Power overshoot absorption and lossless switching self-healing control method and device

Abstract

This specification provides a power supply overshoot energy absorption lossless switching self-healing control method and apparatus. The power supply overshoot energy absorption lossless switching self-healing control method includes: responding to a self-healing switching command issued to a secondary power supply, initializing and generating at least two sets of PID parameters for a PID controller; globally optimizing the at least two sets of parameters using a particle swarm optimization algorithm to obtain an optimized optimal PID parameter combination; acquiring the load voltage of the load in the secondary power supply system at the current moment through the PID controller; simulating and generating simulated voltages of the load at different times within a preset future time period based on the load voltage and the optimal PID parameter combination; calculating the duty cycle of a PWM signal based on the optimal PID parameter combination, the load voltage, and the simulated voltage; switching the main power supply to a standby power supply, and outputting a PWM signal to the standby power supply according to the duty cycle through the PID controller to perform power supply self-healing control.

Description

Technical Field

[0001] This manual relates to the field of aging bench technology, and in particular to a self-healing control method for power overcharge energy absorption and lossless switching. Background Technology

[0002] In the semiconductor chip manufacturing process, aging testing is a crucial step in eliminating early-failure products and ensuring high-yield shipments. Aging tests typically last for hours or even days, during which the power supply system must provide high-precision, uninterrupted power to the device under test. To prevent test interruptions or data loss due to power module failures, the use of two-stage power redundancy technology is of great significance. This involves configuring a main power supply and a backup power supply for the load, and switching to the backup power supply to continue operation when the main power supply experiences a failure warning.

[0003] However, in practical engineering applications, current power switching control technology still has significant technical shortcomings in the switching process: (1) Transient interruption problem of traditional hard switching. The simplest switching method is to use relays or contactors to perform "break-before-connect" or "connect-before-break". However, in the "break-before-connect" mode, there will inevitably be a millisecond-level physical power outage in the power output. At the moment of switching, the main power supply current is suddenly cut off and the backup power supply current suddenly surges in. This drastic current change rate will induce high-frequency noise on the parasitic inductance of the circuit, interfering with the normal operation of the high-frequency chip at the back end. In the "connect-before-break" mode, the jitter and action delay of the mechanical contacts are very likely to generate voltage spikes on the bus, which may damage the precision structure of the chip under test.

[0004] (2) Voltage fluctuations during load transfer. The power supply lines of the aging test bench are usually long and have significant line impedance. During power transfer, as the current transfers from the main power supply path to the backup power supply path, the common bus voltage is prone to drop or overshoot due to the inconsistent impedance characteristics of the two physical paths (such as contact resistance and differences in the on-state resistance of MOSFETs) and the lag in the response of the control loop. Current open-loop control or single voltage loop control cannot compensate for this voltage drop caused by changes in line impedance in real time, causing the actual voltage at the load end to deviate from the rated value, affecting the standardization of aging tests, and in severe cases, even causing test data errors. Therefore, an effective method is urgently needed to solve this problem. Summary of the Invention

[0005] In view of this, embodiments of this specification provide a power supply overcharge energy absorption lossless switching self-healing control method. One or more embodiments of this specification also relate to a power supply overcharge energy absorption lossless switching self-healing control device, a computing device, a computer-readable storage medium, and a computer program, to address the technical deficiencies existing in the prior art.

[0006] According to a first aspect of the embodiments of this specification, a power supply overshoot energy absorption lossless switching self-healing control method is provided, comprising: In response to the self-healing switching command issued to the secondary power supply, at least two sets of PID parameters of the PID controller are initialized and generated. The at least two sets of parameters are then globally optimized using a particle swarm optimization algorithm to obtain the optimal combination of PID parameters. The load voltage of the load in the secondary power supply system at the current moment is collected by a PID controller. Based on the load voltage and the optimal PID parameter combination, the simulated voltage of the load at different times within a preset future time period is generated. Based on the optimal PID parameter combination, the load voltage, and the simulated voltage, the duty cycle of the PWM signal is calculated; The main power supply is switched to the backup power supply, and a PWM signal is output to the backup power supply according to the duty cycle through the PID controller to perform power self-healing control.

[0007] Optionally, the step of globally optimizing the at least two sets of parameters using a particle swarm optimization algorithm to obtain the optimized optimal PID parameter combination includes: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

[0008] Optionally, the power overshoot energy absorption lossless switching self-healing control method further includes: Read the optimal fitness function values ​​corresponding to the three historical time periods prior to the preset future time period; Based on the optimal fitness function values ​​corresponding to the three historical time periods, calculate the average rate of change of the optimal fitness function values; Based on the average rate of change, the inertia weight, individual learning factor, and social learning factor corresponding to the preset future time period are dynamically adjusted.

[0009] Optionally, calculating the average rate of change of the optimal fitness function value based on the optimal fitness function values ​​corresponding to the three historical time periods includes: Based on the optimal fitness values ​​corresponding to the first historical time period and the second historical time period, calculate the first relative rate of change of the optimal fitness function value; Based on the optimal fitness values ​​corresponding to the first and third historical time periods, the second relative rate of change of the optimal fitness function value is calculated. Calculate the mean of the first relative rate of change and the second relative rate of change, and determine the mean as the average rate of change of the optimal fitness function value; Wherein, the first historical time period is adjacent to the second historical time period, the second historical time period is adjacent to the first historical time period and the third historical time period, and the third historical time period is adjacent to the second historical time period and the preset future time period.

[0010] Optionally, the duty cycle of the PWM signal is calculated as follows:

[0011] in, , , These are the three parameters in the optimal PID parameter combination. Duty cycle, Indicates rated voltage With load voltage The deviation.

[0012] Optionally, the expression for the fitness function is:

[0013] Where T is the length of the preset future time period; t is a continuous time variable; The rated voltage of the load; For based on The simulated voltage output curve of the load obtained from parameter simulation. This refers to each set of PID parameters included in the optimization process.

[0014] According to a second aspect of the embodiments of this specification, a power supply overcharge energy absorption lossless switching self-healing control device is provided, comprising: The optimization module is configured to, in response to the self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for the PID controller, and perform global optimization on the at least two sets of parameters using the particle swarm optimization algorithm to obtain the optimal combination of PID parameters after optimization. The acquisition module is configured to acquire the load voltage of the load in the secondary power system at the current moment through a PID controller; The simulation module is configured to generate simulated voltages of the load at different times within a preset future time period based on the load voltage and the optimal PID parameter combination. The calculation module is configured to calculate the duty cycle of the PWM signal based on the optimal PID parameter combination, the load voltage, and the simulated voltage. The control module is configured to switch the main power supply to the backup power supply and output a PWM signal to the backup power supply according to the duty cycle through the PID controller to perform power self-healing control.

[0015] Optionally, the optimization module is further configured to: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

[0016] According to a third aspect of the embodiments of this specification, a computing device is provided, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement any of the steps of the power supply overcharge energy absorption lossless switching self-healing control method.

[0017] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores computer-executable instructions, which, when executed by a processor, implement the steps of any of the power supply overcharge energy absorption lossless switching self-healing control methods described herein.

[0018] According to a fifth aspect of the embodiments of this specification, a computer program is provided, wherein when the computer program is executed in a computer, the computer is instructed to perform the steps of the above-described power overcharge energy absorption lossless switching self-healing control method.

