A voltage control method for a high-reliability high-voltage power supply

Abstract

The present application relates to the technical field of voltage control. Specifically, it discloses a voltage control method for a high-reliability high-voltage power supply. After input filtering and power conversion step-up processing of the DC voltage signal provided by the power supply to obtain a high-voltage AC signal and performing rectification filtering and sampling, a signal processing technology based on deep learning is used to perform CEEMDAN decomposition and intrinsic mode feature extraction on the sampled voltage signal, and the feature saliency modulation dynamic aggregation is performed on each intrinsic mode after the decomposition of the sampled voltage signal to suppress the noise and interference in the signal, obtaining a pure sampled voltage signal. Furthermore, combined with the external control voltage signal, error calculation and MOS transistor drive control are performed to achieve precise regulation of the output voltage of the high-voltage power supply. In this way, the accuracy of voltage control can be effectively improved, the adaptability to complex and variable load conditions and interference environments can be enhanced, thereby significantly improving the stability and reliability of the high-voltage power supply system.

Description

Technical Field

[0001] This application relates to the technical field of voltage control, and more specifically, to a voltage control method for a high-reliability high-voltage power supply. Background Art

[0002] In the field of power electronics, high-reliability high-voltage power supplies are widely used in many key fields such as medical equipment, industrial control, aerospace, and communication equipment. These application scenarios pose extremely high requirements for the output voltage stability, accuracy, and reliability of the power supply.

[0003] However, traditional voltage control methods, such as analog circuit control or simple digital signal processing (DSP) technology, often fail to achieve ideal control effects when facing complex and variable working conditions and external environmental interference. Specifically, the DC voltage signal provided by the power supply is easily affected by various noises and interference during transmission, including but not limited to electromagnetic interference, line loss, and temperature fluctuations. These adverse factors will reduce the purity and stability of the voltage signal. Although traditional filtering techniques can suppress these interference to a certain extent, the effect is limited, and often leads to signal distortion or delay, affecting the accuracy and response speed of voltage control.

[0004] In addition, during the boost process of the high-voltage power supply, due to the non-linear characteristics of power electronic devices and electromagnetic interference during the switching process, a large amount of high-frequency harmonics and noises will be generated. These harmonics and noises will further affect the stability and accuracy of the output voltage. Although traditional rectification and filtering methods can remove some high-frequency components, their effect is not ideal for complex and variable harmonic interference.

[0005] Therefore, an optimized voltage control method for a high-reliability high-voltage power supply is expected. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. An embodiment of the present application provides a voltage control method for a high-reliability high-voltage power supply, which processes the DC voltage signal provided by the power supply through input filtering and power conversion boost processing to obtain a high-voltage AC signal. Then, after rectifying, filtering, and sampling the high-voltage AC signal, a signal processing technology based on deep learning is used to perform CEEMDAN decomposition and extraction of intrinsic mode features on the sampled voltage signal, and the feature saliency modulation of each intrinsic mode after decomposing the sampled voltage signal is dynamically aggregated to suppress the noise and interference in the signal, obtaining a pure sampled voltage signal. Furthermore, an error calculation and MOS transistor drive control are combined with an external control voltage signal to achieve precise regulation of the output voltage of the high-voltage power supply. In this way, the accuracy of voltage control can be effectively improved, the adaptability to complex and variable load conditions and interference environments can be enhanced, and thus the stability and reliability of the high-voltage power supply system are significantly improved.

[0007] According to one aspect of the present application, there is provided a voltage control method for a high-reliability high-voltage power supply, which includes:

[0008] Processing the initial DC voltage signal provided by the power supply through an input filtering circuit electrically connected to the power supply to obtain a DC voltage signal;

[0009] Performing power conversion and boost processing on the DC voltage signal to obtain a high-voltage AC signal;

[0010] Processing the high-voltage AC signal through a rectifying, filtering, and sampling circuit to output a DC voltage, and sampling the DC voltage to obtain a sampled voltage signal, and the sampled voltage signal is fed back to the control circuit;

[0011] The control circuit generates an error signal based on the sampled voltage signal and an external control voltage signal, and generates a PWM drive signal based on the error signal, and the PWM drive signal is transmitted to the MOS transistor drive circuit;

[0012] The MOS transistor drive circuit drives the MOS transistor based on the PWM drive signal to control the level of the output DC voltage.

[0013] In the above voltage control method for a high-reliability high-voltage power supply, the control circuit generates an error signal based on the sampled voltage signal and the external control voltage signal, including: performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal; performing dynamic aggregation based on feature energy distribution on the set of intrinsic modal semantic feature vectors of the sampled voltage signal to obtain a globally significant aggregation representation vector of the sampled voltage signal; performing signal reconstruction based on the globally significant aggregation representation vector of the sampled voltage signal to obtain a pure sampled voltage signal; and inputting the pure sampled voltage signal and the external control voltage signal into an arithmetic unit to obtain the error signal.

[0014] In the above voltage control method for a high-reliability high-voltage power supply, performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal includes: performing CEEMDAN decomposition on the sampled voltage signal to obtain a set of intrinsic modal functions of the sampled voltage signal; and respectively inputting each intrinsic modal function of the set of intrinsic modal functions of the sampled voltage signal into an intrinsic modal feature extractor based on a 1D-CNN model to obtain the set of intrinsic modal semantic feature vectors of the sampled voltage signal.

