Marine ecological environment influence tracking and monitoring method based on offshore wind power engineering

Abstract

The invention provides a marine ecological environment influence tracking and monitoring method based on an offshore wind power project, and relates to the technical field of marine ecological environment, and the method comprises the steps: carrying out the ecological risk index calculation based on an obtained multi-modal feature vector, and obtaining ecological risk data, and configuring a space-time sequence based on the preprocessed comprehensive data source and the ecological risk data, and predicting a pollution diffusion path and migration blocking based on the preprocessed comprehensive data source. By introducing a dynamic evaluation model of an ecological risk index and combining time sequence modeling and spatial distribution modeling, dynamic analysis and future trend prediction of the ecological risk are realized, and by generating a risk thermodynamic diagram in real time, identifying a high-risk area and simulating a pollutant diffusion path and a biological migration blocking condition, the risk of the ecological risk can be predicted. According to the invention, the monitoring strategy can be dynamically adjusted according to the real-time monitoring data, a scientific basis is provided for construction optimization and ecological restoration, and the accuracy and timeliness of monitoring and evaluation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of marine ecological environment, and in particular to a method for tracking and monitoring the impact of marine ecological environment on offshore wind power projects. Background Art

[0002] The construction and operation of offshore wind farms may produce some chemical substances, which will affect the water quality and salinity of the surrounding sea areas, and thus affect the living environment of marine life. The construction of wind farms may also destroy the original seabed ecosystem, such as coral reefs and seagrass beds, leading to changes in the structure and function of biological communities. At the same time, wind power facilities may also become physical obstacles to the migration of marine life, affecting their survival and reproduction. Therefore, during the entire life cycle of offshore wind power projects, including construction, operation and decommissioning, it is necessary to conduct systematic and continuous monitoring and assessment of the various possible impacts on the surrounding marine ecological environment, provide a scientific basis for environmental protection and energy development, and promote the sustainable development of the offshore wind power industry.

[0003] Existing technologies have the following deficiencies in ecological and environmental monitoring of offshore wind power projects: First, the correlation and comprehensiveness of monitoring data are poor. Single-modal data collection methods are usually used, which cannot effectively integrate multi-modal data such as water quality, acoustics, ecology, and sediments, making it difficult to accurately analyze the interactions between various ecological factors and their combined impacts. Second, there is a lack of dynamic evaluation and prediction capabilities. Existing monitoring methods are mostly based on static analysis, which cannot adjust monitoring strategies in real time or predict pollutant diffusion paths and future changes in ecosystems. This results in the inability to detect high-risk areas in a timely manner, and it is difficult to provide accurate support for optimizing construction or ecological restoration. Summary of the Invention

[0004] The purpose of the present invention is to address the shortcomings of the existing technology and provide a method for tracking and monitoring the impact of marine ecological environment based on offshore wind power projects. It aims to realize multimodal data collection and fusion of water quality, noise, ecological activities and sediment distribution by deploying a multimodal sensor network, including water quality monitoring sensors, acoustic detectors, ecological detectors and remote sensing equipment, so as to comprehensively evaluate the ecological environment status. The present invention also introduces a dynamic assessment method based on the ecological risk index and a spatiotemporal modeling algorithm, which can generate regional risk heat maps in real time and dynamically predict the pollution diffusion path and future change trends of the ecosystem. Through the above improvements, the comprehensiveness of monitoring and the timeliness of decision support are significantly improved, providing a strong technical guarantee for the sustainable development of offshore wind power projects. To achieve the above object, the present invention provides the following technical solutions: A method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment includes the following: Step S100: deploy data collection equipment and divide areas.

[0005] Step S200: Based on the data collection after deploying the data collection devices, obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data, and perform data transmission based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data to obtain a comprehensive data source.

[0006] Step S300: Perform preprocessing based on the obtained comprehensive data source, detect low-confidence regions based on the preprocessed comprehensive data source to obtain low-confidence regions, and extract and fuse features based on the preprocessed comprehensive data source to obtain a multi-modal feature vector.

[0007] Step S400: Calculate the ecological risk index based on the obtained multi-modal feature vector to obtain ecological risk data, configure the spatio-temporal sequence based on the preprocessed comprehensive data source and the ecological risk data, and predict the pollution diffusion path and migration obstruction based on the preprocessed comprehensive data source.

[0008] Step S500: Generate optimization suggestions based on the preprocessed comprehensive data source and the ecological risk data, evaluate the repair priority based on the preprocessed comprehensive data source and the risk assessment data, and generate a decision diagram based on the obtained Gaussian distribution data and trend data.

[0009] Step S600: Introduce new sensors to obtain DNA data and positioning data, introduce a reinforcement learning algorithm to optimize the data collection strategy, and generate an ecological report.

[0010] As a preferred solution of the present invention, the step S100 is specifically: Step S100.1: Divide the area and deploy data collection devices.

[0011] Deploy in key areas, and the key areas include: core monitoring points, sensitive ecological area monitoring points, and background reference points.

[0012] The core monitoring points are: within the construction center and within a radius of 2 kilometers, the sensitive ecological area monitoring points are: coral reefs, seagrass beds, and fish migration path areas, and the background reference points are: sea area points far from the construction area.

[0013] The data collection devices include: water quality monitoring sensors, acoustic monitoring devices, ecological monitoring devices, and remote sensing devices.

[0014] The water quality monitoring sensors are: multi-parameter water quality sensors, the acoustic monitoring devices are: submarine acoustic detectors, the ecological monitoring devices are: optical ecological detectors, and the remote sensing devices are: satellite remote sensing image acquisition systems.

