CN-122026712-A - Multi-path PFC harmonic compensation control system and method based on cloud edge cooperation

CN122026712ACN 122026712 ACN122026712 ACN 122026712ACN-122026712-A

Abstract

The cloud-edge-collaboration-based multi-path PFC harmonic compensation control system and method comprises a cloud intelligent optimization layer, an agent model and edge equipment, wherein the cloud intelligent optimization layer and the agent model are used for transmitting strategy parameters to the edge equipment, the cloud intelligent optimization layer comprises a cloud large model, a strategy parameter storage module and a strategy parameter transmitting module, the edge equipment comprises an air conditioning unit, a bottom embedded control layer, a hardware control layer and an edge control layer, the cloud large model comprises an analyzer agent, a strategy agent and a coordinator agent, the cloud large model interacts with the agent model through data transmission, the strategy agent is respectively connected with the analyzer agent and the strategy agent through data transmission, and the cloud large model is connected with the strategy parameter storage module and the strategy parameter transmitting module through data transmission. The harmonic suppression problem of the PFC circuit in the complex power supply environment is solved.

Inventors

HE XIAOLIN
ZHU ZHIQIANG
FAN XIAOKUN
LIU ZHIHUI
FAN SHENGQUAN
Tang Runzhong

Assignees

珠海格力电器股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. A multipath PFC harmonic compensation control system based on cloud edge cooperation is characterized by comprising: the cloud intelligent optimization layer and the intelligent body model are used for sending policy parameters to the edge equipment; The cloud intelligent optimization layer comprises a cloud large model, a strategy parameter storage module and a strategy parameter issuing module, wherein the edge equipment comprises an air conditioning unit, a bottom embedded control layer, a hardware control layer and an edge control layer, the cloud large model comprises an analyzer intelligent agent, a strategy intelligent agent and a coordinator intelligent agent, and the air conditioning unit comprises a PFC circuit; The cloud large model is connected with the strategy parameter storage module and the strategy parameter issuing module through data transmission, and the bottom embedded control layer, the hardware control layer and the edge control layer are all connected with air conditioning unit data transmission; The cloud large model is used for classifying and analyzing various problems which cannot be processed locally and outputting corresponding control strategies, and the intelligent body model is used for controlling the normalization of the PFC circuit of the air conditioning unit and completing real-time feedforward adjustment compensation according to the existing local strategies.
2. The cloud-edge collaboration-based multipath PFC harmonic compensation control system of claim 1, wherein: the analyst agent is the core of the cloud large model and is used for diagnosing the state of an air conditioning unit and mining excellent strategies; The strategic agent is used for deducting and generating a better strategy in the virtual environment; the coordinator agent is used for compressing and refining the complex strategy into a parameter packet which can be issued.
3. The cloud-edge collaboration-based multipath PFC harmonic compensation control system of claim 1, wherein: the analyzer agent comprises a multi-layer long-term and short-term memory network and adopts an encoder-decoder architecture; The encoder uses a multi-layer long-short-term memory network for learning the long-term dependency relationship of the operation time sequence data uploaded from the edge equipment, and the decoder is a deep neural network and is responsible for carrying out fault classification or state prediction according to the encoded information.
4. The cloud-edge collaboration-based multipath PFC harmonic compensation control system of claim 1, wherein: The strategic agent includes a flexible actor-evaluator reinforcement learning algorithm for receiving the output of the analyst agent and generating an optimized feed forward control strategy.
5. The cloud-edge collaboration-based multipath PFC harmonic compensation control system of claim 1, wherein: the coordinator agent adopts knowledge distillation technology to compress and refine the complex strategy generated by the strategic agent into a lightweight parameter update set and takes charge of issuing the lightweight parameter update set to each edge device.
6. An air conditioner, comprising: The cloud edge collaboration-based multi-path PFC harmonic compensation control system of any of claims 1-5.
7. The harmonic suppression method based on the cloud edge collaboration-based multipath PFC harmonic compensation control system as claimed in any one of claims 1 to 5, comprising the following steps: Step 1, initializing a cloud large model and an intelligent body model; Step 2, the intelligent body model receives sensor data and judges whether an abnormal operation state is met, if not, the step 3 is carried out, and if yes, the step 4 is carried out; Step 3, conventionally controlling the intelligent body model according to the pre-trained control strategy parameters, generating PWM compensation instructions, and transferring to step 6; step 4, the intelligent body model performs feedforward optimization according to the existing strategy, and stores the parameters of the optimal control strategy to upload control data for cloud analysis; Step 5, based on the better control strategy parameters stored in the step 4, the air conditioning unit feeds back abnormal control state data to the cloud large model, the cloud large model analyzes data characteristics, simulates fault scenes, calculates theoretical better control strategy parameters, and sends new strategy parameters to the air conditioning unit, and the step 4 is returned; Step 6, adding PWM duty cycle compensation instructions combined with the original conventional control algorithm; and 7, the bottom embedded control layer executes the instruction in the step 6, evaluates and compares the control effect in the intelligent body model, and returns the compared optimal control strategy parameters to the step 4 for storage.
8. The harmonic suppression method as recited in claim 7, wherein: in step 4, storing the better control strategy parameters includes: step 4.1, setting an existing strategy pool by the cloud large model, wherein the strategy pool comprises a strategy 1, a strategy 2, a strategy N; Step 4.2, the cloud large model performs an internal theoretical model test and/or operation data analysis fed back by each air conditioning unit, and then performs a test on new strategy parameter adjustment; step 4.3, issuing the new strategy parameters after the test to a part of local air conditioning units for control, and completing parameter adjustment; Step 4.4, each air conditioning unit adjusts strategy parameters according to the power level of the air conditioning unit to be changed into actual control parameters suitable for the air conditioning unit; Step 4.5, feeding back result data after the local air conditioning unit completes the operation task; step 4.6, the cloud large model judges whether the new strategy parameters have better control effect or not by analyzing the data fed back by the operation of the local air conditioning unit, and returns to the step 4.2; and 4.7, storing policy parameters with better control effect, and updating a policy pool of the cloud large model, wherein the updated policy pool comprises policies 1', 2',and N '.
9. A terminal comprises a processor and a storage medium, and is characterized in that: The storage medium is used for storing instructions; the processor being operative according to the instructions to perform the steps of the method according to any one of claims 7-8.
10. Computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the method according to any of claims 7-8.

