Detection system and detection method for high-altitude falling object

Abstract

The invention discloses a high-altitude falling object detection system and method. The high-altitude falling object detection system comprises a multi-sensor fusion module, an image acquisition module, a data processing module, a track prediction module, an early warning module and a data storage module. Through a multi-sensor fusion technology, vibration, sound waves, radar and image data are combined, so that the detection precision is remarkably improved, and the false alarm rate is reduced; real-time calculation of a falling point of a falling object is realized through a track prediction module, and decision support is provided for evacuation and protection; infrared thermal imaging and a wide-angle camera work cooperatively, so that day and night all-weather monitoring is ensured; and the data storage module supports event backtracking and responsibility determination, and introduction of the block chain technology further guarantees data credibility. In addition, the system can be in linkage with equipment such as elevators and roadblocks, active protection is achieved, and the public safety level is greatly improved.

Description

Technical Field

[0001] This application relates to the technical field of high-altitude falling object detection systems, and particularly to a detection system and method for high-altitude falling objects. Background Art

[0002] In recent years, high-altitude falling object incidents have been frequent, posing great potential safety hazards to the lives of pedestrians and public property. Currently, most of the mainstream high-altitude falling object monitoring technologies in the market rely on a single type of sensor, such as cameras or vibration sensors. However, these technologies have exposed many deficiencies in practical applications.

[0003] Specifically, the single-sensor solution has a problem of high false alarm rate and is difficult to accurately identify real high-altitude falling object incidents. At the same time, they cannot perform real-time positioning and tracking of the movement trajectory of the falling object, making it difficult for relevant departments to timely grasp the dynamics of the falling object. Moreover, in low-light environments such as at night, the detection ability will be greatly reduced. Taking the pure vision monitoring solution as an example, its detection effect is extremely vulnerable to lighting conditions. In insufficient light or complex backgrounds, it is very difficult to accurately distinguish the falling object from other background interference objects, resulting in misjudgment. Although the vibration sensor can sense vibration signals, it is easily interfered by ambient noise and is difficult to accurately trace the source of the falling object.

[0004] Moreover, existing monitoring systems generally lack the ability to predict the movement trajectory of falling objects. This means that after a falling object occurs, the system cannot predict the landing point of the falling object in advance, and it is difficult to issue effective warning information in the first place, and it cannot provide enough time for pedestrians to take shelter. Summary of the Invention

[0005] Therefore, this application provides a detection system and method for high-altitude falling objects to solve the problems of high false alarm rate and poor detection ability existing in the prior art.

[0006] To achieve the above object, this application provides the following technical solutions:

[0007] In a first aspect, a detection system for high-altitude falling objects includes a multi-sensor fusion module, an image acquisition module, a data processing module, a trajectory prediction module, an early warning module, and a data storage module;

[0008] The multi-sensor fusion module is composed of a vibration sensor array, an acoustic wave sensor, and a millimeter-wave radar. The vibration sensor array is arranged on the outer facade of a building or the edge of a high-altitude platform, and the vibration sensor array is used to detect in real time the vibration signals caused by the impact of a falling object or the detachment of an object; the acoustic wave sensor is installed on the same side as the vibration sensor and is used to capture the acoustic wave characteristics generated by the friction of the falling object with the air during the falling process; the millimeter-wave radar is oriented at an inclined angle towards the monitoring area and is used to detect the dynamic displacement and speed of the falling object during the fall;

[0009] The image acquisition module includes a wide-angle camera and an infrared thermal imager. The wide-angle camera covers the vertical falling range of the monitoring area, and the infrared thermal imager is used to capture the outline of falling objects at night or under low light conditions;

[0010] The data processing module is connected to the multi-sensor fusion module and the image acquisition module. The data processing module includes a signal filtering unit, a feature extraction unit, and a pattern recognition unit. The signal filtering unit is used to eliminate environmental noise interference. The feature extraction unit generates a falling object feature vector based on vibration frequency, acoustic spectrum, and radar cross-section data. The pattern recognition unit distinguishes natural falling objects from human-thrown objects through a pre-trained falling object classification model;

[0011] The trajectory prediction module is used to receive the output data of the data processing module, and calculate the movement trajectory by combining the initial position, velocity, and gravitational acceleration of the falling object, and predict the landing point range; The warning module includes an audible and visual alarm and a wireless communication unit. When the distance between the predicted landing point and ground pedestrians or vehicles is less than a preset threshold, the audible and visual alarm is triggered and an alarm message is sent to the management platform through the wireless communication unit;

[0012] The data storage module is used to record the sensor data, image data, and predicted trajectory of the falling object event, and record the associated timestamp and location information.

