Switching loss and ripple current comprehensive optimization method and system suitable for ANPC topology

Abstract

The invention discloses a switching loss and ripple current comprehensive optimization method and a switching loss and ripple current comprehensive optimization system suitable for ANPC topology, and aims to solve the contradiction between loss and ripple optimization in traditional design. And a real-time optimal frequency is obtained based on global optimal solution, and frequency adaptive regulation and control during load change are realized. Pre-calculation is realized based on an interpolation algorithm, and the adaptation degree of the system to a processor is improved. According to the invention, through a multi-level optimization architecture, comprehensive optimization of loss and ripples in a full working condition range is realized on the premise of ensuring electric energy quality, application requirements of different computing power platforms are compatible, and the method has the characteristics of high adaptability and high robustness.

Description

Technical Field

[0001] The present invention belongs to the technical field of power converter optimization, and particularly relates to a comprehensive optimization method and system for switching loss and ripple current applicable to an ANPC topology. Background Art

[0002] With the rapid development of new energy power generation, electric vehicles and other fields, the three-phase medium-voltage ANPC converter, as an important power conversion device, in the new energy power generation system, the ANPC converter can be used to convert the direct current generated by renewable energy into alternating current and integrate it into the power grid. Its multi-level output characteristic can reduce the harmonic distortion of the output waveform and improve the power quality. In the field of electric vehicles, the high power density and high efficiency characteristics of the ANPC converter can meet the requirements of electric vehicles for the compactness and energy-saving of the power system, and at the same time can achieve fast response and precise motor torque control, improving the dynamic performance and driving range of the vehicle.

[0003] Based on this, its operating efficiency and output performance directly affect the overall system efficiency. The traditional fixed switching frequency modulation method has inherent defects: the switching loss increases significantly under high-frequency conditions, resulting in increased device temperature rise, while the output current ripple increases during low-frequency operation, which will affect the power quality and filter design. Most of the existing variable switching frequency technologies adopt a single-objective optimization strategy, or switch between a set of limited switching frequencies based on empirical rules, and it is difficult to achieve the global optimal balance between loss and ripple. Especially when the load current changes dynamically, the existing methods are prone to constraint conflicts, resulting in a decrease in system reliability. In the prior art, the variable frequency control method of the photovoltaic inverter disclosed in the patent with publication number CN103731010B is adjusted based on fixed rules and cannot adapt to complex working condition changes (such as sudden load, grid disturbance) in real time, and only considers two parameters, power and junction temperature, without involving power quality indicators such as current ripple and harmonics; although the patent with publication number CN114499253B has advantages in device-level thermal management, due to its single-objective orientation and discrete control characteristics, it is difficult to meet the comprehensive requirements of complex power electronic systems for efficiency, quality, and lifespan. Therefore, there is an urgent need for an adaptive frequency optimization method that can dynamically coordinate multi-objective constraints and take into account adaptability and robustness to meet the requirements of efficient and reliable operation of power electronic devices under complex working conditions.

[0004] Regarding the inter-phase ripple current of the bridge arm, it reflects the fluctuation of the current. Such ripple current will affect the electromagnetic compatibility, thermal stress, output voltage quality and system efficiency of the system. It refers to the current fluctuation caused by the change of the switching state of different phases and the different load characteristics in the multi-level inverter. In the ANPC circuit, each bridge arm usually consists of multiple switching tubes to achieve multi-level output. Due to the timing difference of the switching actions and the non-linear characteristics of the load, there will be differences in the instantaneous values of the bridge arm currents of different bridge phases. This difference is the inter-phase ripple current of the bridge arm. The existing switching frequency modulation strategies cannot comprehensively consider various indicators of the converter circuit structure, it is difficult to achieve the global optimal balance of multiple indicators, and it cannot effectively adapt to the computing power of different processors. Summary of the Invention

[0005] The present invention focuses on the problems existing in the above-mentioned prior art, and provides a comprehensive optimization method for switching losses and ripple current applicable to the ANPC topology.

