Amin Rezaeizadeh

An Iterative Learning Control Approach for Radio Frequency Pulse Compressor Amplitude and Phase Modulation

Radio Frequency (RF) pulse compressors are used in linear accelerators (Linac) to achieve high power levels by shortening the RF pulse length. In their original form, the phase of the incoming pulse is reversed by 180 ° which generates a high peak power at the output of the pulse compressor, followed by an exponential decay. This pulse shape however is not appropriate with regard to timing stability as well as for having equal energy gain for multi-bunch operation. To achieve flat-topped pulses, a method has been previously proposed that analytically modulates the input phase waveform. In the present contribution an alternative way to producing flat-topped RF pulses is proposed which is based on Iterative Learning Control techniques. This approach manipulates the input waveforms iteratively in order to generate flat-topped amplitude and phase pulses at the output of the pulse compressor.

An adaptive iterative learning control scheme for reducing CO2 emission in gasoline engines

This paper proposes a control algorithm for a Toyota gasoline engine problem that is addressed in a student competition format. The control objective is to minimize the fuel consumption while avoiding specified dangerous situations. The approach develops a feed-forward control based on an adaptive Iterative Learning Control. In this method, the plant is run several times and the controller iteratively updates the actuation inputs in order to generate the desired reference torque profile. The algorithm converges after approximately 10 iterations providing the corresponding locally optimal control trajectories.

MPC based supervisory control design for a Free Electron Laser

This paper develops a Model Predictive Control based approach for controlling multiple Radio Frequency (RF) stations in a Free Electron Laser (FEL). Each RF station includes a klystron which delivers high RF power, of the order of 50 megawatts, to feed the cavities in which the electron beam is accelerated. A local PID controller at each RF station controls the RF voltage measured at the local cavities. The proposed MPC scheme acts as a supervisory control to keep the beam energy constant by adjusting the set points of the local PID loops. The concept has been successfully tested at the SwissFEL injector test facility with three full-scale RF stations.

Robust-based control design for the beam injector facility

This paper develops a robust control design for the electron beam injector facility in the Swiss Free Electron Laser (SwissFEL) machine. The relationship between radio frequency parameters and the electron beam quantities are commonly modeled as a static response matrix. Typically, the response matrix has a high condition number which makes inverse-based control approaches very sensitive to gain uncertainties. This problem is usually addressed by singular value decomposition to invert the response matrix, followed by filtering the singular values, so that the inverse-based control performs better in regards to uncertainties in the model. In this paper, we study a more structured robust MIMO control design that can be applied to all systems that have static response matrix (such as “orbit correction” system). To the best knowledge of the authors, it is the first time that such a robust control scheme is implemented on an accelerator machine. The designed controller has been successfully tested at the SwissFEL injector test facility and the results are compared to the inverse-based control design.

Optimal Radio-Frequency Power Distribution in a Linear Accelerator Using Beam Energy Measurements

A linear accelerator, including several radio-frequency (RF) stations, can be viewed as one virtual RF station with a certain RF voltage (in amplitude and phase). This paper proposes an optimization scheme that, for a specified total beam energy gain, determines the klystrons output powers as well as the modulators high voltages optimally. The algorithm employs the nonlinear static characteristics curves of klystrons to determine analytically the input RF amplitude of the drive chain. The proposed algorithm facilitates the klystron operating point setting with an automatic procedure for the desired energy gain of a linac.

Time-Delayed Feedback Control of Electron Beam Energy Using

In the two-bunch operation mode of the Swiss Free Electron Laser (SwissFEL) machine, two electron bunches, separated by 28 ns, are fired and accelerated every 10 ms. The electron beam is passed through several accelerating cavities that are energized by high-power radio frequency (RF) pulses. The RF voltage generates a specific time-dependant accelerating gradient profile that results in different energy gains for the two bunches. It is often required that the two electron beams experience the same accelerating gradient through the cavities. This can be achieved by shifting the RF pulse in time to change the relative energy of the two beams. However, in the digital RF system used here, the minimum shift is limited by the sampling time of the digital-to-analog converters. In this brief, we develop a control scheme for the energy difference control, which finely tunes the RF pulse timing by a fractional delay filter. The method has been successfully tested at the SwissFEL injector test facility.