Artificial Intelligence-Based Power Switching Control Method and System

Abstract

This application provides an artificial intelligence-based power supply switching control method and system. It uses an AI network model to comprehensively evaluate multiple state parameters of the mains power input, generating mains power health parameters. Real-time monitoring and analysis of the battery pack's operating state parameters generate battery health parameters. Based on the mains power health parameters and the battery health parameters, a power supply switching decision is made, generating a switching enable signal. According to the switching enable signal, the output characteristics of the backup power supply channel are dynamically synchronized with the output characteristics of the main power supply channel. The backup power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the mains power input. When the output characteristics meet preset synchronization conditions, the load power supply path is switched from the main power supply channel to the backup power supply channel, thereby reducing the latency and power supply fluctuations of the communication power supply system's power supply control switching and improving the accuracy of power supply control switching.

Description

Technical Field

[0001] This application relates to the field of communication power supply technology, and more specifically, to a power supply switching control method and system based on artificial intelligence. Background Technology

[0002] Critical communication load scenarios such as communication base stations, server rooms, and data centers have extremely high requirements for power supply continuity. In current communication power systems, AC input power (such as AC-DC rectifier modules) and battery packs (such as DC direct supply) are typically switched via relays, Schottky diodes, or comparators. However, this current power supply control switching suffers from problems such as high switching delay, low accuracy, and large fluctuations.

[0003] Therefore, how to reduce the delay and power supply fluctuation of power supply control switching in communication power supply systems and improve the accuracy of power supply control switching are urgent problems to be solved. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a power supply switching control method and system based on artificial intelligence, so as to reduce the delay and power supply fluctuation of power supply control switching in communication power supply systems and improve the accuracy of power supply control switching.

[0005] In a first aspect, this application provides a power supply switching control method based on artificial intelligence, comprising:

[0006] The system uses an artificial intelligence network model to comprehensively evaluate and process multiple status parameters of the mains power input, generating mains power health parameters.

[0007] Real-time monitoring and analysis of the battery pack's operating status parameters to generate battery health parameters;

[0008] Based on the mains power health parameters and the battery health parameters, a power supply switching decision is made, and a switching enable signal is generated.

[0009] The output characteristics of the backup power supply channel and the main power supply channel are dynamically synchronized according to the switching enable signal. The backup power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the AC power input.

[0010] When the output characteristics meet the preset synchronization conditions, the load power supply path is switched from the main power supply channel to the backup power supply channel.

[0011] Optionally, the step of comprehensively evaluating and processing multiple status parameters of the mains power input through an artificial intelligence network model to generate mains power health parameters includes:

[0012] Collect the multiple state parameters of the mains power input, including at least two of the following: voltage parameter, frequency parameter, voltage change rate parameter, load current parameter, and ambient temperature parameter;

[0013] The multiple state parameters are constructed into a multi-dimensional feature vector. The multi-dimensional feature vector is then processed by a combination model of convolutional neural network and long short-term memory network to extract time-series features and predict fault probability, thereby generating mains power fault probability parameters.

[0014] The mains power health parameter is calculated based on the deviation of the mains power failure probability parameter and the voltage parameter. The mains power health parameter is negatively correlated with the deviation of both the mains power failure probability parameter and the voltage parameter.

[0015] Optionally, the real-time monitoring and analysis of the battery pack's operating status parameters to generate battery health parameters includes:

[0016] The operating status parameters of the battery pack are collected, including battery voltage parameters, battery current parameters, battery internal resistance parameters, and battery temperature parameters.

[0017] The extended Kalman filter algorithm is used to perform state estimation processing on the operating state parameters to generate battery state of charge parameters and battery health state parameters.

[0018] The maximum available power parameters of the battery are calculated based on the battery state-of-charge parameters, the battery internal resistance parameters, and the battery voltage parameters.

[0019] The battery health parameters are generated based on the battery state of charge parameters, the battery voltage parameters, the battery maximum available power parameters, and the battery health status parameters.

[0020] Optionally, the step of performing a power supply switching decision based on the mains power health parameters and the battery health parameters, and generating a switching enable signal, includes:

[0021] The mains power health parameters are compared with preset mains power health thresholds, and the battery health parameters are compared with preset battery health thresholds.

[0022] When the mains power health parameter is lower than the preset mains power health threshold and the battery health parameter is higher than the preset battery health threshold, the switching enable signal is generated. The switching enable signal is used to indicate the start of the synchronous control process of the backup power supply channel.

[0023] Optionally, the step of dynamically synchronizing the output characteristics of the backup power supply channel with the output characteristics of the main power supply channel based on the switching enable signal includes:

[0024] Upon receiving the switching enable signal, the soft-start control process of the backup power supply channel is initiated, and the output voltage parameters of the backup power supply channel are adjusted.

[0025] The output voltage parameters and output voltage change rate parameters of the main power supply channel are collected in real time and used as the output characteristics of the main channel.

[0026] The output voltage parameters of the backup power supply channel are controlled to track the output characteristics of the main channel by a trajectory fitting algorithm, so that the difference between the output voltage parameters of the backup power supply channel and the output voltage parameters of the main power supply channel meets a preset synchronization threshold, and the difference between the output voltage change rate parameters of the backup power supply channel and the output voltage change rate parameters of the main power supply channel meets a preset change rate threshold.

[0027] Optionally, the step of controlling the output voltage parameters of the backup power supply channel to track the output characteristics of the main channel through a trajectory fitting algorithm, so that the difference between the output voltage parameters of the backup power supply channel and the output voltage parameters of the main power supply channel meets a preset synchronization threshold, and the difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel meets a preset change rate threshold, includes:

[0028] Using the output voltage parameters and the output voltage change rate parameters of the main power supply channel as reference trajectories, an objective function for the output trajectory of the backup power supply channel is constructed.

[0029] Based on the output trajectory objective function, the duty cycle parameter of the power conversion circuit of the backup power supply channel is adjusted by a proportional-integral control algorithm.

