Mine monitoring method based on millimeter wave radar multi-feature fusion

Abstract

The invention relates to a mine monitoring method based on millimeter-wave radar multi-feature fusion, and the method comprises the steps: transmitting an FMCW signal through a millimeter-wave radar, receiving a target echo, sequentially carrying out the signal preprocessing, 1DFFT distance measurement, 2DFFT speed measurement, velocity ambiguity resolution and DOA estimation angle measurement, and extracting the distance, velocity, angle and energy variation features of a target point cloud; then millimeter-level deformation is obtained through phase analysis; obtaining a vibration frequency through band-pass filtering and spectrum analysis; fusing the six types of features of the target point cloud, adopting a confidence counter weighting decision mechanism, and triggering early warning when a fusion score exceeds a preset threshold value; according to the invention, through the advantages of high reliability, strong anti-interference performance and high non-contact safety of multi-feature fusion decision, the false alarm rate is significantly reduced, and the real-time accurate early warning capability of the mine dangerous state is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mine safety monitoring, and particularly relates to a mine monitoring method based on millimeter wave radar multi-feature fusion. BACKGROUND

[0002] With the progress of science and technology and the development of the coal mining industry, the mine monitoring algorithm, as an important part of intelligent coal mines, runs through the whole life cycle of coal mine production, and is a key engine for realizing safe, efficient and intelligent operation of coal mines. It changes the traditional passive management mode relying on manual inspection and experience into a modern management system driven by data, intelligent early warning and active prevention and control through deep integration of artificial intelligence, big data, the Internet of Things and sensor technology. Millimeter wave radar is gradually introduced due to its advantages of all-weather, all-day, non-contact and high resolution, and can continuously analyze and deeply mine massive multi-source heterogeneous data such as gas concentration, dust content, temperature, roof pressure and equipment operating state at the bottom of the well for 7*24 hours, accurately identify early weak features of potential risks such as gas overrun, fire hazards, roof instability and equipment failure, and the accuracy and timeliness of its early warning far exceed manpower, thereby gaining valuable time for personnel evacuation and emergency disposal and greatly reducing the probability of major accidents from the source. At the same time, through intelligent optimization and scheduling of data such as mining progress, transportation efficiency, equipment working conditions and personnel positioning, the mine monitoring algorithm can realize the coordinated and efficient operation of various production links such as mining, transportation, ventilation and drainage, reduce invalid waiting and energy consumption, and improve the recovery rate of coal resources and overall production efficiency. However, due to the complex environment in the mine, such as the metal support of the roadway side wall, the cable bridge, the large mechanical and electrical equipment and the wet and high reflectivity surrounding rock, a large number of strong scatterers are formed, causing serious multipath interference and clutter, making it easy to trigger false alarms by single distance or energy features. At the same time, due to the micro-Doppler effect caused by the continuous vibration of underground operation machinery, the start and stop of the belt conveyor and the airflow disturbance of the ventilation machine, it is difficult to distinguish dangerous deformation from normal disturbance by relying only on speed features. In addition, millimeter wave radar reacts to slight movements or insignificant signals such as small vibrations in the mine or the normal movement of personnel and vehicles, which may be misjudged as dangerous situations and trigger false alarms. SUMMARY

[0003] Therefore, the present application provides a mine monitoring method based on millimeter wave radar multi-feature fusion, which can solve the problem of high false alarm rate of millimeter wave radar in complex mine environments caused by the difficulty in distinguishing real signals from interference signals by relying on single features for target detection, and significantly improve the monitoring accuracy.

[0004] To achieve the above-mentioned purpose, the present application provides a mine monitoring method based on millimeter wave radar multi-feature fusion, comprising the following steps: S1, radar signal collection and preprocessing, obtaining echo signal and original radar data; S2, the echo signal is processed, and multi-dimensional features of the target point cloud are extracted, including target point cloud distance, speed, angle, and reflection energy value; S3, the phase change amount of the target point cloud is extracted, and the deformation value and spectrum of the target point cloud are calculated based on the band-pass filtered displacement signal; S4, multi-feature fusion processing is performed according to the extracted features of the target point cloud, and a warning result is output; S401, respectively calculating the distance change amount, speed change amount, angle change amount, reflection energy change amount, deformation value and vibration frequency value of the target point cloud; S402, presetting the warning threshold and confidence counter of the six feature change amounts respectively; S403, counting the confidence counter scores of the six feature change amounts; S404, fusion processing to obtain the confidence counter score , outputting the dangerous warning judgment result, the expression of the fused confidence counter score is: ; Wherein, represents the confidence counter score of the i-th feature change amount, when the fused confidence counter score is greater than or equal to the preset threshold , it is judged that there is danger, and a warning notice is given; otherwise, it is judged that there is no danger and no warning notice is given, and the expression is: ; Wherein, represents the output judgment result.