[0019] This specification's embodiments, in response to a self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for a PID controller. These parameters are then globally optimized using a particle swarm optimization algorithm to obtain the optimal PID parameter combination. The PID controller acquires the load voltage of the load in the secondary power supply system at the current moment. Based on the load voltage and the optimal PID parameter combination, simulated voltages of the load are generated at different times within a preset future time period. Based on the optimal PID parameter combination, load voltage, and simulated voltage, the duty cycle of the PWM signal is calculated. The primary power supply is switched to the backup power supply, and the PID controller outputs a PWM signal to the backup power supply according to the duty cycle for power self-healing control. The method provided in this specification's embodiments can eliminate destructive circulating currents and automatically compensate for line voltage drops to maintain a constant bus voltage during the power handover process, achieving truly "seamless and lossless" soft switching, thereby helping to ensure the stable operation of the secondary power supply system. Attached Figure Description

[0020] Figure 1 This is a flowchart of a power supply overcharge energy absorption lossless switching self-healing control method provided in one embodiment of this specification; Figure 2a This is a schematic diagram of a power overcharge energy absorption lossless switching circuit provided in one embodiment of this specification; Figure 2b This is a schematic diagram illustrating a power supply self-healing control effect provided in one embodiment of this specification; Figure 3 This is a schematic diagram of the structure of a power supply overcharge energy absorption lossless switching self-healing control device provided in one embodiment of this specification; Figure 4 This is a structural block diagram of a computing device provided in one embodiment of this specification.

[0021] Figure 2a The components are as follows: 1. Common output circuit; 2. Main power supply circuit; 21. Main BUCK circuit; 22. First power switch; 23. Buffer circuit; 3. Backup power supply circuit; 31. Backup BUCK circuit; 311. Second power switch; 32. Third power switch; 33. Second RCD snubber circuit; 4. Control circuit; 5. Voltage sensor.

[0022] It should be noted that these accompanying drawings and textual descriptions are not intended to limit the scope of the embodiments in this specification in any way, but rather to illustrate the concepts of the embodiments in this specification to those skilled in the art by referring to specific embodiments. Detailed Implementation

[0023] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0024] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0025] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0026] This specification provides a power supply overcharge energy absorption lossless switching self-healing control method. This specification also relates to a power supply overcharge energy absorption lossless switching self-healing control device, a computing device, a computer-readable storage medium, and a computer program, which will be described in detail in the following embodiments.

[0027] Figure 1 A flowchart of a power supply overshoot energy absorption lossless switching self-healing control method according to an embodiment of this specification is shown, which specifically includes the following steps.

[0028] Step 102: In response to the self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for the PID controller, and perform global optimization on the at least two sets of parameters using the particle swarm optimization algorithm to obtain the optimal combination of PID parameters.

[0029] In one optional implementation, the step of globally optimizing the at least two sets of parameters using a particle swarm optimization algorithm to obtain the optimized optimal PID parameter combination includes: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

[0030] Furthermore, the expression for the fitness function is:

[0031] Where T is the length of the preset future time period; t is a continuous time variable; The rated voltage of the load; For based on The simulated voltage output curve of the load obtained from parameter simulation. This refers to each set of PID parameters included in the optimization process.

[0032] Specifically, in the particle swarm optimization process, there are k particles in total, and for each particle, a PID parameter vector is generated. The system response of the secondary power system is simulated over a period of time T by performing m iterations of optimization and using the PID parameter vector (optimal PID parameter combination) represented by the particle with the optimal fitness function value after iteration.

[0033] To more accurately measure control precision, the embodiments in this specification use the cumulative error integral in the continuous time domain as the fitness function (objective function). .

[0034] The fitness function expression is:

[0035] Where T is the length of the preset future time period; t is a continuous time variable; The rated voltage of the load; For based on The simulated voltage output curve of the load obtained from parameter simulation. This refers to each set of PID parameters included in the optimization process.

[0036] If we define the current time as 0s, then the preset future time period is 0 to Ts. Continuing with the previous example, let's assume there are k particles in total, and for each particle, we have a PID parameter vector... If the optimization is iterated m times, then during the entire optimization process, there will be a total of k×m different PID parameter vectors. For any one of these k×m different PID parameter vectors, its The value is obtained by adjusting the PID parameter. As parameters for the PID controller within that future time period, then combined with the circuit and based on the current load voltage and the PID parameters... The simulation generates the load's voltage at the next time t1 within a preset future time period; then, based on the simulated voltage at the next time t1 and the PID parameters... The simulation generates the load voltage at time t2, which is the next time step after time t1; this process continues until the simulated voltage at time T is generated. The simulated voltages at each time step are then substituted into the fitness function expression to calculate the fitness function value corresponding to the PID parameter vector. For the other k×m-1 PID parameter vectors, their respective fitness function values ​​can be calculated using the same process. The PID parameter combination corresponding to the smallest fitness function value among the calculated k×m fitness function values ​​is then determined as the optimal PID parameter combination.

[0037] In an optional embodiment, the power overshoot energy absorption lossless switching self-healing control method further includes: Read the optimal fitness function values ​​corresponding to the three historical time periods prior to the preset future time period; Based on the optimal fitness function values ​​corresponding to the three historical time periods, calculate the average rate of change of the optimal fitness function values; Based on the average rate of change, the inertia weight, individual learning factor, and social learning factor corresponding to the preset future time period are dynamically adjusted.

[0038] Furthermore, the step of calculating the average rate of change of the optimal fitness function value based on the optimal fitness function values ​​corresponding to the three historical time periods includes: Based on the optimal fitness values ​​corresponding to the first historical time period and the second historical time period, calculate the first relative rate of change of the optimal fitness function value; Based on the optimal fitness values ​​corresponding to the first and third historical time periods, the second relative rate of change of the optimal fitness function value is calculated. Calculate the mean of the first relative rate of change and the second relative rate of change, and determine the mean as the average rate of change of the optimal fitness function value; Wherein, the first historical time period is adjacent to the second historical time period, the second historical time period is adjacent to the first historical time period and the third historical time period, and the third historical time period is adjacent to the second historical time period and the preset future time period.

[0039] Specifically, before the start of a preset future time period (denoted as the nth period), the optimal fitness function values ​​corresponding to the three historical time periods preceding this preset future time period are read from the memory and denoted as follows: , , Then, based on the optimal fitness values ​​corresponding to the first historical time period and the second historical time period respectively. Calculate the first relative rate of change of the optimal fitness function value. Based on the optimal fitness values ​​corresponding to the first and third historical time periods respectively. Calculate the second relative rate of change of the optimal fitness function value. The formula is as follows:

[0040]

[0041] in, To prevent tiny constants with a denominator of zero.

[0042] Next, calculate the average of the two relative rates of change mentioned above. As an evaluation indicator for measuring the stability of the current system control effect:

[0043] In the embodiments of this specification, the calculated average rate of change can be used as a basis. Dynamically adjust the inertia weights for preset future time periods. Individual learning factors Social learning factors The adjustment logic is as follows: Establish the average rate of change The mapping function with the three parameters mentioned above. When When the fitness is relatively small, it indicates that the system fitness is stabilizing or has converged to near the optimal solution. In this case, reducing the fitness value is appropriate. , , The value of is used to reduce the search step size, perform local fine-tuning, and suppress oscillations; when A larger value indicates that the system's operating conditions are fluctuating drastically or that it is in a period of adjustment. In this case, increasing the value is appropriate. , , The value of is adjusted to enhance global search capabilities and quickly capture new optimal particles.

