Time sequence marine ecological environment early warning monitoring data abnormal point detection method

Abstract

The invention discloses a time sequence marine ecological environment early warning monitoring data abnormal point detection method, and relates to the technical field of marine ecological environment analysis, and the method comprises the steps: collecting a plurality of types of marine monitoring indexes, constructing a multi-source and multi-dimensional time sequence data set, and carrying out the preprocessing and unification of the data into a standard input format. And based on the causal relationship and co-evolution characteristics between the factors, constructing a graph structure model, and depicting a coupling propagation path between the factors. Generating a local time sequence model by adopting a sliding window, constructing an anti-fact path, comparing the propagation difference of an original path in a map structure, and identifying a key disturbance moment; and in combination with historical event similarity and space consistency information, the early warning credibility of abnormal points is evaluated, and efficient and interpretable graded early warning response is realized. According to the method, accurate identification and credible early warning of abnormal points are realized, causal modeling, anti-factual reasoning and hierarchical response capabilities are realized, and the intelligence and interpretability of marine ecological monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of marine ecological environment analysis, and in particular to a method for detecting abnormal points in time-series marine ecological environment early warning monitoring data. Background Art

[0002] With the increasingly complex changes in the marine ecological environment, monitoring and intelligent early warning based on multi-source data have become an important research direction in marine ecological management. In recent years, thanks to the development of sensor networks, remote sensing observations, and marine Internet of Things technologies, the marine monitoring system can achieve high-frequency continuous collection of physical, chemical, and ecological factors, forming a multi-dimensional, time-series data structure. In data-driven ecological early warning research, the academic community has gradually explored the use of statistical modeling, machine learning, and deep learning methods to identify key change points and assist in the early identification and risk warning of events such as red tides, hypoxic zones, and abnormal sewage discharges. However, most current methods still rely on single-factor abnormal fluctuation discrimination or black-box model output, suffering from problems such as insufficient causal structure modeling and limited system interpretability.

[0003] Although some studies have attempted to introduce a multi-dimensional index joint modeling mechanism, such as using co-variation trends, feature fusion algorithms, or time series clustering methods for abnormal point detection, the existing methods still have obvious limitations in the following key aspects. First, most methods ignore the complex causal relationships and feedback mechanisms among marine ecological factors and cannot depict the dynamic impact path of the change of a certain monitoring factor on the overall ecological structure of the system, resulting in abnormal identification being biased towards local fluctuation judgment and lacking the logic of system evolution. Second, most existing models use unidirectional time series modeling or fixed threshold discrimination and lack a "counterfactual reasoning" mechanism, making it impossible to effectively measure whether a certain data point is a key perturbation source point for system evolution, restricting the robustness and interpretability of the identification results. In addition, traditional methods are difficult to integrate the knowledge structure of historical ecological events and spatial consistency distribution information, resulting in weak model generalization ability and high false alarm rate, and are difficult to support the needs of large-scale monitoring of multiple marine parameters in real scenarios. In contrast, the invention of our side introduces a feedback propagation and graph attention mechanism based on graph structure modeling, integrating a four-dimensional coupling mechanism of causal modeling, counterfactual generation, historical similarity, and spatial consistency, effectively solving the bottlenecks such as low accuracy of abnormal point identification, poor interpretability, and unclear hierarchical response in the prior art, and realizing a marine ecological early warning method with structural intelligence and evolution prediction ability. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is that the existing methods for detecting abnormal points in the marine ecosystem lack a causal feedback modeling mechanism, cannot perform counterfactual-level perturbation analysis, and the early warning output lacks credibility explanation and hierarchical stratification, as well as the problem of how to achieve structured, interpretable, and credibility-driven automatic identification and early warning response of abnormal points in the marine ecosystem.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A method for detecting abnormal points in time-series marine ecological environment early warning monitoring data, including continuously collecting background interference factors of the marine ecological environment to form a multi-source multi-dimensional time-series monitoring data set of the marine ecological environment, preprocessing the data set, and unifying it into a standard input format; based on the causal relationship and co-evolution characteristics between monitoring factors in the multi-source multi-dimensional time-series monitoring data set of the marine ecological environment, regarding each monitoring factor as a node and the feedback path between monitoring factors as an edge, and comprehensively determining the edge weight by combining time causality, mutual information intensity, feedback propagation, and graph attention mechanism, and constructing an EFMG graph by depicting the coupling propagation relationship between factors; through comparative analysis of the propagation path and structural influence of the EFMG graph, constructing a sliding window local time-series model, generating an excluded counterfactual evolution path for each time point, comparing the feedback propagation difference between the original path and the counterfactual path in the structure of the EFMG graph, identifying the perturbation moment through the counterfactual evolution path, and combining historical event similarity and spatial consistency information to evaluate the early warning credibility and conduct hierarchical early warning response.

