Flying capacitor voltage estimation method and device, storage medium and computer device

Abstract

The flying capacitor voltage estimation method, device, storage medium and computer device provided by the application, when periodically estimating the capacitor voltage of the flying capacitor multilevel converter, first obtain the input variable and the capacitor voltage estimation value of the converter in the last period, and then obtain the output voltage estimation value in the current period; At the same time, the actual value of the output voltage of the converter in the current period is collected, and the actual value is compared with the output voltage estimation value to construct a loss function; Then, taking minimizing the loss function as the target, the capacitor voltage update amount in the current period is accurately calculated through the chain rule, and the capacitor voltage estimation value in the current period can be obtained by subtracting the update amount from the capacitor voltage estimation value in the last period. In this process, the application can continuously iterate and optimize the capacitor voltage estimation value based on historical data and real-time feedback, so that the estimation result is more accurate and reliable, meets the performance requirements of the power electronic converter, and avoids the dependence on a large number of sensors.

Description

Technical Field

[0001] This application relates to the field of power converter technology, and in particular to a method, apparatus, storage medium and computer equipment for estimating the voltage of a flying capacitor. Background Technology

[0002] Power converter technology is widely used in many fields such as grid connection of new energy sources, all-electric propulsion aircraft, and new energy vehicles. In these fields, the performance of power electronic converters often determines the overall system quality. In order to improve the operating voltage and power level of power converters, the academic community has proposed the flying capacitor multilevel converter. With its multilevel structure, it not only improves the power level of the device, but also reduces the harmonic content of the output voltage and increases the power density, which has attracted widespread attention.

[0003] Currently, in the control of flying capacitor voltage, achieving active balancing of the flying capacitor voltage requires real-time and accurate acquisition of the capacitor voltage, which leads to a sharp increase in the number of sensors in the system. For example, an n-level flying capacitor converter requires x(n-2) flying capacitor voltage sensors, where x represents the number of phases. However, the increase in the number of sensors not only significantly increases the system cost and size but may also pose potential reliability risks due to the quality and quantity of the sensors themselves. Summary of the Invention

[0004] The purpose of this application is to at least solve one of the above-mentioned technical defects, especially the technical defect that the acquisition of capacitor voltage in the prior art requires a large number of sensors, which not only greatly increases the system cost and size, but may also bring hidden dangers to the system reliability due to the quality and quantity of the sensors themselves.

[0005] This application provides a method for estimating the voltage of a flying capacitor, the method comprising:

[0006] Obtain the input variables and capacitor voltage estimate of the multilevel converter with flying capacitor in the previous cycle, and determine the output voltage estimate of the current cycle based on the input variables and capacitor voltage estimate.

[0007] The actual output voltage value of the multilevel converter in the current cycle is collected, and the loss function for the current cycle is constructed based on the estimated output voltage value and the actual output voltage value.

[0008] With the goal of minimizing the loss function, the capacitor voltage update amount for the current period is determined by a chain rule;

[0009] The estimated capacitor voltage of the multilevel converter in the current cycle is obtained by subtracting the capacitor voltage update from the estimated capacitor voltage of the previous cycle.

[0010] Optionally, determining the output voltage estimate for the current cycle based on the input variable and the capacitor voltage estimate includes:

[0011] A linear neural network estimation model for the multilevel converter is determined; the linear neural network estimation model is a discrete-time model established based on the contribution relationship between the input variables and the capacitor voltage to the output voltage.

[0012] The input variables and the estimated capacitor voltage are input into the linear neural network estimation model to obtain the output voltage estimate of the current cycle output by the linear neural network estimation model.

[0013] Optionally, the expression for the linear neural network estimation model includes:

[0014]

[0015] In the formula, This represents the estimated output voltage value for the k-th cycle; This represents the estimated capacitor voltage in the (k-1)th period; This represents the input variable in the (k-1)th period; T represents the transpose. This represents the power supply excitation in the (k-1)th cycle.

[0016] Optionally, the expression for the loss function includes:

[0017]

[0018] In the formula, Let represent the loss function for the k-th period; This represents the estimated output voltage value for the k-th cycle; This represents the actual output voltage value in the k-th cycle.