[0044] The following linear dynamic adjustment formula is specifically adopted:

[0045]

[0046]

[0047] in, The mapping function maps the average rate of change to the interval [0,1]. The function chosen is the Gaussian sigmoid function, and its specific form is as follows:

[0048] in, The distribution width coefficient controls the rate of ascent of the curve. The larger the value, the flatter the curve.

[0049] The Gaussian S-shaped function can be used to achieve the average rate of change. When smaller, for , , The impact of parameter changes is relatively small.

[0050] In addition, the embodiments in this specification perform particle swarm iteration and optimal parameter output in the following manner:

[0051]

[0052] in, Let be the position vector, representing the position vector of the i-th particle at the t-th iteration; Let be the velocity vector, representing the adjustment step size of the PID parameters for the i-th particle at the t-th iteration; The optimal solution for an individual particle represents the position with the best fitness found by the i-th particle from the beginning to the t-th iteration. The global optimal solution represents the position with the best fitness among all particles in the entire particle swarm from the beginning to the t-th iteration. , The random number for each iteration follows a uniform distribution within the interval [0,1], used to increase the randomness of particle motion.

[0053] As mentioned earlier, each time period involves m iterations with k particles. After the iterations, the search is performed to find the... The smallest globally optimal particle ,Will The corresponding parameters are used as the optimal PID parameter combination for the current cycle in the secondary power system, and this optimal fitness value is applied. Stored in memory for adaptive parameter calculation in the next cycle.

[0054] Step 104: Collect the load voltage of the load in the secondary power system at the current moment through the PID controller.

[0055] Specifically, after receiving the self-healing switching command, the system can use a PID controller to collect the load voltage of the load in the secondary power system at the current moment. .

[0056] Step 106: Based on the load voltage and the optimal PID parameter combination, simulate the load voltage at different times within a preset future time period.

[0057] Specifically, as mentioned earlier, if the current time is defined as 0 seconds, then the preset future time period is 0 to T seconds. Given a total of k particles, the PID parameter vector represented by each particle is... If the optimization is iterated m times, then during the entire optimization process, there will be a total of k×m different PID parameter vectors. For any one of these k×m different PID parameter vectors, its The value is obtained by adjusting the PID parameter. As parameters for the PID controller within that future time period, then combined with the circuit and based on the current load voltage and the PID parameters... The simulation generates the load's voltage at the next time t1 within a preset future time period; then, based on the simulated voltage at the next time t1 and the PID parameters... The simulation generates the simulated voltage of the load at the next time t2 after the next time t1; and so on, until the simulation generates the simulated voltage at time T.

[0058] Since the optimal PID parameter combination is one of k×m different PID parameter vectors, the process of simulating and generating the simulated voltage at different times within a preset future period based on the optimal PID parameter combination is similar to the aforementioned process of simulating and generating the simulated voltage at different times based on any PID parameter vector, and will not be repeated here.

[0059] Step 108: Calculate the duty cycle of the PWM signal based on the optimal PID parameter combination, the load voltage, and the simulated voltage.

[0060] In one optional implementation, the duty cycle of the PWM signal is calculated as follows:

[0061] in, , , These are the three parameters in the optimal PID parameter combination. Duty cycle, Indicates rated voltage With load voltage The deviation.

[0062] The expression is as follows:

[0063] Step 110: Switch the main power supply to the backup power supply, and output a PWM signal to the backup power supply according to the duty cycle through the PID controller to perform power supply self-healing control.

[0064] Specifically, after determining the optimal PID parameter combination, it is applied to the secondary power system. Specifically, based on the load voltage and the optimal PID parameter combination, the simulated voltage of the load at different times within a preset future time period is generated. Then, based on the optimal PID parameter combination, the load voltage, and the simulated voltage, the duty cycle of the PWM signal is calculated. The main power supply is then switched to the backup power supply, and the PID controller outputs the PWM signal to the backup power supply according to the duty cycle to perform power self-healing control.

[0065] In practical applications, the aforementioned "combined circuit and based on the current load voltage and the PID parameters" "Simulate and generate the simulated voltage of the load at the next time t1 within a preset future time period," specifically, this can be achieved by first modifying the PID parameters... Substitute the values ​​into the aforementioned duty cycle calculation formula, and then calculate the rated voltage of the load. The load voltage at the current moment Substitute the values ​​into the formula to calculate the duty cycle at the current moment, and apply this duty cycle to the circuit for simulation. That is, the PID controller outputs a PWM signal to the backup power supply according to the duty cycle, so that the backup power supply supplies power to the load, thereby simulating the simulated voltage of the load at the next moment t1.

[0066] Figure 2aThis specification provides a schematic diagram of a power supply overshoot absorption and lossless switching circuit, comprising: a common output circuit 1, a main power supply circuit 2, a backup power supply circuit 3, and a control component. The main power supply circuit 2 includes a main BUCK circuit 21 and a first soft-switching circuit, electrically connected. The first soft-switching circuit is electrically connected to the common output circuit 1 and includes a first power switch 22 and a buffer circuit 23 connected in parallel with the first power switch 22. The backup power supply circuit 3 includes a backup BUCK circuit 31, electrically connected to the common output circuit 1, and having a second power switch 311. The input terminal of the control component is electrically connected to the common output circuit 1, and the output terminal of the control component is connected to the second power switch 311, for sending a PWM signal with a variable duty cycle to the second power switch 311.

[0067] This embodiment of the specification, by setting a first soft-switching circuit, forcibly reduces the voltage rise rate when the first power switch 22 is turned off, eliminating the high-frequency oscillation source at its source. This makes the change curve of the common output circuit 1 during the switching process smoother, effectively avoiding load voltage overshoot caused by interference and ensuring the purity of the power supply to the chip under test. The control component can control the backup power supply circuit 3 to perform voltage compensation, thereby significantly reducing the voltage drop during the switching process.

[0068] In some possible implementations, the control components include a control circuit 4 and a voltage sensor 5 connected to the common output circuit 1 for detecting the voltage of the common output circuit 1, and the voltage sensor 5 is electrically connected to the input terminal of the control circuit 4.

[0069] Because the first soft-switching circuit filters out high-frequency spike interference, the electrical signal obtained by the control component from the common output circuit 1 is more realistic and stable. This prevents subsequent backup power supply closed-loop regulation from being falsely triggered by noise, enabling more accurate voltage compensation and significantly reducing the voltage drop during switching. Once the control component detects a slight voltage drop on the common output circuit 1, indicating that the main power supply circuit 2 has been cut off, the control component immediately starts the voltage regulation algorithm. The control component quickly outputs a high duty cycle PWM signal to drive the second power switch 311.

[0070] The backup power supply rapidly injects current into the common output circuit 1 through the backup BUCK circuit 31. Thanks to the stable voltage environment brought about by the "soft shutdown" of the main circuit in the previous stage, the feedback loop can accurately calculate the compensation amount, quickly pull back the voltage of the common output circuit 1 and lock it at the rated value, realizing a seamless switch from "main circuit pass-through" to "backup circuit PWM regulation".

