Automobile anti-flooding online intelligent early warning system and method

Abstract

The invention discloses an automobile anti-flooding online intelligent early warning system and method, and relates to the technical field of vehicle safety monitoring, and the method comprises the following steps: obtaining a water level reflection signal of a road surface in front of a vehicle, environmental rainfall density data and a vehicle inclination angle parameter, and forming a continuous raindrop reflection sequence according to a time sequence; and based on the raindrop reflection sequence, analyzing an echo particle intensity change rule in a high rainfall period, and constructing a reflection distribution diagram for reflecting a rainfall interference form. According to the method, accurate identification of the real water level under rainfall interference is realized by constructing a raindrop reflection sequence and a reflection distribution diagram; the recognition rhythm is dynamically adjusted through the difference mapping table and the time control instruction, the reflection deviation is corrected in cooperation with reverse detection and virtual signal occupation, it is guaranteed that the early warning process is continuous and stable, and the reliability and safety of vehicle wading risk judgment are improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle safety monitoring technology, specifically to an online intelligent early warning system and method for vehicle flooding. Background Technology

[0002] Online intelligent early warning for vehicle flooding refers to the real-time collection of information such as surrounding water level, rainfall intensity, terrain changes, and vehicle posture during the operation or parking of electric vehicles, utilizing onboard sensors and an external environmental perception network. This data is transmitted to an intelligent analysis unit via a communication network for comprehensive judgment, thereby issuing an early warning when the water level approaches a dangerous threshold or when there is a risk of flooding. This early warning process is online, meaning that data collection and analysis are continuous, allowing for dynamic perception and decision-making alerts regarding potential flooding hazards without affecting normal vehicle use. Essentially, it is an active safety technology combining environmental perception, data fusion, and risk prediction, aiming to provide drivers with a basis for risk assessment before wading through water, preventing electrical damage to the vehicle and occupants from being trapped due to sudden flooding or heavy rain.

[0003] The existing technology has the following shortcomings: In existing technologies, vehicle flood warning systems typically rely on onboard radar to identify water levels from echo signals of the road surface ahead. However, in heavy rainfall, the electromagnetic waves emitted by the radar are easily reflected by dense raindrops, creating numerous stray echoes. This results in false, smooth reflective surfaces appearing in the signals received by the system. In such cases, the warning device may mistakenly identify the road ahead as a uniformly shallow area of ​​water, when in reality there may be deep potholes or sudden flooding. Because these false planar signals mask the true water level changes, the system cannot correctly distinguish between shallow water and deep potholes, easily leading to an underestimation of the risk level. This can cause vehicles to continue driving into areas with high water levels, resulting in serious consequences such as engine flooding and electrical short circuits.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an online intelligent early warning system and method for vehicle flooding, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an online intelligent early warning method for vehicle flooding, comprising the following steps: The water level reflection signal on the road surface in front of the vehicle, the ambient rainfall density data, and the vehicle tilt angle parameters are acquired and formed into a continuous raindrop reflection sequence in chronological order for subsequent rain curtain interference characteristic analysis. Based on the analysis of raindrop reflection sequence, the intensity variation of echo particles during periods of high rainfall is analyzed, and a reflection distribution map reflecting the morphology of rainfall interference is constructed to provide spatial distribution basis for subsequent water accumulation identification. Based on the reflection distribution map, the boundary change characteristics of the flat reflection area are detected. Combined with the differences in reflection energy and time delay at different water depths, a difference mapping table between shallow and deep water is established to distinguish different water level characteristics. By using a difference mapping table to adjust the timing of early warning triggers, the execution rhythm of water level identification is determined, and a time control command containing the identification cycle and trigger delay is generated to optimize the identification timing under rainfall interference conditions. Based on time control commands, the system performs recognition adjustments in rainy conditions. By inserting reverse detection windows during rain gaps, alternately skipping interference sections, and releasing virtual warning signals to occupy positions, it continuously corrects reflection deviations in water level recognition, maintaining the continuity and stability of the vehicle flood warning process.

[0007] Preferably, the steps for obtaining the raindrop reflection sequence are as follows: The electromagnetic wave transmitting and receiving device installed at the front of the vehicle continuously scans the road surface in the direction of the vehicle's travel, collects water level reflection signals from the road surface and near-ground air layer, and records information on water surface echo intensity, reflection angle, and echo delay time. Rainfall density data is acquired by an environmental monitoring device installed on the outside of the vehicle body, and the rainfall density data is recorded synchronously with the water level reflection signal acquisition time point to form paired data with time sequence. The vehicle tilt angle parameters are obtained by an attitude measurement device installed at the center of the vehicle body, and the vehicle tilt angle parameters are correlated with the water level reflection signal and rainfall density data of the corresponding time period to correct the reflection angle offset caused by attitude changes. The water level reflection signal, rainfall density data, and vehicle tilt angle parameters are integrated in chronological order to form a continuous raindrop reflection sequence, providing the original input basis for subsequent rain curtain interference characteristic analysis.

[0008] Preferably, in the process of forming the raindrop reflection sequence, the water level reflection signal, rainfall density data and vehicle tilt angle parameters are paired and arranged according to a unified time label. Each time node fully contains reflection information, rainfall intensity information and vehicle attitude information. By continuously combining them in time sequence, a time chain dataset is constructed, so that the raindrop reflection sequence has temporal continuity and spatial correspondence under dynamic rainfall conditions, which is used to improve the accuracy and stability of subsequent reflection feature analysis.

[0009] Preferably, the steps for constructing the reflectance distribution map are as follows: Based on the water level reflection signal data corresponding to each time node in the raindrop reflection sequence, the intensity of the reflection signal in the continuous rainfall period is extracted in layers, and the raindrop reflection sequence is divided into intervals at fixed time intervals to form a continuous energy distribution trajectory. After obtaining the time-intervalized reflection energy distribution, the high rainfall period is determined based on the changing trend of rainfall density data in the raindrop reflection sequence. The time interval of the high rainfall period is analyzed in detail and the echo particle intensity distribution is extracted. After obtaining the intensity distribution of echo particles during periods of high rainfall, a three-dimensional spatial distribution matrix is ​​constructed according to the correspondence between spatial location and time sequence to depict the intensity variation relationship of reflected signals between time and spatial location. After obtaining the spatial distribution matrix, the trajectory of reflected energy change is continuously mapped to form a reflection distribution map, which is used to reflect the pattern of rainfall interference and provide spatial distribution basis for subsequent water accumulation identification.

[0010] Preferably, the steps for establishing the difference mapping table are as follows: Based on the reflection distribution map, the continuous reflection area is spatially divided to identify the initial boundary of the flat reflection area, and the reflection plane under different slope conditions is geometrically corrected in combination with the vehicle tilt angle parameter to ensure accurate spatial positioning. After obtaining the initial boundary range of the flat reflective area, the reflective energy change characteristics on both sides of the boundary are compared to extract the energy difference characteristics of the water depth change, and the rainfall density data is correlated with the reflective energy difference to improve the recognition accuracy. After analyzing the energy difference on both sides of the boundary of the flat reflection area, the echo time delay variation characteristics under the same spatial location are compared to supplement the judgment basis of water depth difference, and the reflection energy and time delay information are superimposed and displayed in a spatial coordinate system. After completing the boundary detection of the flat reflection area and comparing the energy and time delay features, the features are combined according to spatial location, time sequence and reflection energy level to form a difference mapping table between shallow and deep water.