[0005] Preferably, the millimeter wave radar utilizes a high-frequency circuit to generate and transmit a frequency-modulated continuous wave FMCW signal , receives the echo signal of the target object reflection signal , and obtains the original radar data.

[0006] Preferably, the multi-dimensional features of the target point cloud include the following steps: S201, one-dimensional Fourier transform 1DFFT is performed on the original radar data, time domain signal is converted into frequency domain signal, distance information of the target object is extracted, and distance dimension spectrum is obtained; S202, CFAR constant false alarm detection is used to extract target distance units from the distance dimension spectrum, and strong target points are obtained; ​S203, performing coherent processing on the raw radar data, then performing a two-dimensional Fourier transform (2DFFT) to obtain a range-Doppler spectrum, performing velocity deambiguation, and compensating for Doppler phase deviation to obtain an actual velocity; S204: Use digital beamforming (DBF) technology to perform weighted synthesis on the array signal of the radar antenna to form a virtual beam for DOA angle estimation to obtain the target angle.

[0007] Preferably, extracting the phase variation of the target point cloud comprises the following steps: S301, extracting a phase sequence within a target distance unit where the target point cloud is extracted; The phase information of the distance unit corresponding to the target point cloud is extracted by the inverse tangent function, and the phase sequence is subjected to phase unwrapping processing. The phase of the target is extracted once in each frame period, and the phase of the target point cloud obtained by cyclic transmission changes with the value of the frame number, that is, the small displacement of the target point cloud changes with time; S302: Calculate the phase change of the target point cloud, expressed as: ; in, represents the phase change, represents the radar wavelength, Indicates the displacement change caused by the target point cloud; S303, calculating the deformation value of the target point cloud; According to the correspondence between the phase and displacement, the result of the deformation change over time after filtering is obtained, and the deformation characteristics of the distance unit where the target point cloud is located are extracted and analyzed. The deformation value expression of the distance unit where the target point cloud is located is: ; in, Represents the distance unit index, through which the position of the target point cloud in the distance direction can be determined. Indicates the index of a single electromagnetic wave chirp emitted by the radar. represents the radar electromagnetic wave emission period, Represents the radar phase of the target point cloud after filtering; By The distance unit where the target point cloud is continuously extracted within a certain time period is The phase at obtains the vibration frequency value; S304: Perform fast Fourier transform on the filtered phase signal, and obtain the corresponding frequency within a preset time according to the peak value and its harmonic characteristics.

[0008] Preferably, after extracting the phase information of the distance unit corresponding to the target point cloud, the phase information is unwrapped, and according to the distribution range of the target frequency, a band-pass filter is used to distinguish the vibration signal of the target frequency band, so as to obtain continuous phase change.

[0009] Preferably, the distance change amount confidence counter is If the distance change amount is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The speed change amount confidence counter is If the speed change amount is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The angle change amount confidence counter is If the angle change amount is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The reflected energy change amount confidence counter is If the reflected energy change amount is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The deformation value confidence counter is If the deformation value is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The vibration frequency value confidence counter is If the vibration frequency value is greater than or equal to a preset warning threshold , the counter is accumulated , otherwise, it is 0. The confidence counter score expression of the six feature change amounts is: ; Wherein, represents the th feature change amount confidence counter, represents the th feature change amount, represents the th preset feature change amount warning threshold.

[0010] Compared with the prior art, the beneficial effects of the present application are: The present invention constructs a more comprehensive target state perception by fusing six-dimensional features of distance, speed, angle, reflected energy, deformation value and vibration frequency, significantly improving the accuracy and reliability of mine monitoring. Traditional monitoring methods rely on a single feature and are difficult to distinguish between real dangers and environmental noise, effectively overcoming the problem of multiple interferences in complex underground environments. The present invention realizes the precise capture of millimeter-level tiny deformations and weak vibrations by establishing a precise mapping model of phase changes and physical deformations, breaking through the insensitivity of existing technologies to slow deformations. At the same time, the multi-feature fusion decision algorithm based on confidence counter and dynamic weight allocation effectively suppresses false alarms caused by metal reflections, equipment vibrations, etc., providing real-time and reliable intelligent early warning guarantees for mine safety production. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Schematic diagram of the process of the mine monitoring method of multi-feature fusion of the present invention; Figure 2 This is the target point cloud time domain signal diagram of the present invention; Figure 3 This is the target point cloud spectrum estimation diagram of the present invention. DETAILED DESCRIPTION

