Project supported by the National Natural Science Foundation of China (Grant No. 61174192).
Project supported by the National Natural Science Foundation of China (Grant No. 61174192).
† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant No. 61174192).
Generally, a magnetic target can be described with six parameters, three describing the position and three describing the magnetic moment. Due to a lack of sufficient components from one magnetometer, we need more than one magnetometer when locating the magnetic target. Thus, a magnetometer array should be designed. The baseline of the array is an important factor that affects the localization accuracy of the target. In this paper, we focus on the localization of a static target by using a scalar magnetometer array. We present the scalar magnetometer array with a cross-shaped structure. We propose a method of determining the optimal baseline according to the parameters of the magnetometer and detection requirements. In the method, we use the traditional signal-to-noise ratio (SNR) as a performance index, and obtain the optimal baseline of the array by using the Monte Carlo method. The proposed method of determining the optimal baseline is verified in simulation. The arrays with different baselines are used to locate a static magnetic target. The results show that the location performance is better when using the array with the optimal baseline determined by the proposed method.
A target containing ferromagnetic material can generate a magnetic anomaly under the geomagnetic field. The magnetic anomaly can be used to locate the magnetic target. Recently, various magnetic location techniques have been used in many areas, such as unexploded ordnance detection,[1,2] magnetic target tracking,[3–7] and human medical investigation.[8,9] In these location techniques, the magnetic target is considered as a magnetic dipole, which has six parameters, three describing the position and three describing the magnetic moment.[7] In order to calculate the parameters, we should construct at least six nonlinear functions. Thus, an array with the magnetometers is widely used to locate the magnetic target. Wynn[10] proposed a method of magnetic dipole localization based on the magnetic gradient. Nara[11] obtained a localization formula for magnetic dipole localization by magnetic vectors and its spatial gradients. The method requires the measurement of the magnetic anomaly field, and the accuracy of localization is highly sensitive to the noise in the magnetic anomaly field. In order to suppress the effect of the geomagnetic field, magnetic gradient tensor arrays comprising multiple vector magnetometers were proposed to locate the magnetic target.[7,12] Lee[13] presented a gradient-based method of locating and identifying a magnetic object in the presence of a geomagnetic field. Sui[14] proposed a method based on multiple-order magnetic gradient tensors for locating a magnetic dipole. The magnetic gradient tensor can improve the detection resolution of the target. However, there are still some problems in locating the magnetic target.
Comparing with the vector magnetometer, the advantage of the scalar magnetometer is that the measurement is almost not influenced by the orientation of measurement coordinate axes.[15] We can assemble a scalar magnetometer array without considering the orientation. Meanwhile, some kinds of scalar magnetometers such as an optically pumped magnetometer also have high sensitivities. For the localization of a target by using scalar magnetometers, some methods were proposed. Mcfee[16] proposed to measure the magnetic field in a two-dimensional grid and estimate the target localization and magnetic moment. Zalevsky[17] presented a high-resolution automatic detection algorithm based on a Wavelet transform and a scalar sensor array. Nelson[18] designed a multi-sensor towed array to detect the buried unexploded ordnance. The above methods are based on the data fitting. Thus, we need a single magnetic sensor or sensor array to measure the magnetic field with a designed scan routine (scan several lines). Fan[19] proposed the method to locate the magnetic target based on the scalar sensor array. In Ref. [19], an optimization problem was built according to the relationship between the total geomagnetic field measured by magnetometers and their positions, which was calculated by an improved particle swarm optimization algorithm. This method of locating the target can reduce the computing time. Thus, the methods based on the scalar magnetometer array are also very important for locating the magnetic target.
In this paper, we present an array with a cross-shaped structure and propose a method of optimizing the baseline. An appropriate baseline of the array is very important for magnetic anomaly detection with high performance. We use the traditional signal-to-noise ratio (SNR) as a performance index, and obtain the optimal baseline of the array by using the Monte Carlo method.
A magnetic target can be considered as a magnetic dipole when the distance between the target and the sensor is bigger than three times the largest dimension of the target. The magnetic field vector
In practice, the magnetic field measured by the magnetometer includes the ambient geomagnetic field
According to Eq. (
Generally, the broad characteristics of the geomagnetic field are consistent over a local region. We can consider that its effect is the same as the measurement of each magnetometer in the array (shown in Fig.