[0019] Optionally, determining the capacitor voltage update for the current period using a chain rule, with the objective of minimizing the loss function, includes:

[0020] Determine the partial derivative of the loss function with respect to the estimated capacitor voltage using the chain rule;

[0021] The capacitor voltage update amount for the current cycle is calculated based on the preset learning rate and the partial derivative.

[0022] Optionally, the expression for the partial derivative includes:

[0023]

[0024] In the formula, This represents the capacitor voltage in the (k-1)th period. The estimated value; Let i represent the i-th variable of the input variables in the (k-1)-th period.

[0025] This application also provides a flying capacitor voltage estimation device, comprising:

[0026] The data acquisition module is used to acquire the input variables and capacitor voltage estimate of the multilevel converter with flying capacitor in the previous cycle, and determine the output voltage estimate of the current cycle based on the input variables and capacitor voltage estimate.

[0027] The function construction module is used to acquire the actual output voltage value of the multilevel converter in the current cycle, and construct the loss function for the current cycle based on the output voltage estimate and the actual output voltage value;

[0028] The update amount determination module is used to determine the capacitor voltage update amount for the current period by means of a chain rule with the goal of minimizing the loss function.

[0029] The estimation update module is used to obtain the estimated value of the capacitor voltage of the multilevel converter in the current cycle by subtracting the updated amount of the capacitor voltage from the estimated value of the capacitor voltage in the previous cycle.

[0030] Optionally, the data acquisition module includes:

[0031] The model determination submodule is used to determine the linear neural network estimation model of the multilevel converter; the linear neural network estimation model is a discrete-time model established based on the contribution relationship between the input variables and the capacitor voltage to the output voltage.

[0032] The model estimation submodule is used to input the input variables and the estimated capacitor voltage into the linear neural network estimation model to obtain the estimated output voltage of the current cycle output by the linear neural network estimation model.

[0033] This application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the flying capacitor voltage estimation method as described in any of the above embodiments.

[0034] This application also provides a computer device, including: one or more processors, and memory;

[0035] The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the flying capacitor voltage estimation method as described in any of the above embodiments.

[0036] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0037] The flying capacitor voltage estimation method, apparatus, storage medium, and computer equipment provided in this application, when performing periodic capacitor voltage estimation on a flying capacitor multilevel converter, can first obtain the input variables and estimated capacitor voltage values ​​of the converter in the previous cycle, and determine the estimated output voltage value of the current cycle based on the input variables and estimated capacitor voltage values, as the basis data for subsequent capacitor voltage estimation; then, the actual output voltage value of the converter in the current cycle can be collected, and then a loss function for the current cycle can be constructed based on the estimated output voltage value and the actual output voltage value to quantify the degree of deviation between the estimated value and the actual value; next, with the goal of minimizing the loss function, that is, minimizing the difference between the estimated output value and the actual value, this application can accurately calculate the magnitude and direction of the adjustment required to the estimated capacitor voltage value of the previous cycle through the chain rule, as the capacitor voltage update amount for the current cycle; therefore, by subtracting the capacitor voltage update amount from the estimated capacitor voltage value of the previous cycle, the estimated capacitor voltage value of the multilevel converter in the current cycle can be obtained. In this process, this application can continuously iterate and optimize the capacitor voltage estimate based on historical data and real-time feedback, making the estimation results more accurate and reliable, meeting the power electronic converter's requirement for accurate capacitor voltage estimation, while avoiding the many drawbacks of traditional methods that rely on a large number of sensors. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 A flowchart illustrating a method for estimating the voltage of a flying capacitor provided in an embodiment of this application;

[0040] Figure 2 An interactive schematic diagram illustrating a flying capacitor voltage estimation process provided in an embodiment of this application;

[0041] Figure 3 This is a schematic diagram of the structure of a flying capacitor voltage estimation device provided in an embodiment of this application;

[0042] Figure 4 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] Currently, in the control of flying capacitor voltage, achieving active balancing of the flying capacitor voltage requires real-time and accurate acquisition of the capacitor voltage, which leads to a sharp increase in the number of sensors in the system. For example, an n-level flying capacitor converter requires x(n-2) flying capacitor voltage sensors, where x represents the number of phases. However, the increase in the number of sensors not only significantly increases the system cost and size but may also pose potential reliability risks due to the quality and quantity of the sensors themselves.