[0071] In some possible implementations, the control circuit 4 is a PID control circuit 4. Traditional hard switching, at the moment of disconnection, due to the sudden current change (extremely high di / dt), not only generates high voltage across the switching transistor but also excites high-frequency oscillations in the parasitic inductance and stray capacitance of the line. This oscillation is coupled to the load end through the common output circuit 1, causing severe overshoot and noise in the load voltage. This embodiment of the specification, by setting a first RCD absorption circuit, forcibly reduces the voltage rise rate (dv / dt) when the first power switch is turned off, eliminating the high-frequency oscillation source at its source. This makes the voltage change curve of the common output circuit 1 smoother during the switching process, effectively avoiding load voltage overshoot caused by interference and ensuring the purity of the power supply to the tested chip. The control circuit 4 in this embodiment of the specification is a PID control circuit 4, which, combined with the voltage sensor 5, forms a hardware closed-loop feedback loop, enabling real-time acquisition of the voltage of the common output circuit 1 and timely voltage compensation by the backup power supply circuit 3. The control component in this embodiment of the specification is a PID control component (PID controller), which can achieve precise adjustment of the output of the second power switch 311.

[0072] In some possible implementations, the first RCD snubber circuit includes a diode and a capacitor, both of which are connected in parallel with the first power switch 22.

[0073] In the first RCD snubber circuit, the core function of the diode and capacitor is to work together to suppress the voltage spike generated when the first power switch 22 is turned off, and to protect the first power switch 22 from overvoltage damage. The first RCD snubber circuit is a key buffer structure to ensure the stable operation of the system.

[0074] The first RCD snubber circuit includes at least a buffer capacitor and an anti-reverse current diode. When the control component controls the first power switch 22 to suddenly turn off, the current will be transferred to the parallel buffer capacitor, limiting the rate of voltage rise across the first power switch 22 and achieving a physical "soft turn-off".

[0075] When the system decides to switch, the control component sends a signal to quickly turn off the first power switch 22. Due to the presence of the buffer circuit 23 and the main Buck circuit structure, the current in the main power supply circuit does not drop to zero instantaneously, but is slowly discharged through the capacitor (soft turn-off), avoiding voltage spikes. At the same time, the control component detects a decreasing trend in the voltage across the load side of the common output circuit 1, and immediately increases the PWM duty cycle of the switch in the backup power supply circuit 3 based on feedback from the voltage acquisition unit, so that the backup power supply provides energy to maintain the voltage stability of the common output circuit 1.

[0076] The working principle of the power overshoot absorption lossless switching circuit in the embodiments of this specification is mainly based on the synergistic effect of "hardware buffer" and "closed-loop feedback". Its complete operation process is divided into the following four key stages: Normal power supply phase (common output circuit 1 normally on mode) During normal system operation, the control component keeps the first power switch 22 in the main power supply channel fully on, and the energy from the main power supply is transferred to the common output circuit 1 and the load connected to the common output circuit 1 with low loss through the first power switch 22. At this time, the voltage across the buffer circuit 23 connected in parallel across the first power switch 22 is close to zero (short-circuited), and it is in a ready state; while the backup power supply circuit 3 is in standby mode, and its internal second power switch 311 is off or in a very low duty cycle state, and does not output energy to the common output circuit 1.

[0077] Switching between trigger and "soft shutdown" action phases When the control component detects a main power failure warning or receives a manual switching command, it sends a control signal to suddenly disconnect the first power switch 22.

[0078] Physical process: Since the first power switch 22 was previously in a state of continuous high-current conduction, at the moment of sudden physical turn-off, the current in the parasitic inductance of the line and the main inductance has huge inertia. At this time, the current cannot disappear instantly, but automatically changes its path and flows into the parallel buffer circuit 23.

[0079] Soft shutdown principle: The current begins to charge the capacitor on the buffer circuit 23. Utilizing the physical characteristic that the voltage across the capacitor cannot change instantaneously (Vc must rise linearly with charge accumulation), the rate of voltage rise (dv / dt) across the first power switch 22 is forcibly limited.

[0080] This hardware structure transforms the originally extremely steep "hard shutdown" voltage jump into a gently rising slope. This not only prevents the first power switch 22 from being damaged by overvoltage, but more importantly, it blocks the high-frequency oscillations and back electromotive force generated by the sudden current stop, so that the voltage on the common output circuit 1 will not exhibit severe overshoot or high-frequency noise, but will instead show a smooth and predictable downward trend.

[0081] Intermediate transition and energy maintenance phase During the extremely short time window when the first power switch 22 is completely off and the backup power supply has not yet fully established its output, the voltage on the common output circuit 1 maintains the charge stored in the large-capacity filter capacitor connected in parallel with the main power supply circuit 2. Since the main power supply no longer replenishes energy, the voltage of the common output circuit 1 will exhibit a smooth, slight drop, which serves as a signal to trigger the backup power supply to intervene.

[0082] Backup power closed-loop voltage regulation intervention stage Voltage sensor 5 monitors the voltage of common output circuit 1 in real time. Once a slight voltage drop is detected (indicating that the main circuit has been cut off), the control component immediately starts the voltage regulation algorithm.

[0083] Closed-loop regulation: The control component quickly outputs a high duty cycle PWM signal to drive the second power switch 311 (MOS transistor) of the backup power supply circuit 3.

[0084] Seamless takeover: The backup power supply quickly injects current into the common output circuit 1 through the backup BUCK circuit 31. Thanks to the stable voltage environment brought about by the "soft shutdown" of the main circuit in the previous stage, the feedback loop can accurately calculate the compensation amount, quickly pull back the voltage of the common output circuit 1 and lock it at the rated value, realizing a seamless switch from "main circuit pass-through" to "backup circuit PWM regulation".

[0085] A schematic diagram of a power supply self-healing control effect provided in the embodiments of this specification is shown below. Figure 2b As shown. Figure 2b In the figure, the horizontal axis represents the simulation time (unit: sampling points or time step), and the vertical axis represents the output voltage (unit: V). Figure 2b The image shows the signal curves of the voltage across the load during the startup phase, the self-healing switching phase, and the subsequent stable operation phase of different secondary power supply systems.

[0086] At t=1.5×10 6 Switching at this point and comparing the voltage waveform of the load, the three curves are as follows: Yellow curve: Response curve using soft-switching circuit and without closed-loop control; Blue curve: Response curve using hard switching circuit under closed-loop control; Red curve: Response curve using soft-switching circuit and closed-loop control (or response curve after using the method provided in the embodiments of this specification).

[0087] from Figure 2bAs can be seen, after using the power supply self-healing control method provided in the embodiments of this specification, the voltage fluctuation amplitude of the load is significantly reduced during the switching of the main and backup power supplies. Traditional hard switching, due to the sudden current change (extremely high di / dt) at the moment of disconnection, not only generates high voltage across the switching transistor but also excites high-frequency oscillations in the parasitic inductance and stray capacitance of the line. This oscillation is coupled to the load end through the common output circuit 1, causing severe overshoot and noise in the load voltage. However, by setting a buffer circuit 23 in the circuit, the voltage rise rate (dv / dt) when the switching transistor is turned off is forcibly reduced, eliminating the high-frequency oscillation source at its source. This makes the change curve of the common output circuit 1 during the switching process smoother, effectively avoiding load voltage overshoot caused by interference and ensuring the purity of the power supply to the tested chip. Because the buffer circuit 23 filters out high-frequency spike interference, the signal fed back to the control circuit 4 by the voltage sensor 5 is more accurate and stable. This prevents subsequent backup power supply closed-loop regulation from being falsely triggered by noise, enabling more accurate voltage compensation and significantly reducing the voltage drop amplitude during the switching process.