[0030] Obtain a first difference between the output voltage parameter of the backup power supply channel and the output voltage parameter of the main power supply channel, and a second difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel;

[0031] The proportional coefficient and integral coefficient of the proportional-integral control algorithm are dynamically adjusted based on the first difference and the second difference until the first difference satisfies the preset synchronization threshold and the second difference satisfies the preset rate of change threshold.

[0032] Optionally, switching the load power supply path from the main power supply channel to the backup power supply channel includes:

[0033] A synchronization control signal is generated. The rising edge of the synchronization control signal triggers the turn-on operation of the power switching device of the backup power supply channel, and the falling edge triggers the turn-off operation of the power switching device of the main power supply channel.

[0034] The synchronous control signal is converted into a gate drive signal for the power switching device by a drive circuit. The voltage amplitude and pulse width of the gate drive signal meet the turn-on and turn-off characteristics of the power switching device.

[0035] During the turn-off process of the power switching device in the main power supply channel, its drain-source voltage parameters are monitored to ensure that the voltage spike during the turn-off process does not exceed the device's withstand voltage value.

[0036] During the turn-on process of the power switching device in the backup power supply channel, its on-current parameters are monitored to ensure that the current rise rate does not exceed the maximum allowable current change rate of the device.

[0037] When the main power supply channel power switch is completely turned off and the backup power supply channel power switch is completely turned on, the voltage transient during the switching process is suppressed by the output filter circuit to ensure that the load voltage fluctuation meets the preset stability threshold.

[0038] Optionally, the method further includes:

[0039] After the power supply channel switch is completed, the mains power health parameters are continuously monitored;

[0040] When the mains power health parameter recovers to a level higher than the preset mains power health threshold and the duration reaches the preset stable time, an inverse switching enable signal is generated;

[0041] The output characteristics of the main power supply channel and the backup power supply channel are dynamically synchronized according to the reverse switching enable signal, and a reverse switching operation is performed when the preset synchronization conditions are met to switch the load power supply path from the backup power supply channel back to the main power supply channel.

[0042] Secondly, this application provides an artificial intelligence-based power switching control system, which includes a machine-readable storage medium and a processor. The machine-readable storage medium stores machine-executable instructions. When the processor executes the machine-executable instructions, the artificial intelligence-based power switching control system implements the aforementioned artificial intelligence-based power switching control method.

[0043] Thirdly, this application provides an electrical device that uses an artificial intelligence-based power switching control system as described in the second aspect for power control.

[0044] The AI-based power switching control method and system provided in this application comprehensively evaluates and processes multiple state parameters of the mains power input through an AI network model to generate mains power health parameters. Real-time monitoring and analysis of the battery pack's operating state parameters generate battery health parameters. Based on the mains power health parameters and the battery health parameters, a power switching decision is made, generating a switching enable signal. The output characteristics of the backup power supply channel are dynamically synchronized with the output characteristics of the main power supply channel according to the switching enable signal. The backup power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the mains power input. When the output characteristics meet preset synchronization conditions, the load power supply path is switched from the main power supply channel to the backup power supply channel. This ensures that the load is almost unaffected during switching through a highly consistent output voltage trajectory and extremely short switching time, avoiding equipment restarts, bit errors, or link breaks caused by voltage drops and transient fluctuations in traditional switching methods. This significantly reduces the power control switching delay and power fluctuations in the communication power system. Simultaneously, with the accurate prediction and real-time analysis of the AI ​​network model, the mains power condition can be predicted in advance, and optimal switching decisions can be made, further improving the accuracy and reliability of power control switching. Attached Figure Description

[0045] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0046] Figure 1 A flowchart illustrating an artificial intelligence-based power switching control method provided in an embodiment of this application;

[0047] Figure 2 This is a schematic diagram of a power supply switching control system based on artificial intelligence, provided as an embodiment of this application.

[0048] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0049] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0051] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0052] Figure 1 This is a flowchart illustrating an artificial intelligence-based power switching control method provided in an embodiment of this application. It should be understood that in other embodiments, the order of some steps in the artificial intelligence-based power switching control method of this embodiment can be shared according to actual needs, or some steps can be omitted or maintained. Figure 1 As shown, the method may include the following steps:

[0053] S110. The mains power input is comprehensively evaluated and processed through an artificial intelligence network model to generate mains power health parameters.

[0054] This step is a prerequisite for achieving seamless power supply switching. Its core lies in using artificial intelligence technology to accurately assess the mains power input status, providing a reliable basis for subsequent switching decisions. In critical scenarios such as communication base stations and server rooms, the stability of the mains power supply directly affects the operation of communication equipment. Therefore, it is necessary to predict the risk of mains power failure in advance through multi-dimensional parameter collection and intelligent analysis.

[0055] S111. Collect the plurality of state parameters of the mains power input, wherein the plurality of state parameters include at least two of the following: voltage parameter, frequency parameter, voltage change rate parameter, load current parameter, and ambient temperature parameter.

[0056] In this step, the mains power detection and prediction module is equipped with an analog-to-digital converter (ADC), whose sampling frequency can be set to, for example, 1MHz, to achieve high-frequency acquisition of mains power status parameters and ensure that instantaneous voltage fluctuations or frequency shifts are captured.

[0057] Specifically, the voltage parameter V_AC(t) is acquired by a voltage sensor connected in parallel to the mains input circuit; the frequency parameter f_AC(t) is extracted from the voltage signal through a phase-locked loop circuit, accurately reflecting minute changes in the mains frequency; the voltage change rate parameter dV_AC / dt is obtained by differential calculation of the voltage values ​​at two consecutive sampling times, i.e., dV_AC / dt=(V_AC(t)-V_AC(t-Δt)) / Δt, where Δt is the sampling interval (1μs), and the unit is V / s; the load current parameter I_load(t) is acquired by a Hall current sensor connected in series in the mains power supply circuit, used to reflect the current load power consumption; and the ambient temperature parameter T_cmv(t) is acquired by a temperature sensor installed on the casing of the mains rectifier module, used to assess the impact of the environment on the mains power supply equipment. These parameters together constitute a multi-dimensional feature reflecting the quality of the mains power supply, providing a comprehensive data foundation for subsequent health assessments.