[0011] Preferably, the difference mapping table records the energy intensity, echo delay, and water level characteristics of the reflection point using spatial coordinates as an index, and divides the reflection characteristics corresponding to different water level types into intervals in the form of thresholds, so that shallow water areas correspond to distribution areas with concentrated energy and short delay, and deep water areas correspond to distribution areas with dispersed energy and long delay, so as to achieve rapid judgment and identification of water level categories.

[0012] Preferably, the steps for generating time control instructions are as follows: Based on the reflected energy, time delay characteristics, and spatial location distribution of shallow and deep water recorded in the difference mapping table, the continuous change relationship of water features in the time dimension is extracted, and the time distribution of reflected energy changes is corrected by combining rainfall density data and vehicle speed parameters. After obtaining the continuous water level characteristic change relationship in the time dimension, the warning trigger priority is divided according to the reflected energy fluctuation range in the difference mapping table, and the triggering areas are arranged into a continuous time triggering sequence in time order. After determining the warning triggering order and priority, the identification period and triggering delay parameters are set according to the stable interval of the reflection characteristics of each water level type in the difference mapping table, and the identification period and triggering delay are embedded inside the identification period to reduce interference. After setting the recognition cycle and trigger delay parameters, the time parameters and the trend of reflected energy change are combined to generate time control commands to ensure that the recognition action maintains continuity and rhythm consistency under rainfall interference conditions.

[0013] Preferably, the time control instruction includes an identification start time, an execution interval time, a delay trigger time, and a water level type identifier. The identification start time is used to determine the initial execution point of the early warning trigger node, the execution interval time is used to control the time interval between continuous identification actions, the delay trigger time is used to postpone identification execution during rainfall interference, and the water level type identifier is used to indicate the water level status within the corresponding identification cycle, so as to achieve dynamic matching and time synchronization of the identification rhythm.

[0014] Preferably, the recognition adjustment under rainy conditions is performed according to the time control command. The steps to correct the reflection deviation in water level recognition are as follows: inserting a reverse detection window during rainfall intervals, alternately skipping interference sections, and releasing virtual early warning signals to fill gaps. Based on the recognition cycle and trigger delay parameters set in the time control command, the recognition start time and execution window during the rainfall process are determined, and a reverse detection window is inserted when a rainfall gap is detected to capture changes in the echo signal; After the reverse detection window is inserted, the interference segment in the current identification cycle is identified according to the triggering order set in the time control command, and a skip operation is performed when uneven energy distribution of echo signal or abnormally elongated reflection delay curve is detected to maintain the continuous identification rhythm. While performing the interference section skipping operation, the virtual early warning signal is released according to the time control command to maintain the continuous existence of the early warning signal stream and keep the output rhythm stable. After completing reverse detection, skipping interference sections, and virtual early warning signal placement, the operation results are integrated into a new identification sequence in chronological order to achieve continuous correction of reflection deviation and maintain the continuity and stability of the early warning process.

[0015] An online intelligent early warning system for vehicle flooding includes a data acquisition module, an interference feature extraction module, a water accumulation identification module, a time control module, and an identification and correction module. The data acquisition module acquires water level reflection signals from the road surface in front of the vehicle, environmental rainfall density data, and vehicle tilt angle parameters, forming a continuous raindrop reflection sequence in chronological order for subsequent rain curtain interference characteristic analysis. The interference feature extraction module analyzes the variation pattern of echo particle intensity during periods of high rainfall based on raindrop reflection sequence analysis, and constructs a reflection distribution map to reflect the morphology of rainfall interference, providing spatial distribution basis for subsequent water accumulation identification; The water accumulation identification module detects the boundary change characteristics of flat reflection areas based on the reflection distribution map, and establishes a difference mapping table between shallow and deep water accumulation by combining the differences in reflection energy and time delay at different water depths, which is used to distinguish different water level characteristics. The time control module uses a difference mapping table to adjust the timing sequence of early warning triggers, determines the execution rhythm of water level identification, and generates time control instructions containing identification cycles and trigger delays to optimize the identification timing under rainfall interference conditions. The recognition and correction module performs recognition adjustments under rainy conditions according to time control commands. By inserting reverse detection windows during rain gaps, alternately skipping interference sections, and releasing virtual warning signals to occupy positions, it continuously corrects the reflection deviation in water level recognition, maintaining the continuity and stability of the vehicle flood warning process.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention achieves effective separation of stray echo interference in heavy rainfall environments by introducing the dynamic construction of raindrop reflection sequences and spatial feature identification of reflection distribution maps during flood warning processes. This allows water level identification to be based on actual ground reflection signals. By fusing multi-source reflection data with rainfall density information in chronological order, the reflection signals are uniformly expressed in both temporal and spatial dimensions, thereby improving the accuracy of identifying shallow and deep water features under complex rain conditions and providing a stable data foundation for early warning decisions.

[0017] This invention establishes a difference mapping table and dynamically adjusts it using time control commands, enabling real-time regulation of water level recognition rhythm in response to rainfall changes, ensuring that the timing of warning triggering is synchronized with environmental changes. During the recognition and adjustment process, the synergistic effect of the reverse detection window and virtual signal occupancy continuously corrects reflection deviations, maintaining continuous output of warning signals. This method achieves stable operation of the warning process under strong interference conditions, improving the reliability and safety performance of vehicle wading risk assessment. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of an online intelligent early warning method for vehicle flooding according to the present invention.

[0020] Figure 2 This is a schematic diagram of a module of an online intelligent early warning system for vehicle flooding according to the present invention. Detailed Implementation

[0021] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0022] This invention provides, for example Figure 1 The method for online intelligent early warning of water damage to automobiles, as shown, includes the following steps: The water level reflection signal on the road surface in front of the vehicle, the ambient rainfall density data, and the vehicle tilt angle parameters are acquired and formed into a continuous raindrop reflection sequence in chronological order for subsequent rain curtain interference characteristic analysis. To generate a raindrop reflection sequence that reflects the characteristics of environmental rainfall disturbance, multi-dimensional sensing information of the road surface in front of the vehicle is continuously collected and processed in a time series manner. The specific implementation steps are as follows: An electromagnetic wave transmitting and receiving device installed at the front of the vehicle continuously scans the road surface ahead of the vehicle's travel direction, collecting water level reflection signals from the road surface and near-surface atmosphere. This signal contains multi-dimensional information such as water surface echo intensity, reflection angle, and echo delay time. To ensure signal stability under different rainfall scenarios, the electromagnetic wave transmitting device periodically emits pulse beams at a constant frequency, and the receiving end records the corresponding echo amplitude and arrival time in chronological order. Each signal acquisition corresponds to the vehicle's real-time location data, establishing a correspondence between the reflected signals on the time axis and spatial coordinates, providing a foundation for subsequent continuous temporal analysis.