[0088] In some possible implementations, the backup power supply circuit 3 includes a second soft-switching circuit, which is connected to the backup BUCK circuit 31 and the second soft-switching circuit, and is electrically connected to the common output circuit 1. The second soft-switching circuit includes a third power switch 32 and a second RCD snubber circuit 33, which are connected in parallel with the third power switch 32. The second RCD snubber circuit 33 may also include a diode and a capacitor, both of which are connected in parallel with the third power switch 32. The backup power supply circuit 3 is largely the same as the main power supply circuit 2, except that in the backup power supply circuit 3, the second power switch 311 is connected to the output of the control component to receive a PWM signal with a variable duty cycle.

[0089] This specification also provides a semiconductor aging test system, including the aforementioned power overcharge energy absorption lossless switching circuit. The main power supply is electrically connected to the main BUCK circuit 21 of the power overcharge energy absorption lossless switching circuit. The backup power supply is electrically connected to the backup BUCK circuit 31 of the power overcharge energy absorption lossless switching circuit. In this specification embodiment, the power overcharge energy absorption lossless switching circuit achieves soft shutdown by setting an RCD absorption circuit in the main power supply circuit 2, and combines it with a voltage sensor 5 for closed-loop feedback control of the backup power supply for voltage compensation, thereby significantly reducing the voltage drop during the switching process.

[0090] This specification describes an embodiment based on an improved particle swarm optimization algorithm to dynamically determine PID parameters. This method employs a dual dynamic adjustment strategy within each control cycle. First, the algorithm's hyperparameters are adaptively adjusted. Based on the optimal fitness trend of the previous three historical cycles, the convergence state of the current system is evaluated, and the inertia weight and learning factor of the particle swarm optimization algorithm are dynamically set accordingly. Second, the control parameters are continuously optimized. Using the updated particle swarm optimization algorithm, combined with a circuit prediction simulation model, an iterative search is performed with the cumulative error over future time cycles as the objective function. Finally, the optimal PID parameter combination is selected for circuit control in the next cycle.

[0091] This specification's embodiments, in response to a self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for a PID controller. These parameters are then globally optimized using a particle swarm optimization algorithm to obtain the optimal PID parameter combination. The PID controller acquires the load voltage of the load in the secondary power supply system at the current moment. Based on the load voltage and the optimal PID parameter combination, simulated voltages of the load are generated at different times within a preset future time period. Based on the optimal PID parameter combination, load voltage, and simulated voltage, the duty cycle of the PWM signal is calculated. The primary power supply is switched to the backup power supply, and the PID controller outputs a PWM signal to the backup power supply according to the duty cycle for power self-healing control. The method provided in this specification's embodiments can eliminate destructive circulating currents and automatically compensate for line voltage drops to maintain a constant bus voltage during the power handover process, achieving truly "seamless and lossless" soft switching, thereby helping to ensure the stable operation of the secondary power supply system.

[0092] In addition, the integrated circuit power supply power fusion self-healing control method provided in the embodiments of this specification includes the following steps: Step 1: Simultaneously monitor the output voltage and output current of each power module.

[0093] A power architecture is applied to multiple power modules connected in parallel to supply power to a single bus. This power architecture includes a power combiner and an electronic device for executing the integrated circuit power supply power fusion self-healing control method provided in this specification. The electronic device includes an integrated circuit power supply power fusion self-healing control computer program. When the computer program is executed by a processor, it implements the integrated circuit power supply power fusion self-healing control method of the embodiments of this specification.

[0094] In this embodiment, the definition of health is elevated from "current remaining power capacity" to "the ability to maintain voltage quality over a future period." This method achieves a truly forward-looking health assessment by predicting the future output performance of each power module.

[0095] This step involves high-sampling-rate real-time data acquisition, and during execution, the output voltage of each power module is continuously and synchronously monitored. and output current A high sampling rate is a prerequisite for capturing subtle degradation features such as ripple and transient response.

[0096] Step 2: Perform output compression and feature extraction on the output voltage sequences of each monitored power module to obtain voltage feature sequences.

[0097] To address the computational challenges posed by high sampling rate data, a sliding window is used for data compression and feature extraction of the output voltage sequence of each power module.

[0098] In one optional embodiment, the output voltage sequence of each monitored power module is compressed and its features are extracted to obtain a voltage feature sequence, which may include the following sub-steps: Sub-step 1: For each power module, use a sliding window to extract the output voltage sequence of the power module to obtain multiple window sub-voltage sequences.

[0099] The sliding window length is the first number of sampling points, and the sliding step size is the second number of sampling points.

[0100] Sub-step 2: For each window sub-voltage sequence, extract two preset features to form the feature vector of the window sub-voltage sequence.

[0101] Among them, two preset characteristics include: average voltage within the window and peak ripple; One feasible way to construct a feature vector is as follows: Let the window length be There are 1 sampling point, with a step size of 1. One sampling point.

[0102] For each window Two core features are extracted to form the feature vector of the window. : Average voltage This reflects the average output voltage level within the window.

[0103]

[0104] Ripple peak value This reflects the stability and dynamic quality of the output voltage within the window.

[0105]

[0106] This process compresses the data volume to its original size. At the same time, it fully preserves key information such as the long-term drift trend and high-frequency ripple characteristics that characterize the performance degradation of the power module.

[0107] Sub-step 3: Combine the feature vectors of each window sub-voltage sequence into the voltage feature sequence of the power module.

[0108] Step 3: For each power module, determine the target feature vector based on the voltage feature sequence and the preset performance degradation prediction model.

[0109] The target feature vector is used to characterize the future voltage state of the power module.

[0110] In practical implementation, a dedicated "performance degradation predictor" can be deployed for each power module. Its core is a "compression-sensing-fusion" end-to-end neural network used to predict the future voltage output characteristics of the power module.

[0111] In the embodiments of this specification, a performance degradation prediction model is constructed based on a liquid neural network fusion architecture. The preset performance degradation prediction model includes a dual-branch perceptron and a liquid neural network.

[0112] In one optional embodiment, for each power module, determining the target feature vector based on the voltage feature sequence and a preset performance degradation prediction model may include the following sub-steps: Sub-step 1: Input the voltage feature sequence into the dual-branch sensing sub-network to obtain the first feature vector and the second feature vector.

[0113] In this optional embodiment, the dual-branch sensing part will transmit the voltage feature sequence. Input the LSTM branch and TCN branch of the dual-branch perceptron to obtain the first feature vector and the second feature vector.

[0114] The LSTM branch is used for long-term trend capture: it consists of a two-layer LSTM network with 64 hidden layers. It captures the slow drift and long-term degradation trend of the output voltage through its gating mechanisms such as forget gates, input gates, and output gates. Its cell state... Its update mechanism gives it long-range memory capabilities.

