Environment-adaptive data fusion method and system for dredger

Abstract

The application discloses a kind of based on environmental adaptation's dredger's data fusion method and system, the present application relates to the technical field of data fusion, according to environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value, sea state grade quantification value and illumination intensity quantification value determine environmental state change rate, and mark corresponding environmental state change content, improve the accuracy of quantification environmental state change rate.According to the performance content and the current working scene of dredger determine the performance evaluation index of each sensor, the data fusion of each performance evaluation index is carried out, and the environmental adaptive combination is output;If environmental state change rate is higher than preset environmental state change rate threshold, then trigger real-time update of environmental adaptive combination, and adjust the priority of each environment detection part in the environmental adaptive combination, to adapt to the changing environment where dredger is located, and output final environmental sensing result, ensure the accuracy of final environmental sensing result.

Description

Technical Field

[0001] This invention relates to the field of data fusion technology, and in particular to a data fusion method and system for an environment-adaptive dredger. Background Technology

[0002] As the core equipment for dredging operations, dredgers often operate in waters filled with obstacles such as pontoons, pipelines, and auxiliary vessels. During operations, they need to frequently turn and maneuver, and changes in the hull load can lead to significant changes in draft. Shallow spots and reefs near the vessel pose significant safety hazards. To achieve safety situational awareness in the operating waters, dredgers currently rely mainly on the traditional mode of "maritime radar + AIS + human lookout".

[0003] Existing technologies typically rely on single meteorological data or simple visual judgments to assess environmental conditions, lacking comprehensive perception and control of multiple environmental dimensions such as precipitation intensity, visibility, temperature, sea state, and light intensity. This results in the system's inability to accurately capture subtle changes in environmental conditions, making it difficult to determine changes in environmental conditions and affecting the accuracy of the rate of change in environmental conditions, ultimately leading to low accuracy in the final environmental sensing results. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a data fusion method and system for dredgers based on environmental adaptation.

[0005] This invention provides a data fusion method for dredgers based on environment adaptation, including:

[0006] The current location of the dredger is marked, and environmental multi-source sensing is triggered based on the current location of the dredger to collect raw data on precipitation intensity, visibility, temperature and sea state. Based on the mapping of multiple raw data, the corresponding environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value and sea state level quantification value are determined.

[0007] The corresponding quantized value of light intensity is determined by recognizing the images collected by the camera of the dredger. The rate of change of environmental state is determined based on the quantized values ​​of ambient temperature, visibility, precipitation intensity, sea state level and light intensity, and the corresponding environmental state change content is marked.

[0008] The system labels multiple sensors on a dredger and determines the corresponding performance content based on the detection of each sensor. Based on the performance content and the current working scenario of the dredger, the system determines the performance evaluation index of the sensor, fuses the data of each performance evaluation index, and outputs an environment-adaptive combination.

[0009] If the rate of change of the environmental state is higher than the preset threshold for the rate of change of the environmental state, the real-time update of the environmental adaptive combination is triggered, and the priority of each environmental detection part in the environmental adaptive combination is adjusted to adapt to the changing environment in which the dredger is located, and the final environmental sensing result is output.

[0010] This invention provides a data fusion system for dredgers based on environmental adaptation. This system is applied to the aforementioned data fusion method for dredgers based on environmental adaptation. The data fusion system for dredgers based on environmental adaptation includes:

[0011] The quantization module is used to mark the current position of the dredger and trigger multi-source environmental sensing based on the current position of the dredger to collect raw data on precipitation intensity, visibility, temperature and sea state. Based on the mapping of multiple raw data, the corresponding environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value and sea state level quantization value are determined.

[0012] The environmental change module is used to determine the corresponding light intensity quantization value based on the recognition of images collected by the dredger's camera, determine the environmental state change rate based on the environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value, sea state level quantization value and light intensity quantization value, and mark the corresponding environmental state change content.

[0013] The environment adaptive module is used to label multiple sensors of the dredger and determine the corresponding performance content based on the detection of each sensor. Based on the performance content and the current working scenario of the dredger, the module determines the performance evaluation index of the sensor, fuses the data of each performance evaluation index, and outputs the environment adaptive combination.

[0014] The environmental perception result module is used to trigger real-time updates of the environmental adaptive combination if the rate of change of the environmental state is higher than the preset environmental state change rate threshold, and to adjust the priority of each environmental detection part in the environmental adaptive combination to adapt to the changing environment in which the dredger is located, and output the final environmental perception result.

[0015] Compared with the prior art, the beneficial effects of the present invention are:

[0016] (1) Mark the current position of the dredger and trigger multi-source environmental perception based on the current position of the dredger to collect raw data of precipitation intensity, visibility, temperature and sea state, and determine the corresponding environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value and sea state level quantification value based on the mapping of multiple raw data; determine the corresponding light intensity quantification value based on the recognition of the images collected by the dredger's camera, determine the environmental state change rate based on the environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value, sea state level quantification value and light intensity quantification value, and mark the corresponding environmental state change content. The environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value, sea state level quantification value and light intensity quantification value are introduced to perform multi-source perception of various environmental parameters of the dredger and realize the control of multiple environmental dimensions, so as to improve the accuracy of the environmental state change rate.

[0017] (2) Mark multiple sensors of the dredger and determine the corresponding performance content based on the detection of each sensor. Determine the performance evaluation index of the sensor according to the performance content and the current working scenario of the dredger. Fusion of the various performance evaluation indexes and output of the environmental adaptive combination. If the rate of change of the environmental state is higher than the preset environmental state change rate threshold, the real-time update of the environmental adaptive combination is triggered, and the priority of each environmental detection part in the environmental adaptive combination is adjusted to adapt to the changing environment of the dredger. The final environmental sensing result is output to control the environmental changes and realize the data fusion of various performance evaluation indexes, improve the accuracy of the environmental adaptive combination, and ensure the accuracy of the final environmental sensing result. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the data fusion method for an environment-adaptive dredger in an embodiment of the present invention.

[0019] Figure 2 This is a flowchart illustrating step S11 in the data fusion method for an environment-adaptive dredger according to an embodiment of the present invention.

[0020] Figure 3 This is a flowchart illustrating step S12 in the data fusion method for an environment-adaptive dredger according to an embodiment of the present invention.

[0021] Figure 4 This is a flowchart illustrating step S13 in the data fusion method for an environment-adaptive dredger according to an embodiment of the present invention.

[0022] Figure 5 This is a flowchart illustrating step S14 in the data fusion method for an environment-adaptive dredger according to an embodiment of the present invention.