[0023] While collecting water level reflection signals, rainfall density data is acquired using an environmental monitoring device installed on the exterior of the vehicle. Rainfall density is measured by detecting the number of raindrop impacts and the volume of liquid per unit time, and this data is recorded synchronously with the time of each reflection signal acquisition. This synchronous recording process ensures that each water level reflection signal has a corresponding rainfall density identifier, allowing for the assessment of the degree of interference in the echo signal based on rainfall trends during subsequent processing. Furthermore, to enhance the temporal continuity of the data, each set of rainfall density data is paired and stored with the corresponding water level reflection signal data in chronological order, forming a preliminary time series structure.

[0024] After acquiring water level reflection signals and rainfall density data, the vehicle's tilt angle parameters are obtained using an attitude measurement device installed at the center of the vehicle body. These parameters include longitudinal pitch and lateral roll angles, characterizing the vehicle's attitude changes relative to the ground. The introduction of the vehicle tilt angle parameters is to correct for reflection angle offsets caused by changes in vehicle attitude during the time-series processing of the reflection signals and rainfall data. The tilt angle parameters are correlated chronologically with water level reflection signals and rainfall density data for the same time period, ensuring that each data set includes not only environmental information but also geometric correction parameters related to vehicle attitude. This approach makes the time-series of the reflection signals more closely resemble the actual road water surface conditions, avoiding signal distortion caused by vehicle uphill, downhill, or turning that could affect subsequent analysis results.

[0025] After simultaneously acquiring water level reflection signals, rainfall density data, and vehicle tilt angle parameters, these three types of data were integrated in chronological order to form a continuous raindrop reflection sequence. During integration, water level reflection signals, rainfall density data, and tilt angle parameters collected within the same time period were paired and arranged according to time labels to ensure that the data at each time point fully contained reflection information, rainfall intensity information, and vehicle attitude information. Subsequently, these multi-source data were combined into a continuous sequence structure according to chronological order, thereby constructing a time-chained dataset reflecting changes in reflection characteristics under dynamic rainfall conditions. This raindrop reflection sequence not only possesses continuous temporal attributes but also reflects the relationship between signal intensity, reflection angle, and vehicle attitude at different rainfall stages, providing a sufficient raw input foundation for subsequent rain curtain interference feature extraction.

[0026] In the process of forming the raindrop reflection sequence, the system uses a unified on-board clock as the time reference, and adds a collection timestamp to the water level reflection signal, environmental rainfall density data, and vehicle tilt angle parameters, and writes them into the corresponding data buffers in the order of the timestamps. Since the sampling frequency and response period of different sensors are different, when performing multi-source data fusion, the sampling time node of the water level reflection signal is used as the master time node, and the sampling data with the smallest time difference from the master time node is selected from the rainfall density data buffer and the vehicle tilt angle parameter buffer, respectively, to form a combination of reflection information, rainfall information, and attitude information at the same time node.

[0027] When rainfall density data or vehicle tilt angle parameters do not have fully synchronized sampling values ​​at the current master time node, nearest neighbor compensation or interpolation based on adjacent sampling values ​​is used to ensure that each master time node corresponds to a complete data set. To reduce the impact of sensor response delay on the fusion results, the system pre-records the average response delay of the environmental monitoring device and attitude measurement device relative to the electromagnetic wave transmitting and receiving device, and performs unified delay correction on the timestamps of the auxiliary sensors before fusion, ensuring that the three types of data are aligned under the same time reference frame.

[0028] After time alignment, error suppression processing was performed on the three types of raw data. For water level reflection signals, isolated outliers that significantly deviated from the continuous echo trajectory were removed, and the echo energy was smoothed and filtered. For environmental rainfall density data, a moving average method was used to reduce instantaneous fluctuations caused by rainstorm impact. For vehicle tilt angle parameters, a low-pass smoothing method was used to suppress high-frequency attitude jitter caused by vehicle vibration. After the above preprocessing was completed, the water level reflection signals, rainfall density data, and vehicle tilt angle parameters at the same time point were arranged continuously in chronological order to form a raindrop reflection sequence.

[0029] After the raindrop reflection sequence is formed, the continuity of data between adjacent time points is verified. When the change in reflected energy, rainfall density, or vehicle tilt angle at a certain time point exceeds a preset abrupt change threshold, that time point is marked as a low-confidence node. For low-confidence nodes, they are not directly used as the basis for judging the water accumulation category, but are smoothed and corrected by combining the data of the preceding and following time points, or their weight in the subsequent identification process is reduced. Through unified clock calibration, timestamp alignment, delay correction, and error suppression processing, the raindrop reflection sequence maintains temporal continuity and spatial correspondence during actual vehicle driving.

[0030] Based on the analysis of raindrop reflection sequence, the intensity variation of echo particles during periods of high rainfall is analyzed, and a reflection distribution map reflecting the morphology of rainfall interference is constructed to provide spatial distribution basis for subsequent water accumulation identification. To extract the variation patterns of echo particle intensity during periods of high rainfall from raindrop reflection sequences and construct a reflection distribution map that reflects the morphology of rainfall interference, a comprehensive spatial and temporal analysis of multi-source reflection information over a continuous time period is conducted. The specific implementation steps are as follows: Based on the water level reflection signal data corresponding to each time node in the raindrop reflection sequence, the reflection signal within a continuous rainfall period is extracted in intensity layers. The water level reflection signal at each time node includes elements such as reflection energy value, echo arrival time, and reflection angle. By arranging these data continuously in chronological order, a basic curve showing the change of reflection energy over time during rainfall can be obtained. In this process, the raindrop reflection sequence is divided into intervals at fixed time intervals, so that each time interval corresponds to a set of continuous reflection signals. This time intervalization method transforms instantaneous reflection changes into energy distribution trajectories over continuous time periods. To eliminate local signal offsets caused by changes in vehicle attitude, the reflection angle data for each time interval is translated according to the vehicle tilt angle parameters obtained in the previous stage, ensuring that the reflection signals in different time intervals remain consistent in spatial coordinates, thereby guaranteeing the continuity of the reflection energy distribution.

[0031] After obtaining the time-interval-based reflection energy distribution, high-rainfall periods were screened and grouped. High-rainfall periods were determined by the continuous trend of rainfall density data in the raindrop reflection sequence; periods of continuous increase and sustained high rainfall density were defined as high-rainfall periods. During high-rainfall periods, the increased number of raindrops and drastic changes in droplet size cause complex energy scattering and superposition effects on the echo signal. To address this characteristic, each time interval within a high-rainfall period was selected as a key analysis segment, and the echo particle intensity distribution was extracted for each time interval. Here, echo particle intensity refers to the energy density of the reflected signal at different spatial sampling points, used to describe the distribution of the reflected signal after dispersion in the rain. By comparing the particle intensity changes between high-rainfall periods and ordinary rainfall periods, the influence range and interference pattern of raindrop interference on the reflection energy distribution can be identified, thus providing a data foundation for subsequent interference pattern description.

