Indoor fan interference discrimination method based on millimeter wave radar signal autocorrelation

Abstract

The present application belongs to the technical field of radar signal processing, and particularly relates to an indoor fan interference distinguishing method based on millimeter wave radar signal autocorrelation. In order to overcome the problem of poor generalization performance of the current energy threshold method and convolutional neural network method in the field of indoor personnel detection, the present application provides an indoor head-shaking fan and human target distinguishing method based on autocorrelation function. The specific operation is to calculate the maximum correlation coefficient value of radar echo under different distance gates to determine the signal periodicity, and to distinguish human signals and fan signals by using the signal periodicity. The method can be applied to personnel detection methods in different indoor scenes, and can effectively improve the anti-interference ability of the radar system to the head-shaking fan.

Description

Technical Field

[0001] This invention belongs to the field of millimeter-wave radar application technology, specifically relating to a method for distinguishing indoor fan interference based on the autocorrelation of millimeter-wave radar signals. Background Technology

[0002] Millimeter-wave radar, as the name suggests, operates in the millimeter-wave frequency band, specifically the 30-300 GHz range. It features all-weather, all-day operation. According to wave propagation theory, the higher the radar's operating frequency, the shorter its wavelength, resulting in higher resolution and stronger penetration. Therefore, millimeter-wave radar boasts advantages such as high resolution, strong anti-interference capabilities, good directivity, and excellent detection performance. These advantages determine its strong capabilities in indoor personnel detection.

[0003] Indoor person detection based on millimeter-wave radar is closely related to the similarity between the person and the test target. With the development of millimeter-wave radar technology, its applications have become more widespread, including indoor person detection, human vital sign detection, and gesture recognition. However, in indoor person detection, there are many interfering elements in indoor environments, such as fans, plants, and curtains. The clutter from these interferences can cause serious false alarms. Therefore, it is necessary to research better anti-interference indoor person detection algorithms to reduce false alarms and achieve better detection performance in millimeter-wave radar-based indoor person detection.

[0004] The current research on effectively improving the anti-interference performance of radar in indoor scenarios mainly includes: (1) increasing the human detection threshold to extract more obvious human signal features. As the detection threshold increases, low-energy interference signals are effectively suppressed, and human feature signals become more obvious. However, in this process, as the amplitude of human movement decreases, the radar detection system will miss more alarms, which can easily lead to the loss of resting human target signals. (2) using deep learning networks to distinguish between interference signals and human signals. Convolutional neural networks are used to extract human target features and interference signal features in the scene, and finally, the features of each layer are combined and fed into the detection network to realize indoor human presence detection under interference conditions. However, the above research is based on detection methods in the same scene. Whether it is the threshold setting method or the use of networks to realize human target detection, both have the characteristics of poor generalization and cannot be universally applied to the distinction between interference signals and human signals in various indoor scenarios and different radar modules. Summary of the Invention

[0005] The purpose of this invention is to address the aforementioned problems and shortcomings, and to overcome the poor generalization performance of current energy threshold methods and convolutional neural network methods in the field of indoor personnel detection, by providing a method for distinguishing between indoor oscillating fans and human targets: a fan signal distinction method based on autocorrelation function. Applying this method to personnel detection methods in different indoor scenarios can effectively improve the anti-interference capability of radar systems against oscillating fans.

[0006] The technical solution of this invention is as follows:

[0007] A method for distinguishing indoor fan interference based on the autocorrelation of millimeter-wave radar signals includes the following steps:

[0008] S1. Acquire the millimeter-wave radar echo signal, then perform an FFT along the range dimension to obtain M*N 1DFFT data (slow time-range dimension) R(M, N), where M is the number of range gates and N is the number of chirps; then pass R(M, N) through MTI filtering (slow time-linear filtering) to obtain R MTI (M, N) data:

[0009] R MTI (M, N) = R (M, N)-1 / N∑ R (M, N)

[0010] S2. Accumulate the filtered data using a long time window, defining the accumulated data to a length T, and obtain R. MTI (M, T):

[0011]