[0015] In the above voltage control method for a high-reliability high-voltage power supply, performing dynamic aggregation based on feature energy distribution on the set of intrinsic modal semantic feature vectors of the sampled voltage signal to obtain a globally significant aggregation representation vector of the sampled voltage signal includes: determining an initial center vector for intrinsic modal clustering of the sampled voltage signal based on the feature energy distribution of the set of intrinsic modal semantic feature vectors of the sampled voltage signal; and performing explicit dynamic aggregation on the set of intrinsic modal semantic feature vectors of the sampled voltage signal based on the spatial span of each intrinsic modal semantic feature vector of the set of intrinsic modal semantic feature vectors of the sampled voltage signal relative to the initial center vector for intrinsic modal clustering of the sampled voltage signal to obtain the globally significant aggregation representation vector of the sampled voltage signal.

[0016] In the above voltage control method for a high-reliability high-voltage power supply, determining an initial center vector for intrinsic modal clustering of the sampled voltage signal based on the feature energy distribution of the set of intrinsic modal semantic feature vectors of the sampled voltage signal includes: calculating a static energy factor for each intrinsic modal semantic feature vector of the set of intrinsic modal semantic feature vectors of the sampled voltage signal to obtain a set of static energy factors for intrinsic modal of the sampled voltage signal; and selecting the intrinsic modal semantic feature vector corresponding to the maximum value in the set of static energy factors for intrinsic modal of the sampled voltage signal as the initial center vector for intrinsic modal clustering of the sampled voltage signal.

[0017] In the above voltage control method for a high-reliability high-voltage power supply, calculating the static energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode static energy factors includes: calculating the kurtosis of the sampling voltage signal intrinsic mode semantic feature vector and inputting the kurtosis into a sigmoid activation function to obtain the sampling voltage signal intrinsic mode static energy factor.

[0018] In the above voltage control method for a high-reliability high-voltage power supply, based on the spatial span of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors relative to the sampling voltage signal intrinsic mode clustering initial center vector, performing explicit dynamic aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain the sampling voltage signal global significant aggregation representation vector, including: based on the spatial span between each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors and the sampling voltage signal intrinsic mode clustering initial center vector, and the set of sampling voltage signal intrinsic mode static energy factors, calculating the dynamic aggregation energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode dynamic aggregation energy factors; based on the set of sampling voltage signal intrinsic mode dynamic aggregation energy factors, performing significant modulation aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain the sampling voltage signal global significant aggregation representation vector.

[0019] In the above voltage control method for a high-reliability high-voltage power supply, calculating the dynamic aggregation energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode dynamic aggregation energy factors includes: taking the square value of the number of feature vectors separated between the sampling voltage signal intrinsic mode semantic feature vector and the sampling voltage signal intrinsic mode clustering initial center vector as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the sampling voltage signal intrinsic mode semantic feature vector and the static energy factor of the sampling voltage signal intrinsic mode clustering initial center vector and the spatial span coefficient to obtain the sampling voltage signal intrinsic mode dynamic aggregation energy factor.

[0020] In the above voltage control method for a high - reliability high - voltage power supply, based on the set of intrinsic mode dynamic aggregation energy factors of the sampled voltage signal, a significant modulation aggregation is performed on the set of intrinsic mode semantic feature vectors of the sampled voltage signal to obtain the globally significant aggregation representation vector of the sampled voltage signal, including: inputting the set of intrinsic mode dynamic aggregation energy factors of the sampled voltage signal into a gated mask unit to obtain a set of intrinsic mode dynamic aggregation weight factors of the sampled voltage signal; calculating the weighted sum of the set of intrinsic mode semantic feature vectors of the sampled voltage signal based on the set of intrinsic mode dynamic aggregation weight factors of the sampled voltage signal to obtain the globally significant aggregation representation vector of the sampled voltage signal.

[0021] In the above voltage control method for a high - reliability high - voltage power supply, signal reconstruction is performed based on the globally significant aggregation representation vector of the sampled voltage signal to obtain a pure sampled voltage signal, including: inputting the globally significant aggregation representation vector of the sampled voltage signal into a signal reconstruction module based on a decoder to obtain the pure sampled voltage signal.

[0022] Compared with the prior art, the voltage control method for a high - reliability high - voltage power supply provided by this application filters the input DC voltage signal provided by the power supply and performs power conversion and boost processing to obtain a high - voltage AC signal. Then, after rectifying, filtering, and sampling the high - voltage AC signal, a signal processing technology based on deep learning is used to perform CEEMDAN decomposition and intrinsic mode feature extraction on the sampled voltage signal, and feature significance modulation dynamic aggregation is performed on each intrinsic mode after the decomposition of the sampled voltage signal to suppress noise and interference in the signal, obtaining a pure sampled voltage signal. Furthermore, an error calculation and MOS transistor drive control are combined with an external control voltage signal to achieve precise regulation of the output voltage of the high - voltage power supply. In this way, the accuracy of voltage control can be effectively improved, the adaptability to complex and variable load conditions and interference environments can be enhanced, and thus the stability and reliability of the high - voltage power supply system are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above - mentioned and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0024] Figure 1 It is a flowchart of a voltage control method for a high - reliability high - voltage power supply according to an embodiment of the present application.

[0025] Figure 2It is a flowchart of sub-step S4 of the voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application.

[0026] Figure 3 It is a schematic diagram of data flow of sub-step S4 of the voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application.