[0012] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0013] This embodiment provides a mine monitoring method based on millimeter wave radar multi-feature fusion, including the following steps: S1. Radar signal acquisition and preprocessing, configuring millimeter wave radar waveform and initialization parameters, and obtaining echo signals and raw radar data; Millimeter wave radar uses high-frequency circuits to generate and transmit frequency modulated continuous wave (FMCW) signals. When electromagnetic waves encounter a target object, part of the electromagnetic waves will be reflected back and captured by the radar's receiving antenna to receive the echo signal of the target object's reflected signal. , get the original radar data.

[0014] S2. Processing the echo signal to extract multi-dimensional features of the target point cloud, including target point cloud distance, speed, angle, and reflected energy value, includes the following steps: S201, performing a one-dimensional Fourier transform (1DFFT) on the raw radar data to convert the time domain signal into a frequency domain signal, extracting the distance information of the target object, and obtaining a range spectrum; S202, extracting a target distance unit from the range dimension spectrum by using CFAR constant false alarm detection to process a threshold of the input radar signal, and obtaining a strong target point; S203, performing coherent processing on the original radar data, performing two-dimensional Fourier transform 2DFFT, obtaining a range-Doppler spectrum, and performing velocity deblurring to compensate for Doppler phase offset to obtain an actual velocity; This is because the radar speed measurement range is limited, and when the target speed exceeds this range, the "velocity ambiguity" phenomenon occurs, that is, the measured speed is inconsistent with the actual speed; S204, using digital beam forming technology DBF to weight and synthesize the array signal of the radar antenna, calculating the phase difference or time difference between the signals received by different antennas to infer the direction of arrival of the target, forming a virtual beam for DOA angle estimation, obtaining the target angle, and realizing angle estimation of the target.

[0015] S3, extracting the phase change amount of the target point cloud, calculating the deformation value and spectrum of the target point cloud based on the displacement signal after band pass filtering; S301, extracting a phase sequence in the target distance unit where the target point cloud is located; The phase information of the distance unit corresponding to the target point cloud is extracted by an inverse tangent function, the phase sequence is processed by phase unwrapping, the phase of the target is extracted once in each frame period, and the phase of the target point cloud changes with the value of the frame number, that is, the micro displacement of the target point cloud changes with time, which can also be regarded as the relationship between the phase of the target and time, The phase formula principle and derivation include the following steps: First, the radar transmission signal model is established, and the expression is: ; Among them, represents the transmission signal carrier, represents the frequency modulation slope; The expression of the radar echo signal is: ; Among them, represents the delay time of the echo signal, according to the FMCW principle, the mixing of the transmission signal and the received signal obtains the original intermediate frequency signal expression: ; Among them, represents the radar intermediate frequency signal carrier frequency, represents the radar phase, The original intermediate frequency signal can be regarded as a sinusoidal signal with frequency and phase The expression is: ; S302, in order to measure the micro movement of the target, the phase change of the distance index corresponding to the strongest energy of the target point cloud is needed, and the expression is: ; Wherein, represents the phase change amount, represents the wavelength of the radar, represents the displacement change amount caused by the target point cloud; Since the phase value is between , and the actual displacement curve needs to be unfolded, whenever the phase difference between the continuous values is greater than / less than ± , the phase unfolding is performed by subtracting from the phase; The displacement change of the target can cause the phase change of the FMCW radar signal, assuming that the wavelength of the radar is , the micro displacement distance is , and the corresponding phase change ; S303, calculate the deformation value of the target point cloud; According to the correspondence between the phase and the displacement, the result of the deformation after filtering is obtained, the deformation characteristics of the distance unit where the target point cloud is located are extracted and analyzed, and the deformation value expression of the distance unit where the target point cloud is located is: ; Wherein, represents the distance unit index, by which the position of the target point cloud in the distance direction can be determined, represents the index of a single chirp of the radar electromagnetic wave, represents the radar electromagnetic wave transmission period, represents the radar phase of the target point cloud after filtering; The vibration frequency value is obtained by continuously extracting the phase at place of the distance unit where the target point cloud is located within time; S304, fast Fourier transform is performed on the filtered phase signal, and the corresponding frequency within the preset time is obtained according to the peak value and its harmonic characteristics; After extracting the phase information of the distance unit corresponding to the target point cloud, the phase information is unwrapped, the vibration signal of the target frequency band is distinguished by using a band-pass filter according to the distribution range of the target frequency, so as to obtain continuous phase change.

