Intelligent tent data acquisition method and rescue early warning platform

Abstract

The invention discloses an intelligent tent data acquisition method and a rescue early warning platform, and belongs to the technical field of data acquisition. Environmental monitoring, attitude and image multi-source data are acquired through distributed nodes, and through dynamic filtering and time domain alignment preprocessing, data are fused by using a dynamic weight distribution matrix; abnormality is detected in combination with a dynamic threshold model based on historical data and a real-time environment, and transmission is performed through a multi-channel protocol after continuous period early warning is triggered. The rescue early warning platform comprises a data receiving module, a data processing server and a rescue information distribution system with differential broadcasting and receipt retransmission functions. Through multi-source sensor fusion and a dynamic weight distribution matrix, the comprehensiveness and precision of environmental risk monitoring are remarkably improved; and a dynamic threshold model and a redundant transmission protocol are combined, so that the false alarm rate is reduced, and second-level early warning response is realized.

Description

Technical Field

[0001] The present invention relates to the field of data acquisition, and particularly to an intelligent tent data acquisition method and a rescue warning platform. Background Art

[0002] Currently, intelligent tents in the field of outdoor emergencies generally have technical bottlenecks in data acquisition and processing capabilities. Existing systems mostly rely on single-point sensor deployment, making it difficult to achieve multi-dimensional collaborative perception of environmental parameters, structural deformation, and visual information, resulting in limited risk assessment accuracy in complex scenarios. In the data preprocessing link, a fixed-parameter filtering algorithm is used, which cannot adapt to the sampling frequency fluctuation and image noise mutation problems under extreme weather. The data fusion lacks a dynamic weight allocation mechanism and is difficult to optimize the fusion strategy in real time according to the parameter correlation degree, weakening the representation ability of multi-dimensional feature vectors. Anomaly detection relies on a static threshold model and cannot adapt to dynamic environments such as day-night temperature differences and terrain differences, easily resulting in false alarms or missed alarms. The emergency communication module mostly uses a single public network transmission method, with communication blind spots in remote areas and a lack of data encryption mechanism. At the rescue response level, existing platforms rely on manual scheduling, unable to achieve intelligent matching of rescue forces based on three-dimensional terrain, with low information distribution efficiency and a lack of personalized push capabilities. Summary of the Invention

[0003] In view of the above technical problems, the present invention provides an intelligent tent data acquisition method, including: Step S1: Obtain multi-source sensor data through distributed data acquisition nodes. The sensors include an environmental monitoring sensor group, an attitude sensor, and an image acquisition device, where the environmental monitoring sensor group includes temperature, humidity, and pressure sensors.

[0004] Step S2: Preprocess the original sensor data, including selecting a corresponding digital filtering algorithm according to the sensor type. The temperature sensor uses a sliding average filter with dynamic window adjustment, and the image data uses adaptive noise reduction processing based on region segmentation; add time stamp marks to each sensor data through a unified clock source, and use an interpolation algorithm to achieve time domain alignment for non-equidistant sampled data.

[0005] Step S3: Perform fusion processing on the preprocessed multi-source data based on time series analysis methods, construct a dynamic weight allocation matrix including sensor confidence weights and environmental parameter correlation degrees; dynamically adjust the fusion coefficient according to the real-time environmental parameter change rate, and generate a multi-dimensional feature vector including temperature gradient change rate, structural deformation index, and visible visibility value.

[0006] Step S4, an anomaly status detection is performed using a dynamic threshold model, and the dynamic threshold model updates the threshold through a preset mode, where the preset mode includes calculating the probability density distribution mode of each environmental parameter based on historical data and dynamically correcting the threshold by combining the weather type and terrain features collected in real time.

[0007] Step S5, when the same anomaly parameter threshold is triggered within three consecutive detection cycles, a secondary warning signal including geographical coordinates, anomaly parameter type, and confidence score is generated and sent to the rescue warning platform through a multi-channel redundant transmission protocol.

[0008] Furthermore, the sliding average filtering with dynamic window adjustment automatically adjusts the window length according to the current sampling frequency; the image adaptive noise reduction processing includes dividing the image area based on color space features and identifying the noise distribution; a first type of noise reduction algorithm is used for high-noise areas, and a second type of noise reduction algorithm is used for low-noise areas.

[0009] Furthermore, the method for generating the dynamic weight allocation matrix includes Step S301, calculating the real-time signal-to-noise ratio of each sensor data through a filtering algorithm as the confidence weight.