[0027] Figure 4 It is a flowchart of sub-step S41 of the voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application.

[0028] Figure 5 It is a flowchart of sub-step S42 of the voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application. Detailed implementation manners

[0029] As shown in the present application and the claims, unless the context clearly indicates an exception, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0030] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0031] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below are not necessarily executed precisely in sequence. On the contrary, as needed, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0032] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0033] In view of the above technical problems, the present application proposes an optimized voltage control method for a high-reliability high-voltage power supply. It processes the DC voltage signal provided by the power supply through input filtering and power conversion boosting to obtain a high-voltage AC signal. Then, after rectifying, filtering, and sampling the high-voltage AC signal, it uses a deep learning-based signal processing technology to perform CEEMDAN decomposition and inherent mode feature extraction on the sampled voltage signal, and dynamically aggregates the feature saliency of each inherent mode after decomposing the sampled voltage signal to suppress the noise and interference in the signal, obtaining a pure sampled voltage signal. Furthermore, it combines with an external control voltage signal to calculate the error and control the MOS transistor drive, so as to achieve precise regulation of the output voltage of the high-voltage power supply. In this way, the accuracy of voltage control can be effectively improved, and the adaptability to complex and variable load conditions and interference environments can be enhanced, thus significantly improving the stability and reliability of the high-voltage power supply system.

[0034] Figure 1 FIG. is a flowchart of a voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application. As Figure 1 shown, the voltage control method for a high-reliability high-voltage power supply includes the steps of: S1, processing the initial DC voltage signal provided by the power supply through an input filtering circuit electrically connected to the power supply to obtain a DC voltage signal; S2, performing power conversion and boosting on the DC voltage signal to obtain a high-voltage AC signal; S3, processing the high-voltage AC signal through a rectifying, filtering, and sampling circuit to output a DC voltage, and sampling the DC voltage to obtain a sampled voltage signal, and feeding back the sampled voltage signal to a control circuit; S4, the control circuit generating an error signal based on the sampled voltage signal and an external control voltage signal, and generating a PWM drive signal based on the error signal, and transmitting the PWM drive signal to a MOS transistor drive circuit; S5, the MOS transistor drive circuit driving a MOS transistor based on the PWM drive signal to control the level of the output DC voltage.

[0035] In the above voltage control method for a high - reliability high - voltage power supply, in step S1, the input filter circuit electrically connected to the power supply processes the initial DC voltage signal provided by the power supply to obtain a DC voltage signal. It should be understood that when the power supply provides the initial DC voltage signal, it may be affected by factors such as grid fluctuations and load changes, resulting in unstable output voltage, and the high - frequency noise and spike voltage in the signal may damage the electronic components in the subsequent power converter and boost circuit. The input filter circuit can effectively filter out high - frequency noise and smooth the DC voltage signal, thus ensuring the stable operation of the subsequent circuit. In the embodiment of the present application, the input filter circuit adopts an LC filter, and by reasonably designing the parameters of the inductor and capacitor, the purpose of suppressing high - frequency noise and reducing voltage fluctuations is achieved.

[0036] In the above voltage control method for a high - reliability high - voltage power supply, in step S2, the DC voltage signal is subjected to power conversion and boost processing to obtain a high - voltage AC signal. It should be understood that in application scenarios such as medical equipment, industrial control, and communication equipment, a relatively high voltage level is usually required to drive the load or perform specific functions. Through power conversion and boost processing, the DC voltage provided by the power supply can be converted into the required high - voltage AC signal to meet the needs of these applications. In addition, during the power transmission and distribution process, high - voltage transmission can effectively reduce line losses, thereby improving energy utilization efficiency.

[0037] In the above voltage control method for a high - reliability high - voltage power supply, in step S3, the rectifying, filtering, and sampling circuit processes the high - voltage AC signal to output a DC voltage, samples the DC voltage to obtain a sampled voltage signal, and feeds the sampled voltage signal back to the control circuit. It should be understood that electromagnetic interference will be generated during the transmission and use of alternating current. The role of the rectifying, filtering, and sampling circuit is to convert the high - voltage AC signal into a DC voltage signal, and remove the high - frequency noise and ripple through the filtering link, further improving the quality of the output voltage for subsequent voltage control. In an embodiment of the present application, the rectifying, filtering, and sampling circuit can be implemented using a full - bridge rectifier in cooperation with a low - pass filter. The full - bridge rectifier can convert the AC signal into a pulsating DC signal, and the low - pass filter further filters out the high - frequency components in the pulsating DC signal to obtain a smooth DC voltage output. Then, a part of the voltage is further extracted from the rectified and filtered DC voltage as the sampled voltage signal to reflect the real - time state of the output voltage and provide accurate feedback for subsequent voltage control. The sampled voltage signal, as a feedback signal, is sent back to the control circuit. The control circuit adjusts the duty cycle of the PWM drive signal according to the difference between the sampled voltage signal and the external control voltage signal (i.e., the error signal), thereby effectively controlling the MOS - tube drive circuit and further achieving the purpose of adjusting the output voltage level.

[0038] Specifically, in the present application, considering that during the step - up processing of the signal, due to the non - linear characteristics of power electronic devices and electromagnetic interference during the switching process, a large amount of high - frequency harmonics and noise will be generated, and these harmonics and noise will further affect the stability and accuracy of the output voltage. Although traditional rectifying and filtering methods can remove some high - frequency components, their effects are not ideal for complex and variable harmonic interference. At the same time, due to the inevitable parasitic inductance and parasitic capacitance in the circuit, additional high - frequency harmonics will also be generated. Therefore, in order to effectively suppress the influence of these interference components on voltage control, the present application further uses a signal - processing technology based on deep learning to purify the sampled voltage signal.