[0058] S112. Construct the multiple state parameters into a multi-dimensional feature vector, and perform temporal feature extraction and fault probability prediction processing on the multi-dimensional feature vector through a combination model of convolutional neural network and long short-term memory network to generate mains power fault probability parameters.

[0059] In this step, the construction of the multidimensional feature vector x(t) adopts the time window method, that is, the state parameters of 100 consecutive sampling times are selected to form a 5×100 matrix (5 parameters, 100 time series data for each parameter), which is used as the input of the AI ​​network model.

[0060] The Convolutional Neural Network (Conv1D) consists of two convolutional layers: the first layer has 32 kernels, a kernel size of 5 (i.e., it processes 5 consecutive time-series data points at a time), a stride of 1, and uses the ReLU activation function to extract local time-series features, such as short-term voltage fluctuation patterns; the second layer has 64 kernels, a kernel size of 3, a stride of 1, and also uses the ReLU activation function to further extract more abstract local features.

[0061] The feature map output by the convolutional layer is converted into a one-dimensional vector through a flattened layer and then input into a Long Short-Term Memory (LSTM) network.

[0062] The LSTM part consists of two bidirectional LSTM layers, each with 128 hidden units. The first LSTM layer captures temporal dependencies from the past to the present in the forward propagation direction and captures potential trends from the present to the future in the backward propagation direction. The outputs of the two LSTM layers are concatenated to form a 256-dimensional feature vector. Finally, a single value is output through a fully connected layer and then activated by a sigmoid activation function. The process yields the mains power failure probability parameter P_fault(t+Δt) within the future time interval Δt (Δt is set to 50ms, i.e., predicting the failure probability within the next 50ms). The Sigmoid function is expressed as σ(x)=1 / (1+e^(-x)), which maps the output value to the range of 0-1, facilitating probability representation; W_o is the weight matrix of the fully connected layer (dimension 256×1), and b_o is the bias term (scalar).

[0063] These parameters are obtained through optimization using a training dataset, which may include, for example, normal operation and fault data of the mains power over the past three months, to ensure that the model has high prediction accuracy.

[0064] S113. The mains power health parameter is calculated based on the deviation of the mains power failure probability parameter and the voltage parameter. The mains power health parameter is negatively correlated with the deviation of both the mains power failure probability parameter and the voltage parameter.

[0065] In this step, the formula for calculating the mains power health parameter H_AC(t) aims to comprehensively consider the voltage stability and fault risk of the mains power. Its expression is H_AC(t)=1-α |(V_AC(t)-V_nom) / V_nom| β P_fault(t+Δt).

[0066] Where V_nom is the nominal mains voltage, which is 380V (effective line voltage) for three-phase mains; |(V_AC(t)-V_nom) / V_nom| is the relative voltage deviation, reflecting the degree to which the voltage deviates from the nominal value. The larger the value, the more unstable the voltage; α and β are weighting coefficients, both of which are dimensionless parameters. For example, they can be obtained through experimental calibration. For example, α can be set to 0.8 to adjust the weight of the voltage deviation on the health status, and β can be set to 0.5 to adjust the weight of the failure probability.

[0067] For example, assuming the mains voltage is 370V (relative deviation of approximately 2.63%) and the fault probability is 0.3, H_AC(t) = 1 - 0.8 × 0.0263 × 0.5 × 0.3 ≈ 1 - 0.0032 = 0.9968, indicating that the mains voltage is in good health. When the voltage drops to 300V (relative deviation of approximately 21.05%) and the fault probability rises to 0.9, H_AC(t) = 1 - 0.8 × 0.2105 × 0.5 × 0.9 ≈ 1 - 0.0758 = 0.9242, indicating a significant decrease in health. This calculation method allows H_AC(t) to intuitively reflect the overall power supply quality of the mains voltage, and it is negatively correlated with both voltage deviation and fault probability, which is consistent with the actual power supply reliability assessment logic.

[0068] S120: Real-time monitoring and analysis of the battery pack's operating status parameters to generate battery health parameters.

[0069] This step utilizes a Battery Management System (BMS) to comprehensively monitor and evaluate the battery pack's status, ensuring that the battery pack can operate as a reliable backup power source in the event of a mains power failure. This module employs a distributed architecture, with each individual battery cell equipped with voltage and temperature data acquisition points, and a current sensor installed in the battery pack's main circuit to achieve precise sensing of the battery's operating status.

[0070] S121. Collect the operating status parameters of the battery pack, including battery voltage parameters, battery current parameters, battery internal resistance parameters, and battery temperature parameters.

[0071] In this step, the battery voltage parameter V_bat(t) includes the individual cell voltage and the total voltage. The individual cell voltage is collected through a voltage divider resistor network connected in series across each cell, and the total voltage is the sum of all individual cell voltages. The battery current parameter I_bat(t) is collected through a shunt connected in series in the battery pack's charging and discharging circuit, accurately capturing current changes during charging and discharging. The battery internal resistance parameter R_int(t) is measured using the AC injection method, i.e., a small-signal AC current (amplitude 100mA) of 1kHz is injected into the battery pack, and the corresponding AC voltage is measured simultaneously. The internal resistance is then calculated using an impedance analyzer. The battery temperature parameter T_bat(t) is collected through an NTC thermistor attached to the surface of each cell, used to evaluate the impact of battery operating temperature on performance. All these parameters are sampled at a frequency of 1kHz to ensure timely reflection of dynamic changes in battery status.

[0072] S122. The operating state parameters are estimated using the extended Kalman filter algorithm to generate battery state of charge parameters and battery health state parameters.