[0023] Figure 6 This is a schematic diagram of the structural composition of the data fusion system for an environment-adaptive dredger in an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] Please see Figures 1 to 6 An environment-adaptive data fusion method for dredgers is proposed and applied to data fusion scenarios. The environment-adaptive data fusion method for dredgers includes:

[0026] Step S11: Mark the current position of the dredger and trigger multi-source environmental sensing based on the current position of the dredger to collect raw data of precipitation intensity, visibility, temperature and sea state, and determine the corresponding quantitative values ​​of environmental temperature, visibility, precipitation intensity and sea state based on the mapping of multiple raw data.

[0027] Step S12: Based on the recognition of images collected by the dredger's camera, determine the corresponding light intensity quantification value, determine the environmental state change rate based on the environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value, sea state level quantification value and light intensity quantification value, and mark the corresponding environmental state change content.

[0028] Step S13: Mark multiple sensors of the dredger and determine the corresponding performance content based on the detection of each sensor. Determine the performance evaluation index of the sensor according to the performance content and the current working scenario of the dredger. Fusion of the various performance evaluation indexes and output of environmental adaptive combination.

[0029] Step S14: If the rate of change of the environmental state is higher than the preset threshold for the rate of change of the environmental state, the real-time update of the environmental adaptive combination is triggered, and the priority of each environmental detection part in the environmental adaptive combination is adjusted to adapt to the changing environment in which the dredger is located, and the final environmental sensing result is output.

[0030] refer to Figure 2 In step S11, the specific steps are as follows:

[0031] S111: The dredger is equipped with a shipborne meteorological instrument. The environmental detection of the shipborne meteorological instrument is triggered according to the current position of the dredger and the scene in which the dredger is located. The dredger performs multi-source environmental sensing to monitor and collect raw data on precipitation intensity, visibility, temperature and sea state in real time.

[0032] S112: Quantify each raw data point, converting it into a unified, standardized environmental parameter matrix to determine the corresponding quantified values ​​for ambient temperature, visibility, precipitation intensity, and sea state. Simultaneously, for the ambient temperature quantified values, map the air temperature data to a 0-1 range, where 0 represents no temperature influence and 1 represents extreme temperature saturation. For the visibility quantified values, map the visibility data to a 0-1 range, where 0 represents excellent visibility and 1 represents extremely poor visibility. For the precipitation intensity quantified values, map the precipitation intensity data to a 0-1 range, where 0 represents no precipitation and 1 represents extremely heavy precipitation.

[0033] In the embodiments of this application, the dredger is equipped with a shipborne meteorological instrument. The environmental detection of the shipborne meteorological instrument is triggered according to the current position of the dredger and the scene in which the dredger is located. The dredger performs multi-source environmental sensing to monitor and collect raw data of precipitation intensity, visibility, temperature and sea state in real time.

[0034] At this time, the dredger is equipped with a shipborne weather instrument, and the temperature sensor is marked with a range of -40℃ to 60℃ and an accuracy of ±0.2℃; the visibility sensor has a range of 50m to 2000m and an accuracy of ±5%; and the precipitation sensor has a range of 0mm / h to 50mm / h and an accuracy of ±0.1mm. At the same time, the shipborne weather instrument is installed on the top of the stern mast to ensure that the sensor can collect accurate environmental data and to avoid obstruction by the hull or human interference.

[0035] The system can automatically trigger environmental detection based on the dredger's current working scenario. For example: Anchoring operations: When the dredger is anchoring, the system will automatically trigger environmental detection to obtain environmental information near the anchorage, ensuring anchoring safety; Dredging operations: When the dredger is dredging, the system will automatically trigger environmental detection to obtain hydrological and meteorological information of the dredging area, optimizing the dredging plan. Additionally, the shipborne weather instrument has a built-in timer that automatically starts environmental detection periodically, such as once per minute; or, the crew can manually trigger environmental detection as needed, such as in case of sudden weather changes or when entering a new work area. The shipborne weather instrument collects raw data on environmental parameters such as temperature, visibility, precipitation intensity, and sea state in real time; the collected raw data is then transmitted to the shipborne edge computing unit for processing and analysis.

[0036] Specifically, when a dredging vessel is conducting dredging operations at sea and encounters heavy rain and rough sea conditions, the onboard weather instrument is installed on top of the stern mast and is in normal working order. The system automatically triggers environmental detection based on the current dredging operation scenario of the dredging vessel.

[0037] The shipborne meteorological instrument collected the following environmental data in real time: temperature: 25℃; visibility: 200m; precipitation intensity: 15mm / h; sea state level: 4. After analyzing the images, the camera image quality analysis unit determined that the light intensity was low and the visibility was poor. After analyzing the point cloud, the lidar point cloud quality analysis unit determined that the visibility was poor and the precipitation intensity was high. The system successfully collected environmental data of the dredger under heavy rain, providing important input information for subsequent sensor performance evaluation and dynamic weight allocation.

[0038] Furthermore, the raw data are quantified and converted into a unified, standardized environmental parameter matrix to determine the corresponding quantified values ​​for ambient temperature, visibility, precipitation intensity, and sea state. Simultaneously, for the ambient temperature quantified values, air temperature data is mapped to a 0-1 range, where 0 represents no temperature influence and 1 represents extreme temperature saturation. For the visibility quantified values, visibility data is mapped to a 0-1 range, where 0 represents excellent visibility and 1 represents extremely poor visibility. For the precipitation intensity quantified values, precipitation intensity data is mapped to a 0-1 range, where 0 represents no precipitation and 1 represents extremely heavy precipitation. This process incorporates the overall consideration of converting to a unified, standardized environmental parameter matrix, ensuring the accuracy of the corresponding ambient temperature quantified values.

[0039] At this point, the raw data of each environmental parameter are mapped to a range of 0-1, where 0 represents the most favorable environment and 1 represents the worst environment; for the quantified value of environmental temperature ( Based on the temperature data (T, unit: °C), use the following formula to calculate: When T≤T min (Extreme cold threshold) or T≥T max At (high temperature threshold), =1 (saturation at extreme temperatures); when T min <T<T low At (low temperature critical value), =(T min -T) / (T min -T low When T low ≤T≤T high (At the high temperature critical value) =(TT high ) / (T max -T high When T high <T<T max hour, =(TT high ) / (T max -T high ).

[0040] For the measurement of visibility ( Based on visibility data (V, unit: m), calculate using the following formula. When V <V min (At the threshold of extremely poor visible environment) =1 (range visible environment); when V min ≤V≤V max (At the threshold of excellent visibility environment) =(V max -V) / (V max -V min When V>V max hour, =0 (Excellent visibility environment).