We define the baseline in the x direction as bx. We can obtain that Δx1m = bx/2 and Δx2m = −bx/2. Thus, we can rewrite Eq. (
According to Eqs. (
According to the 3σ principle of normal distribution, there are about 99.74% of egx dropping in the range [μ – 3μ, μ + 3μ], which can be considered as a reasonable range of egx. The SNR is expressed as
It is noted that the SNR is a function of array characteristic bx, noise σi, relative position (x,y,z), and properties of the target (mx,my,mz). In order to determine the optimal baseline of the array, Monte Carlo simulation is carried out. The way to determine the baseline by and bz (in the y direction and z direction) is similar to the method to determine bx.
Monte Carlo simulation is a method to simulate a probabilistic based on the use of random number of variables.[22,23] It can be used to simulate the mathematics model many times with randomly choosing a value for each variable at each time. The output values in the simulation are collected and the statistical analysis on those values is obtained. Therefore, when using the Monte Carlo simulation for optimizing the baseline, we first generate a large number of random numbers of each variable according to its probability distribution. Then we calculate the value of SNR of each random number. Finally, we analyze the output values and determine the optimal baseline of the array.
In the baseline optimization, SNR is considered as a performance index. From Eq. (
We substitute the random variables (x,y,z,M,θ,φ) and a certain b into Eq. (
When the baseline b is changed, the numbers (S) change as well. Thus, the present W can be considered as a function of baseline b. We can determine an optimal baseline with the maximum W. The diagram of the optimization using Monte Carlo simulation is shown in Fig.
Measurement of magnetic gradients can effectively minimize regional effects and eliminate temporal magnetic variations.[25] Therefore, the gradient of the magnetic anomaly is commonly used in locating the magnetic target. The gradient of the total magnetic anomaly can be expressed as
Three components of total magnetic anomaly are expressed as
Equations (
It can be seen that equation (
The simulation was designed to verify the effectiveness of the proposed method of optimizing the baseline. Firstly, we used the proposed method to determine the optimal baseline of the array. The Monte Carlo simulation method was used to determine the optimal baseline of the array, according to the requirements. The sensor array was designed for locating a target in a range of magnetic moment from 200 A·m2 to 600 A·m2. The maximum detection range of the array was 15 m, and the SNR should be greater than 15 dB. The result is shown in Fig.
Secondly, the arrays with different baselines were used to locate a magnetic target. The target was located at the point (1 m, 1 m, 1 m) and its magnetic moment was (348.18 A·m2, −61.39 A·m2, 353.55 A·m2). The array moved on the path which was parallel to the X axis, starting at the point (−6 m, 9 m, 0 m) and ending at the point (6 m, 9 m, and 0 m). The sampling interval was 0.5 m. In addition, we considered that the measure noise at each magnetometer was of a Gaussian distribution with a mean of 0 nT and a standard deviation of 0.016 nT.[25] We used the localization method based on the array to locate the target. The results are shown in Figs.
An SUV as the target with a constant velocity (−0.87 m/s, 0 m/s, and 0 m/s) was moved along the plan trajectory in the horizontal plane, starting from the point (17.0 m, 18.25 m, 0 m) and ending at the point (−25.5 m, 18.25 m, 0 m). The magnetic moment of the SUV was about 485.8 Am2.
There were only four scalar magnetometers (CS-L) in the experiment. The array was designed as shown in Fig.
We used the localization method based on the array to locate the target. The results are shown in Figs.
Generally, a sensor array is used to locate a magnetic target. The baseline of the array is an important factor that affects the localization accuracy of the target. Therefore, we present a method of determining the optimal baseline of a scalar sensor array with the Monte Carlo method. According to the detection requirements, the optimal baseline can be determined by the proposed method. The proposed method of determining the optimal baseline is verified by the simulation. In the simulation, the arrays with different baselines are used to locate a static magnetic target. The location performance is better, when the sensor array has the optimal baseline determined by the proposed method.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] |