[0115] The TCN branch is used for short-term dynamic capture: it consists of a three-layer dilated TCN with convolutional kernels. Expansion rate By utilizing its causal convolution and dilated convolution, it can efficiently capture the changing patterns of short-term dynamic characteristics such as ripple abrupt changes and switching noise.

[0116] Sub-step 2: Merge the first feature vector and the second feature vector to obtain the merged feature vector.

[0117] The outputs of LSTM and TCN are concatenated along the feature dimension to obtain a fused feature vector. .

[0118] Sub-step 3: Input the fused feature vector into the liquid neural network to obtain adaptive weights.

[0119] fuse feature vectors The input sequence is fed into a lightweight Liquid Neural Network (LNN) fusion layer. This LNN fusion layer acts as a dynamic arbitrator, generating a pair of adaptive soft weights in real time based on the instantaneous features of the input sequence. The adaptive weights include the first weight of the first feature vector and the second weight of the second feature vector.

[0120] More specifically, the way to obtain adaptive weights by inputting the fused feature vector into the liquid neural network can be as follows: input the fused feature vector into the liquid neural network to generate a compact representation of the global semantics of the fused feature vector; map the compact representation of the global semantics into a two-dimensional vector through a lightweight linear output layer; normalize the two-dimensional vector through the Softmax function to obtain adaptive weights, which are a pair of confidence weights that sum to 1.

[0121] The Softmax function is often used in the output layer of multi-class classification problems. It maps any real vector to a probability distribution, where each element has a value between (0,1) and the sum is 1.

[0122] Sub-step 4: The first and second feature vectors are weighted and fused according to the adaptive weights to obtain the target feature vector.

[0123] The continuous-time dynamics of LNNs are described by the first-order nonlinear ordinary differential equations (ODEs) of the internal neuron states: in: It is a continuous time vector of the hidden states of the LNN, with a dimension of 16. It is a learnable liquid time constant vector, where each element corresponds to a time scale of neuron state decay, which is the key to LNN adapting to dynamics at different frequencies; It is the fused input feature at time t; and These are the input weight matrix and the recursive weight matrix, respectively. It is the bias vector; It is a non-linear activation function (such as...) ).

[0124] In practical deployments, the aforementioned continuous-time ODE needs to be discretized and solved using numerical methods. The forward Euler method is employed, with sampling intervals... Approximating ODE:

[0125] in, Representing element-wise multiplication, this discrete form demonstrates how LNNs utilize their recursive connections and learnable time constants to form a dynamic system with memory.

[0126] From the perspective of model fusion functionality, LNN is a dynamic arbiter embedded after LSTM and TCN. Its design aims to utilize very few additional parameters, with the total number of trainable parameters being less than five percent of the entire model, to continuously output a pair of soft weights that smoothly change with the characteristics of the input voltage sequence in high sampling rate scenarios. This allows for adaptive scheduling of the advantageous regions of the two heterogeneous networks without the need for manually setting thresholds or prior rules.

[0127] Specifically, the controller first uses a voltage segment normalized by a sliding window as the driving signal to drive the internal liquid units, i.e., the liquid neural network, to unfold state evolution along the time axis. Unlike conventional recurrent networks, the update of the hidden states within the liquid units follows the first-order dynamic equations determined by the input, the previous state, and the learnable time constant. The transient response characteristics of this dynamic system can be analyzed by its linearized approximation under small-signal perturbations: Assume the system is at equilibrium. Nearby, Then we have:

[0128] in, It is a function The value of the Jacobian matrix at the equilibrium point. The time response characteristics of the system are determined by the matrix. The eigenvalues ​​determine the learnable features. This allows the network to adaptively adjust its dynamic convergence speed.

[0129] Its numerical integration process can effectively fuse the transient information of local ripples with the slow drift caused by long-term degradation within a limited receptive field, forming a compact representation of the global semantics of the input sequence. Subsequently, this representation is mapped to a two-dimensional logits vector through a lightweight linear output layer. :

[0130] Then, the Softmax function is used to... Normalization, compressed into a pair of confidence weights that sum to 1 :

[0131] The final weighted fusion prediction output, i.e., the target feature vector, is calculated by the following formula:

[0132] in, and These are the prediction outputs for the LSTM branch and the TCN branch, respectively.

[0133] First weight That is, the first weight directly affects the degradation trend estimation of the LSTM output, and the second weight... The second weight is applied to the local ripple estimation of the TCN output. Since the weights are recalculated for each batch and each time window, the controller can rapidly increase the sensing sensitivity of the TCN branch when the input waveform undergoes abrupt changes. Increase, and amplify the long-term memory advantage of LSTM branches when the degradation slope tends to flatten. This increases the accuracy of predictions, thereby enabling seamless transitions across time scales in continuous predicted trajectories.

[0134] During the end-to-end training phase, LNN shares the same mean squared error (MSE) loss function with LSTM and TCN. :

[0135] No additional regularization terms or complex multi-task objectives are required. During backpropagation, the time constant within the liquid cell... Mapping weights And other parameters synchronously receive the gradient signal from the final prediction error. This ensures that the fusion strategy and the backbone network parameters converge to the global optimum.

[0136] The final output of the pre-defined performance degradation prediction model is a future time window. Predicted voltage characteristics within .

[0137] Step 4: Calculate the health of the power module based on the target feature vector, rated voltage, and maximum allowable ripple.

[0138] In the embodiments described in this specification, the health of the power module is no longer based on its current maximum power, but is calculated based on the deviation between its predicted future voltage performance and its ideal performance.

[0139] In one optional embodiment, the health of the power module can be calculated based on the target feature vector, the rated voltage, and the maximum allowable ripple by: calculating a first deviation between the instantaneous voltage value of the target feature vector and the rated voltage; calculating a second deviation between the peak ripple value of the target feature vector and the maximum allowable ripple; and mapping the first deviation and the second deviation to the health of the power module using a normalization function.

[0140] In the specific implementation process, we can first define an ideal reference vector, then calculate the multidimensional deviation, and finally calculate the overall health score.

[0141] Define an ideal reference vector: Set an ideal power module in Reference feature vector at time step ,in, Rated voltage, This represents the maximum allowable ripple of the system.

[0142] Calculating multidimensional deviation, which involves calculating the deviation between the predicted value and the ideal value, can be achieved using the following formula: Voltage amplitude deviation:

[0143] Ripple performance deviation:

[0144] Calculate overall health score: Map deviations to health scores using a normalization function. For example, using an exponential decay form:

[0145] in, and It is a weighting coefficient used to balance the contributions of voltage accuracy and ripple stability to health.

[0146] The above process yields a future-oriented, dynamic health assessment result for a power module. This health status not only reflects the current state of the power module, but also predicts its performance reliability in the near future, providing an advanced and accurate basis for subsequent power deficit prediction and self-healing reconfiguration.

[0147] By repeating or paralleling the above health determination process, the health status of each online power module on the common bus can be determined.

[0148] Step 5: Calculate the predicted power deficit value of the bus based on the health status of each power module and the output current sequence of each power module.

[0149] This step is the process of predicting the future power deficit of the system, which mainly includes the following steps: feature extraction and compression of the high sampling rate sequence of load current, prediction of available power of the system, and prediction of power deficit of the bus.