[0041] Quantification of precipitation intensity ( Based on precipitation intensity data (R, unit: mm / h), calculate e3 using the following formula: When R < 0 (data outlier), =0; when 0≤R≤R max (At the threshold of extremely heavy precipitation) =R / R max When R>R max hour, =1 (extremely heavy precipitation); at the same time, the sea state level quantification value ( ): Calculate using the following formula based on sea state grade data (S, using Douglas sea state grades). =S / S max S max The definition is the highest upper limit of the sea state level in the process of mapping the sea state level quantization value; the quantized value is the upper limit of the sea state level. , , , A 4D environmental parameter matrix is ​​formed as follows: , , , ].

[0042] Specifically, based on the environmental data collected by S111, a quantization formula is used to calculate... , , , ,get: =0 (temperature is within the normal range); =0.84 (poor visibility); =0.75 (relatively high rainfall intensity); =0.44 (bad sea conditions); will , , , The environmental parameter matrix is ​​[0, 0.84, 0.75, 0.44]. The system successfully quantified the environmental data of the dredger under heavy rain into an environmental parameter matrix, providing important input information for subsequent sensor performance evaluation and dynamic weight allocation.

[0043] refer to Figure 3 In step S12, the specific steps are as follows:

[0044] S121: The camera that marks the dredger takes pictures of the surrounding area of ​​the dredger based on the camera to collect corresponding images, performs image recognition on the images, determines multiple light intensity elements during the recognition process, and determines the corresponding light intensity quantization value based on the quantization of multiple light intensity elements.

[0045] S122: Obtain the quantified values ​​of ambient temperature, visibility, precipitation intensity, sea state, and light intensity. Input these values ​​into the same environmental parameter matrix and iterate within the matrix to determine the corresponding rate of change of environmental state and present the corresponding environmental state change content.

[0046] In the embodiments of this application, the camera of the dredger is marked, and the surrounding area of ​​the dredger is photographed based on the camera to collect corresponding images. Image recognition is performed on the images, and multiple light intensity elements are determined during the recognition process. The corresponding light intensity quantization value is determined based on the quantization of multiple light intensity elements, which takes into account the overall consideration of multiple light intensity elements and ensures the accuracy of the corresponding light intensity quantization value.

[0047] At this point, based on the sensor topology of the dredger, the system marks the visual sensors (such as 3 visible light cameras) deployed forward of the compass deck; the system controls these cameras to continuously photograph the key operating waters around the dredger (such as the 240° wide-angle coverage area at the bow); considering the attitude changes of the hull under adverse sea conditions, the shooting command synchronously calls the GNSS positioning module and attitude sensor data to stabilize the image acquisition, ensuring that the acquired image sequence can truly reflect the optical environment of the current waters, rather than the blurry trailing image caused by the violent swaying of the hull.

[0048] After acquiring the real-time image stream, the shipborne edge computing unit initiates the image quality analysis unit to perform deep recognition analysis on the image. Unlike conventional target recognition, the focus here is on extracting key elements representing the ambient light intensity from the image texture. Specifically, the system calculates the global average brightness, contrast, and image noise level of the image. By analyzing the distribution pattern of the image histogram, the grayscale mean is extracted as a brightness benchmark. The contrast is evaluated by calculating the edge gradient intensity, as a sharp drop in contrast often indicates insufficient lighting or strong glare. At the same time, the random noise distribution in the dark areas of the image is detected, because in low-light environments, increasing the sensor gain will lead to a significant increase in noise points.

[0049] Based on the extracted multiple light intensity elements, the system uses a logarithmic mapping algorithm to fuse them and map them into standardized quantized light intensity values. (Value range 0-1); This mapping logic follows the rule that "0 represents extremely strong illumination (most favorable environment), and 1 represents extremely low illumination (most unfavorable environment)"; the specific formula is:

[0050]

[0051] in, I To infer the actual illuminance from the image brightness elements, I min Set to 0.1 lx (extremely low illumination threshold). I max Set to 60000 lx (extremely high illuminance threshold); if the calculated illuminance I Below 0.1 lx, then Set the value to 1 directly, marking it as an extremely low illumination environment; if it is higher than 60,000 lx, then... Setting the value to 0 indicates an environment with extremely high illumination. The use of logarithmic mapping effectively simulates the nonlinear response characteristics of the human eye and sensors to changes in illumination, ensuring the scientific validity of the quantification results.

[0052] Specifically, the system marks the visible light camera facing forward of the dredging deck to take pictures of the dredging operation area in front of the bow; despite the hull swaying due to sea state 4, the camera combines attitude sensor data for electronic image stabilization and collects a set of real-time images that reflect the current sea surface conditions.

[0053] The image quality analysis unit analyzes the image; due to the heavy rain at night, the image shows a large area of ​​dark tones, the average gray level of the image extracted by the system is extremely low, the contrast is weak, and obvious noise is visible in the dark areas of the image under high ISO gain; the system determines that the current ambient light intensity is characterized by "low illumination, low contrast, and high noise".

[0054] The actual illuminance, inferred from the image brightness, is approximately 100 lx; ​​the system uses the logarithmic mapping formula for calculation.

[0055]

[0056] The system successfully quantified the visual environment of the dredger during a rainstorm night into a quantitative value of light intensity. =0.48; This value indicates that the current lighting conditions are at a low to medium level, which will significantly suppress the performance of the visual sensor; This value, together with the temperature, visibility, precipitation, and sea state quantification values ​​determined in step S112, will construct a complete environmental parameter matrix of [0,0.84,0.75,0.48,0.44], providing key input for the accurate calculation of the sensor performance evaluation model in the subsequent step S13, ensuring that the fusion system can correctly reduce the weight of visual data in nighttime rainstorm scenarios.

[0057] Furthermore, quantitative values ​​of ambient temperature, visibility, precipitation intensity, sea state, and light intensity are acquired. These values ​​are then input into the same environmental parameter matrix, and iterative processing is performed within this matrix to determine the corresponding rate of change in environmental state and to present the corresponding environmental state changes. This approach incorporates a holistic consideration of iterative processing within the environmental parameter matrix, ensuring the accuracy of the corresponding rate of change in environmental state. Simultaneously, the introduction of quantitative values ​​for ambient temperature, visibility, precipitation intensity, sea state, and light intensity enables multi-source sensing of various environmental parameters of the dredger and achieves control over multiple environmental dimensions, thereby improving the accuracy of the rate of change in environmental state.