[0045] Based on this, this application proposes the following technical solution, as detailed below:

[0046] In one embodiment, such as Figure 1 As shown, Figure 1 This is a flowchart illustrating a method for estimating the voltage of a flying capacitor provided in an embodiment of this application. The method specifically includes the following:

[0047] S110: Obtain the input variables and capacitor voltage estimate of the multilevel converter with flying capacitor in the previous cycle, and determine the output voltage estimate of the current cycle based on the input variables and capacitor voltage estimate.

[0048] In this step, when performing periodic capacitor voltage estimation on the flying capacitor multilevel converter, the computer equipment can first obtain the input variables and capacitor voltage estimates of the converter in the previous cycle, and determine the output voltage estimate of the current cycle based on the input variables and capacitor voltage estimates, which serves as the basis for subsequent capacitor voltage estimation.

[0049] Specifically, during capacitor voltage estimation in the current cycle, the computer can directly obtain the estimation result from the previous cycle, i.e., the estimated capacitor voltage value, while the input variables can be constituted by a linear combination of the control signals output by the MCP controller in the previous cycle. For example, in an n-level flying capacitor converter, its input variables in cycle k-1... and capacitor voltage estimate The form can be represented as follows:

[0050]

[0051]

[0052] In the formula, This represents the control signal for the i-th flying capacitor; This represents the estimated capacitor voltage of the i-th flying capacitor.

[0053] Based on this, and combined with the discrete-time model of the n-level topology, the computer can perform correlation calculations on these historical data to derive the output voltage estimate for the current cycle. This output voltage estimate not only reflects the real-time cumulative effect of the capacitor voltage in the topology and the voltage balance state, but also serves as key basic data for estimating the capacitor voltage in the current cycle, ensuring continuity and contextual consistency in subsequent estimation processes.

[0054] S120: Acquires the actual output voltage value of the multilevel converter in the current cycle, and constructs the loss function for the current cycle based on the output voltage estimate and the actual output voltage value.

[0055] In this step, after the estimated output voltage value of the current cycle is calculated through step S110, the computer device can also collect the actual output voltage value of the multilevel converter in the current cycle. Then, based on the estimated output voltage value and the actual output voltage value, a loss function for the current cycle can be constructed to quantify the degree of deviation between the estimated value and the actual value.

[0056] Understandably, by using the output voltage estimate and the actual output voltage, the computer device can construct the loss function for the current cycle, thereby making the voltage estimation error explicit in numerical form. This allows subsequent capacitor voltage estimation algorithms to automatically adjust their internal parameters based on the magnitude of the loss function to reduce the estimation error, ensuring that the estimation results maintain high accuracy and robustness under different operating points, different load disturbances, and even different temperature conditions.

[0057] S130: With the goal of minimizing the loss function, the capacitor voltage update amount for the current cycle is determined by the chain rule.

[0058] In this step, after the loss function is constructed through step S120, the computer device can aim to minimize the loss function, that is, to minimize the gap between the output estimate and the actual value as much as possible, and accurately calculate the magnitude and direction of the adjustment to the capacitor voltage estimate of the previous cycle through the chain rule, as the capacitor voltage update amount of the current cycle.

[0059] Specifically, to minimize the loss function, the computer device can use the chain rule to accurately calculate the partial derivative of the loss function with respect to each previous period's capacitor voltage estimate along the reverse path from the loss function to the capacitor voltage estimate. This reveals the sensitivity of the current estimation error to each capacitor voltage and clearly indicates the precise magnitude and correct direction of the adjustment needed to the previous period's capacitor voltage estimate, which is then converted into the capacitor voltage update for the current period.

[0060] Understandably, through the chain rule, this application can automatically fine-tune the estimated capacitor voltage in each cycle, achieving adaptive correction capability, enabling the estimation process to follow the nonlinear fluctuation characteristics of the capacitor voltage under different operating conditions in real time; at the same time, this continuous parameter update can also effectively suppress accumulated errors and significantly improve the capacitor voltage prediction accuracy of the multilevel converter during long-term operation.

[0061] S140: Subtract the capacitor voltage update from the capacitor voltage estimate of the previous cycle to obtain the capacitor voltage estimate of the multilevel converter in the current cycle.