[0039] Figure 2 It is a flowchart of sub - step S4 of the voltage control method for a high - reliability high - voltage power supply according to an embodiment of the present application. Figure 3 It is a schematic diagram of data flow of sub - step S4 of the voltage control method for a high - reliability high - voltage power supply according to an embodiment of the present application. As Figure 2 and Figure 3As shown, step S4 includes steps: S41, performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal; S42, performing dynamic aggregation based on feature energy distribution on the set of intrinsic modal semantic feature vectors of the sampled voltage signal to obtain a globally significant aggregation representation vector of the sampled voltage signal; S43, performing signal reconstruction based on the globally significant aggregation representation vector of the sampled voltage signal to obtain a pure sampled voltage signal; S44, inputting the pure sampled voltage signal and the external control voltage signal into an arithmetic unit to obtain the error signal.

[0040] Specifically, step S41, performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal. Among them, Figure 4 is a flowchart of sub-step S41 of the voltage control method for a highly reliable high-voltage power supply according to an embodiment of the present application. As Figure 4 shown, step S41 includes steps: S411, performing CEEMDAN decomposition on the sampled voltage signal to obtain a set of intrinsic modal functions of the sampled voltage signal; S412, respectively inputting each intrinsic modal function of the set of intrinsic modal functions of the sampled voltage signal into an intrinsic modal feature extractor based on a 1D-CNN model to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal.

[0041] More specifically, step S411, performing CEEMDAN decomposition on the sampled voltage signal to obtain a set of intrinsic modal functions of the sampled voltage signal. It should be understood that in order to more finely analyze the frequency characteristics of the sampled voltage signal, the present application uses the CEEMDAN (Complete Ensemble Empirical Mode Decomposition with Adaptive Noise) technique to decompose the sampled voltage signal into a series of intrinsic modal functions (IMFs) to obtain a set of intrinsic modal functions of the sampled voltage signal. Among them, each intrinsic modal function of the sampled voltage signal respectively represents the information of the sampled voltage signal within a specific frequency range, which helps to reveal the internal structure of the signal and identify the interference components in the signal. It is worth mentioning that CEEMDAN is an improved empirical mode decomposition (EMD) method, which is very suitable for processing non-linear and non-stationary signals. It adds adaptive white noise to the sampled voltage signal, then performs multiple EMD (Empirical Mode Decomposition) decompositions, and finally takes the average to obtain a more stable decomposition result, which can effectively avoid the occurrence of modal aliasing phenomenon, thereby improving the accuracy and stability of the decomposition.

[0042] More specifically, in step S412, each sampling voltage signal intrinsic mode function in the set of sampling voltage signal intrinsic mode functions is input into the intrinsic mode feature extractor based on the 1D-CNN model to obtain the set of sampling voltage signal intrinsic mode semantic feature vectors. It should be understood that in order to further learn the feature patterns of each sampling voltage signal intrinsic mode function, the present application uses a 1D-CNN model as the intrinsic mode feature extractor to process each sampling voltage signal intrinsic mode function respectively, and captures the local dependencies in the sampling voltage signal intrinsic mode function through one-dimensional convolution operations, and excavates the frequency distribution feature patterns of each intrinsic mode function to obtain the set of sampling voltage signal intrinsic mode semantic feature vectors, thereby providing a data basis for subsequent intrinsic mode feature modulation aggregation. It is worth mentioning that the 1D-CNN model can effectively extract local features in the input signal. It can capture the key patterns and features in the function by sliding the convolutional kernel on the sampling voltage signal intrinsic mode function. In addition, through the convolution operation, the 1D-CNN can reduce the dimension of the input signal while retaining important information, effectively reducing the computational complexity of subsequent processing and improving the running efficiency of the model.

[0043] Specifically, in step S42, dynamic aggregation based on the feature energy distribution of the set of sampling voltage signal intrinsic mode semantic feature vectors is performed to obtain the sampling voltage signal global significant aggregation representation vector. It should be understood that in order to effectively suppress the noise and interference in the sampling voltage signal, the present application proposes a dynamic aggregation method based on the feature energy distribution, which can dynamically adjust the contribution degree weights of each sampling voltage signal intrinsic mode semantic feature vector in the aggregation process through feature energy significance evaluation and a specially designed dynamic weight allocation mechanism for each intrinsic mode semantic feature vector, and realizes the strengthening of important features and the suppression of noise components, thereby effectively extracting pure voltage signal features. Among them, Figure 5 It is a flowchart of sub-step S42 of the voltage control method for a high-reliability high-voltage power supply according to an embodiment of the present application. As Figure 5 shown, step S42 includes the steps of: S421, determining the sampling voltage signal intrinsic mode clustering initial center vector based on the feature energy distribution of the set of sampling voltage signal intrinsic mode semantic feature vectors; S422, performing explicit dynamic aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors based on the spatial span of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors relative to the sampling voltage signal intrinsic mode clustering initial center vector to obtain the sampling voltage signal global significant aggregation representation vector.

