CN-121997548-A - Digital twinning-based water-cooled frequency converter high-cold working condition performance optimization and service life prediction method

CN121997548ACN 121997548 ACN121997548 ACN 121997548ACN-121997548-A

Abstract

The invention relates to the field of life prediction, and discloses a digital twin-based water-cooling frequency converter high and cold working condition performance optimization and life prediction method which is used for providing technical support for stable operation of wind power equipment in an extreme environment. The method comprises the steps of generating accurate heat source data matched with the current working condition based on iterative calculation of real-time electric parameters and initial junction temperature, eliminating static model errors, inputting the heat source data and low-temperature viscosity characteristics of cooling liquid into a simulation model, outputting junction temperature, temperature distribution, key part stress and other multi-physical-field state data, combining a stress change process and a damage accumulation model, quantifying key part accumulated damage, dynamically adjusting the rotating speed of a cooling water pump according to the junction temperature and the environment temperature, balancing anti-freezing protection and heat dissipation requirements, fusing damage data with future load prediction, and outputting a residual service life quantification result. The invention obviously improves the reliability, energy efficiency and economy of equipment under the severe cold working condition, and provides technical support for power electronic equipment in the fields of wind power, rail transit and the like.

Inventors

SONG XIANGBIN
YU RUNQUAN
XU HONGWEI
LI SHUHONG
YU YINGCHUN
MA XIAOBO
YANG JINGXIN

Assignees

河北大唐国际丰宁风电有限责任公司

Dates

Publication Date: 20260508
Application Date: 20251223

Claims (9)