[0073] In this step, the Extended Kalman Filter (EKF) algorithm is used to solve the nonlinear problem in battery state estimation. Its core includes the construction of the state equation and observation equation, as well as the iteration of the filtering process. The state equation is SOC(k+1) = SOC(k) - (η_dis) I_bat(k) Δt) / C_nom+w_k, where SOC(k) is the state of charge at time k (dimensionless, 0 represents fully discharged, 1 represents fully charged), η_dis is the discharge efficiency (for lithium-ion batteries, the discharge efficiency is usually 0.95-0.98, for example, 0.97 can be taken here), I_bat(k) is the battery current at time k (positive when discharging, unit: A), Δt is the sampling time interval (1s), C_nom is the nominal capacity of the battery (unit: Ah, such as 100Ah), and w_k is the process noise (following a Gaussian distribution with mean of 0 and variance of 1e-6).

[0074] The observation equation is V_bat(k) = f_OCV(SOC(k)) - I_bat(k) R_int+v_k.

[0075] Where f_OCV(SOC(k)) is the function relationship between open-circuit voltage and SOC, which is obtained through experimental testing (for example, OCV=4.2V when SOC=1, OCV=3.7V when SOC=0.5, and OCV=2.5V when SOC=0; intermediate values ​​are obtained through interpolation), R_int is the battery internal resistance (unit: Ω), and v_k is the measurement noise (following a Gaussian distribution with a mean of 0 and a variance of 1e-4).

[0076] The extended Kalman filter algorithm continuously corrects the estimated value of SOC through a prediction-update iterative process, causing it to converge to the true value.

[0077] The battery health status parameter SOH is calculated by the ratio of the current actual capacity to the nominal capacity. The actual capacity is obtained by charge-discharge cycle test, that is, from SOC=1 to SOC=0.2, the cumulative charge during the discharge process is recorded and divided by (1-0.2) to obtain the actual capacity. SOH = actual capacity / C_nom (dimensionless, between 0 and 1).

[0078] S123. The maximum available power parameter of the battery is calculated based on the battery state-of-charge parameter, the battery internal resistance parameter, and the battery voltage parameter.

[0079] In this step, the maximum power parameter P_bat,max(t) of the battery reflects the maximum power that the battery can output in the current state, and its calculation formula is P_bat,max(t) = V_bat(t)² / (4 R_int(t)).

[0080] The derivation of this formula is based on the equivalent circuit model of the battery (such as the Thévenin model), that is, the battery can be regarded as an ideal voltage source in series with the internal resistance. When the external load resistance equals the internal resistance, the output power reaches its maximum value, which is V² / (4R). For example, when the battery voltage V_bat(t) = 24V and the internal resistance R_int(t) = 0.1Ω, P_bat,max(t) = 24² / (4×0.1) = 576 / 0.4 = 1440W, indicating that the battery pack can currently provide a maximum power of 1440W. This parameter is crucial for determining whether the battery can handle the current load. If P_bat,max(t) is greater than the current load power P_load(t), then the battery has the load-carrying capacity.

[0081] S124. Generate the battery health parameters based on the battery state of charge parameters, the battery voltage parameters, the battery maximum available power parameters, and the battery health status parameters.

[0082] In this step, the battery health parameter H_bat(t) is a comprehensive indicator that integrates multiple parameters through a weighted summation. Its calculation formula can be, for example, as follows:

[0083] H_bat(t)=0.3×SOC+0.2×(V_bat(t) / V_nom_bat)+0.3×(P_bat, max(t) / P_nom_bat)+0.2×SOH.

[0084] Where V_nom_bat is the battery's nominal voltage (e.g., 24V) and P_nom_bat is the battery's nominal power (e.g., 2000W). The weights of each item are determined using the analytic hierarchy process (AHP). The weights of SOC and maximum available power are relatively high because they directly reflect the battery's load-carrying capacity and remaining capacity.

[0085] For example, when SOC=0.8, V_bat(t)=23.5V (relative value 23.5 / 24≈0.979), P_bat,max(t)=1500W (relative value 1500 / 2000=0.75), and SOH=0.9, H_bat(t)=0.3×0.8+0.2×0.979+0.3×0.75+0.2×0.9=0.24+0.1958+0.225+0.18=0. 8408 indicates that the battery is in good health. If the SOC drops to 0.3, V_bat(t) = 21V (relative value 0.875), P_bat,max(t) = 800W (relative value 0.4), and SOH = 0.6, then H_bat(t) = 0.3×0.3 + 0.2×0.875 + 0.3×0.4 + 0.2×0.6 = 0.09 + 0.175 + 0.12 + 0.12 = 0.505, indicating a significant decline in battery health.

[0086] S130. Based on the mains power health parameters and the battery health parameters, perform a power supply switching decision and generate a switching enable signal.

[0087] This step is the core of the power supply switching decision. By comparing the health status of the mains power and the battery, it determines whether the backup power supply channel needs to be activated. Its decision logic must take into account both power supply reliability and battery life to avoid unnecessary switching operations.

[0088] S131. Compare the mains power health parameters with the preset mains power health threshold, and compare the battery health parameters with the preset battery health threshold.

[0089] In this step, the preset mains power health threshold H_th is set according to the communication equipment's tolerance for power outages. For critical equipment such as 5G base stations, H_th is set to 0.7, meaning that when the mains power health parameter is below 0.7, the mains power is considered to have a high risk of failure. The preset battery health threshold H_bat has a minimum setting of 0.6 to ensure that the battery has sufficient load-carrying capacity and remaining lifespan when put into operation. The comparison process is executed by the comparator module in the MCU synchronous controller. This module performs parameter comparisons every 10μs to ensure real-time decision-making.

[0090] S132. When the mains power health parameter is lower than the preset mains power health threshold and the battery health parameter is higher than the preset battery health threshold, the switching enable signal is generated. The switching enable signal is used to indicate the start of the synchronous control process of the backup power supply channel.

[0091] In this step, the switching enable signal can be, for example, a high-level pulse signal (such as 3.3V), with a pulse width of, for example, 100μs, which is then transmitted to the control circuit of the backup power supply channel after being output.

[0092] The generation logic of this signal is as follows: when the state where H_AC(t) < H_th and H_bat(t) > H_bat,min lasts for 5 consecutive sampling periods (i.e., 50 μs), a switching enable signal is generated to avoid mis-switching caused by instantaneous interference.