Description

Multi-path PFC harmonic compensation control system and method based on cloud edge cooperation Technical Field The invention belongs to the technical field of rectifier circuits, and particularly relates to a multipath PFC harmonic compensation control system and method based on cloud edge cooperation. Background In the running process of the existing air conditioner power supply system, circuit harmonic waves are generated due to the fact that a system power supply is interfered by the outside frequently, so that overcurrent or abnormal heating of the system occurs, and the service life of devices is influenced. At present, a totem-pole bridgeless PFC (Power Factor Correction) circuit power supply mode is generally adopted on an external machine power supply mode of an air conditioner, and the power supply mode has the advantages of simplified circuit structure, convenient control and higher electric energy utilization efficiency, but has the defects that when harmonic waves exist on the electric energy at the input side, the adverse effect of the harmonic waves on the output of the circuit cannot be eliminated, so that a good control strategy is needed to solve the problem, the prior art adopts a voltage-current double-closed loop feedforward control strategy on the basis of the prior closed loop control, but the strategy is only suitable for various complex-change harmonic states of an actual power grid under the condition of specific convention, and the harmonic optimization effect of the control strategy is limited. The prior art discloses an air conditioner operation control method and device and an air conditioner, wherein the method and device comprise the steps of obtaining the working state of a PFC module of the air conditioner, enabling the working state to be in a starting state or a closing state, sending a first pipeline control signal to enable refrigerant fluid to be communicated to a radiator of the air conditioner from a compressor of the air conditioner through a first refrigerant pipeline when the working state is in the starting state, and sending a second pipeline control signal to enable the refrigerant fluid to be communicated to the radiator from the compressor through a second refrigerant pipeline or the first refrigerant pipeline when the working state is in the closing state. However, the prior art has the defects that the existing general cloud edge end computing architecture has obvious defects when being applied to a high-performance totem pole PFC circuit: 1) The edge side AI model is generalized, is not optimized for microsecond-level response requirements of the switching frequency of the two-way PFC harmonic power supply, and has poor instantaneity; 2) The cloud model is disjointed with the PFC physical principle, the processed data characteristics and control instructions of the cloud model cannot be deeply related with a circuit harmonic generation mechanism, and the treatment effect is limited; 3) Policy migration capability is weak, and effective technical means are lacked to realize safe and efficient multiplexing of control policies among PFC circuits with different power levels. Therefore, there is an urgent need to provide a system and a method for controlling multipath PFC harmonic compensation based on cloud-edge cooperation. Disclosure of Invention In order to solve the defects in the prior art, the invention discloses a multipath PFC harmonic compensation control system and method based on cloud edge cooperation. The core is to construct a cooperative system consisting of cloud layered cognitive agents and an edge-specific lightweight AI model. The PFC harmonic optimization feedforward control method is used for solving the problems that in the prior art, a blank interval on PFC harmonic management capability is 1, conventional PFC harmonic optimization control capability is limited and limited by bandwidth, 2, an existing PFC harmonic optimization feedforward control strategy cannot meet system stability control under an unstable input state, and 3, an existing PFC control strategy requires technicians to adaptively adjust input and output electric energy requirements of each product, so that short plates which cannot be mutually referred by PFC control strategies among different products are solved. The invention adopts the following technical scheme. The first aspect of the invention provides a multipath PFC harmonic compensation control system based on cloud edge cooperation, which comprises the following components: the cloud intelligent optimization layer and the intelligent body model are used for sending policy parameters to the edge equipment; The cloud intelligent optimization layer comprises a cloud large model, a strategy parameter storage module and a strategy parameter issuing module, wherein the edge equipment comprises an air conditioning unit, a bottom embedded control layer, a hardware control layer and an edge control layer, the cloud large