[0062] In this step, after the capacitor voltage update is calculated in step S130, the computer device subtracts the capacitor voltage update from the capacitor voltage estimate of the previous cycle, and can accurately obtain the capacitor voltage estimate of the multilevel converter in the current cycle without relying on a large number of sensors.

[0063] Understandably, in this way, the computer equipment can use the results of the previous cycle as a basis in each operating cycle and adaptively correct them using the actual measured data of the output voltage in the current cycle, so that the capacitor voltage estimate continuously converges to the true value. Meanwhile, the multilevel converter can fully utilize the continuity of historical data and the structural characteristics of the discrete-time model, while introducing real-time feedback signals into the update process, resulting in estimation results that exhibit high accuracy, high robustness, and fast response.

[0064] In the above embodiments, when performing periodic capacitor voltage estimation on a flying capacitor multilevel converter, the input variables and estimated capacitor voltage values ​​of the converter in the previous cycle can be obtained first. The estimated output voltage value for the current cycle is then determined based on these values, serving as the foundational data for subsequent capacitor voltage estimation. Subsequently, the actual output voltage value of the converter in the current cycle can be acquired. A loss function for the current cycle can then be constructed based on the estimated and actual output voltage values ​​to quantify the degree of deviation between the estimated and actual values. Next, this application aims to minimize the loss function, i.e., to reduce the gap between the estimated and actual output values ​​as much as possible. The chain rule is used to accurately calculate the magnitude and direction of the adjustment required to the estimated capacitor voltage value of the previous cycle, which serves as the capacitor voltage update amount for the current cycle. Therefore, by subtracting this capacitor voltage update amount from the estimated capacitor voltage value of the previous cycle, the estimated capacitor voltage value of the multilevel converter in the current cycle can be obtained. In this process, this application can continuously iterate and optimize the capacitor voltage estimate based on historical data and real-time feedback, making the estimation results more accurate and reliable, meeting the power electronic converter's requirement for accurate capacitor voltage estimation, while avoiding the many drawbacks of traditional methods that rely on a large number of sensors.

[0065] In one embodiment, the process of determining the output voltage estimate for the current cycle based on the input variables and the capacitor voltage estimate in step S110 may include:

[0066] S111: Determine the linear neural network estimation model of the multilevel converter; the linear neural network estimation model is a discrete-time model established based on the contribution relationship between the input variables and the capacitor voltage to the output voltage.

[0067] S112: Input the input variables and capacitor voltage estimates into the linear neural network estimation model to obtain the output voltage estimate of the current cycle output by the linear neural network estimation model.

[0068] In this embodiment, when estimating the output voltage, the computer device can first determine the linear neural network estimation model of the multilevel converter. This model is a discrete-time model established based on the contribution relationship between the input variables and the capacitor voltage to the output voltage. Therefore, after inputting the input variables and the estimated capacitor voltage into the linear neural network estimation model, the output voltage estimate of the current cycle output by the linear neural network estimation model can be obtained.

[0069] Specifically, after the model is determined, the computer device inputs the input variables of the previous cycle and the estimated value of the capacitor voltage into the linear neural network estimation model. The model can then perform layer-by-layer calculations on these inputs according to the preset network parameters and output the estimated value of the output voltage for the current cycle, thereby realizing real-time prediction of the dynamic changes in the output voltage.

[0070] Understandably, by introducing a linear neural network estimation model, this application can not only accurately characterize the complex voltage coupling relationships in multi-level topologies without adding additional sensors, but also adapt to the periodic fluctuations of the flying capacitor voltage with higher computational efficiency, making the output voltage prediction process lighter and more controllable.

[0071] In one embodiment, the expression for the linear neural network estimation model in step S111 may include:

[0072]

[0073] In the formula, This represents the estimated output voltage value for the k-th cycle; This represents the estimated capacitor voltage in the (k-1)th period; This represents the input variable in the (k-1)th period; T represents the transpose. This represents the power supply excitation in the (k-1)th cycle.

[0074] In this embodiment, the linear neural network estimation model introduces a power source excitation as a bias term, which can be directly superimposed on the linear combination result to adjust the output voltage estimate, making the predicted value closer to the true value.

