equal to the adjustment coefficient multiplied by the corresponding component of the orthogonal direction vector. For example, if the orthogonal direction vector has two components, α-axis and β-axis, the α-axis component of the correction vector is the adjustment coefficient multiplied by the α-axis component of the orthogonal vector, and the β-axis component is similarly calculated. The mapping coefficient vector corresponding to the current moment represents a two-dimensional vector used to map the equivalent output voltage step to the voltage space vector increment. It is phase-dependent with the voltage space vector at the actual transition moment, and the vector components correspond to the mapping ratios of the α-axis and β-axis, respectively. The initial value of the mapping coefficient vector is a two-dimensional unit vector (α-axis component 1, β-axis component 0).
[0084] It should be noted that if the absolute value of the equivalent output voltage step is less than 0.01 times the rated phase voltage of the inverter (to avoid the step being too small), then 0.01 times the rated phase voltage is used as the replacement value; if the square of the magnitude of the current voltage space vector is less than 0.01 times the square of the rated voltage (to avoid the voltage crossing to zero), then 0.01 times the square of the rated voltage is used as the replacement value. The preset update step size is initially set to 0.005; if after 3 consecutive corrections, the absolute value of the phase impulse difference continues to decrease (correction is effective), the step size remains unchanged; if the absolute value of the difference increases (correction is overdone), then the step size is multiplied by 0.5 (reduced by half); the minimum step size is not less than 0.001 to avoid correction stagnation. When after 5 consecutive corrections, the absolute value of the phase impulse difference is less than 0.0005 radians (convergence threshold, set based on the allowable range of switching impulses), then the mapping coefficient is determined to have converged, and subsequent corrections are stopped (the current coefficient is maintained); if the deviation exceeds the threshold again during subsequent switching, the correction process is restarted to avoid resource waste caused by excessive coefficient updates.
[0085] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.
[0086] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.
Claims
1. A method for seamless switching of operating modes of a multi-level high-voltage frequency converter, characterized in that, Includes the following steps: Step S101: During several switching cycles before the planned transfer, the three-phase voltages of the inverter and the grid are collected, and the instantaneous phase and instantaneous angular velocity of the inverter and the grid are calculated accordingly. A calculation rule is established to calculate the level reconstruction phase impact within a short time window with the transfer time as the center. Step S102: Based on the level reconstruction information to be performed, the DC voltage of the power unit and the change, determine the equivalent output voltage step, and calculate the predicted level reconstruction phase impulse at the candidate time. Step S103: From the set of candidate transfer times formed by the switching cycle, select the time that minimizes the absolute value of the predicted level reconstruction phase impact as the actual transfer time. Step S104: In the switching cycle before the actual transfer time, apply a phase pre-shaping amount equal to the inverse of the predicted level reconstructed phase impulse to the modulation reference phase, while keeping the modulation reference amplitude unchanged. Step S105: Perform a synchronous transfer action at the actual transfer time, and calculate the measured level reconstruction phase impact amount within a short time window according to the calculation rules; Step S106: Based on the difference between the measured level reconstructed phase impulse and the predicted level reconstructed phase impulse, the switch state change and DC voltage change are mapped to an equivalent output voltage step, and the mapping coefficient is corrected accordingly. The mapping coefficient C is related to the instantaneous phase θ(t0) of the inverter corresponding to the actual transfer time t0. map The formula for calculating (θ(t0)) is as follows: ; Where μ represents the update step size, and e represents the phase impact difference. This represents the square of the inverter voltage space vector magnitude corresponding to the actual transfer time t0. This represents the equivalent output voltage step of a certain phase at the actual transition time t0. The orthogonal direction vector of the inverter space voltage vector corresponding to the actual transfer moment; Determine the discrete sampling time index corresponding to the planned transfer time, and determine the discrete summation interval covered by the short time window with the discrete sampling time index as the center and the number of sampling points contained in a preset multiple of the switching cycle as the half width. Within the discrete summation interval, the difference between the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid is calculated, and the difference is multiplied by the switching cycle to obtain the single-point slip. All single-point slips within the discrete summation interval are accumulated and summed to obtain the level reconstruction phase impulse.
2. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 1, characterized in that, At the discrete sampling time of each switching cycle, the three-phase voltage of the inverter and the three-phase voltage of the grid are synchronously acquired, and the three-phase voltage of the inverter and the three-phase voltage of the grid are multiplied by the linear transformation matrix from the three-phase stationary coordinate system to the two-phase stationary coordinate system respectively to obtain the two-phase orthogonal voltage components of the inverter and the two-phase orthogonal voltage components of the grid. The inverter's two-phase quadrature voltage components and the grid's two-phase quadrature voltage components are respectively taken as independent variables and substituted into the arctangent function with quadrant discrimination to calculate the instantaneous phase of the inverter and the grid. Calculate the difference between the instantaneous phase at the current sampling time and the instantaneous phase at the previous sampling time, and divide the difference by the switching period to obtain the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid.
3. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 2, characterized in that, Determine a set of candidate transition times aligned with the switching cycle, and perform the following calculation steps for each candidate transition time in the set: Step S201: Obtain the level reconstruction information at the candidate transfer time, quantify it into the switching state change of each power unit, and at the same time obtain the current DC voltage value, the switching state before reconstruction, and the DC voltage change of each power unit. Step S202: Calculate the product of the switch state change and the DC voltage value, calculate the product of the switch state and the DC voltage change before reconstruction, add the two products to obtain the unit-level step, and accumulate the unit-level step of all power units in each phase to obtain the equivalent output voltage step. Step S203: Obtain the instantaneous phase of the inverter, calculate the cosine and sine values of the instantaneous phase to form a unit direction vector, multiply the equivalent output voltage step by the unit direction vector to obtain the voltage vector step in the two-phase stationary coordinate system; Step S204: Construct the current voltage space vector using the two-phase orthogonal voltage components of the inverter, and swap the positions of the two-phase orthogonal voltage components and invert one of them to construct a direction vector orthogonal to the current voltage space vector; Step S205: Calculate the dot product of the direction vector and the voltage vector step, calculate the square of the magnitude of the current voltage space vector, divide the dot product by the square of the magnitude, and obtain the predicted level reconstruction phase impulse at the candidate transition time.
4. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 1, characterized in that, Obtain a set of candidate transition times, where each candidate transition time in the set is an integer multiple of the switching period, and each candidate transition time corresponds one-to-one with a predicted level reconstructed phase impulse. For each candidate transition time in the set, the absolute value operation is performed on the corresponding predicted level reconstructed phase impulse to obtain the non-negative absolute value metric of the candidate transition time. Compare the non-negative absolute values of all candidate transition times in the set, select the candidate transition time with the smallest value, and determine that time as the actual transition time.
5. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 1, characterized in that, Determine the pre-shaping time interval with the actual transfer time as the endpoint and a duration of one switching cycle; Obtain the predicted level reconstructed phase impulse corresponding to the actual transfer time, calculate the inverse of the predicted level reconstructed phase impulse, and use the calculation result as the phase pre-shaping amount; For each modulation update time within the pre-shaping time interval, obtain the original modulation reference phase and the original modulation reference amplitude at that time; The original modulation reference phase is added to the phase pre-shaping amount to obtain the updated modulation reference phase, and the original modulation reference amplitude is directly used as the updated modulation reference amplitude. Calculate the cosine and sine values of the updated modulation reference phase, and multiply the cosine and sine values by the updated modulation reference amplitude to obtain the two orthogonal voltage components of the modulation reference in the two-phase stationary coordinate system.
6. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 1, characterized in that, Determine the discrete sampling time index corresponding to the actual transfer time, and determine the number of sampling points covered by the half-width of the short time window according to the preset multiple of the switching period, thereby establishing a short time window sampling interval centered on the actual transfer time and covering the sampling points before and after. And control the hardware to perform synchronous transfer actions at the actual transfer time; For each discrete sampling moment within the short time window sampling interval, the inverter's three-phase voltage and the grid's three-phase voltage are collected. The inverter's three-phase voltage and the grid's three-phase voltage are then multiplied by the linear transformation matrix from the three-phase stationary coordinate system to the two-phase stationary coordinate system to obtain the inverter's two-phase orthogonal voltage components and the grid's two-phase orthogonal voltage components.
7. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 6, characterized in that, Substitute the two-phase quadrature voltage components of the inverter and the two-phase quadrature voltage components of the grid into the arctangent function with quadrant discrimination, respectively, to calculate the instantaneous phase of the inverter and the instantaneous phase of the grid. Calculate the difference between the instantaneous phase at the current sampling time and the instantaneous phase at the previous sampling time, and divide the difference by the switching period to obtain the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid; Within the short-time-window sampling interval, the difference between the instantaneous angular velocity of the inverter and the instantaneous angular velocity of the grid is calculated. The difference is multiplied by the switching cycle to obtain the single-point slip. All single-point slips within the short-time-window sampling interval are accumulated and summed to obtain the measured level reconstruction phase impulse.
8. The seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 1, characterized in that, Obtain the measured level reconstruction phase impulse and the predicted level reconstruction phase impulse corresponding to the actual transfer time. Calculate the difference between the measured level reconstruction phase impulse and the predicted level reconstruction phase impulse to obtain the phase impulse difference. Acquire the switching state changes of each power unit at the actual transfer time, the current DC voltage value, the switching state before the actual transfer time, and the DC voltage changes; Calculate the product of the switch state change and the DC voltage value, calculate the product of the switch state and the DC voltage change before the actual transfer moment, add these two products to obtain the unit-level step value, and accumulate the unit-level step values of all power units in each phase to obtain the equivalent output voltage step.
9. A seamless switching method for operating modes of a multi-level high-voltage frequency converter according to claim 8, characterized in that, The current voltage space vector is constructed by the two-phase quadrature voltage components of the inverter obtained at the actual transfer moment, and the positions of the two-phase quadrature voltage components are swapped and one of them is inverted to construct a direction vector orthogonal to the current voltage space vector. Calculate the square of the magnitude of the current voltage space vector, divide the phase impulse difference by the product of the equivalent output voltage step and the square of the magnitude to obtain the normalized error, and multiply the normalized error by the preset update step size to obtain the adjustment coefficient. The adjustment coefficient is multiplied by the direction vector to obtain the correction vector, and then the correction vector is added to the mapping coefficient vector corresponding to the current time to complete the correction of the mapping coefficient.