An oil spill detection analysis method and system based on multi-source data fusion

Abstract

The application discloses a kind of based on multi-source data fusion's oil spill detection analysis method and system, it is related to oil spill monitoring technology.The method includes data acquisition step, real-time acquisition video image and external environment data to obtain the environmental characteristics of current time;Laser irradiation parameter acquisition step, based on the environmental parameter mapping relationship of pre-experiment calibration, laser irradiation parameter is obtained using environmental characteristics;Execute laser irradiation step, according to laser irradiation parameter control array distribution laser emitter executes irradiation, collects fluorescence spectrum and judges whether there is oil film to extract detection oil film characteristics;Confidence calculation step, according to single point confidence and spatial consistency confidence, obtain final confidence;Oil spill result output step, according to final confidence and detection oil film characteristics, calculate oil spill risk value and compare with preset alarm threshold to trigger alarm.The application realizes high sensitivity, high reliability offshore oil spill real-time monitoring and automatic alarm.

Description

Technical Field

[0001] This invention relates to the field of oil spill monitoring technology, specifically to an oil spill detection and analysis method and system based on multi-source data fusion. Background Technology

[0002] Currently, marine oil spill monitoring mainly relies on radar remote sensing, manual inspection, and single optical detection methods, but these methods all have obvious shortcomings in practical applications.

[0003] Radar remote sensing, such as synthetic aperture radar (SAR), can achieve large-scale, all-weather monitoring, but its ability to identify thin oil films is insufficient, especially those with a thickness of micrometers. Furthermore, it is easily affected by biofilms on the sea surface, low-wind-speed areas, and other similar objects, resulting in a high false alarm rate.

[0004] Manual inspection relies on repeated checks for oil spills, which is inefficient, untimely, and unsuitable for responding to sudden large-scale oil spills. Alternatively, video surveillance systems can provide intuitive images for observation and recording, but image acquisition is easily affected by lighting and weather conditions, and they lack quantitative detection capabilities, such as obtaining oil film thickness or identifying oil type, thus failing to achieve automated early warning.

[0005] In contrast, oil spill detectors based on multi-source data fusion utilize the natural property of petroleum substances, especially polycyclic aromatic hydrocarbons (PAHs), to emit characteristic fluorescence under ultraviolet light excitation. This offers advantages such as non-contact operation, high sensitivity, and fast response. The principle of an oil spill detector is as follows: the sensor emits a specific wavelength of ultraviolet laser light, such as 355nm or 405nm, towards the water surface. When the laser irradiates the oil film, the oil molecules absorb energy and transition to an excited state, subsequently emitting a longer wavelength fluorescence signal, such as 400-600nm. This fluorescence signal is then received by a photomultiplier tube or CCD. Simultaneously, the water body produces Raman scattering light, which can be simply referred to as water Raman scattering. By analyzing the fluorescence intensity, spectral characteristics, and Raman scattering spectral information, micron-level oil film thickness identification, oil type differentiation, and real-time monitoring with second-level response can be achieved.

[0006] For example, patent CN110160452A discloses a method for measuring the thickness of oil film on the water surface based on laser Raman and laser fluorescence. This patent proposes a formula for calculating oil film thickness. By measuring the Raman signal intensity of the water surface at the same distance from the detection system (ignoring the fluorescence background) at the oil-free location and the fluorescence signal intensity at a wavelength of λ at the oil-covered location, the oil film thickness at different distances can be obtained using the oil film thickness calculation formula. Patent CN111141684A discloses a method for detecting the visible / infrared polarization characteristics of marine oil spills. This patent proposes a method combining intensity detection and polarization detection to compare and analyze the visible / infrared polarization characteristics of common marine oil spills, thereby distinguishing the types of oil spilled.

[0007] However, the real-world marine environment is extremely complex, with factors such as surface waves, ocean currents, strong daytime sunlight, water turbidity, and the presence of phytoplankton that can generate fluorescent signals. These factors pose significant challenges to current laser-induced fluorescence technology. Specifically: surface waves cause dynamic changes in the laser's incident angle, spot position, and oil film thickness distribution, resulting in strong fluctuations in the fluorescence signal; ocean currents cause the oil film to move with the water flow, leading to a higher probability of missed detections with current fixed-point detection methods; strong background light from solar radiation can overwhelm weak fluorescent signals and even saturate the detector; high water turbidity reduces laser penetration, attenuating the reach of the laser to the oil film and the reception of the fluorescent signal; and the fluorescence spectral characteristics of phytoplankton, which also produce fluorescence, can interfere with the spectral characteristics during oil film detection.

[0008] Therefore, there is an urgent need to develop an oil spill detection method that can integrate multi-source information and has anti-interference capabilities. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides an oil spill detection and analysis method and system based on multi-source data fusion. The aim is to construct an intelligent oil spill detection technology solution that can adapt to complex marine environments and possess point-by-point reliability assessment and spatial consistency verification capabilities by fusing video image information and laser-induced fluorescence spectral information. This overcomes the technical defects of existing technologies, such as low detection accuracy, high false alarm rate, difficulty in identifying thin oil films, and lack of quantitative confidence assessment mechanisms, caused by factors such as surface waves, ocean current drift, strong light, water turbidity, and fluorescent biological interference. Thus, it achieves high-sensitivity, high-reliability, and robust real-time monitoring and automatic alarm of marine oil spills.

[0010] This invention provides an oil spill detection and analysis method based on multi-source data fusion, the method comprising the following steps: The data acquisition step involves real-time acquisition of video images and external environmental data to obtain the environmental characteristics at the current moment. This step, by collecting environmental features and video images at the current moment, provides a data foundation for subsequent adaptive adjustment of laser parameters and elimination of environmental interference; The laser irradiation parameter acquisition step involves acquiring the laser irradiation parameters using the environmental characteristics based on a preset environmental parameter mapping relationship, wherein the environmental parameter mapping relationship is based on a pre-tested calibration. By obtaining the correlation between environmental characteristics and laser parameters through pre-calibration experiments, the complex online optimization problem is transformed into a table lookup and matching problem, reducing computational complexity, ensuring real-time performance, and utilizing prior calibration knowledge to ensure the scientific validity and effectiveness of parameter adjustment. The laser irradiation step is performed by controlling the array-distributed laser emitters to perform irradiation according to the laser irradiation parameters, collecting the fluorescence spectrum of each irradiation point, and determining whether an oil film exists based on the fluorescence spectrum. If an oil film exists, the characteristics of the oil film are extracted and detected. An array of independently controllable lasers enables multi-point irradiation and information acquisition, increasing the monitoring range. Based on the laser irradiation parameters, laser irradiation is performed to monitor oil spills. The presence of an oil film is initially determined by fluorescence spectroscopy as a preliminary screening condition for subsequent confidence calculations, reducing invalid calculations. The confidence level calculation steps involve calculating and obtaining the final confidence level based on the single-point confidence level and the spatial consistency confidence level. The single-point confidence level reflects the confidence level of the oil film characteristics collected at the corresponding irradiation point, and the spatial consistency confidence level reflects the consistency of the corresponding irradiation point across the distribution of all irradiation points. The confidence level of each irradiation point is obtained by analyzing the measurement reliability of each irradiation point and the consistency of the point in spatial distribution. This constructs a reliability assessment from point to area, thereby improving the accuracy of oil spill risk assessment. The oil spill result output step calculates the oil spill risk value of the irradiated point based on the final confidence level and the corresponding oil film characteristics, and compares the oil spill risk value with the preset oil spill alarm threshold. If the oil spill alarm threshold is exceeded, an oil spill alarm is triggered. By using confidence level as a reliability correction factor for detecting oil film characteristics at each irradiation point, a risk quantification mechanism is achieved through comprehensive analysis of the original criteria and reliability. This effectively suppresses false alarms caused by environmental interference such as waves and light while maintaining sensitivity to the determination of the true oil film, thereby improving alarm accuracy.