[0015] As a preferred solution of the present invention, the step S200 is specifically: Step S200.1: Based on the deployment of data collection devices, collect data to obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0016] The collected data includes: water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0017] The water quality monitoring data includes: salinity, pH value, ammonia nitrogen, phosphate, and heavy metal concentration. The acoustic monitoring data includes: the propagation path, decibel value, and frequency characteristics of the construction area and equipment operation noise. The ecological monitoring data includes: the activity tracks, population quantity, and migration paths of marine organisms. The sediment monitoring data includes: the distribution and concentration change of seabed sediments.

[0018] The water quality monitoring data is obtained through water quality monitoring sensors. The acoustic monitoring data is obtained through acoustic monitoring devices. The ecological monitoring data is obtained through ecological monitoring devices. The sediment monitoring data is obtained through remote sensing devices.

[0019] The sampling frequency of the water quality monitoring sensors is set to: once every minute. The sampling of the acoustic monitoring devices is set to: ten times per second. The sampling of the ecological monitoring devices is set to: covering an area of 1 square kilometer per hour. The remote sensing device is: once a day.

[0020] For the monitoring points in sensitive ecological areas, high-frequency sampling is adopted, which is once per minute. For the background reference points, low-frequency sampling is adopted, which is once every 10 minutes.

[0021] When the water quality parameters or noise parameters at the monitoring points exceed the warning threshold, it will automatically switch to the high-frequency sampling mode, specifically:

[0022] where: T is the sampling interval time, is the time interval of high-frequency sampling, is the time interval of low-frequency sampling, is the real-time monitoring data value, such as the salinity value currently measured at a certain monitoring point, is the reference data value, such as the average salinity value of the monitoring area or the historical normal value, is the warning threshold, that is, the allowable range of data deviation from the reference value.

[0023] When the deviation between the real-time value and the reference value of the collected data exceeds the warning threshold, it will automatically switch to the high-frequency sampling mode. When the deviation between the real-time value and the reference value of the collected data is within the allowable range, it will automatically switch to the low-frequency sampling mode.

[0024] Step S200.2: Perform data transmission based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0025] The collected water quality monitoring data, acoustic monitoring data, ecological monitoring data and sediment monitoring data are uploaded to the cloud, and the 5G network is used for long-distance data transmission to edge computing devices.

[0026] The collected water quality monitoring data, acoustic monitoring data, ecological monitoring data and sediment monitoring data are preliminarily analyzed through edge computing devices.

[0027] Preliminary analysis includes: data compression and local early warning.

[0028] Data compression compresses the obtained data. Local warning generates warning signals for noise or pollutant exceeding the standard, and performs edge computing to obtain a comprehensive data source, specifically:

[0029] Where: R is the data compression ratio, expressed as a percentage, indicating the proportion of data reduction after compression. The raw data size, such as the size of the unprocessed raw data collected by the sensor, is in bytes. It is the size of compressed data, such as the size of data after edge computing processing, in bytes.

[0030] As a preferred solution of the present invention, step S300 is specifically as follows: Step S300.1: Preprocessing is performed based on the obtained comprehensive data source.

[0031] Preprocessing includes: outlier detection, interpolation repair and data normalization.

[0032] It should be noted that, since pre-processing is a technology well known to those skilled in the art, it will not be described in detail here.

[0033] Step S300.2: Perform low-confidence region detection based on the pre-processed integrated data source to obtain a low-confidence region.

[0034] Confidence interval analysis was used to evaluate the combined data sources.

[0035] When the confidence interval width of a monitoring point exceeds the preset threshold, it is marked as a low confidence area.

[0036] Re-deploy sensors using underwater robots and increase sampling frequency in low-confidence areas.

[0037] Step S300.3: Perform feature extraction and fusion based on the pre-processed comprehensive data source.

[0038] Nonlinear feature extraction is performed through an autoencoder, which includes an encoder and a decoder.

[0039] The encoder compresses the input data into a low-dimensional feature vector, and the decoder reconstructs the original data from the low-dimensional feature vector to obtain a water quality feature vector, an acoustic feature vector, an ecological feature vector, and a sediment feature vector.

[0040] Based on the obtained water quality feature vector, acoustic feature vector, ecological feature vector, and sediment feature vector, weighted fusion is performed to obtain a multi-modal feature vector.

[0041] As a preferred solution of the present invention, the step S400 is specifically as follows: Step S400.1: Calculate the ecological risk index based on the obtained multi-modal feature vector.

[0042] The calculation of the ecological risk index includes: risk factor calculation and weight assignment.

[0043] The risk factor calculation is performed according to the deviation degree between the actual comprehensive data source and the safety threshold to calculate the risk value of the risk factor.

[0044] The weight assignment is performed according to the regional sensitivity and factor priority to obtain the risk factor weight.

[0045] Based on the obtained risk factor risk value and risk factor weight, the ecological risk index is calculated to obtain ecological risk data, specifically: ; In the formula: ERI is the ecological risk index, which is used to represent the comprehensive environmental risk level of the region, n is the number of risk factors, including water quality, noise, and biodiversity environmental factors, is the risk value of the i-th risk factor, indicating the environmental risk degree of a certain factor, is the weight of the i-th risk factor, which is used to represent the influence degree of this factor on the overall risk and is dynamically adjusted according to the regional sensitivity.

[0046] Step S400.2: Configure the spatio-temporal sequence based on the preprocessed comprehensive data source and ecological risk data.

[0047] The spatio-temporal sequence includes: time series and spatial distribution.

[0048] The time series uses a long short-term memory network to process the time series data, takes the preprocessed comprehensive data source of history as the input, and trains the long short-term memory network to predict the environmental risk in the next 1 week to 1 month.

[0049] The spatial distribution uses a Gaussian process model to configure the risk value distribution within the region. The water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data of the monitoring points within the region are used as inputs to train the Gaussian process model, predict the risk values of the unmonitored regions, and generate a complete spatial distribution map.

