Research On Temperature Rise Of Dry-type Transformer Based On The Electromagnetic-Thermal Coupled Method

Offshore oil and gas platform power system is responsible for power supply for offshore oil rigs and living areas.In order to ensure oil output,stable operation of the power system is particularly important.Dry-type transformers are the hub equipment that connect power generation and consumption platforms in offshore power systems.Reducing the transformer failure rate is significance for improving system reliability.Overheating faults are one of the common faults for transformers.Lighter will accelerate insulation aging.Severe will lead to fires and explosions,but dry-type transformers are significantly different from the oil-immersed transformers in structural and operating characteristics,and their thermal analysis methods are different.This paper focuses on the temperature rise of dry-type transformers on offshore oil and gas platform,and analysis the temperature characteristic based on multi-physics coupling,the overload capacity based on the thermal network model,and hot-spot temperature prediction.The main work contents are as follows:(1)An electromagnetic-thermal coupled model of epoxy resin cast dry-type transformer is established.Analyzed the coupling relationship between the magnetic field and the circuit,and established a field-circuit coupling finite element model to improve the accuracy of the calculation the core loss and winding loss for the dry-type transformer.The electromagnetic-thermal coupled model was established by indirect coupling with loss as the medium,and the accuracy of the model was verified with five indicators of secondary voltage,current,core loss,winding loss,and temperature,compared to a single temperature field analysis,the electromagnetic-thermal coupled improves the accuracy of temperature.The location of the hot-spot temperature is found,that can guidance for the sensor layout during online monitoring.(2)An analysis method for overload capacity of dry-type transformers based on a thermal network model is proposed.Considering the influence of harmonic currents on the losses of various parts for dry-type transformers,the heat source input in the thermal network model is modified.The double-layer thermal network model that contains the outer surface to the air and the outer surface to hot-spot are established,used to calculate the hotspot temperature through the analytical method,which is more suitable for engineering applications.The thermal resistance and heat capacity parameters in the thermal network model are identified by the least square method,which solves the problem of insufficient physical parameters of the transformer,the parameters of the thermal network model cannot be calculated.The overload time is used as a characterization of the overload capacity,and the relationship between the ambient temperature and the overload time is analyzed,which is conducive to improving the overload capacity of the transformer within a certain temperature rise limit,and using the transformer's remaining capacity.(3)The temperature monitoring based on infrared thermal imager and the hot-spot temperature prediction based on PF-SVR are studied.The installation position of the infrared thermal imager was determined based on the results of the electromagnetic-thermal coupled thermal analysis.And the relationship between the outer surface temperature of the winding and the hot-spot temperature measured by the infrared thermal imager was determined based on the thermal network model.Non-contact and online monitoring was implemented,solved the problem that the transformers already in operation are not convenient to install the embedded temperature sensor.SVR is applied to the hot-spot temperature prediction of dry-type transformers,and the particle filter method is used to optimize the parameters of the support vector regression.This method can dynamically optimize the parameters of SVR to overcome the problem that the optimization effect decreases when the data of the test subset and the training data in the conventional optimization method are significantly different.The actual hot-spot temperature data is used to verify the prediction results based on PF-SVR.Hot-spot temperature prediction can help operation and maintenance personnel to find abnormal temperature conditions in advance,reduce or even avoid overheating failures of dry-type transformers,and improve equipment reliability.