[0011] Furthermore, the strategy for obtaining the environmental parameter mapping relationship includes: By generating waves with various parameters using a wave generator, different water surface ripple amplitudes and water surface drift speeds are simulated. Combined with different light intensities and water turbidity, multiple sets of environmental feature combinations are constructed. For each combination of environmental features, different combinations of values ​​for irradiation density, irradiation angle and irradiation power in the laser irradiation parameters are selected and oil film detection is performed. The laser irradiation parameters corresponding to the highest oil film detection accuracy or the condition that meets the preset accuracy threshold are recorded as the optimal laser irradiation parameters for the current combination of environmental features. All combinations of environmental features and their corresponding optimal laser irradiation parameters are stored as the environmental parameter mapping relationship.

[0012] By replacing repeated trial and error in the field with controlled experimental calibration, the difficulty and cost of system deployment and debugging are reduced. The optimal laser irradiation parameters corresponding to different environmental characteristics are obtained under controlled experiments, and the mapping relationship covers the combination space of multi-dimensional environmental factors such as light, waves, and turbidity, ensuring the adaptability of the oil spill detection method under various sea conditions.

[0013] Furthermore, the single-point confidence level is used to characterize the reliability of the oil film detection data collected at the irradiation point under the current local ripple state and current laser irradiation conditions. The single-point confidence level analysis sub-strategy includes: Obtain the local water surface area corresponding to each illumination point in the video image; Local ripple features of the irradiation point are extracted based on the local water surface area; The measurement reliability of the irradiation point is determined based on the local ripple characteristics and the laser irradiation parameters corresponding to the irradiation point. A single-point confidence level for the irradiation point is generated based on the reliability of the measurement.

[0014] By performing local analysis on each irradiation point using video image information, the ripple characteristics in each local area and the laser parameters of the corresponding point are analyzed as a reliability assessment framework for that irradiation point. This takes into account the dual effects of environmental interference and system settings, making the reliability judgment more accurate.

[0015] Furthermore, the local ripple characteristics include at least one of the following: local ripple amplitude, local ripple change frequency, local water surface drift velocity, and local reflective disturbance intensity. The step of determining the measurement reliability of the irradiation point based on the local ripple characteristics and the laser irradiation parameters corresponding to the irradiation point includes: The ripple stability of the irradiation point is determined based on the local ripple characteristics. Based on the matching between at least one of the actual irradiation angle, actual irradiation power and actual irradiation density of the irradiation point and at least one of the recommended irradiation angle, irradiation power and irradiation density under the current local ripple state, the degree of adaptation of the irradiation parameters of the irradiation point is determined. The measurement reliability of the irradiation point is determined based on the stability of the ripples and the degree of adaptation of the irradiation parameters.

[0016] The reliability of the measurement at each irradiation point is determined by a combination of two dimensions: the stability of the local ripples and the degree of adaptation of the irradiation parameters. This provides a quantifiable evaluation dimension and specific judgment logic for the reliability of the detection results at each irradiation point.

[0017] Furthermore, the step of obtaining the local water surface area corresponding to each irradiation point in the video image specifically involves: using a dedicated ultraviolet band camera to acquire a first image containing the positions of all irradiation points; using a broadband beam splitter to completely overlap the first image with the video image; and superimposing the positions of the irradiation points onto the video image to determine the local water surface area corresponding to each irradiation point.

[0018] The first image of the laser spot location is acquired using a dedicated ultraviolet-band camera. The first image is then perfectly superimposed on the video image using a broadband beam splitter, achieving precise spatial overlay. This solves the problem of accurate correspondence between the laser spot location and the pixel area of ​​the video image, and avoids complex coordinate transformations and calibration error accumulation. It also provides positional assurance for the extraction of local ripple features in single-point confidence analysis.

[0019] Furthermore, the spatial consistency confidence score is used to characterize the degree of continuous support of the detection results of the corresponding irradiation point across multiple irradiation point distributions. The spatial consistency confidence score analysis sub-strategy includes: Based on the spatial location of each irradiation point in the water surface to be tested, construct the adjacency relationship of the irradiation points; For any target irradiation point, obtain the oil film judgment results and detection oil film characteristics of its adjacent irradiation points; Based on the consistency of the oil film judgment results between the target irradiation point and the adjacent irradiation points, and the similarity of the detected oil film features, the neighborhood support degree of the target irradiation point is determined. The connectivity support level of the target irradiation point is determined based on whether the target irradiation point and the surrounding irradiation points form a continuous oil film detection area. The spatial consistency confidence of the target illumination point is generated based on the neighborhood support level and the connectivity support level.

[0020] Based on the constraint of the spatial continuity of the oil film, the adjacency relationship of the irradiation points is constructed to obtain the oil film judgment results and oil film characteristics of the adjacent points. The degree of neighborhood support is determined according to the consistency of the judgment results and the similarity of features between the target irradiation point and the adjacent irradiation points, and the degree of connectivity support is determined by analyzing whether a continuous oil film region is formed. The spatial consistency confidence of the target irradiation point is obtained by comprehensively analyzing the two dimensions of neighborhood support and connectivity support, which takes into account both local similarity and regional integrity, making the evaluation more comprehensive and reliable.

[0021] Furthermore, the closer the spatial distance between the target irradiation point and the adjacent irradiation point, and the higher the similarity of the detected oil film features, the higher the degree of neighborhood support of the adjacent irradiation point for the target irradiation point. When the target irradiation point and multiple adjacent irradiation points have consistent oil film judgment results and form a continuous distribution, the connectivity support of the target irradiation point is improved; When the target irradiation point is an isolated oil film detection point and no corresponding oil film feature is detected at its adjacent irradiation points, the connectivity support of the target irradiation point is reduced.

[0022] Further defining the specific judgment rules in the spatial consistency confidence analysis process, providing specific quantitative rules for enhancing support when points are closer and decreasing support when points are farther away, and for increasing support when points form a connected region and decreasing support when points are isolated, making spatial confidence assessment more accurate.

[0023] Furthermore, following the confidence calculation step, a verification process is also included: If the confidence level of a single point is lower than the preset single-point confidence threshold, the influence weight of the irradiation point in the oil spill risk value determination process is reduced, or the laser emission module is controlled to repeatedly irradiate and detect the irradiation point. If the spatial consistency confidence level of the irradiated point is lower than the preset spatial consistency threshold, the irradiated point will be marked as an isolated suspected point, and the subsequent detection results of its adjacent irradiated points will be used for verification.

[0024] By constructing a verification processing mechanism, different verification strategies are developed for two low-confidence scenarios: single-point unreliability and spatial inconsistency. These strategies have fault tolerance and verification functions, avoiding misjudgments caused by single measurement deviations, thereby improving the reliability of the final decision.

[0025] Further, determining the oil spill risk value at the irradiation point based on the final confidence level and the corresponding detected oil film characteristics includes: The oil film risk value at the irradiation point is determined based on the detected oil film characteristics. The final confidence level is used as the basis for the reliability correction of the oil film risk value to obtain the oil spill risk value at the irradiation point.

[0026] Using confidence level as a reliability correction factor for oil film risk value makes the original oil film characteristics very strong. If the confidence level is low, the final risk value will also be effectively suppressed, significantly reducing the false alarm rate, and forming a complete closed loop from confidence level assessment and correction to final judgment.

[0027] This invention also proposes a detection system employing the above-mentioned oil spill detection and analysis method based on multi-source data fusion, the system comprising: The laser detection module has multiple laser emitters arranged in an array at its laser emitting end. The switching, power, and illumination angle of each emitter can be independently controlled. The module is used to emit a laser beam of preset wavelength to illuminate the water surface and to receive fluorescence signals to collect fluorescence spectra. The environmental acquisition module is used to acquire real-time environmental data of the detection area. The video acquisition module is used to acquire video image data of the detection area in real time; The data processing module is used to filter, denoise, and extract features from external environmental data, video image data, and fluorescence signals. The oil spill identification and alarm module has a built-in confidence assessment module to calculate the final confidence level, and uses the final confidence level to determine the oil spill risk and issue an alarm.