[0093] For example, when the mains power health parameter drops to 0.65 (lower than H_th = 0.7), the battery health parameter is 0.7 (higher than H_bat,min = 0.6), and after this state lasts for 50 μs, a switching enable signal is output to trigger the synchronization control process of the standby channel.

[0094] S140. Control the output characteristics of the standby power supply channel to achieve dynamic synchronization with the output characteristics of the main power supply channel according to the switching enable signal.

[0095] Among them, the standby power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the mains power input.

[0096] This step is the key to achieving seamless switching. By precisely controlling the output voltage and change rate of the standby power supply channel to make it consistent with the main power supply channel, voltage transients and current shocks are avoided during switching, ensuring stable power supply at the load end.

[0097] S141. After receiving the switching enable signal, start the soft start control process of the standby power supply channel and adjust the output voltage parameters of the standby power supply channel.

[0098] In this step, the standby power supply channel can adopt a synchronous Buck topology structure, and its soft start control is implemented by a PWM controller.

[0099] After receiving the switching enable signal, the reference voltage of the PWM controller starts from 0 V and linearly rises to the target voltage (the output voltage of the main power supply channel, such as 24 V) at a preset slope. The slope can be set to 1 V / ms, for example, to avoid voltage overshoot during startup.

[0100] Specifically, the soft start time constant τ_ss = L / R_eq, where L is the inductor in the Buck circuit (which can be taken as 100 μH, for example), and R_eq is the equivalent load resistance (calculated according to the current load current. For example, when the load current is 10 A, R_eq = 24 V / 10 A = 2.4 Ω), then τ_ss = 100 μH / 2.4 Ω ≈ 41.67 μs, ensuring a smooth soft start process.

[0101] Meanwhile, the PWM controller can adjust the output voltage by adjusting the duty cycle D(t). The formula for calculating the duty cycle is D(t) = (V_out(t) + V_MOS) / V_bat(t), where V_out(t) is the output voltage of the standby channel (e.g., 24V), V_MOS is the on-state voltage drop of the MOSFET (e.g., 0.1V), and V_bat(t) is the battery pack voltage (e.g., 25V). The initial duty cycle D(t) = (0 + 0.1) / 25 = 0.004. As V_out(t) increases, the duty cycle gradually increases to (24 + 0.1) / 25 = 0.964, achieving precise adjustment of the output voltage.

[0102] S142. Real-time acquisition of the output voltage parameters and output voltage change rate parameters of the main power supply channel, as the main channel output characteristics.

[0103] In this step, the output voltage parameter V_bus(t) of the main power supply channel is acquired by a voltage sensor installed on the output bus. The sensor uses a differential amplifier circuit to ensure that minute voltage fluctuations are captured. The output voltage change rate parameter dV_bus / dt is obtained by numerically differentiating V_bus(t), such as dV_bus / dt=(V_bus(t)-V_bus(t-Δt)) / Δt.

[0104] Where Δt = 1μs, the unit is V / s. This parameter reflects the rate of voltage change and is an important indicator of synchronous control. For example, when the main channel output voltage rises from 24V to 24.01V in 10μs, then dV_bus / dt = (0.01V) / 10μs = 1000V / s.

[0105] S143. The output characteristics of the backup power supply channel are dynamically synchronized with the output characteristics of the main channel by using a trajectory fitting algorithm.

[0106] S1431. Using the output voltage parameters and the output voltage change rate parameters of the main power supply channel as reference trajectories, construct the objective function of the output trajectory of the backup power supply channel.

[0107] In this step, the objective function for output trajectory aims to minimize the output difference between the backup channel and the main channel. For example, it can be defined as:

[0108] J=k1×(V_bus(t)-V_bat, out(t))²+k2×(dV_bus / dt-dV_bat, out / dt)².

[0109] Where k1 and k2 are weighting coefficients set according to actual needs, adjusting the proportions of voltage deviation and rate of change deviation in the objective function, respectively; V_bat,out(t) is the output voltage of the backup channel; and dV_bat,out / dt is the output voltage rate of change of the backup channel. The physical meaning of this objective function is to make the output voltage trajectory of the backup channel as close as possible to the main channel, while ensuring a consistent rate of change, thus laying the foundation for seamless switching.

[0110] S1432. Based on the output trajectory objective function, the duty cycle parameter of the power conversion circuit of the backup power supply channel is adjusted by a proportional-integral control algorithm.

[0111] In this step, a proportional-integral (PI) controller is used to minimize the objective function J. Its inputs are the voltage deviation e_V = V_bus(t) - V_bat,out(t) and the rate of change deviation e_dV = dV_bus / dt - dV_bat,out / dt. The output is the duty cycle adjustment ΔD. The transfer function of the PI controller is G(s) = K_p + Ki / s, where the proportional coefficient K_p and integral coefficient Ki can be tuned, for example, using the Ziegler-Nichols method. For example, the PI parameters for the voltage deviation can be K_p = 0.1 and Ki = 100, and the PI parameters for the rate of change deviation can be K_p = 0.001 and Ki = 1, ensuring a fast response and small overshoot.

[0112] For example, when e_V=10mV, the proportional term output is 0.1×10mV=1mV, and the integral term is adjusted according to the cumulative amount of deviation, so that the duty cycle gradually increases, thereby reducing the voltage deviation.

[0113] S1433. Obtain the first difference between the output voltage parameter of the backup power supply channel and the output voltage parameter of the main power supply channel, and the second difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel.

[0114] In this step, the first difference ΔV = V_bus(t) - V_bat,out(t), in mV, reflects the static difference between the output voltages of the two channels; the second difference ΔdV = dV_bus / dt - dV_bat,out / dt, in V / s, reflects the difference in the rate of change of the output voltages of the two channels.

[0115] These two differences can be calculated in real time by a subtraction circuit and transmitted to the PI controller as a basis for adjustment. For example, if the main channel output voltage is 24.000V and the standby channel is 23.995V, then ΔV = 5mV; if the main channel's rate of change is 500V / s and the standby channel's is 480V / s, then ΔdV = 20V / s.