[0012] S3, R MTI The autocorrelation function (ACF) sequence is obtained by solving the corresponding autocorrelation function for each distance dimension m of the (M, T) time series. M (T):

[0013] ACF m (T)=f ACF (R MTI (m, T))

[0014] Among them, f ACF The autocorrelation function (ACF) is obtained by iterating through the sequence. m The maximum autocorrelation coefficient value in (T) And its corresponding serial number x, This indicates the reliability of the period length x corresponding to the sequence. The larger the value, the more pronounced the periodicity of the sequence at length x;

[0015] S4. For each distance dimension... Compare, set ACF greater than or equal to 0.6 m The (T) sequence is a sequence with fan interference. ACF below 0.6 m The (T) sequence is a sequence without fan interference.

[0016] The beneficial effects of this invention are that it proposes a fan signal discrimination method based on autocorrelation function, taking into account the echo characteristics of millimeter-wave radar and the periodic motion of a fan. A key innovation is its ability to effectively distinguish fan signal interference in typical indoor scenarios; furthermore, under long-term observation, it can differentiate fan signal interference based on the time window. The signal processing method uses changes in the fan signal to determine whether the fan is on or off. This method is computationally simple and makes full use of the periodicity of the fan signal. It can distinguish between fan interference and human signals in indoor scenes, thereby improving the subsequent detection rate of human targets and reducing false alarms. Attached Figure Description

[0017] Figure 1 The image is a slow-time distance-dimensional MTI image containing fan signals;

[0018] Figure 2 This is a flowchart of the present invention;

[0019] Figure 3 Distance dimension with fan signal value. Detailed Implementation

[0020] This invention aims to improve the ability of millimeter-wave radar to distinguish interference from oscillating fans in indoor detection scenarios. First, the millimeter-wave radar echo signal is obtained through the radar system, and FFT is performed to obtain 1DFFT data (slow time-range dimension). Then, MTI processing is applied to the 1DFFT data. Next, the autocorrelation function of the filtered data is calculated. This requires accumulating the data over time to a length T to obtain R. MTI (M, T), then for R MTI The ACF is obtained by solving the corresponding ACF sequence for each distance dimension of the (M, T) time series. M (T), find the maximum autocorrelation coefficient in each distance-gated ACF sequence. And the corresponding index x, by comparing each distance dimension This value is used to obtain the periodicity of the signal in that distance dimension.

[0021] The fan interference signal 1DFFT diagram of the present invention is attached. Figure 1 As shown, the flowchart is attached. Figure 2As shown in the figure, the experimental results are compared in the appendix. Figure 3 As shown.

[0022] The advantages of this invention are mainly reflected in its applicability to scenarios with periodic interference (such as oscillating fans and air conditioner fan blades). It considers both the characteristics of millimeter-wave radar echoes and effectively utilizes the periodicity of interference signals, enabling it to distinguish between fan interference and human signals from radar echo signals. Furthermore, based on simple misaligned multiplication calculations, it improves detection performance without affecting the real-time performance of the detection system, laying the foundation for the practical application of millimeter-wave radar detection.

Claims

1. A method for distinguishing indoor fan interference based on autocorrelation of millimeter-wave radar signals, wherein the fan is an oscillating fan, characterized in that, Includes the following steps: S1. Acquire millimeter-wave radar echo signals, and then perform FFT along the range dimension to obtain 1DFFT data of size M*N. Where M is the number of distance gates and N is the number of chirps; then... After MTI filtering, the result is... data: ， S2. Accumulate the filtered data using a long time window, defining the accumulated data to a length T, and obtain... : ， S3, to The time series is obtained by solving the corresponding autocorrelation function sequence for each distance dimension m. : ， in, The autocorrelation function is obtained by traversing the sequence to find each... Maximum autocorrelation coefficient value and its corresponding serial number , This indicates the reliability of the period length x corresponding to the sequence. The larger the value, the more pronounced the periodicity of the sequence at length x; S4. For each distance dimension... Compare, set ≥0.6 The sequence is a sequence with fan interference. Below 0.6 The sequence is one without fan interference.

Images