[0028] The beneficial effects of this invention are: This invention establishes a mapping relationship between environmental feature combinations and optimal laser irradiation parameters through pre-experimental calibration. During real-time detection, it dynamically matches the optimal irradiation density, irradiation angle, and irradiation power based on the environmental characteristics at the current moment. This achieves adaptive adjustment of the laser irradiation strategy to complex sea conditions, effectively avoiding the problem of signal quality degradation or even failure when environmental factors such as waves, illumination, and turbidity change in traditional fixed-parameter detection methods. This allows the detection system to maintain optimal excitation conditions under different sea conditions, significantly improving the stability of fluorescence signals and the sensitivity of oil film detection. At the same time, the array-type independently controllable laser emission module can be controlled to perform optimal irradiation according to the laser irradiation parameters.

[0029] This invention introduces a confidence calculation step, quantitatively evaluating the reliability of the detection results at each irradiation point from two dimensions: single-point and spatial distribution. At the single-point dimension, it analyzes local ripple characteristics such as ripple amplitude, ripple change frequency, water surface drift velocity, and reflective disturbance intensity in the corresponding local water surface area. This, combined with the matching degree between the actual laser irradiation parameters at that point and the recommended parameters under the current local ripple state, comprehensively determines the credibility of the oil film detection data collected at that point. At the spatial dimension, it constructs the adjacency relationship between irradiation points and analyzes the consistency and continuity between the target irradiation point and adjacent irradiation points in terms of oil film judgment results and detected oil film characteristics. This effectively identifies and suppresses the interference of isolated noise points on the detection results. This multi-level confidence assessment mechanism, from micro to macro, enables the system to not only provide a qualitative conclusion on whether an oil film has been detected but also a quantitative evaluation of the reliability of the detection result, providing richer and more scientific information support for subsequent oil spill determination decisions.

[0030] This invention uses the final confidence level as a reliability correction factor for detecting oil film characteristics. By fusing the original oil film risk value with the corresponding confidence level, the final oil spill risk value is obtained, realizing a weighted correction of the reliability of the detection results. This allows the system to reasonably adjust the influence weight of the point in the final alarm judgment based on the confidence level, even if the original fluorescence signal fluctuates under adverse detection conditions such as severe waves and strong light. This maintains a high sensitivity to real oil spill events while significantly reducing the false alarm rate caused by environmental interference. Furthermore, the verification process set after the confidence level calculation step provides a fault tolerance mechanism for reducing the weight or repeated irradiation of low confidence points, and provides a tracking and verification mechanism for spatially isolated suspected points to be verified by combining the subsequent detection results of adjacent points. This forms a complete closed-loop processing link from data acquisition, parameter adaptation, confidence level assessment to risk judgment and verification correction, thereby significantly improving the engineering reliability and practical deployment value of the oil spill detection system. Attached Figure Description

[0031] Figure 1 This is a logical schematic diagram of an oil spill detection and analysis method based on multi-source data fusion according to the present invention.

[0032] Figure 2 This is a schematic diagram of the single-point confidence calculation logic in an oil spill detection and analysis method based on multi-source data fusion according to the present invention.

[0033] Figure 3 This is a schematic diagram illustrating the spatial consistency confidence calculation logic in an oil spill detection and analysis method based on multi-source data fusion according to the present invention.

[0034] Figure 4 This is a logical diagram illustrating the verification process in an oil spill detection and analysis method based on multi-source data fusion according to the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0038] In order to overcome the technical defects of traditional oil spill detection technology, such as low detection accuracy, high false alarm rate and difficulty in identifying thin oil films caused by factors such as water surface waves, ocean current drift, strong light, water turbidity and fluorescent biological interference.

[0039] This invention proposes an oil spill detection and analysis method based on multi-source data fusion, such as... Figure 1 As shown, the method includes the following: The data acquisition step involves real-time acquisition of video images and external environmental data to obtain the environmental characteristics at the current moment. These environmental characteristics are a set of quantitative parameters reflecting the water surface state and external conditions of the detection area. These parameters affect the excitation efficiency, transmission quality, and reception reliability of the laser-induced fluorescence signal, serving as the basis for subsequent adaptive adjustment of laser irradiation parameters. In this invention, the environmental characteristics consist of multiple components, including light intensity, water ripple amplitude, water turbidity, and water surface drift velocity, and each component is acquired in real-time using corresponding sensors or video image analysis methods.

[0040] Specifically, light intensity affects the strength of ambient background light, directly impacting the signal-to-noise ratio during fluorescence signal reception. This can be obtained in real-time by installing a light sensor on the detector housing. Ripple amplitude reflects the surface roughness and wave intensity of the detection area. Waves cause dynamic changes in the laser incident angle, spot position, and oil film thickness distribution, resulting in drastic fluctuations in the fluorescence signal. This can be obtained through analysis of acquired continuous video images. Water turbidity refers to the decrease in transparency caused by the scattering and absorption effects of suspended particles on light transmission. This affects the attenuation of the laser beam in water. Turbidity values ​​can be collected in real-time using an optical turbidity sensor. Surface drift velocity refers to the horizontal speed and direction of the surface water within the detection area, reflecting the combined effects of ocean currents, tidal currents, and wind-driven currents. Surface drift velocity affects the trajectory and diffusion rate of the oil film on the water surface. This can be obtained by analyzing the optical flow field from video images to extract the surface motion vector.

[0041] After data acquisition, the laser irradiation parameter acquisition step is entered. This step is based on a preset environmental parameter mapping relationship. According to the current environmental characteristics obtained in the data acquisition step, the laser irradiation parameters that match the characteristics of this step are retrieved from the pre-built environmental parameter mapping relationship and used as the irradiation method under the current sea state conditions. The environmental parameter mapping relationship refers to the data set that establishes a one-to-one correspondence between each possible combination of environmental features in the environmental feature space and the experimentally verified optimal laser irradiation parameters under that combination condition. The environmental feature combination is composed of multiple environmental feature components, such as light intensity, water surface ripple amplitude, water turbidity, and water surface drift velocity. Different values ​​for different environmental feature components form different environmental feature combinations. The laser irradiation parameters include three adjustable parameters: irradiation density, irradiation angle, and irradiation power. Irradiation density refers to the number of laser irradiation points per unit area of ​​water surface; irradiation angle refers to the angle between the laser beam and the water surface normal; and irradiation power refers to the optical power output by the laser transmitter.

[0042] Specifically, the strategy for obtaining the environmental parameter mapping relationship is as follows: The experiment is conducted in a pre-constructed, controllable experimental environment. The establishment of the experimental environment requires the use of a wave tank, wave generator, light simulation system, water turbidity adjustment device, standard oil film laying device, and laser-induced fluorescence detection equipment with the same configuration as the detection system being calibrated. Saltwater with the same salinity as actual seawater is injected into the wave tank to simulate the optical properties of the real marine environment. A wave generator is installed at one end of the tank. The wave generator is mechanically driven to generate waves with different specific parameters in the tank, which can simulate different water surface ripple amplitudes and water surface drift speeds in the tank. The illumination simulation system is used to generate controllable lighting conditions above the experimental water tank to simulate different light intensities. The system can be composed of a set of high-power halogen lamps or an LED array to simulate different light intensity environments. It combines a water turbidity adjustment device to change the turbidity of the water in the tank to simulate the optical transparency of different water bodies. The water turbidity adjustment device consists of a metering pump, a turbidity standard storage tank and a stirrer. By injecting a quantitative amount of turbidity standard into the tank through the metering pump and turning on the stirrer to disperse it evenly, the turbidity of the water can be adjusted, thereby constructing multiple sets of environmental characteristic combinations. A standard oil film spreading device is used to spread a uniform oil film of controllable thickness on the surface of the water in the tank. The device consists of a precision injection pump, a linear translation mechanism and a scraper. It can inject a known volume of oil into the tank and use the linear translation mechanism to drive the scraper to move across the water surface at a constant speed. The scraper spreads the injected oil evenly into an oil film of a predetermined thickness. The specific implementation process of the calibration experiment is as follows: First, determine each set of environmental feature combinations. An orthogonal experimental design method or a uniform design method can be used to select a limited number of representative combination points from the complete five-dimensional environmental features as test points for the calibration experiment. For each set of environmental feature combinations, under this combination condition, different combinations of laser irradiation parameters (irradiation density, irradiation angle, and irradiation power) are selected one by one, and oil film detection is performed. Then, the fluorescence spectra collected at each irradiation point are preprocessed, and characteristic parameters such as fluorescence peak intensity, water Raman signal intensity, and fluorescence peak wavelength are extracted. Based on these characteristic parameters, it is determined whether an oil film exists at the irradiation point. The detection result is compared with the known actual situation, i.e., that there is indeed an oil film covering the point. The proportion of correct judgments among all irradiated points is counted as the oil film detection accuracy under this parameter combination. The accuracy value ranges from 0 to 1, and the closer the value is to 1, the better the detection effect. Different laser irradiation parameter combinations are tested in sequence, and the detection accuracy under each group of parameters is recorded during the test. The laser irradiation parameter with the highest oil film detection accuracy or the laser irradiation parameter corresponding to the preset accuracy threshold condition is recorded as the optimal laser irradiation parameter for this environmental feature combination. All environmental feature combinations and their corresponding optimal laser irradiation parameters are stored as the environmental parameter mapping relationship.