[0010] Step S302, calculating the correlation degree between environmental parameters based on parameter correlation, and automatically increasing the corresponding sensor weight coefficient when the correlation degree exceeds the set correlation degree threshold.

[0011] Step S303, automatically enabling the data of the same type of sensor of adjacent nodes to replace the failed sensor and marking the data source label.

[0012] Furthermore, the fusion processing includes constructing a sensor data weight allocation matrix, dynamically adjusting the fusion coefficient based on the environmental parameter correlation degree; generating a multi-dimensional feature vector including the temperature gradient change rate, structure deformation index, and visible visibility value.

[0013] Furthermore, the anomaly status detection includes: establishing a dynamic threshold adjustment model, automatically updating the decision threshold based on historical data and real-time environmental parameters; generating a secondary warning signal when the threshold is triggered in three consecutive detection cycles.

[0014] A rescue warning platform includes: A data receiving module, including real-time receiving the monitoring data streams of multiple intelligent tents.

[0015] A data processing server, including a data cleaning unit for removing duplicate data and invalid data packets; a feature extraction unit for parsing the key parameters in the multi-dimensional feature vector; a warning judgment unit for performing risk assessment and grading based on a decision tree model; a communication gateway that supports multi-protocol conversion and has a breakpoint resumption function.

[0016] Further, the communication gateway includes a Beidou satellite communication interface for emergency communication in areas without public network coverage, and a data encryption unit that uses national cryptographic algorithms to perform end-to-end encryption on the transmitted data.

[0017] Further, the early warning judgment unit includes a multi-level early warning trigger mechanism that sets three-level response criteria of yellow, orange, and red, and a geographical fence generation module that automatically delimits the affected area based on the coordinates of the early warning tent.

[0018] It also includes a rescue information distribution system, which includes automatically dispatching rescue work orders containing three-dimensional topographic maps to pre-registered rescue agencies, sending disaster avoidance guidance information to affected tent terminals, generating public early warning information and publishing it through designated social media interfaces. The distribution system uses differential broadcast technology to achieve personalized information push for different terminals, and establishes an information receipt confirmation mechanism to initiate a retransmission process for unconfirmed receiving terminals.

[0019] Further, the data processing server also includes a data persistence module that stores raw monitoring data using a time series database, a cache acceleration unit that uses in-memory computing technology to achieve real-time data analysis, and a log audit system that completely records data operation traces and generates compliance reports.

[0020] The beneficial effects of the present invention compared with the prior art are as follows: (1) Through multi-source sensor fusion and dynamic weight allocation matrix, the present invention significantly improves the comprehensiveness and accuracy of environmental risk monitoring; combined with a dynamic threshold model and a redundant transmission protocol, it reduces the false alarm rate and achieves a second-level early warning response. (2) The present invention uses a time series database and in-memory computing technology to break through the bottleneck of multi-terminal data collaborative processing efficiency and provide intelligent and systematic decision-making support for complex rescue scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is an exemplary step flow chart of the data acquisition method of the present invention.

[0022] Figure 2 It is an exemplary step flow chart of the generation method of the dynamic weight allocation matrix of the present invention.

[0023] Figure 3 It is a schematic block diagram of the composition of the early warning platform of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0024] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] Embodiments of the present invention provide a method for collecting intelligent tent data, including, Step S1: Obtain multi-source sensor data through distributed data collection nodes. The sensors include an environmental monitoring sensor group, an attitude sensor, and an image acquisition device. The environmental monitoring sensor group includes temperature, humidity, and air pressure sensors.

[0026] Exemplarily, the environmental parameters of the intelligent tent have multi-modal coupling characteristics. For example, the correlation between temperature gradient changes and air pressure data can reflect the trend of meteorological mutations, and the tilt angle data of the attitude sensor needs to be fused with the visual information of the image acquisition device to accurately identify the risk of tent structure deformation.

[0027] The distributed data collection nodes are deployed at the key stress points of the tent support frame and the peripheral environment according to the topological structure. The environmental monitoring sensor group uses a wireless micro-power consumption module to synchronously collect data on the temperature difference inside and outside the tent, humidity distribution, and air pressure fluctuations. The attitude sensor integrates a three-axis accelerometer and a gyroscope to monitor the displacement and vibration characteristics of the tent skeleton at a sampling frequency of 10 Hz. The image acquisition device is equipped with a wide-angle camera and an infrared thermal imaging module, which are used for visual environment monitoring and nocturnal biological activity perception respectively.