[0150] Feature extraction and compression of high-sampling-rate load current sequences: For the acquired high-sampling-rate load current sequences That is, the output current sequence of each power module is processed by a sliding window of fixed length for dimensionality reduction. Let the window length be... There are 1 sampling point, with a step size of 1. Each sampling point. For each window The data within will no longer be used in its entirety. Instead of extracting one original data point, two key features are extracted.

[0151] System available power prediction: Based on the health status of each power module calculated in step four, predict the total power that all online power modules in the system can provide at the future time Δt, i.e., P_available_pred(t+Δt)=Σ(Hi(t)). In this embodiment, it is assumed that the health status changes negligibly within a very short prediction window Δt, and P_rated_i is the average load current within the window.

[0152] Bus power deficit prediction: Calculate the predicted power deficit ΔP_pred(t+Δt) at the future time Δt = P_load_pred(t+Δt) - P_available_pred(t+Δt).

[0153] P_load_pred(t+Δt) is calculated from the predicted load current I_load_pred(t+Δt) and the rated bus voltage V_bus_nominal.

[0154] Step 6: Predict the state parameters of the bus voltage based on the predicted power deficit value.

[0155] The state parameters of the bus voltage include: instantaneous voltage value and peak ripple.

[0156] This step is the direct basis for triggering the compensation action. In actual implementation, a dynamic model of the system's bus voltage can be established. This model takes the power deficit ΔP as input and the bus voltage V_bus and its ripple peak value ΔV_ripple as output. The predicted power deficit ΔP_pred(t+Δt) at time Δt obtained in step five is input into this dynamic model to predict the instantaneous value of the bus voltage V_bus_pred(t+Δt) and the ripple peak value V_ripple_pred(t+Δt) at the future time Δt.

[0157] Specifically, the model can be simplified to: V_bus_pred(t+Δt)=V_bus_nominal-Z_sys (ΔP_pred(t+Δt) / V_bus_nominal) Where Z_sys is the system equivalent impedance as seen from the output of the power module. Meanwhile, ripple prediction can take into account switching noise and loop response overshoot caused by power surges.

[0158] Step 7: Compare the status parameters of the bus voltage with the preset voltage safety threshold, generate a power combining command based on the comparison result and send it to the power combiner so that the power combiner can perform power compensation according to the power combining command.

[0159] In one optional embodiment, the method for comparing the state parameters of the bus voltage with a preset voltage safety threshold and generating a power combining command based on the comparison result can be as follows: First, compare the instantaneous voltage value with the warning voltage threshold; then compare the ripple peak value with the ripple warning threshold. The preset voltage safety thresholds include: warning voltage threshold and ripple warning threshold; Secondly, if the instantaneous voltage value is less than the warning voltage threshold, or the peak ripple value is greater than the ripple warning threshold, a power combining command is generated. The power combining command carries the injected power value.

[0160] In the specific implementation process, a warning voltage threshold V_warn higher than the traditional undervoltage protection point can be set, for example, 0.5% lower than the rated voltage; a ripple warning threshold ΔV_warn can be set; the judgment logic is set as follows: IF(V_bus_pred(t+Δt)<V_warn)OR(V_ripple_pred(t+Δt)> If ΔV_warn is found, a power combining instruction is immediately generated, which contains the power value to be injected, P_inject_cmd=ΔP_pred(t+Δt); otherwise, the power combiner remains in standby or outputs zero.

[0161] As a highly dynamic and controllable voltage or current source, the power combiner immediately begins to respond upon receiving a power combining command. Its goal is to reach the required power injection level at a future time t+Δt, thereby offsetting the predicted power deficit that occurs at that time.

[0162] In this way, the injection action of the power combiner is synchronized or compensated in advance with the occurrence of system power deficit, thereby avoiding bus voltage drop and ripple increase caused by control delay.

[0163] After the power combiner begins injecting power, the system enters a closed-loop phase. Power is replenished by monitoring the actual bus voltage and comparing it with the rated value until the faulty or degraded power module is replaced in the field, and the system returns to normal operation.

[0164] The integrated circuit power supply power fusion self-healing control method disclosed in the embodiments of this specification synchronously monitors the output voltage and output current of each power module; performs output compression and feature extraction on the monitored output voltage sequence of each power module to obtain a voltage feature sequence; for each power module, a target feature vector is determined based on the voltage feature sequence and a preset performance degradation prediction model; the health of the power module is calculated based on the target feature vector, rated voltage, and maximum allowable ripple; the predicted power deficit value of the bus is calculated based on the health of each power module and the output current sequence of each power module; the state parameters of the bus voltage are predicted based on the predicted power deficit value; the state parameters of the bus voltage are compared with a preset voltage safety threshold, and a power combining command is generated based on the comparison result and sent to the power combiner so that the power combiner performs power compensation according to the power combining command. This solution, firstly, transforms compensation from reactive remediation to proactive prevention by predicting future voltage conditions, fundamentally eliminating the voltage drop problem caused by control delays in existing technologies, requiring only ultra-low bus transient disturbances; secondly, the load is completely unaware of any voltage quality degradation during the entire self-healing reconfiguration process, enabling uninterrupted power supply; and thirdly, since the system can tolerate power module performance degradation and maintain stability through predictive compensation, it is not necessary to disconnect the power module when it first shows slight degradation, thus improving the remaining lifespan utilization of the power module.

[0165] Corresponding to the above method embodiments, this specification also provides an embodiment of a power supply overcharge energy absorption lossless switching self-healing control device. Figure 3 A schematic diagram of a power supply overshoot energy absorption lossless switching self-healing control device according to one embodiment of this specification is shown. Figure 3 As shown, the device includes: The optimization module 302 is configured to, in response to the self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for the PID controller, and perform global optimization on the at least two sets of parameters through the particle swarm optimization algorithm to obtain the optimized optimal PID parameter combination. The acquisition module 304 is configured to acquire the load voltage of the load in the secondary power system at the current moment through a PID controller; The simulation module 306 is configured to simulate and generate the simulated voltage of the load at different times within a preset future time period based on the load voltage and the optimal PID parameter combination. The calculation module 308 is configured to calculate the duty cycle of the PWM signal based on the optimal PID parameter combination, the load voltage, and the simulated voltage. The control module 310 is configured to switch the main power supply to the backup power supply and output a PWM signal to the backup power supply according to the duty cycle through the PID controller to perform power self-healing control.

[0166] Optionally, the optimization module 302 is further configured to: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

[0167] Optionally, the power overcharge energy absorption lossless switching self-healing control device further includes a processing module configured to: Read the optimal fitness function values ​​corresponding to the three historical time periods prior to the preset future time period; Based on the optimal fitness function values ​​corresponding to the three historical time periods, calculate the average rate of change of the optimal fitness function values; Based on the average rate of change, the inertia weight, individual learning factor, and social learning factor corresponding to the preset future time period are dynamically adjusted.

[0168] Optionally, the processing module is further configured to: Based on the optimal fitness values ​​corresponding to the first historical time period and the second historical time period, calculate the first relative rate of change of the optimal fitness function value; Based on the optimal fitness values ​​corresponding to the first and third historical time periods, the second relative rate of change of the optimal fitness function value is calculated. Calculate the mean of the first relative rate of change and the second relative rate of change, and determine the mean as the average rate of change of the optimal fitness function value; Wherein, the first historical time period is adjacent to the second historical time period, the second historical time period is adjacent to the first historical time period and the third historical time period, and the third historical time period is adjacent to the second historical time period and the preset future time period.