[0043] Furthermore, after testing all laser irradiation parameter combinations on a fixed set of environmental features in the calibration experiment and obtaining the detection accuracy for each set of parameters, the data processing module not only records the set of parameters with the highest accuracy as the optimal laser irradiation parameter combination, but also performs statistical analysis on the accuracy corresponding to different irradiation densities, different irradiation angles, and different irradiation power values ​​under the same set of environmental features. By comparing the decrease in accuracy when each of these three parameters deviates from the optimal value, it can be determined which laser irradiation parameter is least affected by environmental interference under the current environmental feature combination. The parameter with the smallest decrease is the most suitable parameter for fine adjustment as the master control parameter under the current environmental conditions. Therefore, in the mapping relationship record corresponding to this environmental feature combination, in addition to storing the optimal laser irradiation parameter combination, the parameter with the least interference is also stored as the master control parameter for fine adjustment.

[0044] In the actual detection process, after obtaining the environmental feature vector at the current moment through the data acquisition steps, the environmental parameter mapping relationship is searched first to find the record that is closest to the current environmental features. The optimal laser irradiation parameter combination stored in the record is read as the starting parameter, and the parameter marker with the least interference marked in the record is read. Then, the adjustment is performed according to the following logic: If the current environmental features are exactly the same as or have a very small deviation from the environmental features recorded in the mapping relationship, the optimal laser irradiation parameter combination in the record is directly used for irradiation without any additional adjustment; if the current environmental features are different from the environmental features recorded in the mapping relationship but are still within the effective range of interpolation, the optimal laser irradiation parameter combination after interpolation is first calculated using the multidimensional linear interpolation method, and then the parameter marked as the least disturbed in the interpolation result is finely adjusted in a small range near its optimal value. That is, while keeping the other two parameters unchanged as the interpolation result, the value of the least disturbed parameter is changed with a small step size and rapid verification irradiation is performed. The value that makes the actual detection signal-to-noise ratio reach the highest is selected as the final irradiation parameter, while the other two parameters remain unchanged as the interpolation result.

[0045] Since this parameter is least affected by environmental interference, small adjustments around its optimal value will not significantly affect the overall detection performance of the system. However, this fine adjustment can compensate for the minor deviations caused by interpolation errors and real-time environmental fluctuations, making the laser irradiation parameters more consistent with the current actual environmental conditions.

[0046] After obtaining the laser irradiation parameters, the laser irradiation step is executed. The optimal laser irradiation parameters output by the laser irradiation parameter acquisition step are used to control the array-distributed laser emitters to perform irradiation. Each laser emitter is independently packaged as an independently controllable irradiation unit. All emitters are installed on the same base and face the water surface.

[0047] The laser irradiation parameters include three controllable dimensions: irradiation density, irradiation angle, and irradiation power. Irradiation density refers to the number of laser irradiation points per unit area within the water surface detection zone, which determines the number of laser emitters to be activated in the laser detection module and the spatial distribution density of these activated emitters on the water surface. Irradiation angle refers to the angle between the laser beam and the normal to the water surface, which determines the direction of the emitted beam from each activated laser emitter. In the laser detection module, each laser emitter is mounted on an independent miniature electronically controlled tilting platform to control its tilt angle. Upon receiving a target irradiation angle command, all activated laser emitters are simultaneously tilted to the target irradiation angle, ensuring that the laser incident angle at all irradiation points remains consistent. Irradiation power refers to the optical power output by each laser emitter, which determines the laser energy density at which the beam reaches the oil film on the water surface. The corresponding drive current value is calculated based on the target irradiation power value, and the same current control signal is sent to the drive circuit of each activated emitter via a digital-to-analog converter, causing all activated emitters to emit lasers at the same power. It should be noted that the spatial distribution density of the illumination density is adjusted by controlling the number of laser emitters that are turned on, while the illumination angle and illumination power are uniformly controlled for all the emitted emitters that are turned on. This is because the environmental characteristics are obtained on a macroscopic scale across the entire detection area, reflecting the overall environmental state of the area rather than the local state of each illumination point. Therefore, the corresponding angle and power compensation should be consistent on a global scale.

[0048] All activated emitters send trigger pulses. After the laser pulse is emitted, the oil film molecules at each irradiation point are stimulated to emit fluorescence signals. The fluorescence signals at each irradiation point are collected by the receiver in the laser detection module. The fluorescence spectrum at each point is obtained by the spectrometer. The raw data collected is preprocessed by the data processing module. The presence of oil film is determined based on the preprocessed spectral data. The presence of typical oil film characteristic peaks in the fluorescence spectrum is used to determine whether there is an oil film covering at the irradiation point. The principle of laser-induced fluorescence technology for determining the presence of oil films is as follows: Analyze the pre-processed spectrum at a certain irradiation point to detect a significant broad peak in the 420 nm to 550 nm band, and if the intensity of this peak is significantly higher than the background noise level of adjacent bands, then an oil film is considered to exist at that irradiation point. Different oil types are distinguished by analyzing the wavelength corresponding to the highest fluorescence intensity. The oil film thickness is analyzed by analyzing the fluorescence intensity and water Raman signal intensity, and the interference of fluorescent microorganisms is distinguished by the difference in the peak wavelength position and half-width at half-maximum of the spectrum.

[0049] The presence of an oil film at each irradiation point is determined based on the fluorescence spectrum. If an oil film is present, the irradiation points with oil films are preliminarily screened for subsequent analysis. The oil film detection characteristics of the corresponding points are extracted. The oil film detection characteristics are a set of quantitative parameters extracted from the fluorescence spectrum. These parameters can characterize the physicochemical properties of the oil film and are used for subsequent confidence assessment and oil spill risk calculation. Specifically, they include four parameters: fluorescence peak intensity, water Raman signal intensity, fluorescence peak wavelength, and spectral half width at half maximum (FWHM).

[0050] After laser irradiation is performed, the obtained detection results and other data are input into the confidence calculation step. This step performs a quantitative assessment of the reliability of the detection results at each irradiation point from two dimensions: single point and spatial distribution. On the one hand, such as Figure 2 As shown, the single-point confidence level of each irradiation point is calculated using a single-point confidence analysis sub-strategy. The basis for this is that water surface waves cause periodic changes in oil film thickness and laser incident angle, resulting in corresponding fluctuations in fluorescence signal intensity. These fluctuations follow specific physical laws. If the oil film characteristics actually measured at a certain irradiation point are consistent with the fluctuation patterns that should exist under the local wave conditions, then the measurement result at that point is reliable; otherwise, it may be affected by interference and therefore unreliable.