[0058] At this point, the system calls the basic environmental quantization value (ambient temperature quantization value) output in step S112. Visibility quantification value Quantitative value of precipitation intensity Quantitative value of sea state level The illumination feature quantization value (illuminance quantization value) output from step S121 is compared with the illumination feature quantization value. The system maps these five heterogeneous dimensions of parameters to the same high-dimensional environmental parameter matrix, constructing a standardized environmental parameter matrix (t) at the current time t = [ , , , , The vector follows a unified "0-1" quantization logic, where 0 represents the most favorable operating environment and 1 represents the worst extreme environment, thereby eliminating the interference of different physical dimensions on the fusion calculation.

[0059] The system presets a sampling time interval Δt = 0.1s. It calculates the difference in each dimension by comparing the current environmental parameter matrix (t) with the historical vector E(t-Δt) from the previous sampling time. The Euclidean norm is used to calculate the rate of change of the environmental state |ΔE|, with the formula:

[0060]

[0061] The calculation process essentially involves solving for the "distance" between two state points in the environmental parameter matrix to characterize the severity of environmental evolution. If the calculated value of |ΔE| is large, it indicates that the environment has experienced drastic fluctuations in a very short period of time (such as sudden rainstorms or dense fog), and the system needs to trigger a rapid response mechanism.

[0062] Based on the calculated rate of change |ΔE| and the fluctuations of each component, the system semantically labels the changes in environmental state. The system presets the threshold for the rate of change of environmental state to 0.02. If |ΔE| ≥ 0.02, the system determines it as an "environmental abrupt change" and marks the current environment as unstable, requiring the triggering of a fast weight update mode (period T1 = 0.5s). If |ΔE| < 0.02, the system determines it as an "environmental stability" and marks it as a steady state, using a conventional update mode (period T2 = 2s) to save computational resources. At the same time, the system will specifically mark the dominant factors that cause the rate of change to exceed the limit (such as abrupt changes in precipitation intensity) to provide a basis for subsequent sensor performance degradation analysis.

[0063] Specifically, the system extracts the outputs of S112 and S121 to construct the environmental parameter matrix at the current time t; among which, the environmental temperature... =0 (normal temperature 25℃), visibility =0.84 (low visibility 200m), precipitation intensity =0.75 (heavy rain 15mm / h), light intensity =0.48 (low light at night), Sea State Class =0.44 (Sea state 4), i.e., E(t)=[0,0.84,0.75,0.48,0.44].

[0064] The system retrieves historical data from the previous sampling time (Δt = 0.1s ago), E(t - Δt) = [0, 0.82, 0.70, 0.48, 0.42]; calculates the differences in each dimension and substitutes them into the formula to calculate the rate of change of environmental state:

[0065]

[0066] The calculated result of 0.057 is significantly less than the preset threshold of 0.02. The system determines that although the current environment is harsh, the change process is stable and there are no sudden changes. Therefore, the system marks the environmental state change as "stable" and determines the weight update cycle as T2=2s. This shows that although the dredger is in a severe rainstorm, the environmental parameters change gradually. The system does not need to start the high-frequency emergency update mode, thus optimizing the computing power consumption of the edge computing unit while ensuring the perception accuracy.

[0067] refer to Figure 4 In step S13, the specific steps are as follows:

[0068] S131: Based on the detection of the dredger, multiple sensors are identified and the autonomous detection of multiple sensors is triggered. In each sensor, the performance content of each performance dimension is determined according to the autonomous detection of the sensor. Based on the performance content and the current working scenario of the dredger, the performance evaluation index of the sensor is determined to mark the performance evaluation index of each sensor.

[0069] S132: Mark the location of each sensor and perform multi-factor fusion based on various performance evaluation indicators to determine the corresponding data combination. Fuse the data combination and construct the corresponding environmental adaptive combination according to the priority of each performance evaluation indicator during the fusion process.

[0070] In the embodiments of this application, multiple sensors are identified based on the detection of the dredger, and autonomous detection of the multiple sensors is triggered. In each sensor, the performance content of each performance dimension is determined according to the autonomous detection of the sensor. Based on the performance content and the current working scenario of the dredger, the performance evaluation index of the sensor is determined to mark the performance evaluation index of each sensor. This takes into account both the performance content and the current working scenario of the dredger, and ensures the accuracy of the sensor performance evaluation index.

[0071] At this point, based on the hardware topology of the dredger, the system identifies five types of core sensors deployed on the hull: visual sensors, lidar, marine radar, infrared thermal imager, and AIS receiver. The system sends autonomous detection commands to each sensor node, triggering the sensors to actively perceive the current aquatic environment. During this process, each sensor independently collects raw observation data, such as optical images from visual sensors, point cloud data from lidar, and echo signals from marine radar, and provides real-time feedback on its operating status parameters (such as signal strength, signal-to-noise ratio, and data integrity), which serve as the basic data source for performance evaluation.

[0072] After each sensor completes its autonomous detection, the system analyzes and determines the performance of each sensor under the current environment based on the environmental parameter matrix and the sensor's physical characteristics. This process essentially quantifies the degree of attenuation of specific performance dimensions of the sensors by environmental factors. The system analyzes the attenuation of each dimension according to a preset sensor performance mapping model.

[0073] For visual sensors, the system focuses on analyzing the visibility attenuation factor (0.8). Image outline blurring caused by precipitation interference factor (0.6) The raindrop noise introduced and the light factor (0.8) The low-light noise caused by the low light intensity was used to determine the degree of degradation in image clarity and target recognition ability.

[0074] For lidar, the system focuses on calculating the visibility scattering factor (0.9). The beam attenuation caused by ) and the precipitation blocking factor (0.5) The decrease in point cloud density caused by this can be used to determine the extent of the decline in ranging accuracy and contour building capability.

[0075] For maritime radar, the system assesses its resistance to rain and fog interference, with a focus on analyzing heavy precipitation (0.2). ) and giant waves (0.2 The weak impact of precipitation on signal stability was investigated to determine its robustness advantage in harsh environments; for infrared thermal imagers, their low-light adaptability was systematically analyzed, taking into account precipitation (0.3). The impact of temperature gradient on detection performance was investigated to determine its effectiveness at night and in rainy / foggy conditions. For AIS, the system only needs to consider extreme temperatures (10.05°C). ) and raging waves (0.1) It has a very slight impact on signal reception.