[0032] Echo particle intensity refers to the energy density of the reflected signal within a unit time interval and a unit spatial sampling area, used to characterize the dispersion and aggregation of echo reflections under rain curtain conditions. The system first divides the raindrop reflection sequence into multiple continuous time intervals at fixed time intervals, and extracts the reflection energy, echo delay, and reflection angle characteristics of the water level reflection signal within each time interval. Simultaneously, it combines this with the corresponding rainfall density data within that time interval to determine whether the current period is a high-rainfall period. Preferably, periods where the rainfall density exceeds a preset threshold for multiple consecutive time intervals, and where the discrete scattered echoes in the near-surface atmosphere region significantly increase, are identified as high-rainfall periods.

[0033] During periods of high rainfall, echo signals in each time interval were segmented and statistically analyzed according to spatial location. The average reflected energy, energy fluctuation amplitude, and energy gradient between adjacent sampling points were calculated for each spatial segment. When the echo mainly originates from actual road surface water, its reflection typically exhibits relatively concentrated energy distribution, more continuous spatial distribution, and relatively gentle energy changes between adjacent sampling points. When the echo is mainly affected by raindrop clutter, its reflection typically exhibits discrete energy distribution, an increased number of local peaks, and frequent energy fluctuations between adjacent sampling points. By comparing the energy concentration, dispersion, and fluctuation frequency of each spatial segment within a continuous time interval, the variation pattern of echo particle intensity during periods of high rainfall was extracted to distinguish between raindrop clutter reflection and ground / water accumulation reflection.

[0034] To improve identification accuracy, a continuity analysis is performed on the echo results of the current time interval and adjacent time intervals. If a high-energy reflection at a certain spatial location appears only in a single short time interval and lacks temporal continuity, it is more likely to be transient raindrop clutter. If a high-energy reflection at a certain spatial location exists stably in multiple consecutive time intervals, and its echo delay changes correspond to changes in the vehicle's spatial position during movement, it is more likely to be reflection from actual road surface water. By judging temporal continuity, effective reflection patterns can be extracted from the mixed raindrop clutter and ground / water reflections.

[0035] The difference mapping table between shallow and deep water is established based on the boundary characteristics of smooth reflection areas, reflection energy characteristics, echo delay characteristics, and the degree of rainfall interference. Specifically, firstly, continuous and smooth reflection areas are identified based on the reflection distribution map and used as candidate water surface reflection areas. Then, the candidate areas are geometrically corrected by combining vehicle tilt angle parameters to eliminate the influence of vehicle pitch and roll on the reflection position. Next, the changes in reflection energy, echo delay, and their temporal trends on both sides of the candidate area boundary are extracted and combined according to spatial location, temporal order, and reflection energy level to form the difference mapping table.

[0036] The difference mapping table records the reflected energy, echo delay, and relative water level characteristics corresponding to each reflection point, indexed by spatial location. It further divides the reflection characteristics corresponding to different water level types according to threshold intervals. Shallow water areas correspond to feature intervals with concentrated energy, short boundary changes, and short echo delays; deep water areas correspond to feature intervals with wider energy distribution, more pronounced boundary changes, and longer echo delays. In actual identification, the currently detected reflected energy, echo delay, boundary change, and rainfall density values ​​are matched with the feature intervals in the difference mapping table. If the current feature combination matches the shallow water feature interval more closely, it is determined to be shallow water; if it matches the deep water feature interval more closely, it is determined to be deep water; and if it falls within the boundary range of the intervals, it is further confirmed by combining subsequent time interval results.

[0037] The time control command is generated based on the water accumulation type change trend, current rainfall density level, and echo stability within a continuous time interval recorded in the difference mapping table. The time control command includes the identification start time, execution interval, delayed trigger time, and water level type identifier, used to determine the execution rhythm of subsequent identification and adjustment. When a period of high rainfall is identified and the echo fluctuation is large, the time control command increases the identification interval and delays the triggering of the identification action; when rainfall interference is identified as weakening and echo continuity is enhanced, the time control command shortens the identification interval, increasing the tracking frequency of water accumulation changes.

[0038] The reverse detection window refers to a supplementary detection period inserted outside the regular identification cycle. It is used during breaks in rainfall or when raindrop scattering weakens to re-detect echo samples that were heavily interfered with during the main identification process. The "reverse" in reverse detection means that the main identification time segment that has just passed is reviewed in reverse chronological order, rather than continuing the detection along the time progression of the main identification cycle. Specifically, when the time control command determines that a break in rainfall is currently in, the system initiates a reverse detection window, shorter than the regular identification cycle. It re-extracts the reflection energy and echo delay of sampling points marked as low confidence in the previous main identification segment and compares the re-detection results with the main identification results. If they match, the confidence of the identification result for that time segment is increased; if the difference is significant, the reverse detection results are used to correct the original identification result, or the final water accumulation type is temporarily withheld. Through the reverse detection window, the identification results that were previously heavily interfered with by the rain curtain can be reviewed during periods of reduced interference, thereby reducing the false positive rate.

[0039] After obtaining the intensity distribution of echo particles during periods of high rainfall, a three-dimensional spatial distribution matrix is ​​constructed according to the correspondence between spatial location and temporal sequence. This matrix depicts the intensity variations of reflected signals at different times and spatial locations. The horizontal axis of this three-dimensional spatial distribution matrix represents the time axis, the vertical axis represents the distance along the vehicle's direction of travel, and the vertical axis represents the reflected energy intensity value. This spatial mapping method transforms the changes in reflected energy over a continuous time period into a spatial distribution pattern, thus intuitively presenting the spatial extent of rainfall interference. During the construction of the spatial distribution matrix, the corrected reflected signal data and rainfall density data from the previous stage are synchronously projected into this spatial coordinate system. This ensures that each spatial point not only corresponds to a reflected intensity value but also carries the associated rainfall density attribute. This spatial distribution representation with environmental parameter identifiers allows the impact of rainfall on reflection characteristics to be recorded in a structured manner, providing a basis for further identifying the boundaries of interference areas and the trend of interference intensity changes.

[0040] A three-dimensional spatial distribution matrix is ​​used to organize echo reflection characteristics and rainfall interference characteristics within a continuous time frame into a unified data structure. This three-dimensional spatial distribution matrix does not perform a full-scene three-dimensional reconstruction of the environment in front of the vehicle, but rather establishes a local spatiotemporal feature matrix for a limited warning area in front of the vehicle. Specifically, during construction, the warning detection area in front of the vehicle is first divided into multiple distance units along the vehicle's direction of travel, and the continuous acquisition process is divided into multiple time units with a fixed time step. Then, using the time unit and distance unit as indices, the echo characteristics at the corresponding distance position within each time unit are written into the matrix, thus forming a two-dimensional basic structure with time and distance dimensions. Within each matrix unit, feature data such as reflection energy, echo delay, rainfall density, and attitude correction markers are recorded, thereby forming a three-dimensional spatial distribution matrix composed of time, space, and feature dimensions.