[0006] To solve the contradiction between loss and ripple optimization in traditional designs, the present invention constructs an objective function that comprehensively considers the loss positively correlated with the switching frequency and the ripple negatively correlated with it, eliminates the dimensional differences of multiple physical quantities through normalization processing, and obtains the real-time optimal frequency based on global optimal solution to achieve frequency adaptive regulation when the load changes. For the scenario where the processor computing power is limited, the present invention proposes a pre-computation optimization mode, that is, establishing the functional relationship between the load current and the optimal frequency through offline fitting to achieve fast frequency switching without real-time calculation. Further, the present invention adds a hybrid frequency modulation strategy, adopting high frequency to suppress ripple during the current peak period and low frequency to reduce loss during the valley period, providing an active trade-off mechanism for total harmonic distortion and loss. Through a multi-level optimization architecture, the present invention realizes the comprehensive optimization of losses and ripple in the full operating condition range while ensuring power quality, and is compatible with the application requirements of different computing power platforms, featuring strong adaptability and high robustness.

[0007] To achieve the above object, the comprehensive optimization method of the present application includes the following steps:

[0008] Step 1: Measure the drain-source voltage U ds of each switching tube, and sample the drain-source current at the switching moment to obtain the current sequence i x (t);

[0009] Step 2: Based on the data sheet of the switching tube, calculate the instantaneous switching loss P sw_inst of the switching tube according to the measurement result of Step 1;

[0010] Step 3: Perform a moving average integral on the instantaneous switching loss in Step 2 to obtain the average switching loss P sw_ave ;

[0011] Step 4: Measure the DC-side voltage and calculate the maximum value Δi of the inter-phase ripple current of the bridge arm according to the empirical formula max ; The inter-phase ripple current of the bridge arm reflects the current fluctuation. This ripple current will affect the electromagnetic compatibility, thermal stress, output voltage quality, and system efficiency of the system. By optimizing modulation strategies, adding filters, synchronizing switch timings, and load balancing, etc., the inter-phase ripple current of the bridge arm can be effectively suppressed, thereby improving the performance and reliability of the system.

[0012] The inter-phase ripple current of the bridge arm refers to the current fluctuation caused by the different switching states of different phases and different load characteristics in a multilevel inverter. In the ANPC circuit, each bridge arm usually consists of multiple switching tubes to achieve multilevel output. Due to the timing difference of the switching actions and the non-linear characteristics of the load, there will be differences in the instantaneous values of the bridge arm currents of different bridge phases. This difference is the inter-phase ripple current of the bridge arm.

[0013] Step 5: Substitute the relevant parameters into the transfer function of the adopted filter to obtain the maximum value Δi of the output ripple current omax .

[0014] Step 6: Determine the reference values P ref , Δi ref , and normalize the reference values to obtain P * and Δi * , and use them to construct the objective function F(f s ).

[0015] Step 7: Perform a global optimal solution for the objective function to obtain the appropriate switching frequency f s_opt , and complete the optimization of the frequency.

[0016] Specifically, Step 2 is to calculate the instantaneous switching loss as shown in Equation (1):

[0017]

[0018] where f s refers to the switching frequency of the switching tube, i x (t) refers to the current sequence of the drain-source current, U ds refers to the drain-source voltage of the switching tube, U ref refers to the reference drain-source voltage used in the measurement of the selected switching tube data sheet, and K v refers to the voltage adjustment coefficient.

[0019] In Equation (1), the calculation method of E sw (i x (t), T j ) is shown in Equation (2):

[0020] E sw (i x (t), T j ) = E sw (i x (t)) * [1 + K T (T j - T ref )] (2)

[0021] In Equation (2), T j is the actual junction temperature of the switching device, r ref is the reference temperature used during measurement, K T is the temperature adjustment coefficient, and E sw (i x (t)) is the function expression obtained by curve fitting based on the E sw - I DS curve in the data sheet.

[0022] The detailed fitting process is as follows:

[0023] On the E sw - I DS curve, several data points (0, 0), (I1, E sw1 ), (I2, E sw2 ),..., (I n , E swn ) are evenly selected. To leave a margin, the maximum value of I n is 1.25 times the maximum instantaneous current value I max that the switching device may have under the given operating conditions.

[0024] The selected data points are used for quadratic function fitting using the least squares method, and the coefficients a and b of the quadratic function are obtained by minimizing the sum of the squared residuals. The mathematical expression for this step is The obtained function expression is

[0025] The average switching loss P sw_ave obtained in Step 3 above is specifically Equation (3):

[0026]

[0027] where T0 is the power frequency period.

