An adaptive full operating condition efficiency optimal control method for solid-state transformer

Abstract

The application relates to the technical field of transformer control, in particular to a self-adaptive full-working-condition efficiency optimal control method of a solid-state transformer; the method comprises the following steps: acquiring a working condition coordinate vector of the solid-state transformer, wherein the working condition coordinate vector comprises a load normalized coordinate representing a current working load degree and a direction normalized coordinate representing a real-time flow direction of electric energy; performing fitting calculation on the working condition coordinate vector based on pre-calibrated discrete control points by using a curved surface fitting algorithm, generating a basic control instruction, so that the basic control instruction realizes smooth transition when crossing different working condition boundaries; generating a bottom layer driving signal based on the basic control instruction, controlling the action of internal electronic switching devices of the solid-state transformer, and performing full-working-condition efficiency control. The application has the effects of considering the real-time response capability of the control strategy and the adaptability under different working conditions.

Description

Technical Field

[0001] This application relates to the field of transformer control technology, and in particular to an adaptive method for optimal efficiency control of solid-state transformers under all operating conditions. Background Technology

[0002] Solid-state transformers are a new type of power conversion device based on power electronic conversion technology to achieve voltage transformation, energy transmission, and grid interaction control. They typically achieve power conversion between the input and output sides through the switching control of high-frequency power devices. Compared to traditional power frequency transformers, solid-state transformers offer advantages such as strong controllability, ease of integration with distributed energy sources, and support for bidirectional energy transmission. Therefore, they can be applied to scenarios such as renewable energy grid connection, smart power distribution, and power electronic grids. In actual operation, the load demand connected to the solid-state transformer and the power generation status of the renewable energy source constantly change, requiring the device to dynamically switch between various operating conditions, including light load, rated load, full load, and forward and reverse power transmission.

[0003] In existing technologies, efficiency optimization control of solid-state transformers is typically achieved based on pre-calibrated control parameter tables or model predictive control methods. Among these, lookup-based control methods usually obtain control parameters under different load conditions through offline testing and then call upon the appropriate parameters based on the current operating conditions during actual operation.

[0004] However, the actual operating state is usually in a continuous process of change, and discrete control parameters are difficult to cover all operating states. When the equipment operates between different operating parameters, the problem of discontinuous switching of control parameters is likely to occur, causing sudden changes in the drive signals of electronic switching devices, which in turn causes current fluctuations and affects the stability of equipment operation.

[0005] Furthermore, although model predictive control methods can dynamically optimize based on real-time conditions, they typically require the establishment of complex models and the execution of numerous matrix operations. In applications where solid-state transformers are controlled in real-time using embedded control chips, the computational power of the control chips limits their ability to meet the demands of high-frequency real-time regulation.

[0006] In summary, existing solid-state transformer efficiency control methods struggle to simultaneously achieve real-time response capability and adaptability under different operating conditions. Summary of the Invention

[0007] To address the challenges in real-time response and adaptability under different operating conditions in solid-state transformer control, this application provides an adaptive full-condition optimal efficiency control method for solid-state transformers.

[0008] Firstly, this application provides an adaptive method for optimal efficiency control of a solid-state transformer under all operating conditions, employing the following technical solution: An adaptive full-condition efficiency optimal control method for solid-state transformers includes: obtaining the operating condition coordinate vector of the solid-state transformer, wherein the operating condition coordinate vector includes load normalized coordinates representing the current working load level and direction normalized coordinates representing the real-time flow direction of electrical energy. Using a surface fitting algorithm, the coordinate vector of the working condition is fitted and calculated based on pre-calibrated discrete control points to generate basic control commands, so that the basic control commands can achieve smooth transition when crossing different working condition boundaries; Based on the basic control instructions, a low-level drive signal is generated to control the operation of the electronic switching devices inside the solid-state transformer and to perform full-condition efficiency control. Among them, generating the underlying drive signal based on the basic control command includes: determining the local fitting correction weight based on the spatial distance attenuation relationship between the working condition coordinate vector and the preset key extreme working condition point; Based on the local fit correction weight, the basic control command and the extreme value target command are dynamically fused to generate the final optimal control command; The underlying drive signals are generated based on the final optimal control command.