[0041] During matrix construction, the echo reflection point within the current time unit is first geometrically corrected based on the vehicle tilt angle parameters. Then, the corrected echo reflection energy is mapped to the corresponding distance unit. Simultaneously, the rainfall density data corresponding to that time unit is written into the same matrix unit as an environmental interference attribute for that spatial location. If multiple echo samples fall into the same distance unit within the same time unit, the reflection energy of these samples is averaged or the maximum reliable value is selected as the energy characteristic value of that matrix unit. If a matrix unit does not obtain a valid echo within the current time unit, it inherits the background value from the previous time unit and is marked as pending update to maintain the continuity of the matrix in the time dimension.

[0042] When further reflection of lateral spatial variations is needed, each range cell can be further subdivided into sub-cells according to the scanning direction or lateral angle. This allows the spatial dimension of the three-dimensional spatial distribution matrix to reflect both the forward range position and the lateral scanning position. In this way, each cell in the matrix corresponds to the fused reflection characteristics at a specific time, distance, and scanning direction. In this manner, the original radar echo signal and rainfall density data are uniformly converted into a structured dataset with temporal, spatial, and environmental attributes.

[0043] After establishing the three-dimensional spatial distribution matrix, a continuity analysis is performed on the changes in reflected energy between adjacent time units and adjacent spatial units. If a spatial unit exhibits high reflected energy with relatively stable changes across multiple consecutive time units, it is determined that a persistent reflective target exists at that location, potentially corresponding to a real water accumulation area. If some high reflected energy appears only sporadically in individual time units or spatial units, it is determined that it is more likely to be transient interference caused by raindrop clutter. Based on this, a continuous mapping is performed on the three-dimensional spatial distribution matrix to form a reflection distribution map. The reflection distribution map is used to characterize the rain curtain interference pattern, stable reflection areas, and areas to be verified, and serves as the input basis for subsequent boundary detection, difference mapping table establishment, and time control command generation.

[0044] The three-dimensional spatial distribution matrix is ​​essentially a spatiotemporal feature organization method for a limited detection area in front of a vehicle. Its construction process includes temporal discretization, spatial segmentation, attitude correction, feature writing, empty cell compensation, and spatiotemporal continuity analysis. Through this matrix structure, the raw echo signal and rainfall data can be converted into structured inputs that facilitate interference identification and water accumulation classification, thus providing a clear data foundation for subsequent identification and processing.

[0045] After obtaining the spatial distribution matrix, the trajectory of reflected energy variation is continuously mapped to form a reflection distribution map. The reflection distribution map is a two-dimensional projection of the three-dimensional spatial distribution matrix at a specific temporal resolution. By plotting the reflected energy intensity equally distributed on the spatial plane, the morphology of rain curtain interference can be clearly displayed. In the reflection distribution map, areas with concentrated reflected energy intensity correspond to areas of concentrated reflected signals, while areas with sparse energy intensity distribution represent areas where signals with weaker interference pass through. By sequentially arranging the time-continuous reflection distribution maps, a dynamic evolution process of reflection morphology can be formed, thereby revealing the changing patterns of echo signals under continuous rainfall conditions. This reflection distribution map not only reflects the influence of rainfall intensity on the spatial distribution of reflected signals but also reflects the signal density fluctuations caused by raindrop scattering, providing an intuitive spatial distribution basis for identifying the reflection characteristics of different types of accumulated water.

[0046] Based on the reflection distribution map, the boundary change characteristics of the flat reflection area are detected. Combined with the differences in reflection energy and time delay at different water depths, a difference mapping table between shallow and deep water is established to distinguish different water level characteristics. Based on the reflection distribution map, the boundary change characteristics of the flat reflection area are detected. A difference mapping table between shallow and deep water is established by combining the differences in reflection energy and time delay at different water depths. This table is used to distinguish different water level characteristics. The specific implementation steps are as follows: Through layer-by-layer analysis and comparison of the spatial characteristics of the reflection signal in the reflection distribution map. Based on the generated reflection distribution map, the continuous reflection regions are spatially divided to identify the initial boundaries of smooth reflection regions. In the reflection distribution map, each spatial point corresponds to a specific reflection energy value, which originates from the reflection signal in the previous raindrop reflection sequence after time and space mapping. To accurately determine the boundary range of the reflection region, regions with stable and continuous energy change gradients are first extracted from the continuous reflection energy regions of the reflection distribution map as candidate smooth reflection regions. Smooth reflection regions typically exhibit relatively uniform energy distribution and gentle changes in time delay. By scanning the reflection distribution map line by line, these candidate regions are compared with adjacent regions in terms of energy. When a sudden change in reflection energy or an inflection point in the delay curve occurs between adjacent regions, the initial location of the reflection boundary can be determined. To make the boundary division more consistent with the actual road environment, the reflection planes under different slope conditions are geometrically corrected based on the correction information of the vehicle tilt angle parameters from the previous analysis. This ensures that the boundary line of the smooth reflection region maintains a consistent correspondence with the actual ground height, guaranteeing accurate spatial positioning for subsequent analysis of water depth differences.

[0047] After obtaining the initial boundary of the smooth reflective area, the energy variation characteristics on both sides of the boundary are compared to extract the energy difference characteristics of water depth variations. The difference in reflected energy stems from the differences in reflection paths caused by different water depths and the varying degrees of absorption by the medium. In shallower water, the radar wave reflection path between the water surface and the ground is shorter, the reflected energy is relatively concentrated, and the echo delay time is shorter. In deeper water areas, the electromagnetic wave propagation path in the water is longer, the reflected energy is partially absorbed or scattered, the echo energy is weakened, and the delay time increases. To accurately reflect this difference, the center line of the smooth reflective area in the reflection distribution map is used as the energy reference line. Energy variation curves are extracted over a certain distance range both upwards (near the vehicle direction) and downwards (near the distance), and arranged in a time series format. This two-sided energy comparison method clearly presents the energy transition pattern at the boundary of the smooth reflective area, thereby identifying the energy distribution differences between shallow and deep water. Meanwhile, the rainfall density data is correlated with the difference in reflected energy during the same period to determine the extent of the impact of rain curtain interference on the change in reflected energy, thereby further improving the accuracy of water depth difference identification.

[0048] After analyzing the energy differences on both sides of the boundary of the flat reflection area, the echo time delay variation characteristics at the same spatial location are compared to supplement the judgment criteria for water depth differences from a temporal perspective. The time delay data of each reflection point comes from the arrival time of the reflected signal. By calculating the time difference, the changing trend of the reflection path length can be obtained. The time delay curve of shallow water areas is usually smooth and has a small variation amplitude, while the time delay curve of deep water areas shows an elongating trend and a step increase at the boundary. In order to accurately describe this variation characteristic in space, each reflection energy point in the reflection distribution map is mapped to its corresponding time delay data in spatial coordinates, so that the reflection energy and time delay information are superimposed and displayed in the same coordinate system. Through the superimposed spatial-temporal joint distribution, the true boundary of water depth variation can be determined at the location where the abrupt change in reflection energy and the time delay change occur simultaneously. At this time, the time delay difference not only represents the degree of extension of the wave propagation path, but also indirectly reflects the relative change in water depth, providing a quantitative basis for establishing the mapping relationship of water depth differences.