[0028] In Step 4 above, the DC side voltage is measured, and the maximum value of the inter-phase ripple current Δi max of the bridge arm is calculated according to the empirical formula. Among them, the empirical formula for calculating Δi max is specifically:

[0029]

[0030] Among them, U ds is the DC-side voltage, L f is the filter inductor parameter, T s is the modulation ratio.

[0031] The specific content of step 5 is as follows: The studied ANPC converter adopts an LCL filter, and the transfer function of the filter is calculated as:

[0032]

[0033] Among them, i g refers to the grid-side current, i s refers to the inverter-side current, R d refers to the damping resistor, C f refers to the filter capacitor, L g refers to the grid-side filter inductor.

[0034] After substituting each parameter, the output current ripple amplitude Δi omax is:

[0035] Δi omax = Δi max *|G(s)| (6)

[0037] The specific calculation formula of the objective function in step 6 is:

[0038] F(f s ) = α*P * +(1-α)*Δi * (7)

[0039] Among them The principle of selecting the reference values P ref , Δi ref is as follows: P ref is the maximum switching loss that can be achieved under the given operating conditions, and Δi ref is the maximum current ripple amplitude that can be achieved under the given operating conditions. Assume that the given operating condition is that the output current amplitude i p varies between 10 - 30 A (i.e., the output current peak-to-peak value i pp varies between 20 - 60 A), and the switching frequency f s varies between 10 - 40 kHz. Then at this time, P ref is the switching loss when f s = 40 kHz and i pp = 30*2 A; Δi ref is the current ripple when f s = 10 kHz and i pp= the current ripple amplitude at 10*2A, and α is a weight with a value between (0, 1).

[0040] The specific content of step 7 is: Let As can be seen from steps 1 to 6, P sw_ave = k1*f s , Δi omax = k2 / f s ; Let the first derivative of F(f s ) be 0, and the solution can be obtained

[0041]

[0042] For the scenario where the processor computing power is limited, the present invention further proposes a pre-computation mode: In the system initialization stage, the mapping relationship between different load currents I and the optimal switching frequency f s_opt is obtained through offline calculation, and interpolation is used to obtain the relationship between f s_opt and the load current I n . During real-time operation, the switching frequency corresponding to the current load current can be directly obtained without calculating the specific values of losses and ripples.

[0043] The core idea of Akima interpolation is to fit data points through three-segment polynomials while ensuring that the slope change of the interpolation curve is smooth at each data point. The specific interpolation method is as follows: Select n load current sampling points I1, I2,..., I n within the rated operating conditions range, calculate the corresponding optimal frequencies f1, f2,..., f n for each sampling point through steps 1 to 7 in it, and use Akima interpolation to obtain the relationship between f opt and the load current I n .

[0044] The core formula of the Akima interpolation method is:

[0045]

[0046] Among them, m i and m i+1 are the slopes of adjacent intervals, w1 = |m i+1 - m i |, w2 = |m i+2 - m i+1 | are the weights of slope changes. When performing Akima interpolation, the interpolation weights are dynamically adjusted through the slopes of adjacent data points, with good accuracy and fast response accuracy. During real-time operation, the program can locate the interval where the current load current I is located and call the corresponding coefficients for interpolation calculation.

[0047] In addition, a hybrid frequency modulation mode is provided: when the hybrid frequency modulation mode is enabled, the instantaneous value i(t) of each phase current of the independent detection current is detected. When |i(t)| > I th the carrier wave of this phase maintains the original calculated frequency f opt , otherwise, the low frequency f low is adopted; where I th is the threshold value set according to the current peak value, and f s_opt is pre-calculated through steps 1 to 7 in this application, and f low = k * f s_opt , where k is an empirical coefficient.

[0048] It also has a comprehensive optimization system for switching losses and ripple current applicable to the ANPC topology, and the system has a switching loss calculation module, an inter-phase ripple current value calculation module for the bridge arm, an objective function setting and solving module, an Akima interpolation module, and a hybrid frequency modulation module.