[0169] Optionally, the duty cycle of the PWM signal is calculated as follows:

[0170] in, , , These are the three parameters in the optimal PID parameter combination. Duty cycle, Indicates rated voltage With load voltage The deviation.

[0171] Optionally, the expression for the fitness function is:

[0172] Where T is the length of the preset future time period; t is a continuous time variable; The rated voltage of the load; For based on The simulated voltage output curve of the load obtained from parameter simulation. This refers to each set of PID parameters included in the optimization process.

[0173] The above is a schematic scheme of a power overcharge energy absorption lossless switching self-healing control device according to this embodiment. It should be noted that the technical solution of this power overcharge energy absorption lossless switching self-healing control device belongs to the same concept as the technical solution of the aforementioned power overcharge energy absorption lossless switching self-healing control method. Details not described in detail in the technical solution of the power overcharge energy absorption lossless switching self-healing control device can be found in the description of the technical solution of the aforementioned power overcharge energy absorption lossless switching self-healing control method.

[0174] Figure 4 A structural block diagram of a computing device 400 according to one embodiment of this specification is shown. The components of the computing device 400 include, but are not limited to, a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and a database 450 is used to store data.

[0175] The computing device 400 also includes an access device 440, which enables the computing device 400 to communicate via one or more networks 460. Examples of these networks include a Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the Internet. The access device 440 may include one or more of any type of wired or wireless network interface (e.g., a Network Interface Card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) interface, a Wi-MAX interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0176] In one embodiment of this specification, the aforementioned components of the computing device 400 and Figure 4 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 4 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0177] The computing device 400 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or PCs. The computing device 400 can also be a mobile or stationary server.

[0178] The processor 420 is used to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-mentioned power overcharge energy absorption lossless switching self-healing control method.

[0179] The above is a schematic representation of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the aforementioned power overshoot energy absorption lossless switching self-healing control method belong to the same concept. Details not described in detail in the technical solution of the computing device can be found in the description of the aforementioned power overshoot energy absorption lossless switching self-healing control method.

[0180] An embodiment of this specification also provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the above-described power overcharge energy absorption lossless switching self-healing control method.

[0181] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium belongs to the same concept as the technical solution of the above-described power overcharge energy absorption lossless switching self-healing control method. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the above-described power overcharge energy absorption lossless switching self-healing control method.

[0182] An embodiment of this specification also provides a computer program, wherein when the computer program is executed in a computer, the computer is instructed to perform the steps of the above-described power overcharge energy absorption lossless switching self-healing control method.

[0183] The above is an illustrative scheme of a computer program according to this embodiment. It should be noted that the technical solution of this computer program and the technical solution of the above-described power overcharge energy absorption lossless switching self-healing control method belong to the same concept. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solution of the above-described power overcharge energy absorption lossless switching self-healing control method.

[0184] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0185] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0186] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0187] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0188] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A self-healing control method for power supply overshoot energy absorption and lossless switching, comprising: In response to the self-healing switching command issued to the secondary power supply, at least two sets of PID parameters of the PID controller are initialized and generated. The at least two sets of parameters are then globally optimized using a particle swarm optimization algorithm to obtain the optimal combination of PID parameters. The load voltage of the load in the secondary power supply system at the current moment is collected by a PID controller. Based on the load voltage and the optimal PID parameter combination, the simulated voltage of the load at different times within a preset future time period is generated. Based on the optimal PID parameter combination, the load voltage, and the simulated voltage, the duty cycle of the PWM signal is calculated; The main power supply is switched to the backup power supply, and a PWM signal is output to the backup power supply according to the duty cycle through the PID controller to perform power self-healing control.

2. The power supply overshoot energy absorption lossless switching self-healing control method according to claim 1, wherein the step of globally optimizing the at least two sets of parameters using a particle swarm optimization algorithm to obtain the optimized optimal PID parameter combination includes: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

3. The power overshoot energy absorption lossless switching self-healing control method according to claim 1 further includes: Read the optimal fitness function values ​​corresponding to the three historical time periods prior to the preset future time period; Based on the optimal fitness function values ​​corresponding to the three historical time periods, calculate the average rate of change of the optimal fitness function values; Based on the average rate of change, the inertia weight, individual learning factor, and social learning factor corresponding to the preset future time period are dynamically adjusted.

4. The power supply overshoot energy absorption lossless switching self-healing control method according to claim 3, wherein calculating the average rate of change of the optimal fitness function value based on the optimal fitness function values ​​corresponding to the three historical time periods includes: Based on the optimal fitness values ​​corresponding to the first historical time period and the second historical time period, calculate the first relative rate of change of the optimal fitness function value; Based on the optimal fitness values ​​corresponding to the first and third historical time periods, the second relative rate of change of the optimal fitness function value is calculated. Calculate the mean of the first relative rate of change and the second relative rate of change, and determine the mean as the average rate of change of the optimal fitness function value; Wherein, the first historical time period is adjacent to the second historical time period, the second historical time period is adjacent to the first historical time period and the third historical time period, and the third historical time period is adjacent to the second historical time period and the preset future time period.

5. The power supply overshoot absorption lossless switching self-healing control method according to claim 1, wherein the duty cycle of the PWM signal is calculated as follows: in, , , These are the three parameters in the optimal PID parameter combination. Duty cycle, Indicates rated voltage With load voltage The deviation.

6. The power overshoot energy absorption lossless switching self-healing control method according to claim 2, wherein the expression for the fitness function is: in, T represents the length of the preset future time period; t represents a continuous time variable. The rated voltage of the load; Based on The simulated voltage output curve of the load obtained from parameter simulation. This refers to each set of PID parameters included in the optimization process.

7. A power supply overcharge energy absorption lossless switching self-healing control device, comprising: The optimization module is configured to, in response to the self-healing switching command issued to the secondary power supply, initialize and generate at least two sets of PID parameters for the PID controller, and perform global optimization on the at least two sets of parameters using the particle swarm optimization algorithm to obtain the optimal combination of PID parameters after optimization. The acquisition module is configured to acquire the load voltage of the load in the secondary power system at the current moment through a PID controller; The simulation module is configured to generate simulated voltages of the load at different times within a preset future time period based on the load voltage and the optimal PID parameter combination. The calculation module is configured to calculate the duty cycle of the PWM signal based on the optimal PID parameter combination, the load voltage, and the simulated voltage. The control module is configured to switch the main power supply to the backup power supply and output a PWM signal to the backup power supply according to the duty cycle through the PID controller to perform power self-healing control.

8. The power overshoot energy absorption lossless switching self-healing control device according to claim 7, wherein the optimization module is further configured as follows: A fitness function is constructed based on the rated voltage of the load in the secondary power system and the simulated voltage of the load at different times within a preset future time period. The fitness function is used as the objective function, and the objective function is solved based on each set of PID parameters included in the optimization process to obtain multiple corresponding solution results; The PID parameter combination corresponding to the solution with the smallest value is determined as the optimal PID parameter combination.

9. A computing device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the power overcharge energy absorption lossless switching self-healing control method according to any one of claims 1 to 6.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the power supply overcharge absorption lossless switching self-healing control method according to any one of claims 1 to 6.

Images