[0075] For example, in a power management system, power excitation can represent an explicit model of external power supply excitation, determined by the DC power supply voltage and the control signal output by the controller. It primarily reflects the impact of other power inputs besides the main control signal on the output. For instance, during battery charging and discharging, power excitation can quantify the correction amount for voltage prediction caused by changes in ambient temperature. The expression for power excitation can be as follows:

[0076]

[0077] In the formula, Indicates the DC power supply voltage; This represents the control signal output by the controller to the (n-1)th flying capacitor.

[0078] In one embodiment, the expression for the loss function in step S120 may include:

[0079]

[0080] In the formula, Let represent the loss function for the k-th period; This represents the estimated output voltage value for the k-th cycle; This represents the actual output voltage value in the k-th cycle.

[0081] In this embodiment, the loss function measures the degree of difference between the estimated value and the actual value. It adopts the form of mean squared error. By calculating the square of the difference between the estimated value and the actual value, and then taking half of it, the loss value is obtained. This can amplify the error and make the subsequent solution process more sensitive to the deviation.

[0082] In one embodiment, step S130, which aims to minimize the loss function and determines the capacitor voltage update for the current period using a chain rule, may include:

[0083] S131: Determine the partial derivative of the loss function with respect to the estimated capacitor voltage using the chain rule.

[0084] S132: Calculate the capacitor voltage update for the current cycle based on the preset learning rate and partial derivatives.

[0085] In this embodiment, after the loss function is constructed, the computer device can determine the partial derivative of the loss function with respect to the estimated capacitor voltage value according to the chain rule, and then calculate the capacitor voltage update amount for the current period according to the preset learning rate and the partial derivative.

[0086] Understandably, computer equipment can use the chain rule to differentiate the loss function, accurately obtaining the partial derivative of the loss function with respect to the estimated capacitor voltage. This clarifies how the capacitor voltage estimation error propagates between different layers within the model and the specific impact of the error on the capacitor voltage estimation process. Based on this partial derivative, the computer equipment can then combine it with a preset learning rate to calculate the amount of capacitor voltage update required for the current cycle. This ensures that the update amount both reduces the estimation error and keeps the correction amplitude within a stable range, avoiding oscillations or over-adjustment.

[0087] Specifically, the expression for calculating the capacitor voltage update can be as follows:

[0088]

[0089] In the formula, This represents the capacitor voltage update of the i-th flying capacitor in the k-th period; Indicates the learning rate; Represents the loss function Estimated value of capacitor voltage The partial derivatives of .

[0090] In one embodiment, the expression for the partial derivative in step S131 may include:

[0091]

[0092] In the formula, This represents the capacitor voltage in the (k-1)th period. The estimated value; Let i represent the i-th variable of the input variables in the (k-1)-th period.

[0093] In this embodiment, the partial derivative allows the present application to measure the rate of change of the loss function with respect to the estimated capacitor voltage, that is, to reflect how the loss function will change accordingly when the estimated capacitor voltage changes slightly, thus providing important gradient information for adjusting the estimated capacitor voltage and improving the estimation accuracy of the capacitor voltage.

[0094] To better explain the flying capacitor voltage estimation method of this application, the following will be conducted... Figure 2 To further illustrate, illustratively, such as Figure 2 As shown, Figure 2 This is an interactive schematic diagram of a flying capacitor voltage estimation process provided in an embodiment of this application.

[0095] Figure 2 In the middle, the controller outputs the control signal of the current cycle to the multilevel converter. At the same time, it can acquire the control signal output by the controller in the previous cycle. And based on control signals The estimated value of the capacitor voltage in the previous cycle is derived. This can then be combined with the power excitation from the previous cycle. The estimated output voltage for the current cycle is calculated. At this time, the multilevel converter receives the control signal. Afterwards, it can be converted into electrical energy, and the output is the actual value of the output voltage for the current cycle. Based on this, computer equipment can estimate values. and actual value Construct the loss function Then, its partial derivatives are obtained by chain descent method. Then generate updated capacitor voltage values Utilize this updated value Estimated capacitor voltage for the previous cycle By updating, the estimated capacitor voltage for the current period can be obtained. .

[0096] The flying capacitor voltage estimation device provided in the embodiments of this application is described below. The flying capacitor voltage estimation device described below can be referred to in correspondence with the flying capacitor voltage estimation method described above.