[0049] This invention solves the contradiction between loss and ripple optimization in traditional designs, constructs an objective function that comprehensively considers losses positively correlated with the switching frequency and ripples negatively correlated, eliminates the dimensional differences of multiple physical quantities through normalization processing, and obtains the real-time optimal frequency based on global optimal solution, realizing frequency adaptive regulation when the load changes. Moreover, the Akima interpolation method can be used to pre-calculate data, improving the adaptability of the system to the computing power of the processor. Description of the Drawings

[0050] Figure 1 is a schematic structural diagram of an active clamped three-level converter (ANPC);

[0051] Figure 2 is a flowchart of a comprehensive optimization method for switching losses and ripple current applicable to the ANPC topology of the present invention;

[0052] Figure 3 is a simulation schematic diagram of the present invention;

[0053] Figure 4 is a loss calculation simulation module of the present invention;

[0054] Figure 5 is a current ripple calculation simulation module of the present invention;

[0055] Figure 6 is a simulation module for solving the switching frequency f s of the present invention;

[0056] Figure 7 is a comparison diagram of the simulation output current waveforms when the method of the present invention is not used and used;

[0057] Figure 8It is a comparison chart of the simulation switch losses when the method of the present invention is not used and used. Detailed implementation manners

[0058] The technologies described below can be subjected to various transformations and can have various embodiments. Here, specific embodiments are described in detail with reference to the accompanying drawings. However, this does not mean that the technologies described below are limited to specific embodiments. It should be understood that the present invention includes all similar modifications, equivalents, and alternatives without departing from the spirit and scope of the technologies described below.

[0059] As Figure 1 shown, it shows the circuit diagram of a three-level ANPC rectifier, which is widely used in fields such as new energy power generation and electric vehicles. How to formulate a suitable switching frequency modulation strategy to improve system efficiency is a relatively important research topic. The existing modulation strategies cannot simultaneously consider the contradiction between switching losses and ripple current. To solve the deficiencies in traditional designs, the present invention constructs an objective function that comprehensively considers the losses positively correlated with the switching frequency and the ripple negatively correlated with it, eliminates the dimensional differences of multiple physical quantities through normalization processing, and obtains the real-time optimal frequency based on global optimal solution to achieve frequency adaptive regulation when the load changes. For the scenario where the processor computing power is limited, the present invention proposes a pre-computation optimization mode: by establishing the functional relationship between the load current and the optimal frequency through offline fitting, fast frequency switching without real-time calculation is realized. Further, the present invention adds a hybrid frequency modulation strategy, uses high frequency to suppress ripple during the current peak period, and uses low frequency to reduce losses during the valley period, providing an active trade-off mechanism for total harmonic distortion and losses. Through a multi-level optimization architecture, the present invention realizes the comprehensive optimization of losses and ripple in the full operating condition range while ensuring power quality, and is compatible with the application requirements of different computing power platforms, having the characteristics of strong adaptability and high robustness.

[0060] As Figure 2 shown, to achieve the above object, the comprehensive optimization method of the present application includes the following steps:

[0061] Step 1: Measure the drain-source voltage U ds of each switching device, and sample the drain-source current at the switching moment to obtain the current sequence i x (t);

[0062] Step 2: Based on the data sheet of the switching device, calculate the instantaneous switching loss P sw_inst of the switching device according to the measurement results in Step 1;

[0063] Step 3: Perform a moving average integral on the instantaneous switching loss in Step 2 to obtain the average switching loss P sv_ave ;

[0064] Step 4: Measure the DC-side voltage and calculate the maximum value Δi of the inter-phase ripple current of the bridge arm according to the empirical formula max ;

[0065] Step 5: Substitute the relevant parameters into the transfer function of the adopted filter to obtain the maximum value Δi of the output-side ripple current omax 。

[0066] Step 6: Determine the reference values P ref 、Δi ref , and normalize the reference values to obtain P * and Δi * , and use them to construct the objective function F(f s ).

[0067] Step 7: Perform a global optimal solution for the objective function to obtain the appropriate switching frequency f s_opt , and complete the optimization of the frequency.

[0068] The specific content of the said Step 2 is to calculate the instantaneous switching loss as shown in Equation (1):

[0069]

[0070] Among them, f s refers to the switching frequency of the switching tube, U ds refers to the drain-source voltage of the switching tube, U ref refers to the reference drain-source voltage used in the measurement of the data sheet of the selected switching tube. K v refers to the voltage adjustment coefficient.