[0050] Step S400.3: Predict the pollution diffusion path and the migration blockage based on the preprocessed comprehensive data source.

[0051] The pollution diffusion is simulated by combining the hydrodynamic equations, specifically: ; In the formula: C is the pollutant concentration, representing the pollutant level at a certain time point and spatial position, is the change rate of the pollutant concentration with time, u and v are the velocity components of the water flow in the x and y directions, , is the change rate of the pollutant concentration with spatial position, D is the diffusion coefficient, representing the diffusion ability of the pollutant in the water body, , is the second-order change rate of the pollutant concentration with spatial position, describing the diffusion behavior.

[0052] The path optimization is used to calculate the optimal migration path of the organism, obtain the migration path, compare the path differences with the existing facilities, and evaluate the blockage situation.

[0053] As a preferred solution of the present invention, the specific steps of step S500 are as follows: Step S500.1: Generate optimization suggestions based on the preprocessed comprehensive data source and ecological risk data.

[0054] Calculate the impact of the current power on the noise according to the preprocessed comprehensive data source, and adjust the power of the construction equipment, specifically: Evaluate the impact of the equipment on the migration path according to the preprocessed comprehensive data source and ecological risk data, and dynamically adjust the operation frequency of the equipment.

[0055] Mark the high-risk areas according to the ecological risk data and the spatial distribution map, and increase the sensor sampling frequency.

[0056] Step S500.2: Evaluate the repair priority based on the preprocessed comprehensive data source and risk assessment data.

[0057] The damaged degree of the ecosystem and the natural recovery time are used to calculate the repair priority.

[0058] According to the spatial distribution map, the high-risk areas are preferentially repaired, including: coral reef transplantation and seagrass bed artificial cultivation.

[0059] Step S500.3: Generate a decision diagram based on the obtained Gaussian distribution data and trend data.

[0060] Visualize the spatial distribution diagram predicted by the Gaussian process model and display the risk levels in different regions. Plot a trend diagram showing the change of regional risk over time for the results processed by the long short-term memory network.

[0061] As a preferred solution of the present invention, the step S600 is specifically as follows: Step S600.1: Introduce a new type of sensor and obtain DNA data and positioning data.

[0062] The new type of sensor includes: a DNA monitor and a high-frequency acoustic positioning sensor.

[0063] The DNA monitor is used to detect the environmental DNA in the water body to identify the diversity and activities of species in the area in real time. The high-frequency acoustic positioning sensor is used to track the migration path of marine organisms in real time.

[0064] Step S600.2: Introduce a reinforcement learning algorithm to optimize the data acquisition strategy and generate an ecological report, specifically: ; In the formula: is the value function of the current state s and action , indicating the long-term return that can be obtained by selecting action in state s, is the learning rate, which controls the update speed of the value function, and its value range is , r is the current reward, indicating the immediate benefit brought by action in state s, is the discount factor, indicating the weight of future rewards, and its value range is , is the current action leading to the next state, is the next state the set of all possible actions in, is the value function of the optimal action in the next state .

[0065] Generate an ecological report through an office document. The content of the report includes: core monitoring points, monitoring points in sensitive ecological areas, background reference points, coral reefs, seagrass beds, migration paths, comprehensive data sources, ecological risk indices, pollutant concentrations, priority areas for high-risk restoration, spatial distribution diagrams, trend diagrams, DNA data, and positioning data.

[0066] The ecological report will be output in the formats of word and PDF and transmitted to the cloud via the 5G network.

[0067] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By constructing a method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment, which comprehensively covers the whole process from the layout of monitoring equipment, data collection and transmission, data processing and analysis to dynamic assessment and optimization feedback, aiming to achieve systematic and continuous monitoring and scientific assessment of the marine ecological environment. The present invention particularly proposes a comprehensive solution to the problems of single monitoring data, incomplete analysis, and lack of dynamic assessment and prediction ability in the prior art, providing technical support for the coordinated development of wind power projects and ecological protection.

[0068] 2. By deploying a variety of sensor devices, including multi-parameter water quality monitoring sensors, subsea acoustic detectors, optical ecological detectors, and satellite remote sensing image acquisition systems, multi-modal data collection of four key ecological factors, namely water quality, acoustics, ecology, and sediment, is realized. At the same time, through autoencoder technology, feature extraction and fusion of multi-modal data are carried out, effectively integrating multi-source data, comprehensively revealing the interactions and comprehensive impacts among various ecological factors, thus significantly improving the comprehensiveness and relevance of the data.

[0069] 3. By introducing a dynamic assessment model of ecological risk index, combining time series modeling and spatial distribution modeling, dynamic analysis of ecological risk and prediction of future trends are realized. By generating a risk heat map in real time, identifying high-risk areas, and simulating the pollutant diffusion path and the situation of blocked biological migration, the present invention can dynamically adjust the monitoring strategy according to real-time monitoring data, providing a scientific basis for construction optimization and ecological restoration, and improving the accuracy and timeliness of monitoring and assessment.

[0070] 4. Through optimizing the feedback mechanism, real-time optimization suggestions for the construction period and operation period are generated, including equipment power adjustment, operation mode optimization, and sensor layout adjustment. Combining the ecological restoration priority assessment method, high-risk areas are preferentially restored. In addition, the present invention supports the long-term upgrade and continuous improvement of the system. By introducing new sensors and reinforcement learning algorithms, the monitoring efficiency and decision-making ability are continuously improved, and at the same time, a long-term ecological database is established to provide support for regional and international ecological research. Description of the Drawings

[0071] Figure 1 It is a flowchart of a method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment provided by an embodiment of the present application. Detailed Embodiments

[0072] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0073] Please refer to Figure 1 , Figure 1 which is a flowchart of a method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment provided by an embodiment of the present application.