[0028] In this embodiment, the construction target can be dynamically adjusted according to the disaster scenario: in the regular camping mode, the basic environmental parameters (temperature, humidity) are the core of data collection; in the extreme weather warning mode, the collaborative work of the air pressure sensor and the image device is activated, and the risk of thunderstorms or strong winds is predicted through the time-series correlation analysis of air pressure mutation data and cloud images.

[0029] Step S2: Preprocess the original sensor data, including selecting the corresponding digital filtering algorithm according to the sensor type. The temperature sensor uses a sliding average filter with dynamic window adjustment, and the image data uses adaptive noise reduction processing based on region segmentation. Add time stamp marks to the sensor data through a unified clock source, and use an interpolation algorithm to achieve time-domain alignment for non-equidistant sampled data.

[0030] Exemplarily, the dynamic window adjustment logic of the temperature sensor is based on the change rate of environmental parameters: when the temperature volatility exceeds the threshold, the window length is automatically shortened to 5 sampling points to improve the response speed, and vice versa, it is extended to 20 sampling points to suppress high-frequency noise. In the image noise reduction process, the non-local mean filter algorithm is used in high-noise areas (such as sandstorm interference areas), and the fast bilateral filter is used in low-noise areas, taking into account both noise reduction efficiency and detail retention.

[0031] Step S3: Based on the time series analysis method, fuse the preprocessed multi-source data to construct a dynamic weight allocation matrix that includes the sensor confidence weight and the environmental parameter correlation degree; dynamically adjust the fusion coefficient according to the real-time environmental parameter change rate to generate a multi-dimensional feature vector that includes the temperature gradient change rate, the structural deformation index, and the visible visibility value.

[0032] Step S4: Use a dynamic threshold model for abnormal state detection. The dynamic threshold model updates the threshold through a preset mode, and the preset mode includes calculating the probability density distribution mode of each environmental parameter based on historical data and dynamically correcting the threshold by combining the real-time collected weather type and terrain features.

[0033] Exemplarily, in the probability density distribution mode, the setting of the temperature threshold follows the 3σ principle: taking the historical data mean ± 3 times the standard deviation as the normal range; in the terrain feature correction mode, the wind speed threshold in the mountainous area scene is reduced by 20%, and the rainfall threshold in the plain area is increased by 15%.

[0034] In this embodiment, the implementation logic of the dynamic threshold correction includes: weather type classifier: training an SVM model based on image data and air pressure sequence to identify 6 types of weather such as thunderstorms and sandstorms in real time; terrain database matching: calling the pre-stored terrain parameters (such as slope and altitude) through GPS coordinates to correct the temperature and wind speed thresholds; time decay factor: introducing an exponential decay weight to the historical data to ensure that the threshold adapts to the long-term change trend of the environment.

[0035] Step S5: When the same abnormal parameter threshold is triggered within three consecutive detection cycles, generate a secondary warning signal that includes geographical coordinates, abnormal parameter type, and confidence score, and send it to the rescue warning platform through a multi-channel redundant transmission protocol.

[0036] In the embodiment, the content specification of the warning signal includes: geographical coordinates: fusing GPS and inertial navigation data, with a positioning accuracy of up to the meter level; abnormal parameter type: encoded as an 8-bit binary field, supporting 256 abnormal combinations; additional information: carrying an environmental data snapshot and the associated image frame hash value of the last 3 minutes. In one embodiment, for the extreme environment without network, the compressed sensing coding technology is enabled: compressing the warning signal to 10% of the original data volume and transmitting it in batches through the Beidou short message; in another embodiment, a feedback-driven adaptive retransmission mechanism is established: dynamically adjusting the packet fragmentation size according to the channel quality, and enabling forward error correction coding when the packet loss rate > 30%.

[0037] The sliding average filtering with dynamic window adjustment automatically adjusts the window length according to the current sampling frequency; the image adaptive noise reduction processing includes dividing the image area based on the color space characteristics and identifying the noise distribution; using the first type of noise reduction algorithm for high-noise areas and the second type of noise reduction algorithm for low-noise areas.

[0038] As Figure 2 shown in the generation method of the dynamic weight allocation matrix of this embodiment, including, Step S301, calculating the real-time signal-to-noise ratio of each sensor data as the confidence weight through a filtering algorithm.