[0071] In Equation (1), the calculation method of E sw (i x (t), T j ) is shown in Equation (2):

[0072] E sw (i x (t), T j ) = E sw (i x (t)) * [1 + K T (T j - T ref )] (2)

[0074] In Equation (2), T j is the actual junction temperature of the switching tube, T ref is the reference temperature used in the measurement, K T is the temperature adjustment coefficient, and E sw (i x (t)) is based on E sw in the data sheet-I DS The function expression obtained by curve fitting.

[0075] The detailed fitting process is as follows:

[0076] At E sw -I DS Select a number of data points evenly on the -I curve: (0, 0), (I1, E sw1 ), (I2, E sw2 ),..., (I n , E swn ). To leave a margin, the maximum value of I n is 1.25 times the maximum instantaneous current value I max of the switching device under the given operating conditions.

[0077] Use the least squares method to perform quadratic function fitting on the selected data points, and minimize the sum of the squares of the residuals to obtain the coefficients a and b of the quadratic function. The mathematical expression for this step is The obtained function expression is

[0078] The average switching loss P sw_ave obtained in step 3 above is specifically Equation (3):

[0079]

[0080] where T0 is the power frequency period.

[0081] In step 4 above, measure the DC side voltage and calculate the maximum value Δi max of the inter-phase ripple current of the bridge arm according to the empirical formula. Among them, the empirical formula for calculating Δi max is specifically:

[0082]

[0083] where U ds is the DC side voltage, L f is the filter inductor parameter, and T s is the modulation ratio.

[0084] Step 5 above is specifically: The ANPC converter under study uses an LCL filter, and the transfer function of the filter is calculated as:

[0085]

[0086] where i g refers to the grid-side current, i s refers to the inverter-side current, R d refers to the damping resistor, C f refers to the filter capacitor, and Lg Refers to the grid-side filtering inductor.

[0087] After substituting each parameter, the output current ripple amplitude Δi can be obtained omax as:

[0088] Δi omax = Δi max *|G(s)| (6)

[0090] The specific calculation formula of the objective function in step 6 is:

[0091] F(f s ) = α * P * +(1 - α) * Δi * (7)

[0092] Where Select the reference value P ref , Δi ref The principle is as follows: P ref is the maximum switching loss that can be achieved under the given operating conditions, and Δi ref is the maximum current ripple amplitude that can be achieved under the given operating conditions. Assume that the given operating condition is that the output current amplitude i p varies between 10 - 30 A (i.e., the output current peak-to-peak value i pp varies between 20 - 60 A), and the switching frequency f s varies between 10 - 40 kHz. Then at this time, P ref is the switching loss when f s = 40 kHz and i pp = 30 * 2 A; Δi ref is the current ripple amplitude when f s = 10 kHz and i pp = 10 * 2 A, and α is a weight with a value in (0, 1).

[0093] Step 7 is specifically: Assume From steps 1 - 6, it can be known that P sw_ave = k1 * f s , Δi omax = k2 / f s ; Let the first derivative of F(f s ) be 0, and the solution can be obtained

[0094]

[0095] From this, the appropriate switching frequency f s_opt is obtained to complete the optimization of the frequency. Figure 6 is the schematic diagram of the simulation module for solving the switching frequency f s_opt of the present invention.

[0096] Based on the above steps 1 - 7, comprehensive optimization of switching loss and ripple current is achieved. As Figure 3 shown, it gives a simulation schematic diagram of steps 1 - 7.

[0097] For the scenario where the processor computing power is limited, the present invention further proposes a pre - calculation mode: in the system initialization stage, the mapping relationship between different load currents I and the optimal switching frequency f s_opt is obtained through offline calculation, and interpolation is used to obtain the relationship between f s_opt and the load current I n . During real - time operation, the switching frequency corresponding to the current load current can be directly obtained without calculating the specific values of losses and ripples.