[0009] By acquiring the solid-state transformer operating condition coordinate vector, which includes load normalized coordinates and direction normalized coordinates, the current operating state of the solid-state transformer is transformed from traditional discrete operating condition parameters into a continuous two-dimensional state representation. Based on pre-calibrated discrete control points, a surface fitting algorithm is used to generate basic control commands, enabling the control commands to be continuously adjusted according to changes in load state and power transmission direction.

[0010] Compared to existing control methods based on discrete control parameter tables, this scheme achieves smooth transitions in control parameters through continuous surface mapping, reducing abrupt changes in the drive signals of electronic switching devices caused by variations in operating conditions, thereby mitigating the impact of current fluctuations on equipment operational stability. Simultaneously, this method can generate corresponding control commands based on real-time operating conditions and further control the operation of the electronic switching devices inside the solid-state transformer. This allows the solid-state transformer to maintain an adaptive control strategy under different operating states, such as light load, full load, and forward and reverse energy transmission, without requiring large-scale matrix calculations to improve real-time response capabilities, thus balancing real-time response and adaptability under different operating conditions.

[0011] Simultaneously, a local fitting correction weight is determined based on the spatial distance attenuation relationship between the operating condition coordinate vector and key extreme operating conditions. This weight is then used to adjust the fusion ratio between basic control commands and extreme target commands. This allows the control process to automatically determine whether it is approaching an efficiency-sensitive region based on the current operating state. When the equipment is in a normal operating range, the smoothness of the basic control commands is maintained. However, when approaching key extreme operating conditions such as extremely light load or full load, the proportion of the extreme target commands is increased, making the control results more closely match the optimal operating parameters under the corresponding conditions.

[0012] Optionally, obtaining the operating condition coordinate vector of the solid-state transformer includes: Obtain the grid-side voltage and grid-side current of the solid-state transformer. The average active power is calculated based on the grid-side voltage and the grid-side current. The direction is decomposed based on the sign of the average active power to extract the forward transmission power, reverse transmission power, and real-time transmission power amplitude. The load normalized coordinates and the direction normalized coordinates are obtained by mapping the forward transmission power, the reverse transmission power, and the real-time transmission power amplitude.

[0013] The system collects the grid-side voltage and current of the solid-state transformer and calculates the average active power. Then, it performs directional decomposition based on the sign of the average active power to obtain the forward transmission power, reverse transmission power, and real-time transmission power amplitude. At the same time, it characterizes the current scale of power transmission and the direction of power flow, enabling the control system to not only identify load changes but also the bidirectional energy flow state, thereby providing more complete operating status information for the generation of subsequent control commands.

[0014] Optionally, calculating the average active power based on the grid-side voltage and the grid-side current includes: The grid-side voltage and the grid-side current are preprocessed; the preprocessed grid-side voltage and grid-side current are multiplied to obtain the instantaneous power. The instantaneous power is filtered using a digital moving average to output the average active power.

[0015] Because voltage and current sampling signals of solid-state transformers may be affected by high-frequency interference and transient fluctuations during operation, directly using the sampled values ​​for control can easily lead to unstable operating condition judgments. Therefore, preprocessing is performed, and moving average filtering is applied to the instantaneous power to reduce the impact of sampling noise on the power calculation results. This allows the obtained average active power to more accurately reflect the current energy transmission status, thereby improving the reliability of subsequent direction judgment, operating condition coordinate construction, and control command generation processes.

[0016] Optionally, directional decomposition is performed based on the sign of the average active power to extract the forward transmission power, reverse transmission power, and real-time transmission power amplitude, including: if the average active power is greater than or equal to zero, the forward transmission power is assigned the value of the average active power, and the reverse transmission power is assigned the value of zero. If the average active power is less than zero, the forward transmission power is set to zero, and the reverse transmission power is set to the absolute value of the average active power. The absolute value of the average active power is used as the real-time transmission power amplitude.