[0044] More specifically, the step S421 further includes: calculating the static energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode static energy factors; selecting the sampling voltage signal intrinsic mode semantic feature vector corresponding to the maximum value in the set of sampling voltage signal intrinsic mode static energy factors as the initial center vector of the sampling voltage signal intrinsic mode clustering.

[0045] In a specific example of the present application, calculating the static energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode static energy factors includes: calculating the kurtosis of the sampling voltage signal intrinsic mode semantic feature vector and inputting the kurtosis into a sigmoid activation function to obtain the sampling voltage signal intrinsic mode static energy factor. That is, based on the kurtosis of each sampling voltage signal intrinsic mode semantic feature vector, its static energy factor is calculated to evaluate its stability and importance in the feature space, and the sampling voltage signal intrinsic mode semantic feature vector corresponding to the maximum static energy factor is selected as the initial center vector of the sampling voltage signal intrinsic mode clustering to determine the cluster core of the global frequency distribution of the sampling voltage signal, thereby providing a stable reference point for subsequent feature aggregation.

[0046] More specifically, the step S422 further includes: calculating the dynamic aggregation energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors based on the spatial span between each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors and the initial center vector of the sampling voltage signal intrinsic mode clustering, and the set of sampling voltage signal intrinsic mode static energy factors to obtain a set of sampling voltage signal intrinsic mode dynamic aggregation energy factors; performing significant modulation aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors based on the set of sampling voltage signal intrinsic mode dynamic aggregation energy factors to obtain the global significant aggregation representation vector of the sampling voltage signal.

[0047] In a specific example of the present application, calculating the dynamic aggregation energy factor of each sampling voltage signal intrinsic mode semantic feature vector in the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain a set of sampling voltage signal intrinsic mode dynamic aggregation energy factors includes: taking the square value of the number of feature vectors separated between the sampling voltage signal intrinsic mode semantic feature vector and the initial clustering center vector of the sampling voltage signal intrinsic mode as the spatial span coefficient, and calculating the weighted ratio between the product of the static energy factor of the sampling voltage signal intrinsic mode semantic feature vector and the static energy factor of the initial clustering center vector of the sampling voltage signal intrinsic mode and the spatial span coefficient to obtain the sampling voltage signal intrinsic mode dynamic aggregation energy factor.

[0048] In a specific example of the present application, based on the set of sampling voltage signal intrinsic mode dynamic aggregation energy factors, performing significant modulation aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors to obtain the sampling voltage signal global significant aggregation representation vector includes: inputting the set of sampling voltage signal intrinsic mode dynamic aggregation energy factors into a gated mask unit to obtain a set of sampling voltage signal intrinsic mode dynamic aggregation weight factors; calculating the weighted sum of the set of sampling voltage signal intrinsic mode semantic feature vectors based on the set of sampling voltage signal intrinsic mode dynamic aggregation weight factors to obtain the sampling voltage signal global significant aggregation representation vector.

[0049] That is, by considering the feature space proximity of each sampling voltage signal intrinsic mode semantic feature vector to the core of the sampling voltage signal global frequency distribution cluster and the importance difference of the intrinsic attributes, more accurately measure its contribution degree in the sampling voltage signal global features. Then, based on the gated mask mechanism, perform non-linear transformation and screening on the obtained dynamic aggregation energy factors to generate corresponding weights, and use this to perform weighted aggregation on the set of sampling voltage signal intrinsic mode semantic feature vectors, so as to enhance the expression of important features and suppress the influence of irrelevant or noise features, in order to obtain the sampling voltage signal global significant aggregation representation vector.

[0050] Correspondingly, the step S42 includes: processing the set of sampling voltage signal intrinsic mode semantic feature vectors with the following feature dynamic aggregation formula to obtain the sampling voltage signal global significant aggregation representation vector, where the feature dynamic aggregation formula is:

[0051] X = {x1, x2,..., x i ,..., x n}

[0052]

[0053] xc = x m

[0054]

[0055] w si = mask(w i )

[0056]

[0057] Where X represents the set of intrinsic modal semantic feature vectors of the sampled voltage signal, x1, x2, x i , x m and x n respectively represent the first, second, i-th, m-th and n-th sampled voltage signal intrinsic modal semantic feature vectors in the set of intrinsic modal semantic feature vectors of the sampled voltage signal, the value of n is the number of intrinsic modal semantic feature vectors of the sampled voltage signal, represents the eigenvalue at the j-th position in the i-th sampled voltage signal intrinsic modal semantic feature vector, μ i and σ i 4 respectively represent the feature mean and the square of the feature variance of the i-th sampled voltage signal intrinsic modal semantic feature vector, E{·} represents calculating the expected value of the set, sigmoid represents the sigmoid activation function, represents the static energy factor of the i-th sampled voltage signal intrinsic modal semantic feature vector, argmax represents the index corresponding to taking the maximum value, m represents the index of the largest static energy factor in the set of static energy factors of the sampled voltage signal intrinsic modal, x c represents the initial center vector of the sampled voltage signal intrinsic modal clustering, represents the static energy factor of the initial center vector of the sampled voltage signal intrinsic modal clustering, a and b are different weight parameters, Count(x i → x m ) represents the number of feature vectors separated between the i-th sampled voltage signal intrinsic modal semantic feature vector and the initial center vector of the sampled voltage signal intrinsic modal clustering, represents the dynamic aggregation energy factor of the i-th sampled voltage signal intrinsic modal semantic feature vector, w i represents the i-th normalized sampled voltage signal intrinsic modal dynamic aggregation energy factor, mask(·) represents masking processing, θ is a preset mask threshold, w si represents the i-th sampled voltage signal intrinsic modal dynamic aggregation weight factor, V represents the global significant aggregation representation vector of the sampled voltage signal.