[0007] As a preferred scheme of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: the preprocessing of the data set includes using the K-nearest neighbor algorithm or Bayesian inference method to fill in missing values in the case of missing time points of monitoring factors, screening initial abnormal points for the time series of each monitoring factor using the Z-score or interquartile range method, completing the time alignment of monitoring factors using a unified time stamp mechanism, and performing unified numerical conversion using the minimum-maximum normalization or standard deviation normalization method.

[0008] As a preferred scheme of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: the construction of the EFMG graph includes representing various monitoring factors as nodes during the construction process, representing the feedback relationship as edges, and the weight of the edges is determined by the fusion of causal analysis, information correlation, time lag, and graph attention mechanism, and the EFMG graph dynamically updates and changes with the monitoring time window.

[0009] As a preferred scheme of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: the EFMG graph is used to represent the feedback path between factors, and during the construction process, a sliding window is used to model the evolution process of indicators, and whether there is a feedback connection between monitoring factors is determined by combining lagged causal relationships and co-information intensity.

[0010] As a preferred embodiment of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: generating the excluded counterfactual evolution path for each time point includes using a sliding window local time-series model to infer and generate an alternative path in the case of non-occurring points based on the information before and after a given time window.

[0011] As a preferred embodiment of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: generating the excluded counterfactual evolution path for each time point includes identifying time points by comparing the propagation results of the original sequence and the counterfactual path in the feedback structure of the EFMG map, constructing a candidate abnormal set to evaluate the early warning credibility, and performing hierarchical early warning responses.

[0012] As a preferred embodiment of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: evaluating the early warning credibility and performing hierarchical early warning responses includes outputting the early warning credibility score of each candidate abnormal point based on four factors: abnormal perturbation score, EFMG map structure change, historical event similarity, and spatial consistency, and dividing into multiple early warning levels according to the score results and triggering corresponding early warning response strategies.

[0013] Another object of the present invention is to provide a system for detecting abnormal points in time-series marine ecological environment early warning monitoring data, which can construct a local time-series model through a sliding window, generate an excluded counterfactual evolution path for each time point, and solve the problems that the current marine ecological abnormal point detection methods cannot perform counterfactual-level perturbation analysis and the early warning output lacks credibility explanation and hierarchical stratification.

[0014] As a preferred embodiment of the system for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to the present invention, wherein: it includes a data acquisition and standardization processing module, an ecological factor - feedback mapping map construction module, and an abnormal identification module based on counterfactual generation and map perturbation; the data acquisition and standardization processing module is used to construct a unified data input basis; the ecological factor - feedback mapping map construction module is used to depict the coupled dynamic relationship between different monitoring indicators in the ecosystem; the abnormal identification based on counterfactual generation and map perturbation is used to identify perturbation points in the monitoring data, judge the impact degree on the ecological feedback system, and form an early warning decision basis.

[0015] A computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for detecting abnormal points in time-series marine ecological environment early warning monitoring data.

[0016] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of a method for detecting abnormal points in time-series marine ecological environment warning and monitoring data.

[0017] Advantages of the present invention: The method for detecting abnormal points in time-series marine ecological environment warning and monitoring data provided by the present invention realizes high-precision identification and intelligent warning of abnormal points in the marine ecological environment by constructing a complete data processing and abnormal detection process. Among them, by continuously collecting various types of monitoring data such as physical, chemical, ecological, and background factors and performing preprocessing, they are uniformly formatted into high-quality standard inputs, providing a robust data basis for subsequent modeling; by constructing an ecological factor-feedback mapping graph (EFMG), it realizes dynamic feedback modeling between monitoring factors based on lagged causality, mutual information, and graph attention mechanism, and can reveal potential systematic evolution paths; combining the sliding window and local modeling methods, counterfactual sequences are generated and the impact of the absence of this point on the propagation of the graph structure is evaluated, thereby quantifying the critical perturbation intensity of a single point; further constructing a counterfactual perturbation scoring function and a graph propagation difference function, and integrating historical event similarity and spatial consistency, an interpretable and discriminative warning credibility quantification factor is designed to realize the determination of the risk level of abnormal points. Finally, through a credibility-driven hierarchical warning mechanism and the output of a graph-counterfactual-historical comparison report, it effectively improves the system's early identification ability for weak ecological perturbations, the interpretability of warning results, and the scientificity of response strategies, and is applicable to the scenarios of ecological anomaly discovery and risk control in complex multi-parameter environments. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0019] Figure 1 It is the overall flowchart of a method for detecting abnormal points in time-series marine ecological environment warning and monitoring data provided by the first embodiment of the present invention. Detailed Embodiments

[0020] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be made in conjunction with the drawings of the specification. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention shall fall within the scope of protection of the present invention.

