Research On The Key System Of ITER Static Magnetic Field-Testing Device

When International Thermonuclear Experimental Reactor(ITER)device is operating,a high magnetic field area is generated around it,which greatly affect the reliable and stable operation of the diagnostic system,water cooling system,power supply system,control system and other key components of ITER.Therefore,ITER organization urgently needs to develop a large size,high magnetic field strength strong magnetic compatibility testing device to study the anti-magnetic field disturbance capability of the equipment around the device.The magnetic field generator system and power supply system studied in this paper are the core components of the strong magnetic compatibility device.Firstly,according to the design requirements of the device,the mathematical model of the magnetic field intensity generated by each coil structure in the space region is established.To obtain a magnetic field generator device with low manufacturing and operating costs,the design optimization objective is to minimize the product of power loss and weight of the magnetic field generator,and the optimization algorithm design of the multi-objective optimization problem of this kind of magnetic field generator is completed under the multi-physical conditions such as electromagnetism,structure,and control.Combined with a specific example,the commonly coil structures are analyzed,and the optimal design of the system is completed.Then,according to the technical indicators of the power supply system,three characteristics of the power supply system are summarized:high power,low output current ripple and high reliability.On this basis,the application of topology schemes in low-voltage and high-current applications is analyzed,and a design topology of the static magnetic field power supply system is proposed.Next,according to the characteristics of high power and low output current ripple of the power supply system,the interleaved parallel control mode is applied,and the influence of phase inductance mismatch on the total output inductance current ripple is mainly studied,and the method to reduce the influence is proposed.The ripple of the output inductance current is mainly composed of ideal part and non-ideal part.The ideal part is obtained by eliminating the ripple of each phase circuit by interleaved parallel control mode,while the non-ideal part is mainly caused by the mismatch of phase inductance.In the case of mismatched inductance of each phase,the key to reducing output current ripple is to reduce the output current at the switching frequency in non-ideal components.For the convenience of analysis,the amplitude and angle of the output current ripple of each phase are expressed in the form of phasor.The method to reduce the total current ripple in the case of inductance mismatch is changing the phasor angle of each phase to minimize the phasor sum of each phase current ripple.The simplest strategy to reduce the output current ripple is to adjust the phase sequence of each phase to minimize the phasor sum of each phase current ripple in the case of inductance mismatch.In addition,according to the requirements of high reliability operation of the power supply system,open circuit fault diagnosis is carried out for three-phase PWM converter.To expand the application scope of this fault diagnosis technology,this paper proposes a neural network fault diagnosis technology based on Inception to solve the problem that traditional fault diagnosis technology is difficult to identify faults in both rectifier and inverter states.The fault diagnosis technology has the following characteristics:By using only the three-phase grid side current as the fault characteristic,21 open circuit faults of single and double power devices with two different fault characteristic quantities can be diagnosed under the conditions of rectifiers and inverters;Compared with the traditional machine learning architecture,Inception network architecture can make better use of computing resources;The transfer learning method is used to train the model with simulation data and experimental data.The model training is more economical;The diagnosis technology does not need to establish a mathematical model and add additional sensors.It has strong robustness,fast diagnosis speed and high diagnosis accuracy.The overall diagnosis effect is good.Finally,the manufacturing process of the main system of ITER static magnetic field-testing device is described.The relevant experiments are carried out.