[0013] Preferably, the vibration sensor array is arranged in a grid form, the distance between adjacent sensors is 0.5 - 1.5 meters, the installation height is 8 - 30 meters from the ground, and each sensor is built-in with a triaxial accelerometer.

[0014] Preferably, the operating frequency of the millimeter-wave radar is 24 GHz or 77 GHz, the detection angle is 15° - 45°, the detection distance is 5 - 50 meters, and the overlap degree between the radar data and the field of view of the image acquisition module is greater than 80%.

[0015] Preferably, the feature extraction unit of the data processing module uses wavelet transform to perform time-frequency analysis on the vibration signal, and extracts the falling object features in the acoustic signal through the matching pursuit algorithm.

[0016] Preferably, it further includes a barometric pressure sensor module. The barometric pressure sensor is installed at the top of the monitoring area, used to detect the barometric pressure when the falling object is falling, and input the barometric pressure data into the trajectory prediction module to correct the influence of air resistance on the trajectory calculation.

[0017] Preferably, the alarm trigger condition of the trajectory prediction module is: the mass of the falling object is greater than 0.5 kg, the falling speed exceeds 5 m / s, or the horizontal distance between the predicted landing point and the protected area is less than 3 meters.

[0018] Preferably, the data storage module uses blockchain technology to encrypt and store the data of the falling object event, and generates an immutable timestamp and location hash value.

[0019] In a second aspect, a detection method for high-altitude falling objects includes the following steps:

[0020] S1. The high-altitude area is monitored in real time by a vibration sensor array and a sound wave sensor. When an abnormal vibration or sound wave signal is detected, the millimeter-wave radar and the image acquisition module are triggered to start.

[0021] S2. The vibration, sound wave, radar, and image data are fused to extract the characteristics of the falling object and classify them.

[0022] S3. Based on the initial position, velocity, and environmental parameters of the falling object, the movement trajectory is calculated, and the landing point range is predicted.

[0023] S4. If the predicted landing point is within the preset warning area, an audible and visual alarm is triggered, and the event data is uploaded to the management platform.

[0024] S5. The complete event data is stored, including the original sensor data, classification results, trajectory prediction results, and alarm records.

[0025] Compared with the prior art, the present application has at least the following beneficial effects:

[0026] Through the multi-sensor fusion technology, combining vibration, sound wave, radar, and image data, the present invention significantly improves the detection accuracy and reduces the false alarm rate; through the trajectory prediction module, the real-time calculation of the landing point of the falling object is realized, providing decision support for evacuation and protection; the coordinated work of the infrared thermal imaging and the wide-angle camera is adopted to ensure round-the-clock monitoring day and night; the data storage module supports event backtracking and liability determination, and the introduction of blockchain technology further guarantees the credibility of the data. In addition, the system can be linked with devices such as elevators and roadblocks to achieve active protection, greatly improving the public safety level. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] To more intuitively illustrate the prior art and the present application, exemplary drawings are given below. It should be understood that the specific shapes and structures shown in the drawings generally should not be regarded as limiting conditions when implementing the present application; for example, those skilled in the art are capable of making routine adjustments or further optimizations to the addition / deletion / attribution division of certain units (components), specific shapes, positional relationships, connection methods, dimensional proportional relationships, etc. based on the technical concept disclosed in the present application and the exemplary drawings.

[0028] Figure 1 It is a module diagram of a detection system and a detection method for high-altitude falling objects of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The present application will be further described in detail below with reference to the accompanying drawings and through specific embodiments.

[0030] As Figure 1 shown, a detection system for high-altitude falling objects, characterized in that it includes a multi-sensor fusion module, an image acquisition module, a data processing module, a trajectory prediction module, an early warning module and a data storage module;

[0031] The multi-sensor fusion module consists of a vibration sensor array, an acoustic wave sensor and a millimeter-wave radar. The vibration sensor array is arranged on the outer facade of a building or the edge of a high-altitude platform, and the vibration sensor array is used to detect in real time the vibration signals caused by the impact of falling objects or the detachment of objects; The acoustic wave sensor is installed on the same side as the vibration sensor and is used to capture the acoustic wave characteristics generated by the friction between the falling object and the air during the falling process; The millimeter-wave radar is oriented towards the monitoring area at an inclined angle and is used to detect the dynamic displacement and speed of the falling object when it is falling;