[0039] Step S302, calculating the correlation degree between environmental parameters based on parameter correlation, and automatically increasing the corresponding sensor weight coefficient when the correlation degree exceeds the set correlation degree threshold.

[0040] Step S303, automatically enabling the data of the same type of sensors of adjacent nodes to replace the failed sensors, and marking the data source mark.

[0041] The fusion processing includes constructing a sensor data weight allocation matrix, dynamically adjusting the fusion coefficient based on the environmental parameter correlation degree; generating a multi-dimensional feature vector including the temperature gradient change rate, the structural deformation index, and the visible visibility value.

[0042] In this embodiment, the construction of the weight allocation matrix adopts an improved entropy weight-TOPSIS algorithm: calculating the standard deviation and mutual information entropy of each sensor data through a sliding window, dynamically allocating the weight coefficient, and activating the cross-modal fusion channel when the mutual information entropy of the air pressure and temperature data exceeds 0.7; the multi-dimensional feature vector extracts time-frequency domain features through wavelet transform, for example, the tent structure deformation index is calculated from the proportion of the energy of the three-layer wavelet packet of the vibration signal of the support rod.

[0043] The abnormal state detection includes establishing a dynamic threshold adjustment model, automatically updating the determination threshold based on historical data and real-time environmental parameters; generating a secondary warning signal when the threshold is triggered for three consecutive detection periods.

[0044] As Figure 3 shown is a rescue warning platform provided by the method of this embodiment, including, A data receiving module, including real-time receiving the monitoring data streams of multiple intelligent tents.

[0045] Exemplarily, an Apache Kafka is used to build a distributed message queue, supporting the processing of 100,000 concurrent data streams per second; the data streams are stored in hash shards according to the tent ID to ensure high throughput and low latency.

[0046] A data processing server, including a data cleaning unit for removing duplicate data and invalid data packets; a feature extraction unit for parsing the key parameters in the multi-dimensional feature vector; a warning judgment unit for performing risk assessment and classification based on a decision tree model; a communication gateway, supporting multi-protocol conversion and having the function of resuming data transfer from the breakpoint.

[0047] The communication gateway includes a Beidou satellite communication interface for emergency communication in areas without public network coverage, and a data encryption unit that uses national cryptographic algorithms to perform end-to-end encryption on the transmitted data.

[0048] The early warning judgment unit includes: a multi-level early warning trigger mechanism that sets three-level response criteria of yellow, orange, and red; and a geographic fence generation module that automatically delimits the affected area based on the coordinates of the early warning tent.

[0049] In this embodiment, the three-level response criteria are as follows: Yellow early warning: Trigger an abnormality of 1 minor parameter (such as humidity > 90%), and notify the local administrator; Orange early warning: 2 major parameter abnormalities (such as wind speed > 20 m / s + deformation index > threshold), start area broadcasting; Red early warning: The vital sign sensor detects that a person is unconscious, and link the emergency attendance of the drone. The geographic fence is generated based on the Delaunay triangulation algorithm, with the early warning tent as the center, to generate an impact area with a radius of 500 meters, and dynamically superimpose the terrain undulation data to generate an evacuation path.

[0050] It also includes a rescue information distribution system, which includes automatically dispatching rescue work orders containing three-dimensional topographic maps to pre-registered rescue agencies, sending disaster avoidance guidance information to the affected tent terminals, generating public early warning information and publishing it through designated social media interfaces.

[0051] The distribution system uses differential broadcasting technology to achieve personalized information push for different terminals, and establishes an information receipt confirmation mechanism to start the retransmission process for unconfirmed receiving terminals.

[0052] The data processing server also includes a data persistence module that stores the original monitoring data using a time series database, a cache acceleration unit that uses in-memory computing technology to achieve real-time data analysis, and a log audit system that completely records the data operation traces and generates a compliance report.

[0053] The above content is only an example and explanation of the present invention. Those skilled in the art of this technology can make various modifications, supplements, or use similar methods to replace the specific embodiments described, as long as they do not deviate from the scope defined by the invention, they should all fall within the protection scope of the present invention.