[0076] Based on the performance characteristics defined above, the system introduces a sensor performance mapping model to calculate the performance evaluation index P for each sensor. i (Value range 0-1); This index is calculated by coupling various environmental attenuation factors, P i =1 represents optimal performance, P i =0 represents complete failure; specifically, the system directly reveals the coupled calculation logic of each environmental attenuation factor through explicit mathematical analytical formulas. For the performance evaluation index P1 of the visual sensor: Substituting into the formula:

[0077] ;

[0078] For the performance evaluation index P2 of the lidar sensor: Substitute into the formula:

[0079] ;

[0080] For the performance evaluation index P3 of marine radar sensors: Substitute into the formula:

[0081] ;

[0082] For the performance evaluation index P4 of the infrared thermal imager: Substitute into the formula:

[0083] ;

[0084] For AIS performance evaluation metric P5: Substitute into the formula:

[0085] ;

[0086] The system adopts a multiplicative decay model to map environmental quantification values ​​into performance evaluation indicators. After the calculation is completed, the system standardizes and marks the performance evaluation indicators of each sensor, and outputs a performance vector P=[P1,P2,P3,P4,P5] containing the real-time performance status of each sensor, providing a quantitative basis for subsequent confidence decay and weight allocation.

[0087] Specifically, when a dredging vessel is conducting dredging operations at sea, it encounters heavy rain, severe sea conditions (Sea State 4), and a low-light environment at night. The environmental parameter matrix has been determined as E=[0,0.84,0.75,0.48,0.44]. The system marks the five types of sensors on board: vision, lidar, marine radar, thermal imager, and AIS, triggering autonomous detection by all crew members. The visual feedback image is dark and obscured by raindrops, the lidar feedback shows sparse point clouds, while the marine radar and AIS signals remain relatively stable.

[0088] Vision: Due to the severe impact of high visibility quantification (0.84) and high precipitation quantification (0.75), the system determines that its "target recognition" and "image clarity" performance dimensions have been severely degraded, and the performance content is marked as "severely hindered"; LiDAR: Due to the severe impact of rain and fog scattering, the visibility and precipitation factors have a significant effect, and the system determines that its "range accuracy" and "point cloud density" performance dimensions have dropped significantly.

[0089] Marine radar: Due to its physical characteristics, it is not sensitive to rain and fog. The system determines that its "signal stability" is minimally affected by the environment, and its performance is marked as "robust". Infrared thermal imager: It is less affected by low light, but it is affected by precipitation and cooling. The system determines that its "temperature difference detection" performance is somewhat reduced. AIS: It is almost unaffected by weather. The system determines that its performance remains "excellent".

[0090] The system calls the performance mapping formula for real-time calculation: Performance evaluation index P1 of the visual sensor: Substitute into the formula:

[0091] ;

[0092] at the same time, =0; It is 0.84; It is 0.75; It is 0.48; The value is 0.44; therefore, P1=(1-0)×(1-0.8×0.84)×(1-0.6×0.75)×(1-0.8×0.48)×(1-0.4×0.44)≈0.091; the system marks it as a low performance indicator, indicating that the reliability of the visual data is extremely low at this time;

[0093] Performance evaluation index P2 of lidar sensor: Substitute into formula:

[0094] ;

[0095] at the same time, =0; It is 0.84; It is 0.75; It is 0.44;

[0096] ;

[0097] The calculated value is P2≈0.133; the system marks this as a low performance indicator.

[0098] Performance evaluation index P3 for marine radar sensors: Substitute into the formula:

[0099] ;

[0100] at the same time, =0; It is 0.84; It is 0.75; It is 0.44;

[0101] P3 = (1-0) × (1-0.1 × 0.84) × (1-0.2 × 0.75) × (1-0.2 × 0.44) ≈ 0.701; the system marks it as a high-performance indicator and confirms it as the current core sensing method; Infrared thermal imager (P4): calculated to be P4 ≈ 0.503;

[0102] For the performance evaluation index P4 of the infrared thermal imager: Substitute into the formula:

[0103] ;

[0104] at the same time, =0; It is 0.84; It is 0.75; It is 0.48; It is 0.44;

[0105] ;

[0106] The system marks it as a medium performance indicator; AIS performance evaluation indicator P5: calculated P5≈0.956; the system marks it as an extremely high performance indicator; for AIS (P5): substituting into the formula:

[0107] ;

[0108] at the same time, =0; It is 0.44; .

[0109] The system successfully generated the sensor performance evaluation vector P=[0.091,0.133,0.701,0.503,0.956] for the dredger during a rainstorm night. This result accurately quantifies the differentiated impact of harsh environments on different sensors, confirming that marine radar and AIS should be the primary trusted data sources in this scenario, while visual and lidar data need to be significantly downweighted.

[0110] Furthermore, the positions of each sensor are marked, and multi-factor fusion is performed in combination with various performance evaluation indicators to determine the corresponding data combination. This data combination is then fused, and during the fusion process, a corresponding environmental adaptive combination is constructed based on the priority of each performance evaluation indicator. This approach takes into account the overall priority of each performance evaluation indicator and ensures the accuracy of the corresponding environmental adaptive combination.

[0111] At this point, based on the sensor topology of the dredger, the system precisely marks the spatial installation location of each sensor, such as marking the visual sensor as being located forward of the compass deck and the marine radar as being located on top of the bridge, to determine the detection field of view and blind zone coverage of each sensor. The system introduces a confidence decay function (improved Sigmoid function) to map the sensor performance evaluation index Pi calculated in step S131 to the data confidence level C. i This mapping process is not a simple linear transformation, but rather based on the attenuation rate coefficients of each sensor. k i With performance threshold P thi Perform nonlinear mapping;

[0112] Specifically, the "improved Sigmoid function" is used as the confidence decay function for nonlinear mapping. The specific formula for the improved Sigmoid function is as follows:

[0113]

[0114] This formula is a deterministic nonlinear mapping model in the field of mathematics, and its mapping logic is: when the sensor performance evaluation value P i Above the threshold P thi At that time, confidence level C i Smoothness approaches 1 (high confidence); when P i Below the threshold P thi At that time, confidence level C i It exhibits an exponential and rapid decay, approaching 0 (low confidence level). k i This is the decay rate coefficient.

[0115] For example, for vision with extremely low performance evaluation values, its confidence level will rapidly decay to near 0; while for AIS with robust performance, its confidence level will remain at a high level close to 1. This process aims to remove sensor data weights with severely degraded performance, providing a reliable foundation for subsequent fusion.