[0098] The core formula of the Akima interpolation method is:

[0099]

[0100] where m i and m i+1 are the slopes of adjacent intervals, w1 = |m i+1 - m i |, w2 = |m i+2 - m i+1 | are the weights of the slope change. The Akima interpolation method can dynamically adjust the interpolation weights through the slopes of adjacent data points, and has good accuracy and fast response accuracy. During real - time operation, the program can locate the interval where the current load current I is located and call the corresponding coefficients for interpolation calculation.

[0101] The specific interpolation method is as follows:

[0102] Step 1: Select n load current sampling points I1, I2,..., I n within the rated operating conditions range, calculate the corresponding optimal frequencies f1, f2,..., f n for each sampling point through the above - mentioned steps 1 - 7, and use Akima interpolation to obtain the relationship between the original calculated frequency f opt and the load current I n .

[0103] Step 2: Traverse the current array to find the interval [current{i}, current{i + 1}] where the input load current I is located.

[0104] Step 3: Calculate the adjacent slopes. Among them,

[0105] m0: the slope of the left - adjacent interval [current{i - 1}, current{i}];

[0106] m1: The slope of the current interval [current{i}, currenti+1}];

[0107] m2: The slope of the right adjacent interval [current{i+1}, currenti+2}];

[0108] Step 4: Calculate the slope weights: w1 = |m1 - m0|; w2 = |m2 - m1|.

[0109] Step 5: Calculate the interpolation result: If the slope does not change (w1 + w2 ≈ 0), it degrades to linear interpolation. If the slope changes, use formula (9) to calculate the interpolation result.

[0110] The Akima interpolation method can dynamically adjust the interpolation weights through the slopes of adjacent data points, with good accuracy and fast response accuracy. During real-time operation, the program can locate the interval where the current load current I is located and call the corresponding coefficients for interpolation calculation.

[0111] In addition, a hybrid frequency modulation mode is also provided: When the hybrid frequency modulation mode is enabled, the instantaneous value i(t) of each phase current of the independent detection current is detected. When |i(t)| > I th the carrier wave of this phase maintains the original calculated frequency f opt , otherwise the low frequency f is adopted low ; where I th is the threshold set according to the current peak value, and f s_opt is pre-calculated by the method described in this application, and f low = k * f s_opt , where k is an empirical coefficient.

[0112] To test the effectiveness of the method proposed in the present invention during load mutation, a simulation model was built in PLECS. During the simulation, the d-axis given current (i.e., the output current / load current) was made to jump from 10 A to 20 A at 0.2 s, and then jump to 30 A at 0.3 s, and the waveforms of each output were observed.

[0113] As Figure 7 can be seen, if a relatively low switching frequency (f s = 10 kHz) is adopted, the total harmonic distortion is relatively high and the ripple is large, which is 0.5 V when the load is light (the load current is 10 A). After adopting this method, the ripple during light load is reduced to 0.05 V; the simulation results are as Figure 8 shown.

[0114] As Figures 7-8It can be seen that by adopting the comprehensive optimization method proposed by the present invention, the switching frequency can be adaptively adjusted when the output power changes, which can not only ensure high waveform quality when the output power is small, but also ensure low switching losses when the output power is large, and is suitable for devices with wide-range output.

[0115] For the scenario where the processor computing power is limited, the calculation result of the pre-computation mode proposed by the present invention can better match the result of the actual calculation, can be used as a substitute for the actual calculation in the scenario where the processor computing power is limited, greatly reduces the computing power cost, and improves the adaptability of different processors.

[0116] In addition, a hybrid frequency modulation mode is also provided: when the hybrid frequency modulation mode is enabled, the instantaneous value i(t) of each phase current of the independent detection current is detected. When |i(t)| > I th the carrier wave of this phase maintains the original calculated frequency f opt , otherwise the low frequency f low is adopted; where I th is a threshold value set according to the current peak value, and f opt is pre-calculated through steps 1 to 7 in this application, and f low = k * f opt , where k is an empirical coefficient.

[0117] It also has a comprehensive optimization system for switching losses and ripple current applicable to the ANPC topology, and the system has a switching loss calculation module, an inter-bridge-arm ripple current value calculation module, an objective function setting and solving module, an Akima interpolation module, and a hybrid frequency modulation module.

[0118] Although the present invention has been described in detail above with general descriptions and specific embodiments, based on the present invention, some modifications or improvements can be made. The above are only the preferred embodiments of the present invention, and do not limit the scope of the present invention. Other changes and modifications made by those skilled in the art without departing from the spirit and protection scope of the present invention are still included in the protection scope of the present invention.

