Simulation Study Of Integrated Optical Waveguide Magnetic Field Sensor

With the increase in electronic technology and the attention of health, evaluation of the electromagnetic field (EMF) surrounding the equipment has become an important issue. Therefore, various sensors for measuring the magnetic field have been developed. This article mainly studied the integrated waveguide magnetic sensor based on Mach-Zehnder interferometer. Because it has little disturbance to the measured EMF, and it can suppress the influence of electric field effectively, therefore it is an appropriate apparatus using in the electromagnetic measurement.First, this paper has introduced the operation principle of the optical waveguide magnetic field sensor, and built an equivalent circuit model of the magnetic field sensor and a system model. The relationship of sensor head size and the frequency response, and the sensitivity of the sensor system have been analyzed. The simulated results show that we can improve the sensitivity by setting the optical bias atπ/2, increasing the overlap integral factor, reducing the insert lose, increasing the loop antenna area, reducing the gap of electrode, increasing the optical power, reducing the wavelength, reducing the relative intensity noise and reducing the receive bandwidth.Because in the integrated optical waveguide magnetic field sensor's loop antenna, the ratio of the length and the thickness is above 1000, this is disadvantageous to the electromagnetism simulation computation's grid division. In order to obtain the precise result, it requires the huge number of the grid, which causes the computation speed to be slow, even be beyond the computer's computing power. Therefore, it is necessary to seek another effective and fast computing method.The Artificial Neural Network (ANN) is an intelligent analysis and the computation technology. The people have already applied it into the computation electromagnetics. The most common ANN -- BP (back propagation) ANN, can obtain the ability of self-study by self-adjusting weight unceasingly. Compared with the finite element method (FEM), it computes quickly in computation electromagnetics. However as a result of own defect, it can not search globally, the result is random, which limits the ANN's application. The genetic algorithm (GA) is a global searching algorithm, so combination of GA and the ANN will be helpful. But the simple GA may cause the precocious easily, and the result is not precise.This article proposes the optimization to the traditional GA, which makes it converge quickly, precisely, and expands the search area. The method combination of this GA and the ANN, overcomes the ANN's disadvantage, and simulates a target function. The result is very precise. This method has been used to simulate the frequency response of the sensor head in the integrated optical waveguide magnetic field sensor, and also obtained the good result by compared with the electromagnetism simulation result. The extendibility is good. the ANN's simulation time and resources are much less than the traditional method. It builds the foundation for the optimization design the components including integrated optical waveguide magnetic field sensor, by using this method.