[0058] Specifically, in step S43, signal reconstruction is performed based on the globally significant aggregated representation vector of the sampled voltage signal to obtain a pure sampled voltage signal. In a specific example of the present application, step S43 includes: inputting the globally significant aggregated representation vector of the sampled voltage signal into a signal reconstruction module based on a decoder to obtain the pure sampled voltage signal. That is, a decoder structure is used to decode the globally significant aggregated representation vector of the sampled voltage signal and map it back to the original space of the voltage signal to restore the original signal form. In the technical solution of the present application, the decoder includes multiple fully connected layers and activation functions. Through layer-by-layer non-linear transformation, the waveform of the voltage signal can be gradually reconstructed while retaining key information such as the amplitude, frequency, and phase of the signal, thereby obtaining a pure sampled voltage signal.

[0059] In a preferred example of the present application, based on the difference in the source frequency domain distribution of the set of intrinsic mode functions of the sampled voltage signal obtained by CEEMDAN decomposition of the sampled voltage signal, after being extracted by an intrinsic mode feature extractor based on a 1D-CNN model, each sampled voltage signal intrinsic mode semantic feature vector in the set of sampled voltage signal intrinsic mode semantic feature vectors will also have a difference in the encoded feature distribution, resulting in different field dynamic aggregation modes in the feature distribution under the encoded feature distribution difference during feature dynamic aggregation, causing the globally significant aggregated representation vector of the sampled voltage signal to have a complex aggregation space structure. Therefore, when inputting the globally significant aggregated representation vector of the sampled voltage signal into a signal reconstruction module based on a decoder to obtain a pure sampled voltage signal, it is desired to improve the decoding generation convergence and generalization effect of the globally significant aggregated representation vector of the sampled voltage signal under the complex aggregation space structure.

[0060] Based on this, when the global significant aggregation representation vector of the sampled voltage signal is input into the signal reconstruction module based on the decoder, this application considers optimizing the global significant aggregation representation vector of the sampled voltage signal. The optimization process includes: calculating the sum of the absolute values of each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal to obtain the first global significant aggregation spatial structure value of the sampled voltage signal; calculating the square root of the sum of the squares of each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal to obtain the second global significant aggregation spatial structure value of the sampled voltage signal; multiplying each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal by the first global significant aggregation spatial structure value and the second global significant aggregation spatial structure value respectively to obtain the first global significant aggregation structure reference value and the second global significant aggregation structure reference value corresponding to each eigenvalue; multiplying each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal by the length of the global significant aggregation representation vector of the sampled voltage signal and the square root of the length respectively to obtain the first global significant aggregation scale transformation value and the second global significant aggregation scale transformation value corresponding to each eigenvalue; dividing the first global significant aggregation structure reference value by the difference between the first global significant aggregation spatial structure value and the first global significant aggregation scale transformation value to obtain the first global significant aggregation transformation adjustment value; dividing the second global significant aggregation structure reference value by the difference between the second global significant aggregation spatial structure value and the second global significant aggregation scale transformation value to obtain the second global significant aggregation transformation adjustment value; calculating the weighted sum of the first global significant aggregation transformation adjustment value and the second global significant aggregation transformation adjustment value to obtain each eigenvalue of the optimized global significant aggregation representation vector of the sampled voltage signal.

[0061] Correspondingly, the optimization process of the global significant aggregation representation vector of the sampled voltage signal is represented by the following optimization formula:

[0062] v 1i =(α×v i ) / (α - L×v i )

[0063]

[0064]

[0065] v i ∈V∈R 1×L

[0066] v1i ∈V1∈R 1×L

[0067] v 2i ∈V2∈R 1×L

[0068] V' = V1 ⊕ (ω ⊙ V2)

[0069] Where V represents the global significant aggregation representation vector of the sampled voltage signal, v i represents the i-th eigenvalue of the global significant aggregation representation vector of the sampled voltage signal, L represents the length of the global significant aggregation representation vector of the sampled voltage signal, R represents the set of real numbers, α represents the global significant aggregation spatial structure value of the first sampled voltage signal, β represents the global significant aggregation spatial structure value of the second sampled voltage signal, v 1i represents the first global significant aggregation scale transformation value corresponding to the i-th eigenvalue in the global significant aggregation representation vector of the sampled voltage signal, V1 is the first global significant aggregation transformation adjustment vector composed of each of the first global significant aggregation scale transformation values, v 2i represents the second global significant aggregation scale transformation value corresponding to the i-th eigenvalue in the global significant aggregation representation vector of the sampled voltage signal, V2 is the second global significant aggregation transformation adjustment vector composed of each of the second global significant aggregation scale transformation values, ω is the weight hyperparameter, ⊕ represents pointwise addition by position, ⊙ represents pointwise multiplication by position, and V′ represents the optimized global significant aggregation representation vector of the sampled voltage signal.