Claims

1. An intelligent tent data acquisition method, characterized in that: including Step S1: Obtain multi-source sensor data through distributed data acquisition nodes. The sensors include an environmental monitoring sensor group, an attitude sensor, and an image acquisition device. The environmental monitoring sensor group includes temperature, humidity, and air pressure sensors. Step S2: Preprocess the original sensor data, including selecting a corresponding digital filtering algorithm according to the sensor type. Among them, the temperature sensor uses a sliding average filter with dynamic window adjustment, and the image data uses an adaptive noise reduction process based on region segmentation. Add timestamp marks to each sensor data through a unified clock source, and use an interpolation algorithm to achieve time domain alignment for non-equidistant sampled data. Step S3: Perform fusion processing on the preprocessed multi-source data based on time series analysis methods, and construct a dynamic weight allocation matrix including sensor confidence weight and environmental parameter correlation. Dynamically adjust the fusion coefficient according to the real-time environmental parameter change rate, and generate a multi-dimensional feature vector including temperature gradient change rate, structural deformation index, and visible visibility value. Step S4: Use a dynamic threshold model for abnormal state detection. The dynamic threshold model updates the threshold through a preset mode. The preset mode includes calculating the probability density distribution mode of each environmental parameter based on historical data and dynamically correcting the threshold by combining the real-time collected weather type and terrain features. Step S5: When the same abnormal parameter threshold is triggered within three consecutive detection cycles, generate a secondary warning signal including geographical coordinates, abnormal parameter type, and confidence score, and send it to the rescue warning platform through a multi-channel redundant transmission protocol.

2. The intelligent tent data acquisition method according to claim 1, wherein: The sliding average filter with dynamic window adjustment automatically adjusts the window length according to the current sampling frequency. The image adaptive noise reduction process includes dividing the image area based on color space features and identifying the noise distribution. Use the first type of noise reduction algorithm for high-noise areas and the second type of noise reduction algorithm for low-noise areas.

3. The intelligent tent data acquisition method according to claim 1, characterized in that: The method for generating the dynamic weight allocation matrix includes: Step S301: Calculate the real-time signal-to-noise ratio of each sensor data as the confidence weight through a filtering algorithm. Step S302: Calculate the correlation between environmental parameters based on parameter correlation. When the correlation exceeds the set correlation threshold, automatically increase the corresponding sensor weight coefficient. Step S303: Automatically enable the data of the same type of sensor of the adjacent node to replace the failed sensor and mark the data source mark.

4. A method for collecting intelligent tent data according to claim 1, characterized in that: The fusion processing includes constructing a sensor data weight allocation matrix, dynamically adjusting the fusion coefficient based on the environmental parameter correlation, and generating a multi-dimensional feature vector including temperature gradient change rate, structural deformation index, and visible visibility value.

5. A method for collecting intelligent tent data according to claim 1, characterized in that: The abnormal state detection includes: establishing a dynamic threshold adjustment model, automatically updating the judgment threshold based on historical data and real-time environmental parameters, and generating a secondary warning signal when the threshold is triggered in three consecutive detection cycles.

6. A rescue warning platform, characterized in that: including: A data receiving module, including real-time receiving of the monitoring data streams of multiple intelligent tents. A data processing server, including a data cleaning unit for removing duplicate data and invalid data packets. A feature extraction unit for parsing the key parameters in the multi-dimensional feature vector. The early warning judgment unit conducts risk assessment and grading based on the decision tree model; The communication gateway supports multi-protocol conversion and has the function of resuming interrupted transmission.

7. The rescue warning platform according to claim 6, characterized in that: The communication gateway includes a Beidou satellite communication interface for emergency communication in areas without public network coverage, and a data encryption unit that uses national cryptographic algorithms to perform end-to-end encryption on the transmitted data.

8. The rescue warning platform according to claim 6, wherein: The early warning judgment unit includes: a multi-level early warning trigger mechanism that sets three-level response criteria of yellow, orange, and red; a geographical fence generation module that automatically demarcates the affected area based on the coordinates of the warning tent.

9. The rescue warning platform according to claim 6, characterized in that: It also includes a rescue information distribution system, which includes automatically dispatching rescue work orders containing 3D topographic maps to pre-registered rescue agencies, and sending disaster avoidance guidance information to the affected tent terminals; Generating public early warning information and publishing it through the specified social media interface; the distribution system uses differential broadcast technology to achieve personalized information push for different terminals; Establishing an information receipt confirmation mechanism and starting a retransmission process for terminals that have not confirmed receipt.

10. A rescue warning platform according to claim 6, characterized in that: The data processing server also includes a data persistence module that stores the original monitoring data using a time series database, a cache acceleration unit that uses in-memory computing technology to achieve real-time data analysis, and a log audit system that completely records the data operation traces and generates a compliance report.

Images