[0116] S1434. Dynamically adjust the proportional coefficient and integral coefficient of the proportional-integral control algorithm according to the first difference and the second difference until the first difference satisfies the preset synchronization threshold and the second difference satisfies the preset rate of change threshold.

[0117] In this step, the preset synchronization threshold δV can be set to 10mV for example, and the preset rate of change threshold ε can be set to 10V / s for example.

[0118] When ΔV>10mV or ΔdV>10V / s, the proportional coefficient K_p of the PI controller is increased by 10% and the integral coefficient K_i is increased by 5% to speed up the adjustment.

[0119] When ΔV < 5mV and ΔdV < 5V / s, K_p is reduced by 10% and K_i is reduced by 5% to avoid system oscillation.

[0120] For example, when ΔV=15mV and ΔdV=15V / s, K_p is adjusted from 0.1 to 0.11 and Ki is adjusted from 100 to 105, increasing the adjustment amount of the duty cycle and thus rapidly reducing the deviation; when ΔV=3mV and ΔdV=3V / s, K_p is adjusted to 0.09 and Ki is adjusted to 95, making the system tend to stabilize.

[0121] This dynamic adjustment mechanism ensures that ΔV ≤ 10mV and ΔdV ≤ 10V / s within 1ms, thus satisfying the preset synchronization conditions.

[0122] S150. When the output characteristics meet the preset synchronization conditions, the load power supply path is switched from the main power supply channel to the backup power supply channel.

[0123] This step is the execution phase of power supply switching. By controlling the on and off of power switching devices, seamless switching of the load power supply path is achieved. The key is to ensure that the switching time is short enough and that voltage transients and current surges during the switching process are within the allowable range.

[0124] S151. Generate a synchronization control signal. The rising edge of the synchronization control signal triggers the turn-on operation of the backup power supply channel power switching device, and the falling edge triggers the turn-off operation of the main power supply channel power switching device.

[0125] In this step, the synchronization control signal is generated by the timer module in the MCU synchronization controller. Its frequency can be, for example, 100kHz, its duty cycle can be, for example, 50%, and the time deviation between the rising edge and the falling edge can be controlled within 10ns.

[0126] Specifically, when ΔV≤10mV and ΔdV≤10V / s are detected, the timer triggers a synchronization control signal. The rising edge (high-level start point) of this signal is used to trigger the drive circuit of the backup channel MOSFET (Q2), and the falling edge (high-level end point) is used to trigger the drive circuit of the main channel MOSFET (Q1), ensuring that Q2 is turned on first and Q1 is turned off later, thus avoiding power supply interruption.

[0127] S152. The synchronous control signal is converted into a gate drive signal for the power switching device through a drive circuit. The voltage amplitude and pulse width of the gate drive signal meet the turn-on and turn-off characteristics requirements of the power switching device.

[0128] In this step, the driving circuit uses a dedicated MOSFET driver chip to convert the synchronization control signal into a 12V gate drive signal. This voltage amplitude ensures that the MOSFET is fully turned on. The pulse width of the gate drive signal is determined based on the gate charge of the MOSFET. For a typical power MOSFET, the gate charge can be, for example, 100nC, and the drive current can be, for example, 1A. Therefore, the pulse width must satisfy t=Q / I=100nC / 1A=100ns to ensure that the gate charge is fully injected or released. Simultaneously, the driving circuit includes an accelerating capacitor to speed up the switching speed of the MOSFET, ensuring that the turn-on time t_on ≤ 50ns and the turn-off time t_off ≤ 50ns.

[0129] S153. During the turn-off process of the power switching device in the main power supply channel, monitor its drain-source voltage parameters to ensure that the voltage spike during the turn-off process does not exceed the device's withstand voltage value.

[0130] In this step, the drain-source voltage V_DS of the main channel MOSFET (Q1) is monitored in real time using a high-voltage probe to capture voltage spikes during the turn-off process. The MOSFET's withstand voltage V_DS_max can be, for example, 500V. During turn-off, due to the inductor's freewheeling current, V_DS will spike. By connecting an RC snubber circuit (e.g., R=100Ω, C=100pF) in parallel across the MOSFET, the voltage spike can be suppressed to within 300V (less than V_DS_max), ensuring device safety.

[0131] For example, when Q1 is turned off, if the inductor energy causes the V_DS spike to reach 350V, the RC snubber circuit will absorb some of the energy, causing the spike to drop to 280V.

[0132] S154. During the turn-on process of the power switching device in the backup power supply channel, monitor its conduction current parameters to ensure that the current rise rate does not exceed the maximum allowable current change rate of the device.

[0133] In this step, the on-current I_D of the standby channel MOSFET (Q2) is monitored by a current transformer with a bandwidth of 1MHz, which can capture rapid changes in current.

[0134] Assuming the maximum allowable current rise rate of the MOSFET is di / dt_max is 5A / μs, by connecting a small inductor (100nH) in series in the circuit, di / dt = V_bat / L = 25V / 100nH = 250A / μs can be limited. Obviously, this is too large, so it is necessary to adjust the gate resistor. Increasing the gate resistor (e.g., 10Ω) can slow down the switching speed and keep di / dt at 3A / μs (less than 5A / μs).

[0135] For example, when Q2 is turned on, the current rise rate is reduced to 2.5 A / μs by adjusting the gate resistance to 15 Ω to meet device requirements.

[0136] S155. When the main power supply channel power switch is completely turned off and the backup power supply channel power switch is completely turned on, the voltage transient during the switching process is suppressed by the output filter circuit to ensure that the load voltage fluctuation meets the preset stability threshold.

[0137] In this step, the output filter circuit can use a second-order LC filter with parameters such as inductance L=10μH, capacitance C=100μF, and damping resistor R=0.1Ω. The cutoff frequency of the filter can be such as f0=1 / (2π√(LC))=1 / (2π√(10μH×100μF))≈15.9kHz, which can effectively suppress high-frequency voltage transients generated during the switching process.