[0016] S4, multi-feature fusion processing is performed according to the extracted characteristics of the target point cloud, and a warning result is output. The radar multi-feature fusion is specifically introduced as follows: the radar multi-feature fusion processing is to fuse multiple features (such as distance, speed, angle, radar energy, deformation value and vibration frequency value) from the radar to form a more powerful and comprehensive feature representation, so as to improve the accuracy and reliability of target detection. The multi-feature fusion provided in the embodiment specifically includes the following steps: S401, respectively calculating a distance change amount, a speed change amount, an angle change amount, a reflection energy change amount, a deformation value and a vibration frequency value of the target point cloud; S402, respectively presetting a warning threshold and a confidence counter of the six feature change amounts. The selection of the confidence is crucial for balancing the contribution of different features to the fusion result, which is determined and adjusted according to the importance, information amount and application scene of the feature; S403, counting the confidence counter scores of the six feature change amounts; The distance change amount confidence counter is If the distance change amount is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The speed change amount confidence counter is If the speed change amount is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The angle change amount confidence counter is If the angle change amount is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The reflection energy change amount confidence counter is If the reflection energy change amount is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The deformation value confidence counter is If the deformation value is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The vibration frequency value confidence counter is If the vibration frequency value is greater than or equal to the preset warning threshold , the counter is added by , otherwise, it is 0; The confidence counter score expression of the six feature change amounts is: ; Among them, represents the confidence counter of the i-th feature change amount, represents the i-th feature change amount, represents the i-th feature change amount, represents the i-th preset feature change amount, represents the i-th preset feature change amount, represents the i-th preset feature change amount, S404, fusion processing obtains a confidence counter score , outputs a dangerous early warning judgment result, and the expression of the fused confidence counter score is: ; Among them, represents the confidence counter score of the i-th feature change amount, represents the i-th feature change amount, When the fused confidence counter score is greater than or equal to a preset threshold , it is judged that there is danger, and early warning notification is performed; otherwise, it is judged that there is no danger, and early warning notification is not performed, and the expression is: ; Among them, represents the output judgment result.

[0017] The method provided in the embodiment performs multi-feature monitoring on a target object by transmitting and receiving millimeter wave signals, utilizes the high frequency and short wavelength characteristics of millimeter waves, transmits a frequency-modulated continuous wave signal to the target, and processes the received signal by using difference frequency and other parameters, so that the ranging, angle measurement and velocity measurement of the target can be realized, and the phase change of the FMCW (Frequency Modulated Continuous Wave) signal on a specific distance gate caused by the slight fluctuation of the target is monitored, so that the slight deformation or vibration frequency of the target is inverted; in the field of radar target detection, a single feature is often difficult to comprehensively describe the target state, and fusion of multiple features can make up for this deficiency; for example, distance, velocity and angle are basic features measured by a radar, which provide the position and motion information of the target in space, and the radar energy reflects the reflection intensity of the radar signal by the target, which is related to the size, shape and material of the target, and advanced features such as deformation value and vibration frequency value provide more detailed physical state information of the target, such as the deformation degree and vibration state of the target; finally, by fusing these features, a more comprehensive feature representation can be formed, which not only contains the basic position and motion information of the target, but also contains the physical state information of the target, so that the radar system can more accurately describe the target state, thereby improving the accuracy and reliability of target detection.

[0018] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with reference to the preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the scope of the technical solution of the present application. Any modification, equivalent change and modification of the above embodiment made according to the technical essence of the present application, as long as it does not deviate from the technical solution of the present application, is still within the scope of the technical solution of the present application.