[0032] The image acquisition module includes a wide-angle camera and an infrared thermal imager. The wide-angle camera covers the vertical falling range of the monitoring area, and the infrared thermal imager is used to capture the outline of the falling object at night or under low-light conditions;

[0033] The data processing module is connected to the multi-sensor fusion module and the image acquisition module. The data processing module includes a signal filtering unit, a feature extraction unit and a pattern recognition unit. The signal filtering unit is used to eliminate environmental noise interference. The feature extraction unit generates a falling object feature vector based on vibration frequency, acoustic wave spectrum and radar cross-section data. The pattern recognition unit distinguishes natural falling objects from human-thrown objects through a pre-trained falling object classification model;

[0034] The trajectory prediction module is used to receive the output data of the data processing module, and calculate the movement trajectory in combination with the initial position, speed and gravitational acceleration of the falling object, and predict the landing point range; The early warning module includes an audible and visual alarm and a wireless communication unit. When the distance between the predicted landing point and pedestrians or vehicles on the ground is less than a preset threshold, an audible and visual alarm is triggered and an alarm message is sent to the management platform through the wireless communication unit;

[0035] The data storage module is used to record the sensor data, image data and predicted trajectory of the falling object event, and record the associated timestamp and location information.

[0036] When the system starts, the vibration sensor array deployed on the building facade continuously detects high-frequency vibration signals. If abnormal vibrations (such as object impacts or detachments) are detected, the acoustic wave sensor and millimeter-wave radar are triggered synchronously. The acoustic wave sensor captures the specific spectral characteristics (such as high-frequency whistling or low-frequency airflow disturbances) generated by the falling object rubbing against the air. The millimeter-wave radar calculates the displacement, velocity, and movement direction of the falling object in real time by emitting electromagnetic waves and receiving the reflected signals. At the same time, the wide-angle camera of the image acquisition module quickly locates the vibration source area and captures the initial form of the falling object. The infrared thermal imager supplements the identification of the thermal radiation contour of the falling object under low-light conditions. The data processing module performs fusion processing on multi-source signals: the signal filtering unit eliminates environmental noises (such as wind sounds and vehicle vibrations), the feature extraction unit generates a multi-dimensional feature vector by combining vibration frequencies, acoustic wave spectra, and radar cross-sections, and the pattern recognition unit distinguishes natural falling objects (such as falling building materials) from human-thrown objects (such as garbage or heavy objects) based on pre-trained machine learning models (such as support vector machines or convolutional neural networks). The trajectory prediction module constructs a kinematic model based on the initial height, velocity, mass, and environmental parameters (such as wind speed) of the falling object, calculates its parabolic trajectory in real time, and predicts the landing range. If the distance between the predicted landing point and pedestrians, vehicles, or protected areas is lower than the preset threshold, the warning module immediately activates the audible and visual alarm and pushes the location, image, and trajectory data to the management platform through wireless communication. The data storage module encrypts and archives the whole-process data, supporting post-event traceability and liability tracing. The entire system significantly reduces the false alarm rate through multi-sensor redundant verification and dynamic trajectory correction, and realizes a closed-loop response from detection to protection.

[0037] The vibration sensor array is deployed in a grid form, with the adjacent sensor spacing being 0.5 - 1.5 meters and the installation height being 8 - 30 meters from the ground. Moreover, each sensor is equipped with a triaxial accelerometer. The sensors are evenly distributed in a grid form on the building facade or the edge of the high-altitude platform, with an adjacent spacing of 0.5 - 1.5 meters, ensuring the precise positioning of local vibrations and avoiding monitoring blind spots. The installation height is set at 8 - 30 meters from the ground, covering the starting heights of common falling objects (such as balconies and windows). Each sensor is equipped with a triaxial accelerometer, which can detect three-dimensional vibration signals (such as horizontal impacts and vertical detachments). By combining multi-node data cross-verification, it can effectively distinguish the vibrations of falling objects from other interferences (such as construction vibrations).

[0038] The operating frequency of the millimeter-wave radar is 24 GHz or 77 GHz. Selecting 24 GHz or 77 GHz for the radar operating frequency takes into account both high resolution and strong penetration ability, enabling stable detection in harsh weather conditions such as rain and fog. The detection angle is 15° - 45°, and the detection angle is set to 15° - 45° to obliquely cover the falling path of the monitoring area and avoid ground clutter interference. The detection distance is 5 - 50 meters, and the overlap between the radar data and the field of view of the image acquisition module is greater than 80%. The detection distance of 5 - 50 meters is adapted to different building heights and overlaps with the camera's field of view by more than 80%, ensuring the spatial alignment of radar data and visual information for easy fusion analysis (such as calibrating the falling object's motion trajectory in the image through radar speed data).