[0051] The strategy is as follows: A dedicated ultraviolet (UV) camera is used to capture UV images containing all irradiated points. This dedicated UV camera is optically connected to a regular video camera via a broadband beam splitter prism. A broadband beam splitter prism is an optical element that can split an incident light beam into different exit directions according to wavelength range. Its working principle is based on the thin-film interference effect. A specially designed multilayer dielectric film is deposited on the cemented surface of the prism. This film system has high reflectivity for UV light beams and high transmittance for visible light beams, or vice versa. When a beam of light from the water surface enters the beam splitter, the ultraviolet (UV) beam is reflected into the imaging path of a dedicated UV camera, while the visible light beam is transmitted through the beam splitter into the imaging path of a regular video camera. Since both paths share the same entrance window and front objective lens, the field of view, imaging scale, and spatial coordinates of the two images are identical, differing only in wavelength. After the laser emitter performs illumination, the dedicated UV camera captures an UV image containing the positions of all illuminated spots. The pixel coordinates of each spot in this image correspond to the spatial position of an illuminated point on the water surface. Based on the fact that the UV image and video image have been completely overlapped using a broadband beam splitter, the pixel coordinates of each spot in the UV image can be directly mapped to the same pixel coordinates in the video image, thus determining the local water surface area corresponding to each illuminated point. In this invention, this local water surface area can be defined as a square window centered on the spot position with sides of a preset number of pixels. After obtaining the local water surface area corresponding to each illumination point, the local ripple features of that illumination point are extracted from the corresponding local water surface area. The local ripple features refer to a set of quantitative parameters describing the wave motion state in the local water surface area. In one embodiment of the present invention, the local ripple features include any one or more combinations of local ripple amplitude, local ripple change frequency, and local reflective disturbance intensity. Specifically, the local ripple amplitude reflects the intensity of the waves in the local location; the local ripple change frequency reflects the speed of the waves in the local area; and the local reflective disturbance intensity reflects the intensity of the change in the intensity of reflected light from the water surface in the local area. Furthermore, the extraction of local ripple features can be achieved by performing optical flow analysis on multiple consecutive frames of video images acquired by the video acquisition module. The ripple stability of an irradiated point is calculated using the local ripple characteristics of each local water surface region. Ripple stability describes the intensity of wave disturbance in the local water surface region at the moment of laser irradiation; a higher value indicates a more stable water surface and better measurement conditions, while a lower value indicates a more turbulent water surface and worse measurement conditions. In one embodiment of the invention, when the local ripple characteristics contain only one quantization parameter, this single quantization parameter is used to calculate the ripple stability of the current irradiated point. If multiple quantization parameters are selected, each quantization parameter needs to be assigned a corresponding weight, and the sum of all weights is 1. Then, the ripple stability of each irradiated point is calculated using a weighted summation of each parameter and its corresponding weight. The calculation formula is: , in, For ripple stability; This is the normalized value of the local ripple amplitude. , This represents the measured value of the local ripple amplitude. This is a preset reference value for the maximum ripple amplitude. If the ratio is greater than 1, then... Take 1; This represents the normalized value of the local ripple variation frequency. , This represents the measured value of the local ripple frequency variation. This is a preset reference value for the maximum ripple variation frequency. If the ratio is greater than 1, then... Take 1; The intensity of local reflective disturbance is obtained by normalizing the variance of the gray value of a local pixel region in a video image across consecutive frames, and the value ranges from 0 to 1. , , For the preset weighting coefficients, satisfy + + =1; Then, the matching degree of the illumination parameters at each illumination point is determined based on the matching between the actual laser illumination parameters at each illumination point and the recommended laser illumination parameters under the current local ripple state. The degree of illumination parameter matching describes whether the currently used laser illumination parameters are suitable for the local wave state at the illumination point. The larger the value, the better the parameter matching and the better the measurement conditions; the smaller the value, the worse the parameter matching and the worse the measurement conditions. The basis for this step is that different local ripple states have different optimal requirements for laser illumination parameters. For example, in areas with large waves, a smaller illumination angle should be used to maintain the stability of the light spot, while in areas with small waves, a larger illumination angle can be used to obtain a stronger signal. If the local wave state at a certain illumination point differs from the global average state, and the actual illumination parameters at that point are uniformly set based on the global environmental characteristics, then the illumination parameters at that point may deviate from the local optimal parameters at that point. This deviation will reduce the measurement reliability at that point. The calculation of the irradiation parameter fit requires the pre-establishment of a mapping relationship between the local ripple state and the recommended laser irradiation parameters. This mapping relationship is also obtained using a water tank calibration experiment. In the calibration experiment, the detection accuracy is tested under different irradiation angles, irradiation powers, and irradiation densities for different local ripple amplitudes and ripple change frequencies. The laser irradiation parameters that achieve the highest accuracy or meet the accuracy threshold are recorded as the recommended laser irradiation parameters for that local ripple state. The recommended laser irradiation parameters are only used as a reference benchmark in single-point confidence assessment. In one embodiment of the present invention, for each irradiation point, the corresponding recommended irradiation angle, recommended irradiation power, and recommended irradiation density are found in the calibration mapping relationship based on its local ripple amplitude and local ripple change frequency. Then, the relative deviation between the actual value and the recommended value is calculated. Specifically, the deviation between the actual value and the recommended value of each parameter in the laser irradiation parameters is calculated, and the product of this deviation with the pre-assigned weight for each parameter is summed to calculate the irradiation parameter fit. The calculation formula is: , in, The value ranges from 0 to 1, representing the degree of adaptation of the irradiation parameters. The relative deviation of the illumination angle reflects the degree of deviation of the actual illumination angle from the recommended illumination angle. , This is the actual illumination angle. To recommend the illumination angle, The protection factor is a very small positive number used to prevent division by zero errors when the recommended angle is 0. The relative deviation of the irradiation power reflects the degree of deviation between the actual irradiation power and the recommended irradiation power. , This is the actual irradiation power. Recommended irradiation power; The relative deviation of irradiation density reflects the degree of deviation between the actual irradiation density and the recommended irradiation density. , This represents the actual irradiation density. Recommended irradiation density; and if , or If the calculated value is greater than 1, then take 1 to ensure... It always stays within the range of 0 to 1; , , For the preset weighting coefficients, satisfy + + =1; The single-point confidence level is calculated by combining the quantitative results of ripple stability and irradiation parameter fit. This value is used to characterize the reliability of the oil film detection data collected at the irradiation point. The single-point confidence level ranges from 0 to 1, with values ​​closer to 1 indicating more reliable measurement results and values ​​closer to 0 indicating less reliable results. In one embodiment of the invention, the formula for calculating the single-point confidence level is as follows: ,in, For single-point confidence, and For the preset weighting coefficients, satisfy + =1.

[0052] On the other hand, such as Figure 3 As shown, the spatial consistency confidence analysis sub-strategy is used to evaluate the degree of continuous support of the detection results of each irradiation point in spatial distribution. The basis for this is that oil film presents a continuous regional distribution on the water surface. Therefore, if an irradiation point is determined to have an oil film, its surrounding adjacent irradiation points should also have similar oil film judgment results and similar detection oil film characteristics. Conversely, if the oil film judgment result of an irradiation point is inconsistent with the judgment results of its surrounding adjacent irradiation points, or if its detection oil film characteristics are significantly different from those of its adjacent irradiation points, then the detection result of that point may be affected by local interference and is unreliable.

[0053] The specific content of the spatial consistency confidence analysis sub-strategy is as follows: The adjacency relationship of each irradiation point is constructed based on its spatial location in the water surface to be detected. The adjacency relationship refers to the adjacent relationship between irradiation points in terms of spatial location. Two irradiation points are adjacent to each other if and only if the spatial distance between them is less than the preset adjacency distance threshold. The setting of the adjacency distance threshold should ensure that each irradiation point contains at least three adjacent irradiation points in its adjacency range, while not including irradiation points that are too far away in the adjacency range, which would lead to the loss of the physical meaning of spatial consistency. For any target irradiation point, the consistency of its oil film judgment results with those of adjacent irradiation points and the similarity of its detected oil film characteristics are analyzed to determine the neighborhood support level of the target irradiation point. The oil film judgment result refers to the binary result obtained during the laser irradiation step based on the fluorescence spectrum to determine whether an oil film exists at the irradiation point, with a value indicating either the presence or absence of an oil film. The detected oil film characteristics refer to quantitative parameters extracted from the fluorescence spectrum, such as fluorescence peak intensity, water Raman signal intensity, fluorescence peak wavelength, and spectral half-width. The neighborhood support level describes the degree to which the detection result of the target irradiation point is supported by its adjacent irradiation points; a higher value indicates greater consistency between the detection result of that point and the detection results of its surrounding neighboring points, and thus higher reliability.