[0070] That is, for the spatial structure information of the feature set of the global significant aggregation representation vector of the sampled voltage signal in the high-dimensional space, by using the class norm space structured representation of the global significant aggregation representation vector of the sampled voltage signal as a reference window to perform scale-based box transformation on each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal, and implementing box attention weight adjustment based on the spatial structure for each eigenvalue of the global significant aggregation representation vector of the sampled voltage signal, to ensure the spatial transformation invariance of the global significant aggregation representation vector of the sampled voltage signal under feature space interaction, thereby improving the convergence and generalization effects of the decoding generation of the feature set of the global significant aggregation representation vector of the sampled voltage signal under complex spatial structure representation, and improving the signal quality of the pure sampled voltage signal obtained by inputting it into the signal reconstruction module based on the decoder.

[0071] Specifically, in step S44, the pure sampled voltage signal and the external control voltage signal are input into an arithmetic unit to obtain the error signal. In a specific example of the present application, the arithmetic unit performs a differential operation on the pure sampled voltage signal and the external control voltage signal to calculate the difference between the two, thereby obtaining the error signal. The error signal is then sent to a PWM controller, which adjusts the duty cycle of the PWM waveform according to the magnitude and direction of the error signal, and transmits the generated PWM drive signal to the MOS transistor drive circuit to control the switching state of the MOS transistor.

[0072] In the above voltage control method for a high-reliability high-voltage power supply, in step S5, the MOS transistor drive circuit drives the MOS transistor based on the PWM drive signal to control the level of the output DC voltage. It should be understood that as a high-speed switching device, the MOS transistor has a low on-resistance and a fast switching speed, and is very suitable for precisely controlling the output voltage of a high-voltage power supply. Under the control of the PWM drive signal, the MOS transistor drive circuit can adjust the on-time of the MOS transistor according to the pulse width of the PWM drive signal, thereby controlling the magnitude of the output voltage. Since the switching frequency of the MOS transistor can be very high, rapid adjustment of the output voltage can be achieved, real-time response to load changes and external interference, thus ensuring the stability and accuracy of the output voltage and providing strong technical support for the stable operation of the power system.

[0073] In summary, the voltage control method for a high-reliability high-voltage power supply based on the embodiments of the present application is elucidated. By performing input filtering and power conversion boost processing on the DC voltage signal provided by the power supply to obtain a high-voltage AC signal, then, after rectifying, filtering, and sampling the high-voltage AC signal, a signal processing technique based on deep learning is used to perform CEEMDAN decomposition and intrinsic mode feature extraction on the sampled voltage signal, and by dynamically aggregating the feature saliency modulation of each intrinsic mode after the decomposition of the sampled voltage signal, the noise and interference in the signal are suppressed to obtain a pure sampled voltage signal, and then combined with the external control voltage signal for error calculation and MOS transistor drive control to achieve precise adjustment of the output voltage of the high-voltage power supply. In this way, the accuracy of voltage control can be effectively improved, the adaptability to complex and variable load conditions and interference environments can be enhanced, and thus the stability and reliability of the high-voltage power supply system are significantly improved.

[0074] The basic principles of the present invention have been described in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present invention are only examples and not limitations. It cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and easy understanding, rather than limitations. The above details do not limit the present invention to necessarily adopt the above specific details for implementation.

[0075] In the above embodiments, the descriptions of each embodiment have their own emphases. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed method can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the unit division is only a logical function division, and there can be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0076] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the claims in the present invention. Any associated drawing marks in the claims should not be regarded as limiting the claimed rights.

[0077] In addition, obviously, the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0078] Finally, it should be noted that the above description has been given for the purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A voltage control method for a high-reliability high-voltage power supply, characterized in that: include: Processing an initial DC voltage signal provided by the power supply through an input filter circuit electrically connected to the power supply to obtain a DC voltage signal; Performing power conversion and voltage boosting on the DC voltage signal to obtain a high-voltage AC signal; The high-voltage AC signal is processed by a rectifier, filter and sampling circuit to output a DC voltage, and the DC voltage is sampled to obtain a sampled voltage signal, and the sampled voltage signal is fed back to the control circuit; The control circuit generates an error signal based on the sampled voltage signal and the external control voltage signal, and generates a PWM drive signal based on the error signal, and the PWM drive signal is transmitted to the MOS transistor drive circuit; The MOS transistor driving circuit drives the MOS transistor based on the PWM driving signal to control the level of the output DC voltage; The control circuit generates an error signal based on the sampled voltage signal and the external control voltage signal, comprising: performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal; performing dynamic aggregation based on feature energy distribution on the set of intrinsic modal semantic feature vectors of the sampled voltage signal to obtain a global significant aggregation representation vector of the sampled voltage signal; performing signal reconstruction based on the global significant aggregation representation vector of the sampled voltage signal to obtain a pure sampled voltage signal; and inputting the pure sampled voltage signal and the external control voltage signal into an operator to obtain the error signal. The method comprises the following steps: performing dynamic aggregation based on characteristic energy distribution on the set of intrinsic modal semantic feature vectors of the sampled voltage signals to obtain a globally significant aggregated representation vector of the sampled voltage signals, including: determining an initial center vector of the intrinsic modal clustering of the sampled voltage signals based on the characteristic energy distribution of the set of intrinsic modal semantic feature vectors of the sampled voltage signals; performing explicit dynamic aggregation on the set of intrinsic modal semantic feature vectors of the sampled voltage signals based on the spatial span of each intrinsic modal semantic feature vector of the sampled voltage signals relative to the initial center vector of the intrinsic modal clustering of the sampled voltage signals to obtain a globally significant aggregated representation vector of the sampled voltage signals; and performing dynamic aggregation based on the spatial span of each intrinsic modal semantic feature vector of the sampled voltage signals in the set of intrinsic modal semantic feature vectors of the sampled voltage signals relative to the initial center vector of the intrinsic modal clustering of the sampled voltage signals. Among them, signal reconstruction is performed based on the global significant aggregation representation vector of the sampled voltage signal to obtain a pure sampled voltage signal, including: inputting the global significant aggregation representation vector of the sampled voltage signal into a decoder-based signal reconstruction module to obtain the pure sampled voltage signal.