[0017] In traditional control methods, the control strategy may change abruptly during the switching between forward and reverse power. However, this solution establishes forward and reverse power variables separately, which allows the power direction change process to be continuously described, avoiding confusion in direction judgment due to power crossing zero.

[0018] Optionally, the load-normalized coordinates and the direction-normalized coordinates are obtained by mapping the forward transmission power, the reverse transmission power, and the real-time transmission power amplitude, including: Divide the real-time transmission power amplitude by the rated transmission power of the solid-state transformer to obtain the load normalized coordinates; Calculate the difference between the forward transmission power and the reverse transmission power, and calculate the sum of the forward transmission power, the reverse transmission power, and the preset power protection constant; divide the difference by the sum of the three to obtain the directional normalized coordinates.

[0019] Transforming the actual power state to a unified coordinate space enables subsequent surface fitting algorithms to generate control commands based on standardized input parameters, reducing the limitation of the control model's applicability due to differences in equipment capacity.

[0020] Optionally, the step of using a surface fitting algorithm to fit and calculate the working condition coordinate vector based on pre-calibrated discrete control points to generate basic control commands includes: The B-spline surface fitting algorithm is used to calculate the load basis function corresponding to the load normalized coordinates and the direction basis function corresponding to the direction normalized coordinates. The basic control command is generated by weighting and superimposing the load basis function, the direction basis function, and the control point height value corresponding to the discrete control point.

[0021] A B-spline surface fitting algorithm is used to generate basic control commands based on load-normalized coordinates, direction-normalized coordinates, and pre-calibrated control point height values, enabling discrete control data to be converted into continuously changing control surfaces.

[0022] Optionally, based on the spatial distance attenuation relationship between the operating condition coordinate vector and the preset key extreme operating condition points, a local fitting correction weight is determined, including: Calculate the squared Euclidean distance between the working condition coordinate vector and the extreme coordinate vectors of the multiple key extreme working conditions; Divide the square of the Euclidean distance by the square of the sensitive area determination threshold, and take the negative exponent to obtain the attenuation weight corresponding to each critical extreme condition point. The maximum value among all the attenuation weights is selected as the local fitting correction weight.

[0023] The correction weights are continuously adjusted based on the spatial proximity between the current operating state and the extreme operating condition. As the operating state approaches the critical efficiency optimization region, the correction weights automatically increase, causing the control strategy to gradually move closer to the extreme target. Conversely, as the operating state moves away from the extreme region, the correction effect automatically weakens, avoiding frequent corrections that could negatively impact overall operational stability. Therefore, this method enables continuous switching of the control strategy from a globally smooth mode to a locally precise optimization mode.

[0024] Optionally, based on the local fit correction weights, the basic control commands and the extreme value target commands are dynamically fused to generate the final optimal control commands, including: Among the multiple key extreme operating conditions, extract the extreme target instruction corresponding to the maximum value of the local fitting correction weight; Using the local fitting correction weight and the difference between 1 and the local fitting correction weight as complementary weights, the basic control command and the extreme target command are weighted and fused to generate the final optimal control command.

[0025] Optionally, the grid-side voltage and the grid-side current are preprocessed, including: performing low-pass filtering on the grid-side voltage and the grid-side current.

[0026] This application has the following technical advantages: This application constructs a working condition coordinate system that simultaneously characterizes the load state and the direction of energy flow, and uses a continuous surface fitting method to replace the traditional discrete control parameter switching method, realizing the smooth adjustment of control commands as the operating state changes. This solves the problem that existing solid-state transformer efficiency control methods cannot simultaneously achieve control accuracy, real-time response capability, and adaptability to different operating conditions. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of an adaptive full-condition efficiency optimization control method for solid-state transformers according to an embodiment of this application. Detailed Implementation

[0028] This application discloses an adaptive method for optimal efficiency control of solid-state transformers under all operating conditions.