[0054] In one embodiment of the present invention, the neighborhood support degree is calculated as follows: from all adjacent irradiation points of the target irradiation point, the adjacent irradiation points whose oil film judgment results are the same as those of the target irradiation point are selected, and these adjacent irradiation points are called consistent adjacent irradiation points. Then, the similarity of the detected oil film features between the target irradiation point and each consistent adjacent irradiation point is calculated respectively. Finally, the similarity of the detected oil film features of all consistent adjacent irradiation points is weighted and summed according to the spatial distance to obtain the neighborhood support degree.

[0055] The calculation method for the similarity of oil film features uses the Gaussian radial basis function, a nonlinear kernel function commonly used to calculate similarity. The calculation formula is as follows:

[0056]

[0057] In the formula, Indicates the target illumination point and adjacent irradiation points The similarity of oil film features between the two is dimensionless and ranges from 0 to 1. It is a natural exponential function; Indicates the target illumination point and adjacent irradiation points The Euclidean distance between them is dimensionless. This represents the total dimension of the detected oil film feature vectors; it is dimensionless. Indicates the target illumination point The One detection oil film characteristic component; Indicates adjacent irradiation points The Each oil film characteristic component is detected; For the first The standard deviation of each feature component is used to normalize each feature component, and its dimensions are the same as those of the corresponding feature components to ensure... It is a dimensionless quantity; The similarity scale parameter is dimensionless, and its value can be calibrated according to actual usage in this invention. The formula for calculating the spatial distance weight corresponding to each uniformly adjacent illumination point is:

[0058]

[0059] In the formula, Indicates the target illumination point and adjacent irradiation points The spatial distance weight between them is dimensionless and ranges from 0 to 1. It is a natural exponential function; Indicates the target illumination point and adjacent irradiation points The spatial Euclidean distance between them, with the dimension of length; and Target irradiation points The horizontal and vertical coordinates on the water surface are measured in units of length. and Adjacent irradiation points The horizontal and vertical coordinates on the water surface are measured in the corresponding units of length. This is a distance scale parameter, with the dimension being the corresponding length unit. In this invention, its value can be calibrated according to actual usage. The formula for calculating the neighborhood support of the target illumination point is:

[0060] in, Indicates the target illumination point The degree of neighborhood support is dimensionless and ranges from 0 to 1; Indicates the target illumination point The set of consistent adjacent irradiation points, that is, the set of indices of all adjacent irradiation points whose oil film judgment results are the same as the target irradiation point; Represents a set Summing all adjacent irradiated points; Target irradiation point With the The spatial distance weight between uniformly adjacent illumination points is dimensionless. Target irradiation point With the The similarity of detected oil film features between each of the five uniformly adjacent irradiated points is dimensionless; the numerator is the sum of the products of the distance weights and feature similarities of all uniformly adjacent irradiated points, dimensionless; the denominator is the sum of the distance weights of all uniformly adjacent irradiated points, dimensionless, used to normalize the weighted similarity so that the neighborhood support is not affected by the number of adjacent irradiated points. In addition, it is necessary to set the target irradiated point... The oil film judgment results of all its adjacent irradiation points are inconsistent; that is, when the number of adjacent consistent points is 0, directly set... =0.

[0061] After obtaining the neighborhood support level of each target irradiation point, the connectivity support level of that target irradiation point is determined based on whether it forms a continuous oil film monitoring region with surrounding irradiation points. The connectivity support level describes whether the target irradiation point belongs to a continuous spatially connected oil film detection region; the higher the value, the better the connectivity of the region where the point is located, and the higher the reliability.

[0062] In one embodiment of the present invention, the method for calculating the degree of connectivity support is as follows: taking all irradiation points as vertices of a graph, establishing edge connections between adjacent irradiation points with the same oil film judgment result to form several subgraphs, each subgraph corresponding to a connected region; for a target irradiation point, calculating the size of its connected region, i.e., the total number of irradiation points contained therein, and the compactness of the connected region; and then using the normalized value of the connected region size and the compactness to calculate the degree of connectivity support. Specifically, the formula for calculating connectivity support is:

[0063] in, Indicates the target illumination point The degree of connectivity support is dimensionless and ranges from 0 to 1. The larger the value, the better the continuity and compactness of the connected region to which the target illumination point belongs, and the higher the spatial reliability. Indicates the target illumination point The total number of irradiated points contained in the connected region is dimensionless. This value is obtained by the connected region labeling algorithm in graph theory. That is, starting from the target irradiated point, a depth-first search or breadth-first search is performed along the adjacency relationship with the same oil film judgment result, traversing all reachable irradiated points and counting them. This represents the maximum number of illumination points contained in all connected regions of the current detection frame. It is dimensionless and serves as the normalization denominator. The ratio falls within the range of 0 to 1. When the target illumination point belongs to the largest connected region, the ratio is 1, and when it belongs to the isolated region, the ratio is close to 0. Indicates the target illumination point The average spatial distance from all irradiated points within a connected region to the center point of that region, expressed in units of length. The calculation formula is:

[0064] In the formula and The first one in each connected region The x and y coordinates of each irradiation point are expressed in units of length. and These are the arithmetic mean of the x-coordinates and y-coordinates of all illuminated points within the connected region, respectively, with units of length, and represent the geometric center of the connected region. This represents summing over all illuminated points within a connected region. Indicates the first The Euclidean distance from each irradiation point to the center of the region is expressed in units of length. This is a preset maximum radius reference value, its dimension is length, and its function is to serve as a normalization denominator, making... The ratio falls within a reasonable range. When the connected region is very compact, the ratio is close to 0; when the connected region is very loose, the ratio may be greater than 1. In this case, the ratio is determined by the denominator. Compactness factor Falling between 0 and 1 The smaller the value, the closer the compactness factor is to 1. The larger the value, the closer the compactness factor is to 0.

[0065] Based on the calculated neighborhood support and connectivity support To generate the spatial consistency confidence of the target illumination point.

[0066] In one embodiment of the present invention, spatial consistency confidence The calculation formula is:

[0067] in, Indicates the target illumination point Spatial consistency confidence, dimensionless, ranges from 0 to 1. The larger the value, the higher the degree of continuous support obtained by the detection results of the irradiation point in the spatial distribution, and the higher the reliability. Conversely, the closer the value is to zero, the less continuous support the detection results of the irradiation point in the spatial distribution is, and the lower the reliability. , The weight coefficients for neighborhood support and connectivity support are respectively, satisfying... + =1.

[0068] Finally, the single-point confidence and spatial consistency confidence obtained for each illumination point based on the above strategy are weighted and fused according to the preset weight coefficients of the two dimensions to obtain the final confidence of each illumination point. The calculation formula is as follows: ,in, Indicates the target illumination point The final confidence level; Indicates the target illumination point Single-point confidence level; Indicates the target illumination point Spatial consistency confidence; , The weighting coefficients for single-point confidence and spatial consistency confidence are respectively, satisfying... + =1; This final confidence level reflects the reliability of the single-point measurement and the continuity of the spatial distribution, providing a more comprehensive and scientific basis for subsequent oil spill risk value calculation.

[0069] And a verification process is performed after the confidence level calculation, such as... Figure 4 As shown, this step is to verify and correct the output of the confidence calculation step, further improving the reliability of the final oil spill judgment result and avoiding false alarms caused by accidental interference in a single measurement or isolated anomalies in spatial distribution.

[0070] The verification process requires two parallel branches: verification of low single-point confidence illumination points and verification of low spatial consistency confidence illumination points.