[0116] In obtaining confidence level C i Based on this, the system introduces a second weighting factor, the inherent reliability coefficient βi of the sensor. This coefficient reflects the technical advantages of the sensor hardware itself and its adaptability to the dredging vessel's operating scenario. The reliability coefficient βi is a preset value, reflecting the technical advantages and operational adaptability of the sensor hardware. This means that βi is not a dynamically calculated value during operation, but a pre-calibrated static weight based on the adaptability of the sensor's physical characteristics to the specific operating scenario of the dredging vessel. In this case, the low reliability range (βi < 1.0) has values ​​such as β1 = 0.8 (visual); the baseline reliability range (βi = 1.0) has values ​​such as β2 = 1.0 (LiDAR). This is usually used as the baseline for system evaluation. The high reliability range (βi > 1.0) has values ​​such as β3 = 1.2 (maritime radar), β4 = 1.1 (thermal imager), and β5 = 1.3 (AIS).

[0117] For example, marine radar is set to β3=1.2 due to its strong anti-interference capability; AIS is set to β5=1.3 due to its extremely strong data stability; and vision is set to β1=0.8 due to its susceptibility to environmental interference. The system calculates the basic weight Wi′=Ci×βi and normalizes it to ensure that the sum of the weights of all sensors is 1. Thus, the system determines the optimal data combination under the current environment and the initial contribution ratio of each data source, forming a weighted data set.

[0118] The system constructs an "environmentally adaptive combination" based on normalized weights. To prevent sudden weight changes (i.e., "step switching") caused by minor environmental fluctuations, the system introduces a first-order low-pass filter algorithm as a smoothing constraint mechanism. The calculation formula is: Wi′(t) = α × Wi(t) + (1-α) × Wi(t-1), where the smoothing coefficient α is set to 0.3. Wi′(t) is the weight actually output after smoothing at the current time. Wi(t) is the weight actually output after smoothing at the current time. Wi(t-1) is the fusion weight actually used by the sensor in the previous update cycle (i.e., the previous sampling time). This mechanism limits the weight change rate to ≤10% / s to ensure a smooth and continuous transition of the fusion weights. The system output includes the "environmentally adaptive combination" containing the real-time fusion weights of each sensor and the corresponding data priorities. This combination directly determines the distribution of observation noise variance of each data source in the subsequent Kalman filter fusion algorithm.

[0119] Specifically, the dredger was conducting dredging operations at sea when it encountered heavy rain, severe sea conditions (Sea State 4), and a low-light environment at night. The sensor performance evaluation index was determined to be: P = [0.091, 0.133, 0.701, 0.503, 0.956]. The system identifies the marine radar located on top of the bridge and the AIS on the stern mast as the current core detection sources. For the visual sensor, the performance evaluation index P1 = 0.091 is far below its performance threshold P. th1 =0.3, the system calls the improved Sigmoid function to calculate a confidence level C1≈0.109, indicating that its data is almost unreliable; conversely, AIS's P5=0.956 is far higher than the performance threshold P th5 =0.1, and the calculated confidence level C5≈0.994. At this point, the performance threshold P of the visual sensor is... th1 =0.3, AIS performance threshold P th5 =0.1.

[0120] The specific formula for the improved Sigmoid function is as follows:

[0121]

[0122] This formula is a deterministic nonlinear mapping model in the field of mathematics, and its mapping logic is: when the sensor performance evaluation value P i Above the threshold P thi At that time, confidence level C i Smoothness approaches 1 (high confidence); when P i Below the threshold P thi At that time, confidence level Ci It exhibits an exponential and rapid decay, approaching 0 (low confidence level). k i This is the decay rate coefficient.

[0123] The system calculates the basic weights based on the inherent reliability coefficients. Maritime radar, due to its strong anti-interference capability (β3=1.2) and current good performance, has a significantly increased basic weight. In contrast, vision (β1=0.8), due to its extremely low performance and generally poor inherent reliability, has its basic weight greatly compressed. After normalization, the system determines the optimal data combination weight allocation as follows: AIS accounts for approximately 35.9%, maritime radar accounts for approximately 32.8%, thermal imager accounts for approximately 26.6%, while vision and lidar combined account for only approximately 4.7%.

[0124] Assuming the visual weight was 5% in the previous moment, the current calculated weight drops sharply to 2.4%. To prevent sudden changes in perception results, the system activates a smoothing constraint mechanism. Through first-order low-pass filtering, the visual weight at the current moment is adjusted to W1(t) = 0.3 × 0.024 + 0.7 × 0.05 ≈ 4.2%. The system constructs an "environmentally adaptive combination" adapted to the rainy night environment. This combination assigns the highest data priority to AIS and marine radar, ensuring that the dredger can still output stable and reliable environmental perception results in harsh environments, avoiding the risk of false alarms caused by sudden changes in visual data.

[0125] refer to Figure 5 In step S14, the specific steps are as follows:

[0126] S141: Collect a preset environmental state change rate threshold, compare the preset environmental state change rate threshold with the environmental state change rate, if the environmental state change rate is higher than the preset environmental state change rate threshold, then present a mismatch event between the environmental adaptive combination and the current scene of the dredger, and trigger real-time updates of the environmental adaptive combination based on the mismatch event.

[0127] S142: When the dredger encounters environmental changes in the current scene, the changed environment in which the dredger is located is marked, the changed environment in which the dredger is located is fused with the environment adaptive combination, and the final environmental sensing result is determined in the fusion process.

[0128] In the embodiments of this application, a preset environmental state change rate threshold is collected, and the preset environmental state change rate threshold and the environmental state change rate are compared. If the environmental state change rate is higher than the preset environmental state change rate threshold, a mismatch event between the environmental adaptive combination and the current scene of the dredger is presented, and the real-time update of the environmental adaptive combination is triggered based on the mismatch event. The priority of each environmental detection part in the environmental adaptive combination is introduced to change with the current scene of the dredger.

[0129] At this point, the system retrieves the preset environmental state change rate threshold (set to |ΔE|=0.02) from the software parameter configuration. This threshold is the critical threshold for determining whether the environment is in a "gradual" or "abrupt" state. The system compares the real-time environmental state change rate |ΔE| calculated in step S122 with this threshold in real time. When the real-time change rate is lower than the preset threshold (|ΔE|<0.02), the system determines that the environment is in a relatively stable gradual change state. Conversely, if it is higher than the threshold, it is determined to be an environmental abrupt change.