[0039] The feature extraction unit of the data processing module uses wavelet transform to perform time-frequency analysis on the vibration signal. In view of the non-stationary characteristics of the vibration signal, wavelet transform is used for time-frequency analysis to extract the frequency distribution characteristics of transient impact signals (such as specific frequency bands of building material fractures); the matching pursuit algorithm is used to extract the falling object features in the acoustic signal. For the acoustic signal, the matching pursuit algorithm is used to separate the transient acoustic fingerprint of the falling object from the ambient noise (such as the sharp frequency spectrum of glass breaking). By fusing the time-frequency characteristics of vibration and sound waves, a highly discriminative falling object feature vector is constructed. After inputting into the pattern recognition unit, the type of falling object can be accurately classified (such as distinguishing between tile detachment and throwing objects), and the interference of similar signals (such as thunder or mechanical noise) can be suppressed.

[0040] It further includes a barometric pressure sensor module. The barometric pressure sensor is installed at the top of the monitoring area and is used to detect the barometric pressure when a falling object drops, and input the barometric pressure data into the trajectory prediction module to correct the influence of air resistance on trajectory calculation. This sensor is installed at the top of the monitoring area to detect the barometric pressure data in real time when a falling object occurs. After the barometric pressure data is input into the trajectory prediction module, the influence of air resistance on the acceleration of the falling object is calculated in combination with the fluid mechanics model, and the trajectory prediction result is dynamically corrected (such as correcting the drift error of light objects).

[0041] In some implementation schemes, the alarm trigger conditions of the trajectory prediction module are: the mass of the falling object is greater than 0.5 kg, the falling speed exceeds 5 m / s, or the horizontal distance between the predicted landing point and the protection area is less than 3 meters.

[0042] Of course, the above threshold data can also be other values and can be freely adjusted according to the needs of users.

[0043] The data storage module uses blockchain technology to encrypt and store the data of falling object events and generates an immutable timestamp and location hash value.

[0044] When a falling object incident occurs, the system encrypts sensor data, images, and predicted trajectories and uploads them to the blockchain node, generating an immutable record containing a timestamp and a location hash value. For example, the original waveform of the vibration signal and the classification result are jointly packaged into a data block and distributedly stored through a consensus mechanism to ensure the integrity and legal validity of the data during retrospective investigation. This design effectively prevents data tampering (such as artificially deleting alarm records) and provides a credible evidence chain for liability determination (such as locking the time of the falling object in a dispute between the property management and the owner).

[0045] By fusing high-time-resolution event streams with deep learning feature enhancement techniques, dynamic analysis and accurate identification of falling object events in the complex environment of the RX plant are achieved. The microsecond-level displacement changes of falling objects are effectively captured, and the feature extraction of small targets is strengthened by combining spatio-temporal feature enhancement modules, effectively distinguishing falling objects from background interference. Currently, the algorithm detection has achieved an accuracy greater than 95%, a precision greater than 90%, and a recall rate greater than 88%.

[0046] The system combines an RGB vision sensor to quickly locate the spatio-temporal position of the falling object. During the trial operation of the overhaul, the system effectively overcomes complex environmental interference factors such as vibration, lighting, temperature, and electromagnetic fields at the RX plant site, solves operation interference factors such as equipment hoisting, refueling operations, and personnel movement, and integrates technologies such as dynamic vision sensors, AI video algorithms, and edge computing.

[0047] A detection method for high-altitude falling objects includes the following steps:

[0048] S1. The high-altitude area is monitored in real time through a vibration sensor array and a sound wave sensor. When an abnormal vibration or sound wave signal is detected, the millimeter-wave radar and the image acquisition module are triggered to start;

[0049] S2. The vibration, sound wave, radar, and image data are fused to extract the characteristics of the falling object and classify it;

[0050] S3. Based on the initial position, velocity, and environmental parameters of the falling object, the motion trajectory is calculated to predict the landing point range;

[0051] S4. If the predicted landing point is within the preset warning area, an audible and visual alarm is triggered and the event data is uploaded to the management platform;

[0052] S5. The complete event data is stored, including the original sensor data, classification results, trajectory prediction results, and alarm records.