1. A digital twinning-based water-cooled frequency converter high and cold working condition performance optimization and life prediction method is characterized by comprising the following steps: performing iterative calculation based on the real-time operation electric parameters of the water-cooled frequency converter and the initial junction temperature data to obtain accurate heat source data matched with the current junction temperature state; Inputting the accurate heat source data and the low-temperature viscosity characteristic data of the cooling liquid into a simulation model, and outputting multi-physical-field state data after operation, wherein the multi-physical-field state data comprise junction temperature data of each power module, temperature distribution data of a radiator and key part stress data; inputting key part stress data in the multi-physical field state data and a change process thereof into a damage accumulation model to generate key part accumulated damage quantification data; Outputting an optimized control instruction for adjusting the rotating speed of the cooling water pump in real time according to junction temperature data and environmental temperature data in the multi-physical-field state data; and outputting the residual service life prediction data of the water-cooled frequency converter based on the accumulated damage quantification data of the key part and the prediction data of the future load working condition.
2. The digital twinning-based water-cooled frequency converter high and cold condition performance optimization and life prediction method of claim 1, comprising the steps of: generating first loss distribution data according to the real-time operation electric parameters and the first junction temperature data; Taking the first loss distribution data as an input condition, and executing thermal simulation of the water-cooled frequency converter and a heat dissipation system thereof to obtain second junction temperature data; Generating loss correction coefficient data according to the difference between the second junction temperature data and the first junction temperature data; Correcting the first loss distribution data by using the loss correction coefficient data to generate second loss distribution data; and updating the first loss distribution data by using the second loss distribution data, updating the first junction temperature data by using the second junction temperature data, returning to execute thermal simulation until the junction temperature data variation obtained by two adjacent thermal simulations is smaller than a preset threshold value, and taking the loss distribution data finally obtained by the current iteration as accurate heat source data.
3. The digital twinning-based water-cooled frequency converter high and cold condition performance optimization and life prediction method of claim 2, comprising the steps of: Establishing dynamic fluid network resistance parameters based on the low-temperature viscosity characteristic data of the cooling liquid and the geometric structure data of a water cooling system pipeline; Inputting the accurate heat source data and the dynamic fluid network resistance parameters into a fluid and heat transfer joint simulation program to obtain flow distribution data of each parallel branch in the whole cooling loop and flow velocity distribution data of each micro-channel in the radiator; Based on the flow distribution data, the flow velocity distribution data and the accurate heat source data, transient heat transfer calculation is carried out to obtain transient junction temperature data of each power chip and temperature field distribution data of a radiator substrate and a cold plate; based on the temperature field distribution data and the time-dependent change history thereof, the thermal expansion coefficient data of each layer of packaging material in the power module are combined to obtain stress strain data of the key welding layer and the connecting part, which are generated by mismatch of thermal expansion.
4. The digital twinning-based water-cooled frequency converter high and cold condition performance optimization and life prediction method as claimed in claim 3, comprising the steps of: Processing the stress data of the key part, and obtaining first-class stress intensity factor data based on preset defect characteristic size data; According to the temperature distribution data in the multi-physical-field state data, material fatigue strength data and material fracture resistance data corresponding to the current temperature state are obtained; Calculating to obtain damage increment data of the current time step based on the change rate of the stress data of the key part, the first type stress intensity factor data, the material fatigue intensity data and the material fracture resistance data; And accumulating the damage increment data generated by each time step in the operation time process to obtain an accumulated damage value from the initial operation time to the current time, and outputting the accumulated damage quantification data as the accumulated damage quantification data of the key part.
5. The method for optimizing high and cold working condition performance and predicting service life of the water-cooled frequency converter based on digital twinning according to claim 4, wherein local temperature data corresponding to the stress data of the key part are obtained, and brittle risk factor data are generated according to prestored low-temperature brittle transition temperature data of materials; According to the brittle risk factor data, dynamically adjusting the proportionality coefficient data used for representing the relative weight of a fatigue damage mechanism and a brittle fracture mechanism in a damage evolution equation; And recalculating damage increment data of the current time step by using the adjusted proportionality coefficient data, the change rate of the stress data of the key part, the first type stress intensity factor data, the material fatigue intensity data and the material fracture resistance data, and updating accumulated damage quantification data of the key part according to the damage increment data.
6. The method for optimizing and predicting the service life of the water-cooled frequency converter based on digital twinning according to claim 4, wherein the calculating and outputting the optimizing control command for adjusting the rotation speed of the cooling water pump in real time according to the junction temperature data and the environmental temperature data in the multi-physical-field state data comprises the following steps: calculating the minimum safe circulation flow required for preventing the cooling liquid from phase-change solidification in the pipeline according to the environmental temperature data and the physical attribute parameters of the cooling liquid, and generating antifreezing flow threshold data; Analyzing the discrete degree and the highest value of the junction temperature data of each power module in the multi-physical-field state data, and calculating theoretical heat dissipation flow required for maintaining all the power modules at safe and uniform temperature levels by combining a preset junction temperature safety limit value to generate target heat dissipation flow data; Establishing a decision function which contains the antifreezing flow threshold data as a constraint condition and takes approaching the target heat dissipation flow data as an optimization target, and solving the function to obtain global optimal total flow demand data of the system under the current working condition; And mapping according to the global optimal total flow demand data and characteristic curve data of the water pump to obtain a corresponding target water pump rotating speed, and generating an optimal control instruction for driving the speed regulation of the physical water pump.
7. The method for optimizing high and cold working condition performance and predicting service life of a water-cooled frequency converter based on digital twinning according to claim 6, wherein the environmental temperature data is monitored in real time, and a system preheating start flag signal is generated when the environmental temperature data is lower than a preset extremely low temperature threshold value; According to the system preheating starting sign signal and rheological property data of the cooling liquid at extremely low temperature, calculating the safe preheating flow required for preventing the water pump from overload and ensuring the flow of the cooling liquid in the preheating stage, and generating a water pump rotating speed instruction in the preheating stage; executing the water pump rotating speed instruction in the preheating stage, and continuously monitoring the average temperature data of the main loop of the cooling liquid until the average temperature data reaches a preset low-temperature safe operation threshold; and when the average temperature data reaches the low-temperature safe operation threshold, the system preheating start flag signal is withdrawn, and the optimized control instruction is generated by switching.
8. The method for optimizing and predicting service life of a water-cooled frequency converter based on digital twinning according to claim 6, wherein calculating and outputting residual service life prediction data of the water-cooled frequency converter based on the accumulated damage quantification data of the key part and the prediction data of the future load condition comprises: generating virtual wind speed time series data based on long-term weather statistical characteristics of a target wind field; according to the power characteristic curve of the wind driven generator, converting the virtual wind speed time series data into virtual long-term power load series data corresponding to the water-cooling frequency converter; Inputting the virtual long-term power load sequence data into a twin model to obtain thermal stress sequence data experienced by a key part under the virtual load; inputting the thermal stress sequence data into a damage accumulation model, executing damage increment calculation and accumulation processes, and simulating to obtain damage accumulation process data under the future virtual running condition from the current moment; And taking the accumulated damage quantification data of the key part as an initial state, superposing the accumulated damage quantification data with the damage accumulation process data until the accumulated total amount reaches a preset failure damage threshold value, recording the virtual running time from the current moment to the time when the accumulated total amount reaches the threshold value, and outputting residual service life prediction data.
9. The digital twinning-based water-cooled frequency converter high and cold condition performance optimization and life prediction method of claim 1, further comprising: arranging a temperature sensor at a designated position of a physical water-cooling frequency converter, and acquiring actual temperature measurement data of key monitoring points; Comparing the theoretical temperature data of the corresponding position generated by the twin model simulation with the actual temperature measurement data to generate temperature deviation data; based on the temperature deviation data, adjusting undetermined parameters which are preset in the twin model and related to heat dissipation efficiency through a parameter identification algorithm, and generating updated model calibration parameters; And loading the updated model calibration parameters into the twin model for subsequent performance simulation and life prediction calculation.