[0130] Although an environmental change rate below the threshold indicates that no catastrophic environmental change has occurred, continuous minor fluctuations in environmental parameters (such as a slight increase in rainfall or a slight decrease in visibility) can still cause slight deviations between the sensor weight allocation in the current "environmentally adaptive combination" and the real-time scene of the dredger. This deviation is defined as a "mismatch event." The system identifies such events with decreased matching degree by comparing the theoretical weights calculated at the current moment with the actual weights at the previous moment. Once a mismatch event is detected, the system triggers a real-time update mechanism. Under stable environmental conditions, in order to ensure efficient use of computing resources and continuity of perception results, the system sets the weight update cycle to a longer period (such as T2=2 seconds) rather than the fast update cycle (T1=0.5 seconds) during non-abrupt changes.

[0131] After triggering a real-time update, the system recalculates the confidence and weight of each sensor based on subtle changes in the environmental state, thereby adjusting the priority of each detection component in the environmental adaptive combination. This process follows the principle of "weight decay when the environment deteriorates and weight recovery when the environment improves." The system uses a first-order low-pass filtering algorithm to smoothly correct the weights, ensuring that the priority adjustment process is stable and without jumps. This allows the fusion system to always maintain a high degree of adaptation to the current operating scenario of the dredger (such as reduced rain and changes in wave height), ensuring the continuous reliability of the sensed data.

[0132] Furthermore, when the dredger encounters environmental changes in the current scenario, the changed environment of the dredger is marked, and the changed environment of the dredger is fused with the environmental adaptive combination. During the fusion process, the final environmental sensing result is determined. This process takes into account the overall considerations in the fusion process, ensuring the accuracy of the final environmental sensing result. At the same time, environmental changes are controlled, and data fusion of various performance evaluation indicators is achieved, which improves the accuracy of the environmental adaptive combination and ensures the accuracy of the final environmental sensing result.

[0133] At this time, the system monitors the dynamic environment in which the dredger is located in real time, capturing changes in environmental conditions. When environmental parameters fluctuate (such as a slight increase in rainfall intensity or a larger wave), the system compares the real-time environmental parameter matrix at the current moment with that at the previous moment, accurately marking the changing environmental characteristics of the dredger. For example, if quantified values ​​of visibility are available... The system marks the current scene as “visibility continuously deteriorating scene” when the value increases from 0.82 to 0.84. This marking process clarifies the trend of the current environment’s impact on sensor performance and provides a basis for noise variance allocation in the subsequent fusion process.

[0134] The system performs deep fusion of the labeled changing environmental parameters with the "environmentally adaptive combination" output from step S141. This process is not a simple data superposition, but rather employs a weighted Kalman filter algorithm to map the real-time fusion weights Wi of each sensor in the environmentally adaptive combination to the inverse of the observation noise variance. In the fusion calculation, sensors with higher weights (such as AIS and marine radar) have a higher covariance matrix P of their observation data. i Assigning smaller values ​​to sensors means that the system has a higher "trust" in its observations; while sensors with low weights (such as vision and lidar) are assigned larger values ​​to their covariance matrices, thereby reducing their interference with the final result in the state update equation.

[0135] After weighted fusion calculation, the system outputs the final environmental sensing result. This result includes the target state vector, covering key information such as target orientation, distance, category, and speed, and calculates the collision risk level in combination with the collision risk model. Due to the introduction of a smoothing constraint mechanism, the final output sensing result is continuous and smooth in time series, effectively avoiding abrupt changes in the sensing result caused by environmental fluctuations, and providing accurate and reliable data support for the intelligent operation decision-making system of dredgers.

[0136] Please see Figure 6 , Figure 6 This is a schematic diagram of the structural composition of the environmentally adaptive dredger data fusion system in an embodiment of the present invention; the environmentally adaptive dredger data fusion system is applied to the above-mentioned environmentally adaptive dredger data fusion method; the environmentally adaptive dredger data fusion system includes:

[0137] The quantization module 21 is used to mark the current position of the dredger and trigger multi-source environmental sensing based on the current position of the dredger to collect raw data of precipitation intensity, visibility, temperature and sea state, and determine the corresponding environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value and sea state level quantization value based on the mapping of multiple raw data.

[0138] The environmental change module 22 is used to determine the corresponding light intensity quantization value based on the recognition of images collected by the camera of the dredger, determine the environmental state change rate based on the environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value, sea state level quantization value and light intensity quantization value, and mark the corresponding environmental state change content.

[0139] The environment adaptive module 23 is used to mark multiple sensors of the dredger and determine the corresponding performance content based on the detection of each sensor. Based on the performance content and the current working scenario of the dredger, the module determines the performance evaluation index of the sensor, fuses the data of each performance evaluation index, and outputs the environment adaptive combination.

[0140] The environmental perception result module 24 is used to trigger real-time updates of the environmental adaptive combination if the rate of change of the environmental state is higher than the preset environmental state change rate threshold, and adjust the priority of each environmental detection part in the environmental adaptive combination to adapt to the changing environment in which the dredger is located, and output the final environmental perception result.