[0074] In this embodiment, a method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment may include step S100, step S200, step S300, step S400, step S500, and step S600; Step S100: Deploy data collection devices and divide regions.

[0075] Step S200: Based on the deployed data collection devices, collect data to obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data. Based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data, perform data transmission to obtain a comprehensive data source.

[0076] Step S300: Based on the obtained comprehensive data source, perform preprocessing. Based on the preprocessed comprehensive data source, detect low-confidence regions to obtain low-confidence regions. Based on the preprocessed comprehensive data source, perform feature extraction and fusion to obtain a multi-modal feature vector.

[0077] Step S400: Based on the obtained multi-modal feature vector, calculate the ecological risk index to obtain ecological risk data. Based on the preprocessed comprehensive data source and ecological risk data, configure the time-space sequence. Based on the preprocessed comprehensive data source, predict the pollution diffusion path and migration obstruction prediction.

[0078] Step S500: Generate optimization suggestions based on the preprocessed comprehensive data source and ecological risk data. Evaluate the repair priority of the preprocessed comprehensive data source and risk assessment data. Generate a decision diagram based on the obtained Gaussian distribution data and trend data.

[0079] Step S600: Introduce new sensors to obtain DNA data and positioning data. Introduce a reinforcement learning algorithm to optimize the data collection strategy and generate an ecological report.

[0080] As a preferred solution of the present invention, step S100 is specifically: Step S100.1: Divide regions and deploy data collection devices.

[0081] Deploy in key areas, including core monitoring points, sensitive ecological area monitoring points, and background reference points.

[0082] The core monitoring points are: within the construction center and within a radius of 2 kilometers. The sensitive ecological area monitoring points are: coral reefs, seagrass beds, and fish migration path areas. The background reference points are: sea area points far from the construction area.

[0083] The data acquisition equipment includes: water quality monitoring sensors, acoustic monitoring equipment, ecological monitoring equipment, and remote sensing equipment.

[0084] The water quality monitoring sensors are: multi-parameter water quality sensors. The acoustic monitoring equipment is: seabed acoustic detectors. The ecological monitoring equipment is: optical ecological detectors. The remote sensing equipment is: satellite remote sensing image acquisition systems.

[0085] In some specific embodiments, the step S200 is specifically as follows: Step S200.1: Based on the deployment of data acquisition equipment, collect data to obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0086] The collected data includes: water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0087] The water quality monitoring data includes: salinity, pH value, ammonia nitrogen, phosphate, and heavy metal concentrations. The acoustic monitoring data includes: the propagation paths, decibel values, and frequency characteristics of construction area and equipment operation noises. The ecological monitoring data includes: the activity trajectories, population numbers, and migration paths of marine organisms. The sediment monitoring data includes: the distribution and concentration changes of seabed sediments.

[0088] The water quality monitoring data is obtained through water quality monitoring sensors. The acoustic monitoring data is obtained through acoustic monitoring equipment. The ecological monitoring data is obtained through ecological monitoring equipment. The sediment monitoring data is obtained through remote sensing equipment.

[0089] The sampling frequency of the water quality monitoring sensors is set to: once per minute. The sampling of the acoustic monitoring equipment is set to: ten times per second. The sampling of the ecological monitoring equipment is set to: covering 1 square kilometer area per hour. The remote sensing equipment is: once per day.

[0090] High-frequency sampling is taken for the sensitive ecological area monitoring points at once per minute, and low-frequency sampling is taken for the background reference points at once per 10 minutes.

[0091] When the water quality parameters or noise parameters at the monitoring points exceed the warning threshold, it will automatically switch to the high-frequency sampling mode, specifically as follows:

[0092] Where: T is the sampling interval time, is the time interval of high-frequency sampling, is the time interval of low-frequency sampling, is the real-time monitoring data value, such as the salinity value currently measured at a certain monitoring point, is the reference data value, such as the average salinity value of the monitoring area or the historical normal value, is the early warning threshold, that is, the allowable range of data deviation from the reference value.

[0093] When the deviation between the real-time value of the collected data and the reference value exceeds the early warning threshold, it will automatically switch to the high-frequency sampling mode. When the deviation between the real-time value of the collected data and the reference value is within the allowable range, it will automatically switch to the low-frequency sampling mode.

[0094] Step S200.2: Perform data transmission based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data.

[0095] Upload the collected water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data to the cloud, and use the 5G network for long-distance data transmission to the edge computing device.

[0096] Perform preliminary analysis on the collected water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data through the edge computing device.

[0097] The preliminary analysis includes: data compression and local early warning.

[0098] Data compression is to compress the obtained data. Local early warning is to generate early warning signals for noise or pollutant over-standard events and perform edge computing to obtain a comprehensive data source. Specifically:

[0099] Where: R is the data compression rate, expressed as a percentage, indicating the proportion of the reduced data volume after compression, is the size of the original data, such as the size of the unprocessed original data collected by the sensor, in bytes, is the size of the compressed data, such as the size of the data after edge computing processing, in bytes.

[0100] In some specific embodiments, the step S300 is specifically: Step S300.1: Perform preprocessing based on the obtained comprehensive data source.

[0101] The preprocessing includes: outlier detection, interpolation repair, and data standardization.

[0102] It should be noted here that since preprocessing is a well-known technology in the field, it will not be elaborated here.

[0103] Step S300.2: Detect low-confidence regions based on the preprocessed comprehensive data source to obtain low-confidence regions.

[0104] Evaluate the comprehensive data source using confidence interval analysis, specifically:

[0105] In the formula: CI is the confidence interval, which is used to represent the range where the data mean may appear. is the mean of the data, Z is the standard score, and at the 95% confidence level, Z = 1.96. is the standard deviation of the data, n is the sample size, and the smaller the width of the confidence interval, the higher the confidence level of the data.