[0029] Reference Figure 1 An adaptive method for optimal efficiency control of solid-state transformers under all operating conditions includes steps S1-S3.

[0030] S1: Obtain the operating condition coordinate vector of the solid-state transformer. The operating condition coordinate vector includes load normalized coordinates representing the current working load level and direction normalized coordinates representing the real-time flow direction of electrical energy.

[0031] For any sampling time Obtain the grid-side voltage at the grid-side point of common coupling (PCC) of the solid-state transformer. and grid-side current .

[0032] Extracting electrical characteristics from grid-side interfaces for computation can reflect the global energy interaction state, thereby reducing the concurrent throughput pressure on the algorithm. Simultaneously, the rated transmission power specified on the manufacturer's nameplate is obtained. .

[0033] After acquiring the aforementioned multi-source signals, outlier removal is performed. It is determined whether the current grid-side voltage and current exceed their corresponding physical limits. If they do, they are identified as abnormal signals, and the valid values ​​from the previous time point are used to replace them.

[0034] Subsequently, a cutoff frequency of was adopted. Hardware low-pass filter for grid-side voltage and grid-side current Filtering is performed to suppress high-frequency electromagnetic glitches. In this embodiment, the preferred range of the low-pass filter cutoff frequency is [range missing]. to Low-pass filtering is a conventional technique in this field, and the cutoff frequency can be adjusted based on the experience of those skilled in the art.

[0035] based on Sampling frequency, using a hardware timestamp alignment mechanism to synchronize grid-side voltage. and grid-side current To ensure that the reference point is the same at the same time.

[0036] For any sampling time The aligned grid-side voltage With grid-side current Perform a product operation to obtain the instantaneous power. Apply a window size of [size missing] to this instantaneous power. The digital moving average filter outputs the average active power. The preferred range for the size of the digital sliding window is... to If the window is too small, low-frequency ripple suppression will be incomplete; if the window is too large, the command output will exhibit phase lag. Set to... It can balance smoothness and dynamic tracking speed.

[0037] For any sampling time, the corresponding average active power Directional decomposition of the sign: If Determine that electrical energy is flowing in the forward direction and transfer power in the forward direction. Assigned value and reverse transmission power Assigned value ;like It determines that electrical energy is flowing in the reverse direction and will transfer power in the forward direction. Assigned value and reverse transmission power Assigned value At the same time, Defined as real-time transmission power amplitude .

[0038] For any sampling time The ratio of real-time transmission power amplitude to rated power amplitude is used as the normalized load coordinate.

[0039] The formula for calculating the direction mapping relationship can be expressed as follows: ; In the formula, Represents the direction-normalized coordinates. Represents forward transmission power. Represents reverse transmission power. This represents the power protection constant.

[0040] In the formula, as the value of the independent variable, the positive transmission power, increases, the numerator becomes positive and expands, causing the normalized coordinate direction of the dependent variable to shift. Approximation; conversely, as the reverse transmission power of the independent variable increases, the numerator becomes negative, causing the normalized coordinates of the dependent variable to shift. Approaching. In real-world scenarios, this reflects a continuous physical transition between the direction of electrical energy and the load state between forward full load and reverse full load.

[0041] Power protection constant The preferred range is the rated transmission power. Five per thousand to two percent. In this embodiment, it is set to... If the value is too small, the floating-point precision of the operation will be limited and a division-by-zero error will easily occur; if the value is too large, it will affect the direction recognition resolution under medium and light load conditions.

[0042] For example: assuming rated transmission power Power protection constant At a certain sampling time The electrical energy is in a reverse grid-connected state, and the power is transmitted in the forward direction. Reverse transmission power Substitute the normalized coordinates into the calculation load. Substitute into the formula to calculate the normalized coordinates of the direction. Finally, the working condition coordinate vector at that moment is constructed. .

[0043] S2: Using a surface fitting algorithm, the working condition coordinate vector is fitted and calculated based on pre-calibrated discrete control points to generate basic control commands, so that the basic control commands can achieve a smooth transition when crossing different working condition boundaries.