[0138] The formula for calculating the voltage transient ΔV_trans can be, for example, ΔV_trans=(di / dt)×R_eq / ω0. Here, di / dt is the rate of change of current (e.g., 3A / μs), R_eq is the equivalent resistance (e.g., 0.1Ω), and ω0=2πf0≈100krad / s. Therefore, ΔV_trans=3e6A / s×0.1Ω / 1e5rad / s=3V, which is clearly too large. In practice, by increasing the damping resistor and optimizing the capacitor selection, ΔV_trans can be suppressed to within 8mV, meeting the preset stability threshold (10mV).

[0139] S210. After the power supply channel switching is completed, the mains power health parameters are continuously monitored.

[0140] In this step, the mains power detection and prediction module maintains a sampling frequency of 1MHz and continues to collect status parameters such as V_AC(t) and f_AC(t). The mains power health parameter H_AC(t) is updated every 10ms and stored in the cache for evaluation of whether the mains power has returned to normal.

[0141] S220. When the mains power health parameter recovers to a level higher than the preset mains power health threshold and the duration reaches the preset stable time, an inverse switching enable signal is generated.

[0142] In this step, it is assumed that the mains power health threshold can be preset to H_th=0.7, and the preset stabilization time is set to 5s, so as to ensure that the reverse switch is performed only after the mains power recovers to a stable state.

[0143] For example, when the mains power health parameter gradually rises from 0.65 to 0.75 and remains above 0.75 for 5 seconds, it is determined that the mains power has been stabilized and restored. At this time, an inverse switching enable signal is generated. This signal can also be a 3.3V high-level pulse with a width of 100μs, which is transmitted to the control circuit of the main power supply channel.

[0144] S230. According to the reverse switching enable signal, control the output characteristics of the main power supply channel and the output characteristics of the backup power supply channel to achieve dynamic synchronization, and perform reverse switching operation when the preset synchronization conditions are met to switch the load power supply path from the backup power supply channel back to the main power supply channel.

[0145] The reverse switching process in this step is similar to the forward switching process, but in the opposite direction:

[0146] First, after receiving the reverse switching enable signal, the main power supply channel initiates a soft start, and its output voltage V_bus(t) gradually rises from 0V to be consistent with the output voltage V_bat,out(t) of the backup channel. The soft start slope is 1V / ms.

[0147] Then, the trajectory fitting algorithm is used to control V_bus(t) and dV_bus / dt to synchronize with V_bat,out(t) and dV_bat,out / dt until ΔV≤10mV and ΔdV≤10V / s.

[0148] Finally, a synchronization control signal is generated. Its rising edge triggers the main channel MOSFET (Q1) to turn on, and its falling edge triggers the backup channel MOSFET (Q2) to turn off. The switching time is also controlled within 10μs. The output filter circuit suppresses voltage transients to ensure stable power supply to the load. After the reverse switching is completed, the system returns to the mains power supply mode, and the battery pack enters the float charging state, waiting for the next switching demand.

[0149] The method provided in this application comprehensively evaluates and processes multiple state parameters of the mains power input using an artificial intelligence network model to generate mains power health parameters. Real-time monitoring and analysis of the battery pack's operating state parameters generate battery health parameters. Based on the mains power health parameters and the battery health parameters, a power supply switching decision is made, generating a switching enable signal. The output characteristics of the backup power supply channel are dynamically synchronized with the output characteristics of the main power supply channel according to the switching enable signal. The backup power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the mains power input. When the output characteristics meet preset synchronization conditions, the load power supply path is switched from the main power supply channel to the backup power supply channel. This ensures that the load is almost unaffected during the switching process through a highly consistent output voltage trajectory and an extremely short switching time, avoiding equipment restarts, bit errors, or link breaks caused by voltage drops and transient fluctuations in traditional switching methods. This significantly reduces the power supply control switching delay and power supply fluctuations in the communication power system. Simultaneously, with the accurate prediction and real-time analysis of the artificial intelligence network model, the mains power condition can be predicted in advance, and optimal switching decisions can be made, further improving the accuracy and reliability of power supply control switching.

[0150] Figure 2 This is a schematic diagram of the structure of an artificial intelligence-based power switching control system 100 provided in an embodiment of this application. Figure 2 As shown, the processor 120 can be used in the AI-based power switching control system 100 and to perform the functions in this invention.

[0151] The AI-based power switching control system 100 can be a general-purpose server or a special-purpose server; both can be used to implement the AI-based power switching control method of this invention. Although only one server is shown in this invention, for convenience, the functions described in this invention can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0152] For example, the AI-based power switching control system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the AI-based power switching control system 100 may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present invention can be implemented according to these program instructions. The AI-based power switching control system 100 also includes an input / output (I / O) interface 150 between the computer and other input / output devices.

[0153] For ease of explanation, only one processor is described in the AI-based power switching control system 100. However, it should be noted that the AI-based power switching control system 100 of this invention may also include multiple processors, and therefore the steps executed by one processor described in this invention may also be executed jointly by multiple processors or individually. For example, if the processor of the AI-based power switching control system 100 executes steps A and B, it should be understood that steps A and B may also be executed jointly by two different processors or individually by one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0154] This invention discloses an electrical device that uses the artificial intelligence-based power switching control system described in the foregoing embodiments for power control. This electrical device may be, for example, a communication base station, server power supply equipment, industrial Ethernet equipment, or 5G transmission equipment.

[0155] This invention discloses a computer read storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to execute the steps in the artificial intelligence-based power supply switching control method of the aforementioned embodiments.

[0156] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the artificial intelligence-based power switching control method of the aforementioned embodiments.

[0157] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0158] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electronically erasable rewritable read-only memory (EEPROM), compact optical disc (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to have or store data.