[0049] After completing the boundary detection of smooth reflective areas and comparing energy and time delay features, these features are combined according to spatial location, temporal order, and reflective energy level to form a difference mapping table for shallow and deep water accumulation. The difference mapping table uses spatial coordinates as an index and records the energy intensity, echo delay, and relative water level characteristics of each reflective point. This mapping method transforms continuous smooth reflective areas in the reflective distribution map into a structured dataset with water level feature classifications. Shallow water accumulation areas are represented in the mapping table as regions with concentrated energy and short time delays, while deep water accumulation areas correspond to distribution areas with dispersed energy and long time delays. To ensure the difference mapping table can be directly referenced in subsequent identification processes, the reflective features corresponding to different water level types are recorded in the table in the form of thresholds, with each threshold interval representing a specific water level state. This structured data representation allows for rapid determination of water level categories based on real-time reflective data in subsequent identification stages, thus providing an accurate reference for water accumulation identification.

[0050] By using a difference mapping table to adjust the timing of early warning triggers, the execution rhythm of water level identification is determined, and a time control command containing the identification cycle and trigger delay is generated to optimize the identification timing under rainfall interference conditions. The timing sequence of early warning triggers is adjusted using a difference mapping table to determine the execution rhythm of water level identification and generate time control instructions containing the identification cycle and trigger delay. This optimizes the identification timing under rainfall interference conditions. The specific implementation steps are as follows: The difference mapping table is used for time-series correlation, dynamic scheduling, and rhythm coordination. Based on the reflection energy, time delay characteristics, and spatial distribution of shallow and deep water accumulation recorded in the difference mapping table, the continuous change relationship of various water accumulation characteristics in the time dimension is extracted. Each data point in the difference mapping table corresponds to the reflection state at a specific time node, including energy intensity, time delay span, and water level classification information. To determine the basic logic for the warning triggering sequence, the reflection energy change trajectory of the same spatial location at different time periods is first arranged vertically using the time axis as a reference, so that the conversion trend between shallow and deep water accumulation is continuously expressed in the time dimension. In this process, the temporal distribution of reflection energy change is corrected by combining rainfall density data and vehicle speed parameters, so that the water accumulation state at different time nodes corresponds synchronously with the actual driving environment. Through this time correlation method, the evolution rhythm of different water level characteristics in time can be clearly defined, thus providing a data foundation for determining the identification cycle.

[0051] After obtaining the continuous water level characteristic change relationship over time, the warning trigger priority is determined based on the reflected energy fluctuation range in the difference mapping table. The warning trigger priority is based on the water level change rate and energy fluctuation amplitude, reflecting the potential impact of different water accumulation states on vehicle driving safety. To ensure the trigger sequence meets the actual needs under dynamic rainfall conditions, sections with larger energy change rates are defined as high-priority trigger zones, and sections with gradual energy changes are defined as low-priority trigger zones. Subsequently, these priority trigger zones are arranged into a continuous time-based trigger sequence. This sequence can dynamically reflect the potential risk level of the road surface water level ahead based on the water accumulation trend. Combining the spatial information in the reflection distribution map, the time-based trigger sequence is paired with spatial coordinates, so that each trigger node corresponds to a specific road location and water depth state, thus establishing a mapping relationship between time-based triggering and spatial distribution. In this way, the warning system can automatically adjust the trigger rhythm at different times based on spatial location and water level characteristic changes, avoiding reaction delays or excessively frequent warnings caused by a single time-based trigger.

[0052] After determining the warning triggering order and priority, the identification cycle and trigger delay parameters are set based on the stable range of reflection characteristics for each water level type in the difference mapping table. The identification cycle defines the time interval between consecutive water level identification actions, while the trigger delay is used to adjust the timing of identification actions under high interference conditions. To ensure that the identification cycle setting conforms to the continuous characteristics of actual rainfall interference, the average duration of shallow and deep water accumulation during periods of high rainfall is first statistically analyzed, and this duration is used as the initial reference value for the identification cycle. Then, combined with the trend of rainfall density changes, the identification cycle is appropriately adjusted so that the identification rhythm can cover the range of changes throughout the rainfall process. The trigger delay setting is based on the average of the duration of raindrop interference and the recovery time of reflected energy. By embedding the trigger delay within the identification cycle, the identification action can be temporarily suspended during the strongest interference phase, and identification can be initiated again after the signal stabilizes, thereby reducing the impact of interference noise. In this process, each identification action is time-correlated with the previous trigger moment, making the entire identification process form a continuous and stable rhythm chain.

[0053] After setting the identification cycle and trigger delay parameters, these time parameters are combined with the trend of reflected energy changes to generate time control instructions. These instructions include the identification start time, execution interval, delay trigger time, and corresponding water level type identifier, guiding subsequent identification and adjustment processes. When generating time control instructions, the aforementioned identification cycle and trigger delay parameters are sequentially embedded into each early warning trigger node, using the time axis as the main thread, giving each node independent time control attributes. Thus, under rainfall interference conditions, identification actions can be executed sequentially according to the preset time control instructions, ensuring that identification actions at different stages are interconnected in time and continuously covered in space. Simultaneously, the water level type identifier is embedded in the time control instructions, allowing subsequent identification and adjustment processes to automatically match the corresponding execution rhythm based on different water level characteristics, achieving dynamic time control for different rainfall intensities and water accumulation states. In this way, the identification process can maintain consistent rhythm and continuous execution in complex rainfall environments, ensuring synchronization between the early warning trigger timing and the actual environmental conditions.

[0054] Based on time control instructions, the system performs recognition adjustments in rainy conditions. By inserting reverse detection windows during rain gaps, alternately skipping interference sections, and releasing virtual warning signals to occupy positions, it continuously corrects reflection deviations in water level recognition, maintaining the continuity and stability of the vehicle flood warning process. Based on time control commands, the system performs recognition adjustments under rainy conditions. By inserting reverse detection windows during rainfall intervals, alternately skipping interference sections, and releasing virtual warning signals to fill gaps, it continuously corrects reflection deviations during water level recognition and maintains the continuity and stability of the vehicle flood warning process. The specific implementation steps are as follows: Based on the recognition cycle and trigger delay parameters set in the time control command, the recognition start time and execution window during rainfall are determined. In the time control command, each time node corresponds to the trend of reflected energy change and the state of water accumulation. Therefore, when performing recognition adjustment, it is first determined whether there is a rainfall gap based on the delay setting of the current time node. When a rainfall gap is detected, the insertion process of the reverse detection window is immediately initiated. Inserting the reverse detection window refers to opening an additional short-duration, high-frequency reverse detection cycle outside the regular recognition cycle to capture changes in the echo signal during the rainfall gap. Through reverse detection, reflected signals from the road surface and low-altitude areas can be received from angles opposite to the main recognition direction, enabling the system to obtain cleaner echo information during periods of reduced raindrop scattering. The duration and trigger frequency of the reverse detection window are limited by the cycle parameter in the time control command, ensuring a continuous time alternation structure between the reverse detection process and the main recognition process, thereby avoiding time blind spots during recognition.