[0044] Based on the output of the aforementioned steps, the working condition coordinate vector The basic control commands are calculated using a B-spline surface fitting mechanism. .

[0045] Traditional discrete lookup table methods can cause abrupt parameter jumps when crossing discrete operating condition boundaries. This embodiment constructs a continuous surface using a multi-dimensional basis function weighted superposition mechanism, enabling control parameters to smoothly transition along the surface's gradual change trend when crossing different operating conditions, eliminating strategy blind spots. Specifically, its formula can be expressed as: ; In the formula, Representative moment The basic control commands, in the physical control chain of a solid-state transformer, are not abstract software logic planes, but directly correspond to the physical action reference values ​​of the underlying electronic switching devices (e.g., the basic duty cycle percentage of a pulse width modulation signal or the basic phase shift angle of a full-bridge circuit). The numerical values ​​of the basic control commands characterize the optimal parameters for driving the switching transistors, determined by the global smooth surface, before the application of local extremum corrections. and These represent the number of load control points and the number of direction control points, respectively, and are obtained through offline calibration before the equipment is actually put into operation. Representative moment The load normalized coordinates, Representative moment Direction-normalized coordinates Represents the load basis function. Represents the directional basis functions. Representing the control point height value, it can be understood that in the three-dimensional spatial mapping of discrete control points into a continuous and smooth control surface, the load-normalized coordinates and direction-normalized coordinates constitute a two-dimensional basis (X-axis and Y-axis), and this control point height value corresponds to the geometric height of the spatial surface on the Z-axis. Physically, this height value maps to the height predetermined through offline efficiency testing experiments at the [missing information - likely a specific point or value]. The load point and the first The optimal operating parameters of the hardware at the intersection of the points in each direction (i.e., the measured duty cycle or phase shift angle values ​​that can minimize the internal switching losses of the solid-state transformer under specific operating conditions). An index variable representing the direction of the load. Index variable representing the directional dimension. Represents the degree of the load basis function. This represents the degree of the directional basis function.

[0046] Changes in the normalized coordinates of the independent variables, load and direction, alter the local activation weights of each control point. The closer the load condition coordinates are to a control point, the larger the product of the load basis function and direction basis function at that point, causing the dependent variable's basic control command to favor the height value of that control point. In practical scenarios, this reflects that the equipment's action commands are determined by the joint weighting and smooth fusion of multiple surrounding physical parameters.

[0047] Number of load control points Number of directional control points The reasonable range is to The preferred value is If the value is too small, the surface construction resolution will be insufficient; if the value is too large, the computational load will compromise real-time response capabilities. (Basis function degree) and The reasonable range is to Preferred This is to maintain the continuity of the second derivative of the surface.

[0048] For example: Suppose the current working condition coordinates are affected by four adjacent control points, whose control point height values ​​are respectively , , , The two-dimensional weights are obtained through basis function operations. , , , Substituting the values ​​into the formula yields the basic control commands. .

[0049] S3: Generate underlying drive signals based on the basic control commands to control the operation of the electronic switching devices inside the solid-state transformer and perform full-condition efficiency control.

[0050] Understandably, to ensure global smoothness, basic control commands may compromise near extreme points, leading to a loss of accuracy and preventing maximum efficiency from being achieved under full load or very light load conditions. Therefore, by introducing a spatial distance attenuation algorithm, deviations can be compensated and corrected in sensitive areas.

[0051] ; In the above formula, Representative moment Local fitting correction weight, This represents the total number of critical extreme operating points. Index variables representing extreme operating conditions. Representative moment The working condition coordinate vector, Represents the extreme value coordinate vector. This represents the threshold for determining sensitive areas. Represents the symbol for Euclidean distance calculation. This represents an exponential function with the natural constant as its base.