Claims

1. A comprehensive optimization method for switching losses and ripple current applicable to the ANPC topology, characterized in that including: Step 1: Measure the drain-source voltage U of each switching transistor ds and sample the drain-source current at the switching moment to obtain the current sequence i x (t); Step 2: Based on the data sheet of the switching transistor, calculate the instantaneous switching loss P of the switching transistor according to the measurement results of Step 1 sw_inst ; Step 3: Perform a moving average integration on the instantaneous switching loss in Step 2 to obtain the average switching loss P sw_ave ; Step 4: Measure the DC-side voltage and calculate the maximum value Δi of the inter-phase ripple current of the bridge arm according to the empirical formula max ; Step 5: Substitute relevant parameters into the transfer function of the filter used in the circuit to obtain the maximum value of the output ripple current Δi omax ; Step 6: Determine the reference value P according to the constraint conditions ref , Δi ref , and normalize the reference value to obtain P * and Δi * , construct the objective function F(f s ); Step 7: Perform global optimal solution for the objective function to obtain the appropriate switching frequency f s_opt , and complete the optimization of the frequency.

2. The method according to claim 1, characterized in that, In step 2, the calculation formula for the instantaneous switching loss of the switching transistor is: Among them, f s refers to the switching frequency of the switching transistor, U ds refers to the drain-source voltage of the switching transistor, U ref refers to the reference drain-source voltage used in the measurement of the selected switching transistor data sheet, K v refers to the voltage adjustment coefficient, E sw (i x (t), T j ) The calculation formula is: E sw (i x (t),T j ) = E sw (i x (t)) * [1 + K T (T j -T ref )] (2) Among them, T j is the actual junction temperature of the switching transistor, and T ref is the reference temperature used during measurement. K T is the temperature adjustment coefficient. E sw (i x (t)) is the functional expression obtained by curve fitting according to the E sw -I DS curve in the data sheet.

3. The method according to claim 1, characterized in that, The average switching loss P in the step 3 sw_ave is calculated by the following formula: where T0 is the power frequency period.

4. The method according to claim 1, wherein The empirical formula for the maximum value of the inter-phase ripple current of the bridge arm in step 4 is: Among them, U ds is the DC-side voltage, L f is the filter inductor parameter, and T s is the modulation ratio.

5. The method according to claim 1, wherein The filter in the step 5 is an LCL filter, where the relationship between the input current i g and the output current i s is as follows: Among them, R d refers to the damping resistor, C f refers to the filter capacitor, L g refers to the grid-side filter inductor.

6. The method according to claim 5, wherein The calculation formula for the maximum value of the output terminal ripple current is: △i omax = △i max * |G(s)| (6).

7. The method according to claim 1, characterized in that, In the said step 6, P is obtained by normalizing using the reference value * and Δi * Specifically: The calculation formula for the objective function is: F(f s ) = α * P * + (1 - α) * △i * (7) Among them, P ref is the maximum switching loss that can be achieved under a given operating condition, and Δi ref is the maximum current ripple amplitude that can be achieved under a given operating condition, and α is a weight with a value between (0, 1).

8. The method according to claim 7, wherein In step 7, the global optimal solution of the objective function is specifically: Let From Steps 1 to 6, it can be seen that P sw_ave = k1 * f s , Δi omax = k2 / f s ; Let the first derivative of F(f s ) be 0, and the solution can be obtained as follows: Based on this, solve to obtain the optimal switching frequency f s_opt .

9. The method according to claim 1, characterized in that It also has a pre-computation step, specifically: in the system initialization stage, the mapping relationship between different load currents I and the optimal switching frequency f is obtained through off-line calculation, and the Akima interpolation is used to obtain the relationship between the calculated frequency and the load current. s_opt ​ 10. A system for implementing the comprehensive optimization method of switching loss and ripple current applicable to the ANPC topology according to any one of claims 1-9, characterized in that, The system includes a switching loss calculation module, an inter-phase ripple current value calculation module of the bridge arm, an objective function setting and solving module, an Akima interpolation module, and a hybrid frequency modulation module.

Images