[0106] When the width of the confidence interval at the monitoring point exceeds the preset threshold, it is marked as a low-confidence region.

[0107] Re-deploy sensors through an underwater robot to increase the sampling frequency in the low-confidence regions.

[0108] Step S300.3: Perform feature extraction and fusion based on the preprocessed comprehensive data source.

[0109] Perform non-linear feature extraction through an autoencoder, and the autoencoder includes: an encoder and a decoder.

[0110] The encoder compresses the input data into a low-dimensional feature vector, and the decoder reconstructs the original data from the low-dimensional feature vector to obtain a water quality feature vector, an acoustic feature vector, an ecological feature vector, and a sediment feature vector, specifically:

[0111] In the formula: L is the reconstruction error of the autoencoder, X is the original input data. is the reconstructed data decoded by the autoencoder. is the second norm, which represents the squared error between the reconstructed data and the original data.

[0112] Based on the obtained water quality feature vector, acoustic feature vector, ecological feature vector, and sediment feature vector, perform weighted fusion to obtain a multi-modal feature vector, specifically:

[0113] In the formula: is the feature vector after multi-modal data fusion. is the weight of each modal feature, which is dynamically adjusted according to the monitoring priority. is the water quality feature vector. is the acoustic feature vector. is the ecological feature vector. is the sediment feature vector. Feature fusion is to perform weighted combination on the feature vectors from different data sources to generate a unified input.

[0114] In some specific embodiments, the step S400 is specifically as follows: Step S400.1: Calculate the ecological risk index based on the obtained multi-modal feature vectors.

[0115] The calculation of the ecological risk index includes: risk factor calculation and weight assignment.

[0116] The risk factor calculation is based on the degree of deviation between the actual comprehensive data source and the safety threshold to calculate the risk value of the risk factor, specifically:

[0117] In the formula: is the risk value of the risk factor, indicating the degree of deviation of a certain monitoring parameter from the safety threshold. is the parameter value of real-time monitoring, such as the salinity value of the current monitoring point. is the safety threshold of the corresponding factor, such as the environmental safety standard value of salinity. is the absolute deviation between the monitoring value and the safety threshold.

[0118] The weight assignment is based on the regional sensitivity and factor priority to assign the weight of the risk factor, specifically:

[0119] In the formula: is the weight of the i-th risk factor. is the regional sensitivity score of factor i, indicating the potential impact degree of this factor on the environment. is the sum of the sensitivity scores of all factors, which is used for normalization calculation.

[0120] Calculate the ecological risk index based on the obtained risk factor risk value and risk factor weight to obtain the ecological risk data, specifically: ; In the formula: ERI is the ecological risk index, which is used to represent the regional comprehensive environmental risk level. n is the number of risk factors, including water quality, noise and biodiversity environmental factors. is the risk value of the i-th risk factor, indicating the environmental risk degree of a certain factor. is the weight of the i-th risk factor, which is used to represent the impact degree of this factor on the overall risk and is dynamically adjusted according to regional sensitivity.

[0121] Step S400.2: Configure the spatio-temporal sequence based on the preprocessed comprehensive data source and ecological risk data.

[0122] The spatio-temporal sequence includes: time series and spatial distribution.

[0123] The time series uses a long short-term memory network to process time series data. Taking the preprocessed comprehensive data source in history as the input, train the long short-term memory network to predict the environmental risk in the next 1 week to 1 month. Specifically:

[0124] In the formula: is the hidden state at time t, representing the understanding of the long short-term memory network of the current time. is the input data at time t, such as the water quality feature vector at this time point. is the hidden state at the previous time point, representing the memory of past information. is the input weight matrix, which is used to adjust the influence of the input data. is the weight matrix of the hidden state, which is used to adjust the influence of the previous time point on the current prediction. is the bias term, which corrects the linear combination result of the weight matrix. f is the activation function, introducing non-linear characteristics.

[0125] The spatial distribution uses a Gaussian process model to configure the risk value distribution within the region. Taking the water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data of the monitoring points within the region as the input, train the Gaussian process model to predict the risk values of the unmonitored regions and generate a complete spatial distribution map. Specifically:

[0126] In the formula: is the risk value distribution at a certain point x. GP is the Gaussian process, representing the probability distribution of data points. is the mean function, representing the average risk value at point x. is the kernel function, representing the similarity between any two points x and between.

[0127] Step S400.3: Predict the pollution diffusion path and migration blockage prediction based on the preprocessed comprehensive data source.

[0128] Simulate the pollution diffusion by combining the hydrodynamic equation. Specifically: ; Where: C is the pollutant concentration, representing the pollutant level at a certain time point and spatial position, is the change rate of the pollutant concentration over time, and u, v are the velocity components of the water flow in the x and y directions, , is the change rate of the pollutant concentration over spatial position, and D is the diffusion coefficient, representing the diffusion ability of the pollutant in the water body, , is the second-order change rate of the pollutant concentration over spatial position, describing the diffusion behavior.

[0129] Use path optimization to calculate the biologically optimal migration path, obtain the migration path, compare the path differences between the migration path and the existing facilities, and evaluate the blocked situation.

[0130] In some specific embodiments, the step S500 is specifically as follows: Step S500.1: Generate optimization suggestions based on the preprocessed comprehensive data source and ecological risk data.