[0071] The first review processing branch deals with irradiation points where the single-point confidence level is lower than a preset single-point confidence threshold. When the single-point confidence level is low, the reliability of the measurement results is insufficient. Therefore, the review processing can employ two methods. The first method is to reduce the influence weight of the irradiation point in the oil spill risk value determination process. In one embodiment of the invention, by reading the single-point confidence value of the irradiation point, a weight reduction coefficient is calculated based on the difference between the value and the preset single-point confidence threshold. Then, this coefficient is multiplied by the original influence weight of the irradiation point to obtain the corrected influence weight used in the actual oil spill risk value calculation. The second method is to control the laser detection module to repeatedly irradiate and detect the irradiation point. In one embodiment of the invention… In the example, the irradiation point is first marked as a point to be verified. At the end of the current detection cycle or the beginning of the next detection cycle, the control module sends a separate trigger command to the laser emitter corresponding to the irradiation point, and performs laser irradiation and fluorescence spectrum acquisition on the point again with the same laser irradiation parameters. The fluorescence spectra obtained from multiple acquisitions are averaged, and the oil film characteristics are re-extracted and the single-point confidence of the point is recalculated. If the recalculated single-point confidence increases to above the preset single-point confidence threshold, the point is treated as a normal point; if the recalculated single-point confidence is still below the preset single-point confidence threshold, the point is marked as a persistently low-confidence point, and its influence weight is reduced in the oil spill risk value calculation or it is directly excluded. It should be noted that the choice between the two methods for verification processing can be determined according to the real-time and reliability requirements of the actual system. For example, for scenarios with high real-time requirements, the method of reducing the weight is preferred, and for scenarios with high reliability requirements, the method of repeated irradiation is preferred.

[0072] The second review branch targets irradiation points with spatial consistency confidence scores below a preset spatial consistency threshold. For irradiation points with low spatial consistency confidence scores, but whose single-point confidence scores are not below the threshold, the lack of consistency between their detection results and those of neighboring irradiation points suggests that this point may be an isolated anomaly rather than part of a true oil film area. Therefore, it needs to be marked and tracked for review. In one embodiment of the invention, the system marks the irradiation point as an isolated suspected point and records complete information such as its spatial coordinates, detected oil film characteristics, single-point confidence score, and spatial consistency confidence score. In subsequent preset detection cycles, a comprehensive review is performed by combining the subsequent detection results of neighboring irradiation points. In each subsequent detection cycle, the system first checks whether an oil film signal is still detected at the location of the isolated suspected point. If no oil film signal is detected for several consecutive cycles, the isolated suspected point is determined to be a transient interference or noise point, removed from the suspected point list, and no longer triggers an alarm. If the isolated suspected point continues to detect an oil film signal in subsequent cycles, further checks are performed. Check the detection results of adjacent irradiation points in subsequent cycles. If adjacent irradiation points also begin to detect oil film signals in subsequent cycles and the detected oil film characteristics gradually become consistent with the isolated suspected point, it indicates that there is indeed an oil film in the area. The previous isolation was due to the oil film just spreading to the area and the adjacent area not yet being covered. At this time, the isolated suspected point is upgraded to a normal oil film point and participates in the oil spill alarm judgment. If the isolated suspected point continues to detect oil film signals in subsequent cycles but its adjacent irradiation points never detect oil film signals, the point is judged as a permanently isolated abnormal point, which may correspond to a small piece of floating debris on the water surface or fixed noise from the sensor. It is removed from the suspected point list and no longer triggers an alarm.

[0073] Finally, the oil spill result output step is executed. Based on the detected oil film characteristics, the oil film risk value of each irradiation point is determined. By extracting the detected oil film characteristics from the fluorescence spectrum in the laser irradiation step, the oil film risk value of each irradiation point is calculated using the detected oil film characteristics. The detected oil film characteristics refer to a set of quantitative parameters extracted from the fluorescence spectrum, including fluorescence peak intensity, water Raman signal intensity, fluorescence peak wavelength, and spectral half width at half maximum. These parameters directly reflect the physical characteristics of stimulated emission fluorescence of the oil film and are the original objective evidence for judging whether the oil film exists. Therefore, the specific calculation needs to comprehensively consider these spectral characteristic components. First, the characteristic value of each characteristic component in the detected oil film characteristics and its corresponding characteristic component reference value are normalized. Then, multiple characteristic components are weighted and fused according to the pre-set weight coefficients to calculate the oil film risk value. The final confidence level obtained from the confidence level calculation step is used as the basis for the reliability correction of the oil film risk value to obtain the oil spill risk value for each irradiation point. The specific calculation method is as follows: the oil spill risk value is equal to the oil film risk value multiplied by the final confidence level. Here, the oil film risk value reflects the possibility of the existence of oil film at the spectral feature level, and the final confidence level reflects the reliability of the measurement result. The oil spill risk value obtained by multiplying the two takes into account both objective evidence and subjective reliability. The oil spill risk value of each irradiation point is compared with the preset oil spill alarm threshold. The oil spill alarm threshold is a value preset by the staff according to the actual application scenario. If the oil spill risk value is greater than or equal to the alarm threshold, the irradiation point is determined to be an alarm point. If the oil spill risk value is less than the alarm threshold, the irradiation point is determined to be a non-alarm point.

[0074] This invention completes the entire decision-making chain from detecting oil film characteristics and final confidence level to outputting an oil spill alarm through the above steps. By correcting the final confidence level, it ensures that the measurement reliability has a reasonable impact on the decision-making process. Thus, while maintaining a high sensitivity response to real oil spills, it effectively suppresses false alarms caused by environmental interference through the discounting mechanism of the final confidence level, thereby achieving an accurate and reliable automatic oil spill alarm function.

[0075] This invention also proposes a detection system that employs the above-mentioned oil spill detection and analysis method based on multi-source data fusion. The system consists of five modules: a laser detection module, an environmental acquisition module, a video acquisition module, a data processing module, and an oil spill identification and alarm module. The modules interact and work collaboratively with each other through a data bus and a control bus.

[0076] Specifically, the laser detection module is used to emit laser beams to irradiate the water surface and receive fluorescence signals to collect fluorescence spectra. The laser emitting end of this module consists of multiple laser emitters arranged in an array. This array arrangement means that multiple laser emitters are arranged in a regular geometric grid on the same mounting base, and each laser emitter is independently packaged as a complete irradiation unit. Each laser emitter is mounted on an independent miniature electronically controlled tilting platform to control the tilt angle of the laser emission, i.e., the irradiation angle. The on / off state, irradiation power, and irradiation angle of each laser emitter can be independently controlled by the system. Sending control commands to any one emitter in the array does not affect the state of other emitters, allowing the laser detection module to selectively activate some emitters based on the received laser irradiation parameter commands. Based on environmental requirements, the illumination angle and transmission power of all activated transmitters are uniformly set. The receiving end of this module consists of a receiving telescope, fiber optic bundle, spectrometer, and detector array. The optical axis of the telescope is parallel to the reference optical axis of the laser emission module, and the baseline distance between them is precisely calibrated to ensure that the field of view of the receiving telescope can cover the area where all illumination points are located. It is responsible for collecting the weak fluorescence signal generated by the oil film on the water surface after being excited by ultraviolet light. The fiber optic bundle is used to transmit the fluorescence signal collected by the telescope to the spectrometer with low loss. The spectrometer performs spectral processing on the incoming fluorescence signal, decomposing the composite light into monochromatic light of different wavelengths to form a characteristic spectrum. The detector array receives the spectral signal and converts it into an electrical signal. Subsequent algorithms analyze the spectrum to identify the type of oil and determine whether there is an oil spill.

[0077] The environmental acquisition module is used to acquire real-time environmental data of the detection area, including using photoelectric sensors to monitor the light intensity in the environment of the detection area and using optical turbidity sensors to monitor the turbidity of the water in the detection area.

[0078] The video acquisition module is used to acquire video image data of the detection area in real time. This module uses a wide-spectrum beam splitter to fuse a dedicated ultraviolet camera and a visible light camera. The two cameras share the same front optical system. The beam splitter splits the incident light into ultraviolet and visible light paths, which enter the ultraviolet camera and the visible light camera respectively. The visible light camera is used to acquire continuous frame images of the water surface in the detection area. By analyzing the continuous frame images, the amplitude of water surface ripples and the water surface drift speed are obtained. The dedicated ultraviolet camera is used to acquire the first image containing all laser spots when the laser is irradiated. The first image is superimposed with the water surface image to extract the ripple features of the local water surface area corresponding to each irradiation point to realize subsequent confidence calculation and verification.