[0097] In one embodiment, such as Figure 3 As shown, Figure 3This is a schematic diagram of a flying capacitor voltage estimation device provided in an embodiment of this application. This application also provides a flying capacitor voltage estimation device, including a data acquisition module 210, a function construction module 220, an update quantity determination module 230, and an estimate update module 240, specifically comprising the following:

[0098] The data acquisition module 210 is used to acquire the input variables and capacitor voltage estimates of the multilevel converter with flying capacitor in the previous cycle, and to determine the output voltage estimate of the current cycle based on the input variables and capacitor voltage estimates.

[0099] The function construction module 220 is used to acquire the actual output voltage value of the multilevel converter in the current cycle, and construct the loss function for the current cycle based on the output voltage estimate and the actual output voltage value.

[0100] The update quantity determination module 230 is used to determine the capacitor voltage update quantity for the current period by means of a chain rule with the goal of minimizing the loss function.

[0101] The estimation update module 240 is used to obtain the estimated value of the capacitor voltage of the multilevel converter in the current cycle by subtracting the capacitor voltage update amount from the estimated value of the capacitor voltage in the previous cycle.

[0102] In the above embodiments, when performing periodic capacitor voltage estimation on a flying capacitor multilevel converter, the input variables and estimated capacitor voltage values ​​of the converter in the previous cycle can be obtained first. The estimated output voltage value for the current cycle is then determined based on these values, serving as the foundational data for subsequent capacitor voltage estimation. Subsequently, the actual output voltage value of the converter in the current cycle can be acquired. A loss function for the current cycle can then be constructed based on the estimated and actual output voltage values ​​to quantify the degree of deviation between the estimated and actual values. Next, this application aims to minimize the loss function, i.e., to reduce the gap between the estimated and actual output values ​​as much as possible. The chain rule is used to accurately calculate the magnitude and direction of the adjustment required to the estimated capacitor voltage value of the previous cycle, which serves as the capacitor voltage update amount for the current cycle. Therefore, by subtracting this capacitor voltage update amount from the estimated capacitor voltage value of the previous cycle, the estimated capacitor voltage value of the multilevel converter in the current cycle can be obtained. In this process, this application can continuously iterate and optimize the capacitor voltage estimate based on historical data and real-time feedback, making the estimation results more accurate and reliable, meeting the power electronic converter's requirement for accurate capacitor voltage estimation, while avoiding the many drawbacks of traditional methods that rely on a large number of sensors.

[0103] In one embodiment, the data acquisition module 210 may include:

[0104] The model determination submodule is used to determine the linear neural network estimation model of the multilevel converter; the linear neural network estimation model is a discrete-time model established based on the contribution relationship between the input variables and the capacitor voltage to the output voltage.

[0105] The model estimation submodule is used to input the input variables and capacitor voltage estimates into the linear neural network estimation model to obtain the output voltage estimate of the current cycle output by the linear neural network estimation model.

[0106] In one embodiment, the expression for the linear neural network estimation model in the model determination submodule may include:

[0107]

[0108] In the formula, This represents the estimated output voltage value for the k-th cycle; This represents the estimated capacitor voltage in the (k-1)th period; This represents the input variable in the (k-1)th period; T represents the transpose. This represents the power supply excitation in the (k-1)th cycle.

[0109] In one embodiment, the expression for the loss function in the function construction module 220 may include:

[0110]

[0111] In the formula, Let represent the loss function for the k-th period; This represents the estimated output voltage value for the k-th cycle; This represents the actual output voltage value in the k-th cycle.

[0112] In one embodiment, the update amount determination module 230 may include:

[0113] The partial derivative determination submodule is used to determine the partial derivative of the loss function with respect to the capacitor voltage estimate according to the chain rule.

[0114] The update calculation submodule is used to calculate the capacitor voltage update amount for the current cycle based on the preset learning rate and partial derivative.

[0115] In one embodiment, the expression for the partial derivatives in the partial derivative determination submodule may include:

[0116]

[0117] In the formula, This represents the capacitor voltage in the (k-1)th period. The estimated value; Let i represent the i-th variable of the input variables in the (k-1)-th period.

[0118] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the flying capacitor voltage estimation method as described in any of the above embodiments.