[0131] Calculate the impact of the current power on the noise according to the preprocessed comprehensive data source, and adjust the power of the construction equipment. Specifically: Evaluate the impact of the equipment on the migration path according to the preprocessed comprehensive data source and ecological risk data, and dynamically adjust the equipment operation frequency. Specifically:

[0132] Where: is the optimized equipment operation frequency, used to reduce the impact of the equipment on the environment, is the original equipment operation frequency, that is, the working frequency of the equipment in the default mode, is the adjustment factor, dynamically calculated according to the ecological sensitivity and risk index of the monitoring area. The adjustment factor is usually less than 1 to reduce the operation intensity.

[0133] Mark the high-risk areas according to the ecological risk data and spatial distribution map, and increase the sensor sampling frequency.

[0134] Step S500.2: Evaluate the repair priority of the preprocessed comprehensive data source and risk assessment data.

[0135] The damaged degree of the ecosystem and the natural recovery time Calculate the repair priority. Specifically:

[0136] Where: is the repair priority value, used to quantify the repair urgency of different ecological regions, is the degree of damage to the ecosystem, usually quantified based on monitoring data. is the natural recovery time of the ecosystem, which represents the time required for the ecosystem to recover to a healthy state without intervention, and the unit is years.

[0137] Prioritize the restoration of high-risk areas according to the spatial distribution map, including: coral reef transplantation and seagrass bed artificial cultivation.

[0138] Step S500.3: Generate a decision-making map based on the obtained Gaussian distribution data and trend data.

[0139] Visualize the spatial distribution map predicted by the Gaussian process model and display the risk levels of different regions. Plot the trend graph of the regional risk changing with time for the results processed by the long short-term memory network.

[0140] In some specific embodiments, the step S600 is specifically as follows: Step S600.1: Introduce a new type of sensor and obtain DNA data and positioning data.

[0141] The new type of sensor includes: a DNA monitor and a high-frequency acoustic positioning sensor.

[0142] The DNA monitor is used to identify the diversity and activities of species in the region in real time by detecting environmental DNA in the water body. The high-frequency acoustic positioning sensor is used to track the migration path of marine organisms in real time.

[0143] Step S600.2: Introduce a reinforcement learning algorithm to optimize the data acquisition strategy and generate an ecological report, specifically: ; In the formula: is the value function of the current state s and the action , which represents the long-term reward that can be obtained by selecting the action in the state s. is the learning rate, which controls the update speed of the value function, and its value range is . r is the current reward, which represents the immediate benefit brought by the action in the state s. is the discount factor, which represents the weight of future rewards, and its value range is . is the current action that leads to the next state . is the set of all possible actions in the next state . is the value function of the optimal action in the next state

[0144] Generate an ecological report through an office document. The report content includes: core monitoring points, sensitive ecological area monitoring points, background reference points, coral reefs, seagrass beds, migration routes, comprehensive data sources, ecological risk indices, pollutant concentrations, high-risk areas for priority restoration, spatial distribution maps, trend charts, DNA data, and positioning data.

[0145] The ecological report will be output in: word and PDF formats and transmitted to the cloud via the 5G network.

[0146] In the above content, in actual application, first delimit a monitoring area in the waters around the offshore wind power project, including core monitoring points, sensitive ecological area monitoring points, and background reference points. The core monitoring points are set as the waters within 2 kilometers around the construction center and its periphery. The sensitive ecological area monitoring points cover coral reefs, seagrass beds, and fish migration route areas. The background reference points are selected as the sea area points far from the construction influence range. Deploy a variety of monitoring devices in the above areas, including multi-parameter water quality monitoring sensors, sub-bottom acoustic sounders, optical ecological detectors, and satellite remote sensing image acquisition systems, which are respectively used to collect relevant data on water quality, acoustics, ecology, and sediment. All devices are installed at the monitoring points through a precise positioning system, and the layout density and sampling frequency are adjusted according to the regional characteristics to ensure the integrity of the data collection coverage area.

[0147] Collect data from four dimensions: water quality, acoustics, ecology, and sediment. The water quality monitoring sensor is used to measure salinity, pH value, ammonia nitrogen concentration, phosphate concentration, and heavy metal concentration. The sub-bottom acoustic sounder records the decibel value, frequency characteristics, and propagation path of equipment operation and construction noise. The optical ecological detector tracks the activity trajectories, population numbers, and migration routes of marine organisms. The satellite remote sensing image acquisition system is used to monitor the distribution and concentration changes of sub-bottom sediments. All monitoring devices operate in a high-frequency sampling mode. Among them, the sensitive ecological area monitoring points are sampled once per minute, and the background reference points are sampled once every 10 minutes. During the collection process, the devices upload the data to the edge computing device through the 5G communication network to achieve the functions of preliminary data compression and local warning.

[0148] Perform preprocessing operations on the collected multi-modal data, including outlier detection, interpolation repair, and data standardization, to eliminate sensor errors and data biases. Use an autoencoder to extract features from the preprocessed data. Among them, the encoder compresses the input data into a low-dimensional feature vector, and the decoder reconstructs the low-dimensional feature vector to generate a feature data set with the minimum reconstruction error. In this way, water quality feature vectors, acoustic feature vectors, ecological feature vectors, and sediment feature vectors are extracted. The above feature data are weighted and fused according to the monitoring area priority to generate a multi-modal feature vector, providing input for subsequent ecological risk assessment.

[0149] Calculate the ecological risk index through multi-modal feature vectors. The calculation of risk factor values is based on the degree of deviation between monitoring data and safety thresholds. Weight factors are assigned according to regional sensitivity and environmental impact priorities. The comprehensive risk index is calculated through a weighted formula and is used to quantify the overall ecological risk level of the region. On this basis, time series modeling is used to predict the future change trend of ecological risks, and Gaussian process modeling is combined to draw a heat map of risk distribution to mark high-risk areas. The above dynamic assessment results can be used to optimize sensor layout and risk response strategies.