[0021] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a method for detecting abnormal points in time-series marine ecological environment early warning monitoring data, comprising: S1: Continuously collect background interference factors to form a multi-source and multi-dimensional time series monitoring data set, pre-process the data set, and unify it into a standard input format.

[0022] Furthermore, various monitoring indicators in the marine ecological environment are continuously collected, including but not limited to: Physical factors: water temperature (Temp), salinity (Sal), flow rate (Current); Chemical factors: dissolved oxygen (DO), ammonia nitrogen (NH3), pH; Ecological factors: chlorophyll a (Chla), COD, suspended solids concentration; Background interference factors: tidal information, wind speed and direction, ocean current distribution, etc.

[0023] It should be noted that the collected data form a multi-source and multi-dimensional time series data set, which is standardized into a unified format after preprocessing. The processing steps include missing value interpolation (using KNN or Bayesian inference), initial screening of abnormalities (based on Z-score or IQR), time alignment (timestamp unification) and normalization (such as Min-Max or Z-score standardization).

[0024] S2: Based on the causal relationship and co-evolution characteristics between monitoring factors, a graph structure model is constructed, with each monitoring factor as a node and the feedback path between factors as an edge. The edge weight is determined by combining temporal causality, mutual information strength, feedback propagation and graph attention mechanism to characterize the coupled propagation relationship between factors.

[0025] Furthermore, in order to extract the dynamic coupling relationship between indicators and establish a causal transmission path, the present invention designs an ecological factor-feedback mapping graph (EFMG) and establishes a graph structure in the following way: The nodes are the indicator variables; Edges are feedback relationships between indicators; The edge weight is a comprehensive feedback score that has undergone time-delay causal analysis, mutual information calculation, time position correction, and graph attention mechanism optimization.

[0026] The complete edge weight scoring function is expressed as: in, For Node For Node The ecological feedback strength score, is the local time window, is the maximum interval of lag time, For the node The lag causal effect on the node Exponential weight function of Is the Granger coefficient smoothing parameter Indicates at the time delay The Granger regression coefficient under Is the time position correction function, where Is the period length constant Is the joint probability density function Are respectively the nodes And the node At the moment Marginal probability density of Is the graph attention expansion function Is the trainable vector parameter Is the node time series mapping matrix Is the node feature representation; Indicates the lag causal weight function based on Granger, defined as: Indicates the time position correction coefficient, defined as: Indicates the mutual information enhancement term; Indicates the graph attention expansion factor, defined as , Indicates the node To Ecological feedback impact probability score of, where: When : Indicates no significant impact; When : There is a weak or medium coupling relationship; When : There is a significant ecological feedback chain connection, which needs to be focused on.

[0027] S3: Construct a local time series model through a sliding window, generate the excluded counterfactual evolution path for each time point, compare the feedback propagation differences between the original path and the counterfactual path in the graph structure, identify the perturbation moment, and combine the historical event similarity and spatial consistency information to evaluate the early warning credibility and achieve a hierarchical early warning response.

[0028] Furthermore, after the construction of the ecological factor - feedback mapping graph (EFMG), the present invention further proposes an outlier recognition method that integrates local modeling, counterfactual generation, and graph perturbation evaluation, which is particularly suitable for capturing micro - perturbation time - series outliers in the ecosystem that are difficult to judge by traditional statistical indicators.

[0029] By comparing the real ecological evolution path with the counterfactual evolution path after "excluding a certain point" in different time windows, and combining the EFMG graph structure to compare the differences in their feedback propagation paths, it is determined whether each monitoring moment is a key outlier.

[0030] Time - series segmentation and local modeling: The original time series of each index is divided into local segments of length by means of a sliding window , and each segment is considered as a micro - evolution cycle of an ecosystem.

[0031] For each segment, a local state - space model is constructed inside, such as linear fitting, autoregressive (AR), or encoded into a low - dimensional dynamic state representation by a deep sequence model (such as LSTM) , as the normal evolution representation.

[0032] Counterfactual path generation: For each moment in each time segment , a corresponding counterfactual time series is constructed, that is, the evolution path of "assuming that this point did not occur".

[0033] The method of the counterfactual generation network (CounterfactualGenerator, CFG) is used to generate this sequence, and the generation mechanism is expressed as: where is the trained CFG network function, is the parameter set. This network adopts a multi - layer Transformer structure and combines the attention information of nodes in the EFMG to reasonably construct the filling points.