[0055] After the reverse detection window is inserted, the system identifies and skips interference segments in the current identification cycle according to the triggering sequence set in the time control command. The determination of interference segments is based on the reflection distribution map and difference mapping table formed in the previous raindrop reflection sequence. When the time control command reaches the time node corresponding to the interference segment, the system judges the interference intensity based on the reflection energy fluctuation state at that time node. When uneven echo signal energy distribution or abnormally elongated reflection delay curve is detected in the interference segment, water level identification is not immediately performed. Instead, the identification action for that segment is temporarily stopped according to the skipping strategy of the time control command, and restarted at the next time node. Skipping interference segments effectively prevents signal reflection chaos caused by heavy rainfall from interfering with the identification results, while maintaining the continuity of the identification rhythm through the skipped time window. During the skipped segment period, the system continuously records the reflection energy fluctuation information within that time period, providing a reference for subsequent identification adjustments, enabling a smooth transition based on continuous data from previous and subsequent time periods when re-identifying in the next identification cycle.

[0056] While performing the interference section skipping operation, to ensure the integrity of the identification timing and the continuity of the warning process, virtual warning signal placeholders are released according to time control commands. Virtual warning signal placeholders refer to maintaining the continuous existence of the warning signal stream by generating a set of placeholder signals corresponding to time nodes during the time period when the actual reflected signal cannot be effectively identified. This virtual signal is not actual reflected data, but rather time placeholder data generated based on the warning triggering sequence and delay parameters recorded in the time control commands, used to maintain the integrity of the time sequence during identification. When subsequent time nodes re-enter the identifiable state, the virtual signal is replaced by the new identification result, thus achieving seamless transition. In this way, the temporal continuity of the warning signal output is maintained, and even during brief signal interruptions caused by heavy rainfall, the warning system can still maintain a stable output rhythm. The release of virtual warning signals not only ensures the integrity of the signal sequence on the timeline but also allows subsequent identification adjustments to compensate and correct based on the continuous time trajectory, providing continuous reference clues for reflection deviation correction.

[0057] After completing reverse detection, skipping interference sections, and virtual early warning signal placement, the execution results of these operations are integrated into a new recognition sequence in chronological order, thereby continuously correcting reflection deviations in water level recognition. During integration, the execution rhythm of the time control command is used as the main thread, aligning the reflection data of each reverse detection window, the recorded data of each skipped section, and the time nodes of each virtual early warning signal placement, ensuring continuity of the recognition process in the time dimension. By comparing the difference between the reflection signal obtained from reverse detection and the main recognition signal, the magnitude of reflection deviation caused by raindrop interference can be identified. Combining the trend of reflection energy changes before and after skipping sections, the execution time of subsequent recognition cycles is adjusted to avoid peak interference periods. During the continuous execution of multiple recognition cycles, the correspondence between reflection signals and time nodes is dynamically corrected to ensure that the recognition rhythm remains synchronized with environmental changes, thus achieving stable operation of the water level recognition process in heavy rainfall environments.

[0058] This invention achieves effective separation of stray echo interference in heavy rainfall environments by introducing the dynamic construction of raindrop reflection sequences and spatial feature identification of reflection distribution maps during flood warning processes. This allows water level identification to be based on actual ground reflection signals. By fusing multi-source reflection data with rainfall density information in chronological order, the reflection signals are uniformly expressed in both temporal and spatial dimensions, thereby improving the accuracy of identifying shallow and deep water features under complex rain conditions and providing a stable data foundation for early warning decisions.

[0059] This invention establishes a difference mapping table and dynamically adjusts it using time control commands, enabling real-time regulation of water level recognition rhythm in response to rainfall changes, ensuring that the timing of warning triggering is synchronized with environmental changes. During the recognition and adjustment process, the synergistic effect of the reverse detection window and virtual signal occupancy continuously corrects reflection deviations, maintaining continuous output of warning signals. This method achieves stable operation of the warning process under strong interference conditions, improving the reliability and safety performance of vehicle wading risk assessment.

[0060] This invention provides, for example Figure 2 The illustrated automotive flooding online intelligent early warning system includes a data acquisition module, an interference feature extraction module, a water accumulation identification module, a time control module, and an identification correction module. The data acquisition module acquires water level reflection signals from the road surface in front of the vehicle, environmental rainfall density data, and vehicle tilt angle parameters, forming a continuous raindrop reflection sequence in chronological order for subsequent rain curtain interference characteristic analysis. The interference feature extraction module analyzes the variation pattern of echo particle intensity during periods of high rainfall based on raindrop reflection sequence analysis, and constructs a reflection distribution map to reflect the morphology of rainfall interference, providing spatial distribution basis for subsequent water accumulation identification; The water accumulation identification module detects the boundary change characteristics of flat reflection areas based on the reflection distribution map, and establishes a difference mapping table between shallow and deep water accumulation by combining the differences in reflection energy and time delay at different water depths, which is used to distinguish different water level characteristics. The time control module uses a difference mapping table to adjust the timing sequence of early warning triggers, determines the execution rhythm of water level identification, and generates time control instructions containing identification cycles and trigger delays to optimize the identification timing under rainfall interference conditions. The recognition and correction module performs recognition adjustments under rainy conditions according to time control commands. By inserting reverse detection windows during rain gaps, alternately skipping interference sections, and releasing virtual warning signals to occupy positions, it continuously corrects the reflection deviation in water level recognition, maintaining the continuity and stability of the vehicle flood warning process.

[0061] The present invention provides an online intelligent early warning method for vehicle flooding, which is implemented through the above-mentioned online intelligent early warning system for vehicle flooding. For details of the specific method and process of the online intelligent early warning system for vehicle flooding, please refer to the above-mentioned embodiment of the online intelligent early warning method for vehicle flooding, which will not be repeated here.

[0062] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for online intelligent early warning of vehicle flooding, characterized in that, Includes the following steps: The water level reflection signal on the road surface in front of the vehicle, the ambient rainfall density data, and the vehicle tilt angle parameters are acquired and formed into a continuous raindrop reflection sequence in chronological order. Based on the analysis of raindrop reflection sequence, the variation pattern of echo particle intensity during periods of high rainfall is analyzed, and a reflection distribution map reflecting the morphology of rainfall interference is constructed. Based on the reflection distribution map, the boundary change characteristics of the flat reflection area were detected. Combined with the differences in reflection energy and time delay at different water depths, a difference mapping table between shallow and deep water was established. The timing sequence of early warning triggers is adjusted using a difference mapping table to determine the execution rhythm of water level identification and generate time control instructions that include the identification cycle and trigger delay. Based on time control instructions, the system performs recognition adjustments in rainy conditions. By inserting reverse detection windows during rainfall intervals, alternately skipping interference sections, and releasing virtual early warning signals to occupy positions, the system continuously corrects the reflection deviation in water level recognition.