2. The voltage control method for a high-reliability high-voltage power supply according to claim 1, wherein: Performing signal feature extraction based on modal decomposition on the sampled voltage signal to obtain a set of intrinsic modal semantic feature vectors of the sampled voltage signal includes: Performing CEEMDAN decomposition on the sampled voltage signal to obtain a set of intrinsic mode functions of the sampled voltage signal; Each of the sampled voltage signal intrinsic modal functions in the set of the sampled voltage signal intrinsic modal functions is input into an intrinsic modal feature extractor based on a 1D-CNN model to obtain a set of the sampled voltage signal intrinsic modal semantic feature vectors.

3. The voltage control method for a high-reliability high-voltage power supply according to claim 2, wherein: Determining an initial center vector of the intrinsic modal clustering of the sampled voltage signal based on a characteristic energy distribution of a set of intrinsic modal semantic feature vectors of the sampled voltage signal includes: Calculating the static energy factor of each sampled voltage signal intrinsic modal semantic feature vector in the set of sampled voltage signal intrinsic modal semantic feature vectors to obtain a set of sampled voltage signal intrinsic modal static energy factors; The sampled voltage signal intrinsic modal semantic feature vector corresponding to the maximum value in the set of the sampled voltage signal intrinsic modal static energy factors is selected as the sampled voltage signal intrinsic modal clustering initial center vector.

4. The voltage control method for a high-reliability high-voltage power supply according to claim 3, wherein: Calculating the static energy factor of each sampled voltage signal intrinsic modal semantic feature vector in the set of sampled voltage signal intrinsic modal semantic feature vectors to obtain a set of sampled voltage signal intrinsic modal static energy factors includes: The kurtosis of the intrinsic modal semantic feature vector of the sampled voltage signal is calculated, and the kurtosis is input into a si gmoid activation function to obtain the intrinsic modal static energy factor of the sampled voltage signal.

5. The voltage control method for a high-reliability high-voltage power supply according to claim 4, wherein: Based on the spatial span of each sampled voltage signal intrinsic modal semantic feature vector in the set of the sampled voltage signal intrinsic modal semantic feature vectors relative to the initial center vector of the sampled voltage signal intrinsic modal cluster, the set of the sampled voltage signal intrinsic modal semantic feature vectors is explicitly dynamically aggregated to obtain a globally significant aggregated representation vector of the sampled voltage signal, including: Based on the spatial span between each sampled voltage signal intrinsic modal semantic feature vector in the set of sampled voltage signal intrinsic modal semantic feature vectors and the initial center vector of the sampled voltage signal intrinsic modal clustering, and the set of sampled voltage signal intrinsic modal static energy factors, calculating the dynamic aggregation energy factor of each sampled voltage signal intrinsic modal semantic feature vector in the set of sampled voltage signal intrinsic modal semantic feature vectors to obtain a set of sampled voltage signal intrinsic modal dynamic aggregation energy factors; Based on the set of dynamic aggregation energy factors of the sampled voltage signal intrinsic modalities, a saliency modulation aggregation is performed on the set of semantic feature vectors of the sampled voltage signal intrinsic modalities to obtain a global saliency aggregation representation vector of the sampled voltage signal.

6. The voltage control method for a high-reliability high-voltage power supply according to claim 5, characterized in that: Calculating the dynamic aggregation energy factor of each sampled voltage signal intrinsic modal semantic feature vector in the set of sampled voltage signal intrinsic modal semantic feature vectors to obtain a set of sampled voltage signal intrinsic modal dynamic aggregation energy factors includes: The square value of the number of eigenvectors between the intrinsic modal semantic feature vector of the sampled voltage signal and the initial center vector of the intrinsic modal clustering of the sampled voltage signal is used as the spatial span coefficient, and a weighted ratio between the product of the static energy factor of the intrinsic modal semantic feature vector of the sampled voltage signal and the static energy factor of the initial center vector of the intrinsic modal clustering of the sampled voltage signal and the spatial span coefficient is calculated to obtain the intrinsic modal dynamic aggregation energy factor of the sampled voltage signal.

7. The voltage control method for a high-reliability high-voltage power supply according to claim 6, wherein: Based on the set of dynamic aggregation energy factors of the inherent modalities of the sampled voltage signals, performing saliency modulation aggregation on the set of inherent modal semantic feature vectors of the sampled voltage signals to obtain a global saliency aggregation representation vector of the sampled voltage signals, including: Inputting the set of the sampled voltage signal intrinsic modal dynamic aggregation energy factors into a gated mask unit to obtain a set of sampled voltage signal intrinsic modal dynamic aggregation weight factors; A weighted sum of a set of intrinsic modal semantic feature vectors of the sampled voltage signals is calculated based on a set of intrinsic modal dynamic aggregation weight factors of the sampled voltage signals to obtain a global salient aggregation representation vector of the sampled voltage signals.

Images