Claims

1. A mine monitoring method based on millimeter wave radar multi-feature fusion, characterized in that: The following steps are involved: S1. Radar signal acquisition and preprocessing to obtain echo signals and raw radar data; S2. Processing the echo signal to extract multi-dimensional features of the target point cloud, including target point cloud distance, speed, angle, and reflected energy value; S3. extracting the phase change of the target point cloud, and calculating the deformation value and spectrum of the target point cloud based on the displacement signal after bandpass filtering; S4, performing multi-feature fusion processing based on the extracted features of the target point cloud and outputting a warning result; S401, respectively calculating the distance change, velocity change, angle change, reflected energy change, deformation value and vibration frequency value of the target point cloud; S402, presetting warning thresholds and confidence counters for six feature changes respectively; S403, counting the confidence counter scores of the six feature changes; S404: Fusion processing to obtain confidence counter score , output the danger warning judgment result, the confidence counter score after fusion The expression is: ； in, Indicates the The confidence counter score of the feature change, When the fused confidence counter score Greater than or equal to the preset threshold , then it is judged that there is danger and an early warning notification is issued; otherwise, it is judged that there is no danger and no early warning notification is issued. The expression is: ； in, Indicates the output judgment result.

2. The mine monitoring method based on millimeter wave radar multi-feature fusion according to claim 1 is characterized in that: Millimeter wave radar uses high-frequency circuits to generate and transmit frequency modulated continuous wave (FMCW) signals. , receiving the echo signal of the target object's reflected signal , get the original radar data.

3. The mine monitoring method based on millimeter wave radar multi-feature fusion according to claim 1 is characterized in that: Extracting multi-dimensional features of the target point cloud includes the following steps: S201, performing a one-dimensional Fourier transform (1DFFT) on the raw radar data to convert the time domain signal into a frequency domain signal, extracting the distance information of the target object, and obtaining a range spectrum; S202, extracting target range units from the range-dimensional spectrum using CFAR constant false alarm detection to obtain strong target points; S203, performing coherent processing on the raw radar data, then performing a two-dimensional Fourier transform (2DFFT) to obtain a range-Doppler spectrum, performing velocity deambiguation, and compensating for Doppler phase deviation to obtain an actual velocity; S204: Use digital beamforming (DBF) technology to perform weighted synthesis on the array signal of the radar antenna to form a virtual beam for DOA angle estimation to obtain the target angle.

4. The mine monitoring method based on millimeter wave radar multi-feature fusion according to claim 3 is characterized in that: Extracting the phase variation of the target point cloud includes the following steps: S301, extracting a phase sequence within a target distance unit where the target point cloud is extracted; The phase information of the distance unit corresponding to the target point cloud is extracted by the inverse tangent function, and the phase sequence is subjected to phase unwrapping processing. The phase of the target is extracted once in each frame period, and the phase of the target point cloud obtained by cyclic transmission changes with the value of the frame number, that is, the small displacement of the target point cloud changes with time; S302: Calculate the phase change of the target point cloud, expressed as: ； in, represents the phase change, represents the radar wavelength, Indicates the displacement change caused by the target point cloud; S303, calculating the deformation value of the target point cloud; According to the correspondence between the phase and displacement, the result of the deformation change over time after filtering is obtained, and the deformation characteristics of the distance unit where the target point cloud is located are extracted and analyzed. The deformation value expression of the distance unit where the target point cloud is located is: ； in, Represents the distance unit index, through which the position of the target point cloud in the distance direction can be determined. Indicates the index of a single electromagnetic wave chirp emitted by the radar. represents the radar electromagnetic wave emission period, Represents the radar phase of the target point cloud after filtering; By The distance unit where the target point cloud is continuously extracted within a certain time period is The phase at obtains the vibration frequency value; S304: Perform fast Fourier transform on the filtered phase signal, and obtain the corresponding frequency within a preset time according to the peak value and its harmonic characteristics.

5. The mine monitoring method based on millimeter wave radar multi-feature fusion according to claim 4 is characterized in that: After extracting the phase information of the distance unit corresponding to the target point cloud, the phase information is unwrap- ed and, based on the distribution range of the target frequency, a band-pass filter is used to distinguish the vibration signal of the target frequency band to obtain continuous phase changes.

6. The mine monitoring method based on millimeter wave radar multi-feature fusion according to claim 1 is characterized in that: The distance change confidence counter is , if the distance change Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The speed change confidence counter is , if the speed change Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The angle change confidence counter is , if the angle change Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The reflected energy variation confidence counter is , if the reflected energy change Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The deformation value confidence counter is , if the deformation value Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The vibration frequency value confidence counter is , if the vibration frequency value Greater than or equal to the preset warning threshold , the counter accumulates , otherwise 0; The confidence counter score expressions of the six feature changes are: ； in, Indicates the feature change confidence counters, Indicates the The feature variation, Indicates the A pre-set warning threshold for feature changes.

Images