[0053] The technical features of the above embodiments can be combined arbitrarily (as long as there is no contradiction in the combination of these technical features). For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described; these embodiments not explicitly written should also be considered to be within the scope described in this specification.

Claims

1. A detection system for high-altitude falling objects, characterized in that It includes multi-sensor fusion module, image acquisition module, data processing module, trajectory prediction module, early warning module and data storage module; The multi-sensor fusion module is composed of a vibration sensor array, an acoustic wave sensor and a millimeter wave radar. The vibration sensor array is arranged on the facade of the building or the edge of the high-altitude platform. The vibration sensor array is used to detect the vibration signal caused by the impact of falling objects or the detachment of objects in real time; the acoustic wave sensor is installed on the same side of the vibration sensor to capture the acoustic wave characteristics generated by the friction between the falling objects and the air during the falling process; The millimeter-wave radar is tilted toward the monitoring area to detect the dynamic displacement and speed of falling objects; The image acquisition module includes a wide-angle camera and an infrared thermal imager. The wide-angle camera covers the vertical falling range of the monitoring area, and the infrared thermal imager is used to capture the outline of falling objects at night or in low light conditions. The data processing module is connected to the multi-sensor fusion module and the image acquisition module. The data processing module includes a signal filtering unit, a feature extraction unit and a pattern recognition unit. The signal filtering unit is used to eliminate environmental noise interference. The feature extraction unit generates a falling object feature vector based on vibration frequency, sound wave spectrum and radar reflection cross-section data. The pattern recognition unit distinguishes between natural falling objects and man-made thrown objects through a pre-trained falling object classification model. The trajectory prediction module is used to receive the output data of the data processing module, and calculate the motion trajectory in combination with the initial position, speed and gravity acceleration of the falling object, and predict the range of the landing point; the early warning module includes an audible and visual alarm and a wireless communication unit, and when the distance between the predicted landing point and the pedestrian or vehicle on the ground is less than a preset threshold, the audible and visual alarm is triggered and the alarm information is sent to the management platform through the wireless communication unit; The data storage module is used to record sensor data, image data and predicted trajectory of the falling object event, and record associated timestamps and location information.

2. The detection system for high-altitude falling objects according to claim 1, characterized in that, The vibration sensor array is arranged in a grid form, with a spacing of 0.5-1.5 meters between adjacent sensors, an installation height of 8-30 meters from the ground, and each sensor has a built-in triaxial accelerometer.

3. The detection system for high-altitude falling objects according to claim 1, characterized in that, The operating frequency of the millimeter wave radar is 24GHz or 77GHz, the detection angle is 15°-45°, the detection distance is 5-50 meters, and the overlap between the radar data and the field of view of the image acquisition module is greater than 80%.

4. A detection system for high-altitude falling objects according to claim 1, characterized in that, The feature extraction unit of the data processing module uses wavelet transform to perform time-frequency analysis on the vibration signal, and extracts the falling object features in the sound wave signal through a matching pursuit algorithm.

5. The detection system for high-altitude falling objects according to claim 1, characterized in that, It also includes an air pressure sensor module, which is installed on the top of the monitoring area and is used to detect the air pressure when the falling object falls, and input the air pressure data into the trajectory prediction module to correct the influence of air resistance on trajectory calculation.

6. The detection system for high-altitude falling objects according to claim 1, characterized in that, The alarm triggering conditions of the trajectory prediction module are: the mass of the falling object is greater than 0.5kg, the falling speed exceeds 5m / s, or the horizontal distance between the predicted landing point and the protection area is less than 3 meters.

7. The detection system for high-altitude falling objects according to claim 1, characterized in that, The data storage module uses blockchain technology to encrypt and store the falling object event data, and generate an unalterable timestamp and location hash value.

8. A detection method for high-altitude falling objects, characterized in that, The following steps are involved: S1. Monitor the high-altitude area in real time through the vibration sensor array and acoustic wave sensors. When abnormal vibration or acoustic wave signals are detected, trigger the millimeter-wave radar and image acquisition module to start; S2. Integrate the vibration, acoustic wave, radar and image data, extract the characteristics of falling objects and classify them; S3. Calculate the movement trajectory based on the initial position, velocity and environmental parameters of the falling object, and predict the landing range; S4. If the predicted landing point is within the preset warning area, trigger an audible and visual alarm and upload the event data to the management platform; S5. Store the complete event data, including the original sensor data, classification results, trajectory prediction results and alarm records.

Images