[0150] Based on the collected pollutant concentration data and regional water flow information, combined with hydrodynamic equations, simulate the diffusion path of pollutants in the monitoring area. The migration path prediction model uses ecological monitoring data to calculate the optimal migration path of marine organisms such as fish, and evaluates the possible migration obstruction caused by wind power facilities. By comparing the path differences with and without facilities, a migration path impact map is generated. The above prediction results provide an intuitive basis for evaluating the actual impact of wind power projects on the marine ecosystem.

[0151] Based on the dynamic assessment and prediction results, generate optimization suggestions for the construction period and operation period, including adjusting the operating power of equipment, optimizing the construction time period, and increasing the monitoring frequency of sensitive areas. At the same time, formulate an ecological restoration plan according to the risk assessment results, and give priority to restoring the affected areas such as coral reefs and seagrass beds. Restore ecological functions through transplantation, artificial substrate construction, etc. For long-term improvement, monitoring data will be stored in a distributed ecological database, and new sensors and algorithms will be introduced regularly to continuously improve monitoring efficiency and assessment accuracy to support regional and international ecological collaborative protection.

[0152] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made in these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An ocean ecological environment impact tracking and monitoring method based on offshore wind power projects, characterized in that, It includes the following steps: S100. Deploy data acquisition devices and divide regions; S200. Based on the data acquisition after deploying data acquisition devices, obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data. Based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data, conduct data transmission to obtain a comprehensive data source; S300. Based on the obtained comprehensive data source, conduct preprocessing. Based on the preprocessed comprehensive data source, conduct low-confidence region detection to obtain low-confidence regions. Based on the preprocessed comprehensive data source, conduct feature extraction and fusion to obtain a multi-modal feature vector; S400. Based on the obtained multi-modal feature vector, calculate the ecological risk index to obtain ecological risk data. Based on the preprocessed comprehensive data source and ecological risk data, configure the spatio-temporal sequence. Based on the preprocessed comprehensive data source, predict the pollution diffusion path and migration obstruction prediction; S500. Based on the preprocessed comprehensive data source and ecological risk data, generate optimization suggestions. Based on the preprocessed comprehensive data source and risk assessment data, conduct repair priority assessment. Based on the obtained Gaussian distribution data and trend data, generate a decision-making diagram; S600. Introduce new sensors to obtain DNA data and positioning data. Introduce a reinforcement learning algorithm to optimize the data acquisition strategy and generate an ecological report.

2. The marine ecological environment impact tracking and monitoring method based on an offshore wind power project according to claim 1, wherein, The specific content of S100 is as follows: S100.

1. Divide regions and deploy data acquisition devices; Deploy in key regions, and the key regions include: core monitoring points, sensitive ecological area monitoring points, and background reference points; The core monitoring points are: within the construction center and within a radius of 1 - 3 kilometers. The sensitive ecological area monitoring points are: coral reefs, seagrass beds, and fish migration path regions. The background reference points are: sea area points far from the construction area; The data acquisition devices include: water quality monitoring sensors, acoustic monitoring devices, ecological monitoring devices, and remote sensing devices; The water quality monitoring sensors are: multi-parameter water quality sensors. The acoustic monitoring devices are: submarine acoustic detectors. The ecological monitoring devices are: optical ecological detectors. The remote sensing devices are: satellite remote sensing image acquisition systems.

3. The marine ecological environment impact tracking and monitoring method based on the offshore wind power project according to claim 1, wherein The specific content of S200 is as follows: S200.

1. Based on the data acquisition after deploying data acquisition devices, obtain water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data; The collected data includes: water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data; The water quality monitoring data includes: salinity, pH value, ammonia nitrogen, phosphate, and heavy metal concentration. The acoustic monitoring data includes: the propagation path, decibel value, and frequency characteristics of the construction area and equipment operation noise. The ecological monitoring data includes: the activity trajectories, population numbers, and migration paths of marine organisms. The sediment monitoring data includes: the distribution and concentration changes of submarine sediments; The water quality monitoring data is obtained through water quality monitoring sensors. The acoustic monitoring data is obtained through acoustic monitoring devices. The ecological monitoring data is obtained through ecological monitoring devices. The sediment monitoring data is obtained through remote sensing devices; The sampling frequency of the water quality monitoring sensor is set to: collect once every 1 - 2 minutes, the sampling of the acoustic monitoring device is set to: collect ten times in 1 - 2 seconds, the sampling of the ecological monitoring device is set to: cover an area of 1 square kilometer every 1 - 2 hours, and the remote sensing device is: once every 1 - 2 days; For the monitoring points in sensitive ecological areas, high - frequency sampling is adopted, which is to collect 1 - 2 times per minute, and for the background reference points, low - frequency sampling is adopted, which is to collect once every 5 - 10 minutes; When the water quality parameters or noise parameters at the monitoring points exceed the warning threshold, it will automatically switch to the high - frequency sampling mode; S200.

2. Transmit data based on the obtained water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data; Upload the collected water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data to the cloud, and use the 5G network for long - distance data transmission to the edge computing device; Conduct preliminary analysis on the collected water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data through the edge computing device; The preliminary analysis includes: data compression and local warning; Data compression is to compress the obtained data. Local warning is to generate warning signals for noise or pollutant over - standard events and perform edge computing to obtain a comprehensive data source.

4. A method for tracking and monitoring the impact on the marine ecological environment based on an offshore wind power project according to claim 1, characterized in that, The specific content of S300 is as follows: S300.

1. Conduct pre - processing based on the obtained comprehensive data source; The pre - processing includes: outlier detection, interpolation repair, and data standardization; S300.

2. Detect low - confidence regions based on the pre - processed comprehensive data source to obtain low - confidence regions; Evaluate the comprehensive data source using confidence interval analysis; When the width of the confidence interval at the monitoring point exceeds the preset threshold, it is marked as a low - confidence region; Re - deploy sensors through an underwater robot and increase the sampling frequency for low - confidence regions; S300.