[0141] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A data fusion method for dredgers based on environment adaptation, characterized in that, include: The current location of the dredger is marked, and environmental multi-source sensing is triggered based on the current location of the dredger to collect raw data on precipitation intensity, visibility, temperature and sea state. Based on the mapping of multiple raw data, the corresponding environmental temperature quantification value, visibility quantification value, precipitation intensity quantification value and sea state level quantification value are determined. The corresponding quantized value of light intensity is determined by recognizing the images collected by the camera of the dredger. The rate of change of environmental state is determined based on the quantized values ​​of ambient temperature, visibility, precipitation intensity, sea state level and light intensity, and the corresponding environmental state change content is marked. This process involves labeling multiple sensors on a dredger and determining corresponding performance metrics based on the detection data of each sensor. Based on these performance metrics and the current operating environment of the dredger, performance evaluation metrics for each sensor are determined. These performance evaluation metrics are then fused together, and an environment-adaptive combination is output. This includes: identifying multiple sensors based on the dredger's detection and triggering autonomous detection by these sensors; determining the performance content of each performance dimension for each sensor based on its autonomous detection; determining the performance evaluation metrics for each sensor based on this performance content and the current operating environment of the dredger; labeling the performance evaluation metrics of each sensor; and marking the location of each sensor and performing multi-factor fusion based on the various performance evaluation metrics to determine the optimal performance evaluation metrics for each sensor. The system combines the corresponding data and fuses them, constructing an environmentally adaptive combination based on the priority of each performance evaluation index during the fusion process. After each sensor completes autonomous detection, the system analyzes and determines the performance of each sensor in the current environment based on the environmental parameter matrix and the sensor's physical characteristics. This process essentially quantifies the degree of attenuation of specific performance dimensions of the sensors by environmental factors. The system analyzes the attenuation of each dimension according to a preset sensor performance mapping model. The quantified environmental values ​​are mapped to performance evaluation indices. The system standardizes and labels the performance evaluation indices of each sensor, outputting a performance vector containing the real-time performance status of each sensor. The performance vector includes each performance evaluation index. The performance evaluation indicators of each sensor are mapped to data confidence levels. This mapping process is not a simple linear transformation, but a nonlinear mapping based on the attenuation rate coefficient and performance threshold of each sensor. If the rate of change of the environmental state exceeds a preset threshold, the real-time update of the environmental adaptive combination is triggered, and the priority of each environmental detection component in the environmental adaptive combination is adjusted to adapt to the changing environment of the dredger. The final environmental sensing result is then output, including: collecting the preset threshold for the rate of change of the environmental state; comparing the preset threshold for the rate of change of the environmental state with the rate of change of the environmental state; if the rate of change of the environmental state exceeds the preset threshold for the rate of change of the environmental state, a mismatch event between the environmental adaptive combination and the current scene of the dredger is presented, and the real-time update of the environmental adaptive combination is triggered based on the mismatch event. The system uses a first-order low-pass filtering algorithm to smoothly correct the weights, ensuring that the priority adjustment process is smooth and without jumps, so that the fusion system always maintains a high degree of adaptation to the current working scene of the dredger, and ensures the continuity and reliability of the sensing data. When a dredger encounters environmental changes in the current scenario, the system marks the changed environment in which the dredger is located. This changed environment is then fused with an adaptive environment combination, and the final environmental sensing result is determined during the fusion process. The real-time fusion weights of each sensor in the adaptive environment combination are mapped to the reciprocal of the observation noise variance. In the fusion calculation, sensors with higher weights have smaller covariance matrices, indicating a higher level of trust in their observations. Conversely, sensors with lower weights have larger covariance matrices, thus reducing their interference with the final result in the state update equation. After weighted fusion calculation, the system outputs the final environmental sensing result. This result includes a target state vector, covering key information such as target orientation, distance, category, and velocity, and incorporates a collision risk model to calculate the collision risk level. Due to the introduction of a smoothing constraint mechanism, the final sensing result is continuous and smooth in time series.

2. The data fusion method for dredgers based on environmental adaptation according to claim 1, characterized in that, The system marks the current position of the dredger and triggers multi-source environmental sensing based on this position to collect raw data on precipitation intensity, visibility, air temperature, and sea state. It then determines corresponding quantified values ​​for environmental temperature, visibility, precipitation intensity, and sea state based on the mapping of multiple raw data sets, including: The dredger is equipped with an onboard meteorological instrument. The environmental detection of the onboard meteorological instrument is triggered according to the current position of the dredger and the scene in which the dredger is located. The instrument performs multi-source environmental sensing for the dredger to monitor and collect raw data on precipitation intensity, visibility, temperature and sea state in real time.

3. The data fusion method for dredgers based on environmental adaptation according to claim 2, characterized in that, The system marks the current position of the dredger and triggers multi-source environmental sensing based on this position to collect raw data on precipitation intensity, visibility, air temperature, and sea state. It also determines corresponding quantified values ​​for environmental temperature, visibility, precipitation intensity, and sea state based on the mapping of multiple raw data sets. The system further includes: The raw data are quantified and converted into a unified, standardized environmental parameter matrix to determine the corresponding quantified values ​​for ambient temperature, visibility, precipitation intensity, and sea state. For ambient temperature, the air temperature data is mapped to a 0-1 range, where 0 represents no temperature effect and 1 represents extreme temperature saturation. For visibility, the visibility data is mapped to a 0-1 range, where 0 represents excellent visibility and 1 represents extremely poor visibility. For precipitation intensity, the precipitation intensity data is mapped to a 0-1 range, where 0 represents no precipitation and 1 represents extremely heavy precipitation.

4. The data fusion method for dredgers based on environmental adaptation according to claim 1, characterized in that, The quantized value of light intensity is determined by recognizing images captured by the camera on the dredger. The rate of change of environmental state is determined based on the quantized values ​​of ambient temperature, visibility, precipitation intensity, sea state, and light intensity, and the corresponding environmental state changes are marked, including: The camera on the dredger is marked, and the surrounding area of ​​the dredger is photographed based on the camera to collect corresponding images. Image recognition is performed on the images, and multiple light intensity elements are determined during the recognition process. The corresponding light intensity quantization value is determined based on the quantization of multiple light intensity elements.

5. The data fusion method for dredgers based on environmental adaptation according to claim 4, characterized in that, The method involves identifying the corresponding quantized value of light intensity based on the images captured by the dredger's camera, determining the rate of change of environmental state based on the quantized values ​​of ambient temperature, visibility, precipitation intensity, sea state, and light intensity, and marking the corresponding environmental state changes. This also includes: The system acquires quantitative values ​​for ambient temperature, visibility, precipitation intensity, sea state, and light intensity. These values ​​are then input into the same environmental parameter matrix, and the system iterates within the matrix to determine the corresponding rate of change in the environmental state and presents the corresponding environmental state changes.

6. A data fusion system for an environment-adaptive dredger, characterized in that, The environmentally adaptive dredger data fusion system is applied to the environmentally adaptive dredger data fusion method as described in any one of claims 1-5; The environmentally adaptive dredger data fusion system includes: The quantization module is used to mark the current position of the dredger and trigger multi-source environmental sensing based on the current position of the dredger to collect raw data on precipitation intensity, visibility, temperature and sea state. Based on the mapping of multiple raw data, the corresponding environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value and sea state level quantization value are determined. The environmental change module is used to determine the corresponding light intensity quantization value based on the recognition of images collected by the dredger's camera, determine the environmental state change rate based on the environmental temperature quantization value, visibility quantization value, precipitation intensity quantization value, sea state level quantization value and light intensity quantization value, and mark the corresponding environmental state change content. The environment adaptive module is used to label multiple sensors of the dredger and determine the corresponding performance content based on the detection of each sensor. Based on the performance content and the current working scenario of the dredger, the module determines the performance evaluation index of the sensor, fuses the data of each performance evaluation index, and outputs the environment adaptive combination. The environmental perception result module is used to trigger real-time updates of the environmental adaptive combination if the rate of change of the environmental state is higher than the preset environmental state change rate threshold, and to adjust the priority of each environmental detection part in the environmental adaptive combination to adapt to the changing environment in which the dredger is located, and output the final environmental perception result.

Images