[0159] Finally, it should be noted that the above-disclosed embodiments are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An artificial intelligence-based power supply switching control method, characterized by, The method comprises: comprehensive evaluation and processing of multiple state parameters of the mains input by an artificial intelligence network model to generate a mains health parameter; real-time monitoring and analysis of the operating state parameters of the battery pack to generate a battery health parameter; based on the mains health parameter and the battery health parameter, performing power supply switching decision judgment to generate a switching enable signal, comprising: comparing the mains health parameter with a preset mains health threshold, and comparing the battery health parameter with a preset battery health threshold; when the mains health parameter is lower than the preset mains health threshold and the battery health parameter is higher than the preset battery health threshold, generating the switching enable signal, which is used to indicate the starting of a synchronous control process of a backup power supply channel; controlling the output characteristics of the backup power supply channel and the output characteristics of the main power supply channel to realize dynamic synchronization according to the switching enable signal, comprising: after receiving the switching enable signal, starting a soft start control process of the backup power supply channel and adjusting the output voltage parameter of the backup power supply channel; real-time acquisition of the output voltage parameter and the output voltage change rate parameter of the main power supply channel as the main channel output characteristics; controlling the output voltage parameter of the backup power supply channel to track the main channel output characteristics by a trajectory fitting algorithm, so that the difference between the output voltage parameter of the backup power supply channel and the output voltage parameter of the main power supply channel satisfies a preset synchronization threshold, and the difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel satisfies a preset change rate threshold; wherein the backup power supply channel supplies power to the battery pack, and the main power supply channel supplies power to the mains input; when the output characteristics satisfy a preset synchronization condition, switching the load power supply path from the main power supply channel to the backup power supply channel. 2.The AI-based power supply switching control method of claim 1, wherein, The comprehensive evaluation and processing of the multiple state parameters of the mains input by the artificial intelligence network model to generate the mains health parameter comprises: acquiring the multiple state parameters of the mains input, the multiple state parameters comprising at least two of a voltage parameter, a frequency parameter, a voltage change rate parameter, a load current parameter and an environmental temperature parameter; constructing the multiple state parameters into a multi-dimensional feature vector, performing time series feature extraction and fault probability prediction processing on the multi-dimensional feature vector by a convolutional neural network and a long short-term memory network combined model to generate a mains fault probability parameter; calculating the mains health parameter based on the deviation degree of the mains fault probability parameter and the voltage parameter, the mains health parameter being negatively correlated with the deviation degree of the mains fault probability parameter and the voltage parameter. 3.The AI-based power supply switching control method of claim 2, wherein, The real-time monitoring and analysis of the operating state parameters of the battery pack to generate the battery health parameter comprises: acquiring the operating state parameters of the battery pack, the operating state parameters comprising a battery voltage parameter, a battery current parameter, a battery internal resistance parameter and a battery temperature parameter; performing state estimation processing on the operating state parameters by an extended Kalman filter algorithm to generate a battery state of charge parameter and a battery health state parameter; calculating a battery maximum available power parameter based on the battery state of charge parameter, the battery internal resistance parameter, and the battery voltage parameter; generating the battery health parameter according to the battery state of charge parameter, the battery voltage parameter, the battery maximum available power parameter, and the battery state of health parameter. 4.The AI-based power supply switching control method of claim 1, wherein, The output voltage parameter of the backup power supply channel tracks the output characteristics of the main channel by the trajectory fitting algorithm, so that the difference between the output voltage parameter of the backup power supply channel and the output voltage parameter of the main power supply channel meets the preset synchronization threshold, and the difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel meets the preset change rate threshold, including: Taking the output voltage parameter and the output voltage change rate parameter of the main power supply channel as the reference trajectory, an output trajectory target function of the backup power supply channel is constructed; Based on the output trajectory target function, the duty cycle parameter of the power conversion circuit of the backup power supply channel is adjusted by a proportional integral control algorithm; The first difference between the output voltage parameter of the backup power supply channel and the output voltage parameter of the main power supply channel, and the second difference between the output voltage change rate parameter of the backup power supply channel and the output voltage change rate parameter of the main power supply channel are obtained; The proportional coefficient and the integral coefficient of the proportional integral control algorithm are dynamically adjusted according to the first difference and the second difference until the first difference meets the preset synchronization threshold and the second difference meets the preset change rate threshold. 5.The AI-based power supply switching control method of claim 4, wherein, The load power supply path is switched from the main power supply channel to the backup power supply channel, including: A synchronization control signal is generated, the rising edge of the synchronization control signal triggers the conduction operation of the backup power supply channel power switching device, and the falling edge triggers the turn-off operation of the main power supply channel power switching device; The synchronization control signal is converted into a gate drive signal of the power switching device by a driving circuit, and the voltage amplitude and pulse width of the gate drive signal meet the conduction and turn-off characteristics of the power switching device; During the turn-off process of the main power supply channel power switching device, its drain-source voltage parameter is monitored to ensure that the voltage spike during the turn-off process does not exceed the device withstand voltage value; During the conduction process of the backup power supply channel power switching device, its conduction current parameter is monitored to ensure that the current rise rate does not exceed the maximum current change rate allowed by the device; After the main power supply channel power switching device is completely turned off and the backup power supply channel power switching device is completely turned on, the voltage transient during the switching process is suppressed by an output filter circuit to ensure that the load end voltage fluctuation meets the preset stability threshold. 6.The AI-based power supply switching control method of claim 1, wherein, The method further includes: After completing the power supply channel switching, the utility health parameter is continuously monitored; When the utility health parameter recovers to be higher than the preset utility health threshold and the duration reaches the preset stability time, an inverse switching enable signal is generated; According to the reverse switching enabling signal, the output characteristics of the main power supply channel and the output characteristics of the backup power supply channel are dynamically synchronized, and when a preset synchronization condition is met, a reverse switching operation is performed to switch the load power supply path from the backup power supply channel back to the main power supply channel.

7. An artificial intelligence-based power supply switching control system, characterized by, The application also provides a computer readable storage medium, wherein a computer executable instruction is stored in the computer readable storage medium, and the computer executable instruction is executed by a computer to implement the artificial intelligence based power supply switching control method in any one of claims 1-6.

8. An electric device, characterized by The power utilization device adopts the artificial intelligence based power supply switching control system in claim 7 for power utilization control.

Images