Description

Digital twinning-based water-cooled frequency converter high-cold working condition performance optimization and service life prediction method Technical Field The invention relates to the field of life prediction, in particular to a digital twin-based water-cooling frequency converter high and cold working condition performance optimization and life prediction method. Background As global energy structures are transformed into renewable energy, wind power generation is an important form of clean energy, and its installed capacity continues to increase rapidly. The water-cooled frequency converter is used as core power conversion and control equipment of the wind power generation system, and the operation reliability and the service life of the water-cooled frequency converter directly influence the power generation efficiency and the economic benefit of a wind power plant. Particularly in high and cold areas, wind power generation equipment often faces extremely low-temperature environments, which provides serious challenges for performance optimization and life prediction of the water-cooled frequency converter. Under the high and cold working condition, the water-cooled frequency converter needs to simultaneously cope with two major core problems: the low temperature leads to the obvious increase of the viscosity of the cooling liquid, increases the fluid resistance, and is easy to cause uneven flow distribution, local overheating or solidification of the cooling liquid, thereby causing the overrun of the junction temperature of the power module or the freezing loss of the system; in a low-temperature environment, high stress is generated by the power module packaging material due to mismatch of thermal expansion coefficients, meanwhile, the brittle transition temperature of the material is increased, and fatigue damage and brittle fracture risks are superposed, so that the service life of equipment is obviously shortened. Therefore, a digital twin-based water-cooled frequency converter high and cold working condition performance optimization and service life prediction method is provided for solving the problems. Disclosure of Invention The invention provides a digital twin-based water-cooling frequency converter high and cold working condition performance optimization and service life prediction method, which is used for providing technical support for stable operation of wind power equipment in an extreme environment. The invention provides a digital twin-based water-cooling frequency converter high-cold working condition performance optimization and service life prediction method, which comprises the steps of performing iterative computation based on real-time operation electric parameters and initial junction temperature data of a water-cooling frequency converter to obtain accurate heat source data matched with a current junction temperature state, inputting the accurate heat source data and cooling liquid low-temperature viscosity characteristic data into a simulation model, outputting multi-physical-field state data after operation, wherein the multi-physical-field state data comprises junction temperature data of each power module, temperature distribution data of a radiator and key-part stress data, inputting key-part stress data in the multi-physical-field state data and a change process of the key-part stress data into a damage accumulation model to generate key-part accumulated damage quantized data, outputting an optimization control instruction for adjusting the rotation speed of a cooling water pump in real time according to the junction temperature data and environment temperature data in the multi-physical-field state data, and outputting residual service life prediction data of the water-cooling frequency converter based on the key-part accumulated damage quantized data and future load working condition predicted data. Optionally, in a first implementation manner of the first aspect of the present invention, the method includes generating first loss distribution data according to the real-time operation electrical parameter and first junction temperature data, performing thermal simulation of a water-cooled frequency converter and a heat dissipation system thereof by using the first loss distribution data as an input condition, obtaining second junction temperature data, generating loss correction coefficient data according to a difference between the second junction temperature data and the first junction temperature data, correcting the first loss distribution data by using the loss correction coefficient data, generating second loss distribution data, updating the first loss distribution data by using the second loss distribution data, updating the first junction temperature data by using the second junction temperature data, returning to perform thermal simulation until a junction temperature data variation obtained by two adjacent thermal simulations is smaller than a preset thres