[0052] As the spatial Euclidean distance between the independent variable's working condition coordinate vector and the preset extreme value coordinate vector decreases, the fractional part of the exponential component approaches zero, causing the local fit correction weight of the dependent variable to increase and tend towards... Conversely, if the Euclidean distance increases, the local fitting correction weight decreases exponentially. .

[0053] In real-world scenarios, when the solid-state transformer is in a normal load zone, the correction weight is reduced; when the operating condition is extremely close to the extreme value zone, the constraint of the curved surface on the extreme point is immediately strengthened to recover the local efficiency loss.

[0054] Total number of critical extreme operating points Reasonable range to Preferred Sensitive area determination threshold The preferred range is to Preferred If the threshold is too low, the locking range will be too narrow and prone to missing the target; if the threshold is too high, it will disrupt the large-scale global smoothness of the base surface.

[0055] Finally, the fusion outputs the ultimate optimal control command. The reason for constructing this formula is that it is necessary to ensure that the instructions sent to the underlying execution chip are smooth during switching. By dynamically fusing the convex combination model, not only is the complementary normalization of weights maintained, but the infinite smoothness characteristics brought by Gaussian exponential decay are also preserved, thus avoiding the generation of high-frequency instruction glitches.

[0056] ; In the above formula, Representative moment The final optimal control command, Representative moment Local fitting correction weight, Representative moment Basic control commands, Representative moment The extreme value target instructions are developed by R&D personnel during the equipment manufacturing or commissioning phase. For solid-state transformers under several key physical boundary states (i.e., critical extreme operating points, such as extremely light load forward, extremely light load reverse, full load forward, and full load reverse), they use physical experiments or high-precision efficiency tests to identify the theoretically optimal control parameters that enable the equipment to achieve the lowest losses under these extreme conditions. These parameters (such as the optimal duty cycle or phase shift angle) are recorded and stored as an extreme value target library.

[0057] The increase in the weight of the local fitting correction of the independent variable increases the weight of the extreme value target instruction, while the dominance of the basic control instruction decreases by an equal amount. This causes the final optimal control instruction of the dependent variable to shift towards the extreme value instruction. In practical scenarios, this reflects the gradual transition of control instructions for underlying devices from a globally smooth mode to a locally precise locking mode, which protects electronic switching devices while achieving optimal efficiency.

[0058] Finally, based on the ultimate optimal control command This is configured as a pulse waveform control signal to directly drive the hardware. Specifically, the final optimal control command is passed as a duty cycle percentage to the Pulse Width Modulation (PWM) generator to precisely control the switching time ratio of the level states, driving the on and off of the electronic switching devices inside the solid-state transformer in real time. Because this command comprehensively considers the real-time dynamic response of the workload and power flow, the operating rhythm of the electronic switches can automatically adapt to complex two-dimensional changing environments. Regardless of whether the equipment is in a standby state to maintain the system or in a state of extreme overload, it can issue smooth and non-jumping operating parameters to ensure that the solid-state transformer continues to operate within the operating range with the most energy-saving and minimal impact.

[0059] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An adaptive full-condition efficiency optimal control method for solid-state transformers, characterized in that, include: Obtain the operating condition coordinate vector of the solid-state transformer, which includes load-normalized coordinates representing the current working load level and direction-normalized coordinates representing the real-time flow direction of electrical energy. Using a surface fitting algorithm, the coordinate vector of the working condition is fitted and calculated based on pre-calibrated discrete control points to generate basic control commands, so that the basic control commands can achieve smooth transition when crossing different working condition boundaries; Based on the basic control instructions, a low-level drive signal is generated to control the operation of the electronic switching devices inside the solid-state transformer and to perform full-condition efficiency control. Among them, generating the underlying drive signal based on the basic control command includes: determining the local fitting correction weight based on the spatial distance attenuation relationship between the working condition coordinate vector and the preset key extreme working condition point; Based on the local fit correction weight, the basic control command and the extreme value target command are dynamically fused to generate the final optimal control command; The underlying drive signals are generated based on the final optimal control command.