3. Conduct feature extraction and fusion based on the pre - processed comprehensive data source; Non - linear feature extraction is performed through an auto - encoder, and the auto - encoder includes: an encoder and a decoder; The encoder compresses the input data into a low - dimensional feature vector, and the decoder reconstructs the original data from the low - dimensional feature vector to obtain a water quality feature vector, an acoustic feature vector, an ecological feature vector, and a sediment feature vector; Based on the obtained water quality feature vector, acoustic feature vector, ecological feature vector, and sediment feature vector, weighted fusion is performed to obtain a multi - modal feature vector.

5. A method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment according to claim 1, characterized in that, The specific content of S400 is as follows: S400.

1. Calculate the ecological risk index based on the obtained multi - modal feature vector; The calculation of the ecological risk index includes: calculation of risk factors and weight assignment; The calculation of risk factors is based on the deviation degree between the actual comprehensive data source and the safety threshold to calculate the risk factor risk value; Weight assignment is based on regional sensitivity and factor priority to assign, and obtain the risk factor weight; Based on the obtained risk factor risk value and risk factor weight, calculate the ecological risk index to obtain ecological risk data, specifically: ； Where: ERI is the ecological risk index, which is used to represent the comprehensive environmental risk level of the region. n is the number of risk factors, including water quality, noise, and biodiversity environmental factors. is the risk value of the i-th risk factor, indicating the environmental risk degree of a certain factor. is the weight of the i-th risk factor, which is used to represent the influence degree of this factor on the overall risk and is dynamically adjusted according to the regional sensitivity. S400.

2. Configure the spatio - temporal sequence based on the pre - processed comprehensive data source and ecological risk data; The spatio - temporal sequence includes: time series and spatial distribution; The time series uses a long short-term memory network to process time series data. Taking the preprocessed comprehensive data source of history as the input, the long short-term memory network is trained to predict the environmental risk in the next 1 week to 1 month. For the spatial distribution, a Gaussian process model is used to configure the risk value distribution within the region. Taking the water quality monitoring data, acoustic monitoring data, ecological monitoring data, and sediment monitoring data of the monitoring points within the region as the input, the Gaussian process model is trained to predict the risk values of the unmonitored regions and generate a complete spatial distribution map. S400.

3. Predict the pollution diffusion path and the migration obstruction based on the preprocessed comprehensive data source. Simulate the pollution diffusion by combining the hydrodynamic equations. Specifically: ; Where: C is the pollutant concentration, representing the pollutant level at a certain time point and spatial location, is the rate of change of the pollutant concentration with time, and u, v are the velocity components of the water flow in the x and y directions, , is the rate of change of the pollutant concentration with spatial location, and D is the diffusion coefficient, representing the diffusion ability of the pollutant in the water body, , is the second-order rate of change of the pollutant concentration with spatial location, describing the diffusion behavior; Use path optimization to calculate the optimal migration path of organisms, obtain the migration path, compare the path differences between the migration path and the existing facilities, and evaluate the obstruction situation.

6. The marine ecological environment impact tracking and monitoring method based on the offshore wind power project according to claim 1, wherein, The specific content of S500 is as follows: S500.

1. Generate optimization suggestions based on the preprocessed comprehensive data source and ecological risk data. Calculate the impact of the current power on the noise according to the preprocessed comprehensive data source and adjust the power of the construction equipment. Specifically: Evaluate the impact of the equipment on the migration path according to the preprocessed comprehensive data source and ecological risk data, and dynamically adjust the operation frequency of the equipment. Mark the high-risk regions according to the ecological risk data and the spatial distribution map, and increase the sensor sampling frequency. S500.

2. Evaluate the repair priority based on the preprocessed comprehensive data source and risk assessment data. The degree of damage to the ecosystem and the natural recovery time Calculate the repair priority; Repair the high-risk regions preferentially according to the spatial distribution map, including: coral reef transplantation and seagrass bed artificial cultivation. S500.

3. Generate a decision-making map based on the obtained Gaussian distribution data and trend data. Visualize the spatial distribution map predicted by the Gaussian process model and display the risk levels of different regions. Draw a trend chart of the regional risk changing with time for the results processed by the long short-term memory network.

7. A method for tracking and monitoring the impact of offshore wind power projects on the marine ecological environment according to claim 1, characterized in that, The specific content of S600 is as follows: S600.

1. Introduce new sensors and obtain DNA data and positioning data. The new sensors include: a DNA monitor and a high-frequency acoustic positioning sensor. The DNA monitor is used to detect the environmental DNA in the water body in real time to identify the species diversity and activity in the region. The high-frequency acoustic positioning sensor is used to track the migration path of marine organisms in real time. S600.

2. Introduce a reinforcement learning algorithm to optimize the data acquisition strategy and generate an ecological report. Specifically: ; where: is the value function of the current state s and action indicating the long-term return that can be obtained by choosing action in state s, is the learning rate that controls the update speed of the value function, and its value range is , r is the current reward, indicating the immediate benefit brought by action in state s, is the discount factor that represents the weight of future rewards, and its value range is , is the next state caused by the current action , is the set of all possible actions in the next state , is the value function of the optimal action in the next state ; Generate an ecological report through an office document. The report content includes: core monitoring points, sensitive ecological area monitoring points, background reference points, coral reefs, seagrass beds, migration paths, comprehensive data sources, ecological risk indices, pollutant concentrations, high-risk regions to be repaired preferentially, spatial distribution maps, trend charts, DNA data, and positioning data. The ecological report will be output in the formats of word and PDF and transmitted to the cloud through the 5G network.

Images