[0119] In one embodiment, this application also provides a computer device storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the flying capacitor voltage estimation method as described in any of the above embodiments.

[0120] Indicatively, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the internal structure of a computer device 300 provided in an embodiment of this application. The computer device 300 can be provided as a server. (Refer to...) Figure 4 The computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions, such as application programs, that can be executed by the processing component 302. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 302 is configured to execute instructions to perform the flying capacitor voltage estimation method of any of the above embodiments.

[0121] The computer device 300 may also include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate on an operating system stored in memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0122] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0123] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0124] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0125] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for estimating the voltage across a flying capacitor, characterized in that, The method includes: The input variables and capacitor voltage estimates of the multilevel converter with flying capacitors in the previous cycle are obtained, and the voltage is estimated using a preset linear neural network estimation model based on the input variables and capacitor voltage estimates to obtain the output voltage estimate for the current cycle. The actual output voltage value of the multilevel converter in the current cycle is collected, and the loss function for the current cycle is constructed based on the estimated output voltage value and the actual output voltage value. With the goal of minimizing the loss function, the capacitor voltage update amount for the current period is determined by the chain rule; The estimated capacitor voltage of the multilevel converter in the current cycle is obtained by subtracting the capacitor voltage update from the estimated capacitor voltage of the previous cycle. The expression for the linear neural network estimation model includes: In the formula, This represents the estimated output voltage value for the k-th cycle; This represents the estimated capacitor voltage in the (k-1)th period; This represents the input variable in the (k-1)th period; T represents the transpose. This represents the power supply excitation in the (k-1)th cycle; The expression for the loss function includes: In the formula, Let represent the loss function for the k-th period; This represents the estimated output voltage value for the k-th cycle; This represents the actual output voltage value in the k-th cycle; The step of determining the capacitor voltage update for the current period using a chain rule, with the goal of minimizing the loss function, includes: Determine the partial derivative of the loss function with respect to the estimated capacitor voltage using the chain rule; The capacitor voltage update amount for the current cycle is calculated based on the preset learning rate and the partial derivative. The expression for the partial derivative includes: In the formula, Represents the capacitor voltage in the (k-1)th period. The estimated value; Let i represent the i-th variable of the input variables in the (k-1)-th period.

2. A device for estimating the voltage of a flying capacitor, characterized in that, include: The data acquisition module is used to acquire the input variables and capacitor voltage estimates of the multilevel converter with flying capacitor in the previous cycle, and to perform voltage estimation using a preset linear neural network estimation model based on the input variables and capacitor voltage estimates to obtain the output voltage estimate for the current cycle. The function construction module is used to acquire the actual output voltage value of the multilevel converter in the current cycle, and construct the loss function for the current cycle based on the output voltage estimate and the actual output voltage value; The update amount determination module is used to determine the capacitor voltage update amount for the current period by means of a chain rule with the goal of minimizing the loss function. The estimation update module is used to obtain the estimated value of the capacitor voltage of the multilevel converter in the current cycle by subtracting the updated amount of the capacitor voltage from the estimated value of the capacitor voltage in the previous cycle. The expression for the linear neural network estimation model in the data acquisition module includes: In the formula, This represents the estimated output voltage value for the k-th cycle; This represents the estimated capacitor voltage in the (k-1)th period; This represents the input variable in the (k-1)th period; T represents the transpose. This represents the power supply excitation in the (k-1)th cycle; The expression for the loss function in the function construction module includes: In the formula, Let represent the loss function for the k-th period; This represents the estimated output voltage value for the k-th cycle; This represents the actual output voltage value in the k-th cycle; The update amount determination module includes: Determine the partial derivative of the loss function with respect to the estimated capacitor voltage using the chain rule; The capacitor voltage update amount for the current cycle is calculated based on the preset learning rate and the partial derivative. The expression for the partial derivative includes: In the formula, Represents the capacitor voltage in the (k-1)th period. The estimated value; Let i represent the i-th variable of the input variables in the (k-1)-th period.

3. A storage medium, characterized in that: The storage medium stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the flying capacitor voltage estimation method as described in claim 1.

4. A computer device, characterized in that, include: One or more processors, and memory; The memory stores computer-readable instructions that, when executed by the one or more processors, perform the steps of the flying capacitor voltage estimation method as described in claim 1.

Images