2. The online intelligent early warning method for vehicle flooding according to claim 1, characterized in that, The steps for obtaining the raindrop reflection sequence are as follows: The electromagnetic wave transmitting and receiving device installed at the front of the vehicle continuously scans the road surface in the direction of the vehicle's travel, collects water level reflection signals from the road surface and near-ground air layer, and records information on water surface echo intensity, reflection angle, and echo delay time. Rainfall density data is acquired by an environmental monitoring device installed on the outside of the vehicle body, and the rainfall density data is recorded synchronously with the water level reflection signal acquisition time point to form paired data with time sequence. The vehicle tilt angle parameters are obtained by an attitude measurement device installed at the center of the vehicle body, and the vehicle tilt angle parameters are correlated with the water level reflection signal and rainfall density data for the corresponding time period. The water level reflection signal, rainfall density data, and vehicle tilt angle parameters are integrated in chronological order to form a continuous raindrop reflection sequence.

3. The online intelligent early warning method for vehicle flooding according to claim 2, characterized in that, In the process of forming the raindrop reflection sequence, the water level reflection signal, rainfall density data and vehicle tilt angle parameters are paired and arranged according to a unified time label. Each time node fully contains reflection information, rainfall intensity information and vehicle attitude information. By continuously combining the time sequence, a time chain dataset is constructed, so that the raindrop reflection sequence has temporal continuity and spatial correspondence under dynamic rainfall conditions.

4. The online intelligent early warning method for vehicle flooding according to claim 2, characterized in that, The steps for constructing a reflection distribution map are as follows: Based on the water level reflection signal data corresponding to each time node in the raindrop reflection sequence, the intensity of the reflection signal in the continuous rainfall period is extracted in layers, and the raindrop reflection sequence is divided into intervals at fixed time intervals to form a continuous energy distribution trajectory. After obtaining the time-intervalized reflection energy distribution, the high rainfall period is determined based on the changing trend of rainfall density data in the raindrop reflection sequence. The time interval of the high rainfall period is analyzed in detail and the echo particle intensity distribution is extracted. After obtaining the echo particle intensity distribution during periods of high rainfall, a three-dimensional spatial distribution matrix is ​​constructed according to the correspondence between spatial location and temporal order. After obtaining the spatial distribution matrix, the trajectory of reflected energy change is continuously mapped to form a reflection distribution map.

5. The online intelligent early warning method for vehicle flooding according to claim 4, characterized in that, The steps to create a difference mapping table are as follows: Based on the reflection distribution map, the continuous reflection area is spatially divided to identify the initial boundary of the flat reflection area, and the reflection plane under different slope conditions is geometrically corrected in combination with the vehicle tilt angle parameter to ensure accurate spatial positioning. After obtaining the initial boundary range of the flat reflective area, the reflective energy change characteristics on both sides of the boundary are compared to extract the energy difference characteristics of the water depth change, and the rainfall density data is correlated with the reflective energy difference to improve the recognition accuracy. After analyzing the energy difference on both sides of the boundary of the flat reflection area, the echo time delay variation characteristics under the same spatial location are compared to supplement the judgment basis of water depth difference, and the reflection energy and time delay information are superimposed and displayed in a spatial coordinate system. After completing the boundary detection of the flat reflection area and comparing the energy and time delay features, the features are combined according to spatial location, time sequence and reflection energy level to form a difference mapping table between shallow and deep water.

6. The online intelligent early warning method for vehicle flooding according to claim 5, characterized in that, The difference mapping table records the energy intensity, echo delay, and water level characteristics of the reflection point using spatial coordinates as an index. It also divides the reflection characteristics corresponding to different water level types into intervals using thresholds, so that shallow water areas correspond to distribution areas with concentrated energy and short delays, while deep water areas correspond to distribution areas with dispersed energy and long delays.

7. The online intelligent early warning method for vehicle flooding according to claim 5, characterized in that, The steps for generating time control instructions are as follows: Based on the reflected energy, time delay characteristics, and spatial location distribution of shallow and deep water recorded in the difference mapping table, the continuous change relationship of water features in the time dimension is extracted, and the time distribution of reflected energy changes is corrected by combining rainfall density data and vehicle speed parameters. After obtaining the continuous water level characteristic change relationship in the time dimension, the warning trigger priority is divided according to the reflected energy fluctuation range in the difference mapping table, and the triggering areas are arranged into a continuous time triggering sequence in time order. After determining the warning triggering order and priority, the identification period and triggering delay parameters are set according to the stable interval of the reflection characteristics of each water level type in the difference mapping table, and the identification period and triggering delay are embedded inside the identification period to reduce interference. After setting the identification cycle and trigger delay parameters, the time parameters and the trend of reflected energy change are combined to generate time control commands.

8. The online intelligent early warning method for vehicle flooding according to claim 7, characterized in that, The time control command includes the identification start time, execution interval time, delay trigger time, and water level type identifier. The identification start time is used to determine the initial execution point of the early warning trigger node, the execution interval time is used to control the time interval between continuous identification actions, the delay trigger time is used to postpone the identification execution during the rainfall interference phase, and the water level type identifier is used to indicate the water level status within the corresponding identification period.

9. The online intelligent early warning method for vehicle flooding according to claim 7, characterized in that, The recognition adjustment under rainy conditions is performed according to time control instructions. The steps to correct the reflection deviation in water level recognition are as follows: inserting reverse detection windows during rainfall intervals, alternately skipping interference sections, and releasing virtual early warning signals to fill gaps. Based on the recognition cycle and trigger delay parameters set in the time control command, the recognition start time and execution window during the rainfall process are determined, and a reverse detection window is inserted when a rainfall gap is detected to capture changes in the echo signal; After the reverse detection window is inserted, the interference segment in the current identification cycle is identified according to the triggering order set in the time control command, and a skip operation is performed when uneven energy distribution of echo signal or abnormally elongated reflection delay curve is detected to maintain the continuous identification rhythm. While performing the interference section skipping operation, the virtual early warning signal is released according to the time control command to maintain the continuous existence of the early warning signal stream and keep the output rhythm stable. After completing reverse detection, skipping interference sections, and virtual early warning signal placement, the operation results are integrated into a new identification sequence in chronological order to achieve continuous correction of reflection deviation and maintain the continuity and stability of the early warning process.

10. A vehicle flooding online intelligent early warning system, used to implement the vehicle flooding online intelligent early warning method according to any one of claims 1-9, characterized in that, It includes a data acquisition module, an interference feature extraction module, a water accumulation identification module, a time control module, and an identification and correction module: The data acquisition module acquires water level reflection signals from the road surface in front of the vehicle, environmental rainfall density data, and vehicle tilt angle parameters, forming a continuous raindrop reflection sequence in chronological order. The interference feature extraction module analyzes the variation pattern of echo particle intensity during periods of high rainfall based on raindrop reflection sequence analysis, and constructs a reflection distribution map to reflect the morphology of rainfall interference. The water accumulation identification module detects the boundary change characteristics of flat reflection areas based on the reflection distribution map, and establishes a difference mapping table between shallow and deep water accumulation by combining the differences in reflection energy and time delay at different water depths. The time control module uses a difference mapping table to adjust the timing sequence of early warning triggers, determines the execution rhythm of water level identification, and generates time control instructions that include the identification cycle and trigger delay. The recognition and correction module performs recognition adjustments under rainy conditions according to time control instructions. It continuously corrects the reflection deviation in water level recognition by inserting reverse detection windows during rainfall intervals, alternately skipping interference sections, and releasing virtual early warning signals to occupy positions.

Images