2. The adaptive full-condition efficiency optimal control method for solid-state transformers according to claim 1, characterized in that, The process of obtaining the operating condition coordinate vector of the solid-state transformer includes: Obtain the grid-side voltage and grid-side current of the solid-state transformer. The average active power is calculated based on the grid-side voltage and the grid-side current. The direction is decomposed based on the sign of the average active power to extract the forward transmission power, reverse transmission power, and real-time transmission power amplitude. The load normalized coordinates and the direction normalized coordinates are obtained by mapping the forward transmission power, the reverse transmission power, and the real-time transmission power amplitude.

3. The adaptive full-condition efficiency optimal control method for solid-state transformers according to claim 2, characterized in that, The average active power is calculated based on the grid-side voltage and the grid-side current, including: The grid-side voltage and the grid-side current are preprocessed; the preprocessed grid-side voltage and grid-side current are multiplied to obtain the instantaneous power. The instantaneous power is filtered using a digital moving average to output the average active power.

4. The adaptive full-condition efficiency optimal control method for a solid-state transformer according to claim 2, characterized in that, Directional decomposition is performed based on the sign of the average active power to extract forward transmission power, reverse transmission power, and real-time transmission power amplitude, including: if the average active power is greater than or equal to zero, the forward transmission power is assigned the value of the average active power, and the reverse transmission power is assigned the value of zero. If the average active power is less than zero, the forward transmission power is set to zero, and the reverse transmission power is set to the absolute value of the average active power. The absolute value of the average active power is used as the real-time transmission power amplitude.

5. The adaptive full-condition efficiency optimal control method for a solid-state transformer according to claim 2, characterized in that, The load-normalized coordinates and the direction-normalized coordinates are obtained by mapping the forward transmission power, the reverse transmission power, and the real-time transmission power amplitude, including: Divide the real-time transmission power amplitude by the rated transmission power of the solid-state transformer to obtain the load normalized coordinates; Calculate the difference between the forward transmission power and the reverse transmission power, and calculate the sum of the forward transmission power, the reverse transmission power, and the preset power protection constant; divide the difference by the sum of the three to obtain the directional normalized coordinates.

6. The adaptive full-condition efficiency optimal control method for a solid-state transformer according to claim 1, characterized in that, The method utilizes a surface fitting algorithm to fit and calculate the operating condition coordinate vector based on pre-calibrated discrete control points, generating basic control commands, including: The B-spline surface fitting algorithm is used to calculate the load basis function corresponding to the load normalized coordinates and the direction basis function corresponding to the direction normalized coordinates. The basic control command is generated by weighting and superimposing the load basis function, the direction basis function, and the control point height value corresponding to the discrete control point.

7. The adaptive full-condition efficiency optimal control method for solid-state transformers according to claim 1, characterized in that, Based on the spatial distance attenuation relationship between the operating condition coordinate vector and the preset key extreme operating condition points, the local fitting correction weight is determined, including: Calculate the squared Euclidean distance between the working condition coordinate vector and the extreme coordinate vectors of the multiple key extreme working conditions; Divide the square of the Euclidean distance by the square of the sensitive area determination threshold, and take the negative exponent to obtain the attenuation weight corresponding to each critical extreme condition point. The maximum value among all the attenuation weights is selected as the local fitting correction weight.

8. The adaptive full-condition efficiency optimal control method for a solid-state transformer according to claim 7, characterized in that, Based on the local fit correction weight, the basic control command and the extreme value target command are dynamically fused to generate the final optimal control command, including: Among the multiple key extreme operating conditions, extract the extreme target instruction corresponding to the maximum value of the local fitting correction weight; Using the local fitting correction weight and the difference between 1 and the local fitting correction weight as complementary weights, the basic control command and the extreme target command are weighted and fused to generate the final optimal control command.

9. The adaptive full-condition efficiency optimal control method for a solid-state transformer according to claim 3, characterized in that, Preprocessing the grid-side voltage and the grid-side current includes: performing low-pass filtering on the grid-side voltage and the grid-side current.

Images