[0079] The data processing module is used to filter, denoise, and extract features from external environmental data, video image data, and fluorescence signals. After receiving data transmitted from other modules, the module first performs filtering to eliminate sensor noise, removes outliers during the denoising process, and finally performs feature extraction.

[0080] The oil spill identification and alarm module is used for confidence calculation, oil spill risk assessment, and alarm output. It has a built-in confidence assessment module, which is a functional unit specifically used to calculate the final confidence of each irradiation point. It also has a built-in verification processing subunit to perform verification processing steps such as reducing the weight or marking low-confidence irradiation points for verification. Then, using the final confidence of each irradiation point after verification processing and the detected oil film characteristics, the oil spill risk value of each irradiation point is calculated to complete the oil spill risk assessment and alarm. In this invention, the alarm forms include alarm data packets sent to the remote monitoring platform, alarm windows popping up on the local display interface, and activated audible and visual alarm signals. The alarm data packets contain information such as the location, size, and confidence of the oil spill area.

[0081] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An oil spill detection and analysis method based on multi-source data fusion, characterized in that, The method includes the following steps: The data acquisition process involves real-time acquisition of video images and external environmental data to obtain the environmental characteristics at the current moment. The laser irradiation parameter acquisition step involves acquiring the laser irradiation parameters using the environmental characteristics based on a preset environmental parameter mapping relationship, wherein the environmental parameter mapping relationship is based on a pre-tested calibration. The laser irradiation step is performed by controlling the array-distributed laser emitters to perform irradiation according to the laser irradiation parameters, collecting the fluorescence spectrum of each irradiation point, and determining whether an oil film exists based on the fluorescence spectrum. If an oil film exists, the characteristics of the oil film are extracted and detected. The confidence level calculation steps involve calculating and obtaining the final confidence level based on the single-point confidence level and the spatial consistency confidence level. The single-point confidence level reflects the confidence level of the oil film characteristics collected at the corresponding irradiation point, and the spatial consistency confidence level reflects the consistency of the corresponding irradiation point across the distribution of all irradiation points. The oil spill result output step calculates the oil spill risk value of the corresponding point based on the final confidence level of the irradiation point and the corresponding oil film characteristics, and compares the oil spill risk value with the preset oil spill alarm threshold. If the oil spill alarm threshold is exceeded, an oil spill alarm is triggered.

2. The oil spill detection and analysis method based on multi-source data fusion according to claim 1, characterized in that, The strategy for obtaining the environmental parameter mapping relationship includes: By generating waves with various parameters using a wave generator, different water surface ripple amplitudes and water surface drift speeds are simulated. Combined with different light intensities and water turbidity, multiple sets of environmental feature combinations are constructed. For each combination of environmental features, different combinations of values ​​for irradiation density, irradiation angle and irradiation power in the laser irradiation parameters are selected and oil film detection is performed. The laser irradiation parameters corresponding to the highest oil film detection accuracy or the condition that meets the preset accuracy threshold are recorded as the optimal laser irradiation parameters for the current combination of environmental features. All combinations of environmental features and their corresponding optimal laser irradiation parameters are stored as the environmental parameter mapping relationship.

3. The oil spill detection and analysis method based on multi-source data fusion according to claim 1, characterized in that, The single-point confidence level is used to characterize the reliability of oil film detection data collected at the irradiation point under the current local ripple state and current laser irradiation conditions. The single-point confidence level analysis sub-strategy includes: Obtain the local water surface area corresponding to each illumination point in the video image; Based on the local water surface area, extract the local ripple features of the corresponding irradiation point; The measurement reliability of the irradiation point is determined based on the local ripple characteristics and the laser irradiation parameters corresponding to the irradiation point. A single-point confidence level for the irradiation point is generated based on the reliability of the measurement.

4. The oil spill detection and analysis method based on multi-source data fusion according to claim 3, characterized in that, The local ripple characteristics include at least one of the following: local ripple amplitude, local ripple change frequency, local water surface drift velocity, and local reflective disturbance intensity. The step of determining the measurement reliability of the irradiation point based on the local ripple characteristics and the laser irradiation parameters corresponding to the irradiation point includes: The ripple stability of the irradiation point is determined based on the local ripple characteristics. Based on the matching between at least one of the actual irradiation angle, actual irradiation power and actual irradiation density of the irradiation point and at least one of the recommended irradiation angle, irradiation power and irradiation density under the current local ripple state, the degree of adaptation of the irradiation parameters of the irradiation point is determined. The measurement reliability of the irradiation point is determined based on the stability of the ripples and the degree of adaptation of the irradiation parameters.

5. The oil spill detection and analysis method based on multi-source data fusion according to claim 3, characterized in that, The specific steps for obtaining the local water surface area corresponding to each illumination point in the video image are as follows: a first image containing the positions of all illumination points is obtained using a dedicated ultraviolet band camera; the first image is completely superimposed on the video image using a broadband beam splitter; and the positions of the illumination points are superimposed on the video image to determine the local water surface area corresponding to each illumination point.

6. The oil spill detection and analysis method based on multi-source data fusion according to claim 3, characterized in that, The spatial consistency confidence score is used to characterize the degree of continuous support of the detection results at the corresponding irradiation point across multiple irradiation point distributions. The spatial consistency confidence score analysis sub-strategy includes: Based on the spatial location of each irradiation point in the water surface to be tested, construct the adjacency relationship of the irradiation points; For any target irradiation point, obtain the oil film judgment results and detection oil film characteristics of its adjacent irradiation points; Based on the consistency of the oil film judgment results between the target irradiation point and the adjacent irradiation points, and the similarity of the detected oil film features, the neighborhood support degree of the target irradiation point is determined. The connectivity support level of the target irradiation point is determined based on whether the target irradiation point and the surrounding irradiation points form a continuous oil film detection area. The spatial consistency confidence of the target illumination point is generated based on the neighborhood support level and the connectivity support level.

7. The oil spill detection and analysis method based on multi-source data fusion according to claim 6, characterized in that, The closer the spatial distance between the target irradiation point and the adjacent irradiation point, and the higher the similarity of the detected oil film features, the higher the degree of neighborhood support of the adjacent irradiation point for the target irradiation point. When the target irradiation point and multiple adjacent irradiation points have consistent oil film judgment results and form a continuous distribution, the connectivity support of the target irradiation point is improved; When the target irradiation point is an isolated oil film detection point and no corresponding oil film feature is detected at its adjacent irradiation points, the connectivity support of the target irradiation point is reduced.

8. The oil spill detection and analysis method based on multi-source data fusion according to claim 1, characterized in that, Following the confidence calculation step, a verification step is also included: If the confidence level of a single point is lower than the preset single-point confidence level threshold, the influence weight of the irradiation point in the process of determining the oil spill risk value is reduced, or the laser emission module is controlled to repeatedly irradiate and detect the irradiation point. If the spatial consistency confidence level of the irradiated point is lower than the preset spatial consistency threshold, the irradiated point will be marked as an isolated suspected point, and the subsequent detection results of its adjacent irradiated points will be used for verification.

9. The oil spill detection and analysis method based on multi-source data fusion according to claim 1, characterized in that, The process of determining the oil spill risk value at the irradiation point based on the final confidence level and the corresponding detected oil film characteristics includes: The oil film risk value at the irradiation point is determined based on the detected oil film characteristics. The final confidence level is used as the basis for the reliability correction of the oil film risk value to obtain the oil spill risk value at the irradiation point.

10. An oil spill detection system based on multi-source data fusion, employing the oil spill detection and analysis method described in any one of claims 1-9, characterized in that, The system includes: The laser detection module has multiple laser emitters arranged in an array at its laser emitting end. The switching, power, and illumination angle of each emitter can be independently controlled. The module is used to emit a laser beam of preset wavelength to illuminate the water surface and to receive fluorescence signals to collect fluorescence spectra. The environmental acquisition module is used to acquire real-time environmental data of the detection area. The video acquisition module is used to acquire video image data of the detection area in real time; The data processing module is used to filter, denoise, and extract features from external environmental data, video image data, and fluorescence signals. The oil spill identification and alarm module has a built-in confidence assessment module to calculate the final confidence level, and uses the final confidence level to determine the oil spill risk and issue an alarm.

Images