[0034] The counterfactual path construction is expressed as: In order to quantify the perturbation effect of this monitoring point on the feedback structure of the ecosystem, the following counterfactual difference scoring function is introduced, which is used to measure the difference between the real path and the counterfactual path in the propagation result in the graph structure and is expressed as: where is in the window the The counterfactual perturbation score of a point is the true sliding window segment of length ; is the counterfactual evolution sequence generated after removing ; represents the ecological feedback edge weight generated by the true or counterfactual sequence is the path response factor reflecting the combined weight of the state gradient difference and semantic distance between nodes represents node 's feature change gradient under the current segment is the Euclidean distance between node vectors is a small constant to prevent the denominator from being zero is the time-sensitive adaptive anomaly threshold

[0035] : The graph propagation score constructed based on the true sequence (i.e., the EFMG edge weight); : The graph edge weight score recalculated under the counterfactual sequence; : The response weighting factor of the propagation path, defined as: Finally, if is higher than the adaptive threshold , then it is determined that is a strong perturbation anomaly point and is recorded in the candidate set .

[0036] , represents the perturbation intensity of the absence of a certain point on the overall ecological feedback structure; where: When : This point does not affect the system feedback propagation and is a normal point; When : It may be a slight perturbation point and enters the observation set; When : It is regarded as a strong anomaly point and enters the strong candidate set , as the target of subsequent key early warning analysis

[0037] Design a factor function for comprehensive anomaly credibility calculation , which considers the following four influencing sub-factors and constructs through the non-linear fusion function including: The counterfactual perturbation score ; The graph propagation perturbation score ; The historical event matching score ; Spatial consistency relative score ; The early warning credibility quantization factor is expressed as: The counterfactual perturbation score is expressed as: The graph propagation perturbation score is expressed as: Among them, is the belief propagation distribution under the true graph, is the propagation result under the counterfactual path.

[0038] The historical background event matching score is expressed as: Among them, is the event feature vector of the current anomaly point, is the central feature of historical events.

[0039] The spatial consistency relative score is expressed as: Among them, is the anomaly point at the th adjacent site's consistency score (probabilistic processing), is the number of adjacent sites.

[0040] The confidence amplitude enhancement factor is expressed as: Among them, is the cumulative value of the offset rate under the historical anomaly confidence window (obtained from the sliding window history), is the early warning credibility score of the th candidate anomaly point, is the counterfactual perturbation score, is the confidence difference under graph propagation, is the cosine similarity score between the current event and historical typical anomaly events, is the spatial consistency information entropy value, is the historical confidence trend offset enhancement factor, and are the graph feedback values of the true and counterfactual sides respectively, is the feedback perturbation weighting factor, and are the belief propagation results under the true and counterfactual respectively, and They are the vector representations of the current abnormal event and historical events respectively, is the abnormal point At the consistency probability of the adjacent points, is the historical confidence cumulative drift intensity, is the adjustment constant of the similarity non-linear enhancement factor.

[0041] If : It is determined as the low-risk warning level (LevelI), indicating the existence of potential local fluctuations, and it is recommended that the operation and maintenance personnel pay attention; if : It is determined as the high-risk warning level (LevelII), and the system automatically triggers an alarm and conducts regional expansion analysis; if : It is determined as the extremely high-risk level (LevelIII), which is a core alarm event and requires an immediate response, and it is necessary to compare and trace back with the historical database.

[0042] This warning stratification strategy combines spectral perturbation, spatial consistency and historical event similarity, and has good resolution and decision-making stability.

[0043] It should be noted that the basic information of the abnormal point includes: The name of the abnormal index (such as: dissolved oxygen, chlorophyll a, etc.); The timestamp of the abnormal occurrence ; The number of the monitoring station and the geographical location coordinates to which it belongs ; The abnormal intensity score , and the confidence score .

[0044] Call the adjacency matrix of the graph structure at the current time in the ecological feedback mapping graph (EFMG) module , and extract the high-impact links in the first-order and second-order paths around the abnormal node to form a feedback chain path graph for identifying its systematic propagation possibility.

[0045] Call out the counterfactual generation path , and make a graphical comparison with the original sequence before and after the abnormal point (such as error heat map, trend difference map) to externally display the possible system evolution deviation that could have been avoided if this abnormal point had not occurred.

[0046] Calculate the cosine similarity between the current event feature vector and all the known ecological event center vectors in the historical label database , and extract the top historical events (such as red tides, hypoxia zones, sewage discharges), and output their reference impact durations, areas, and recommended response measures to construct a two-way association warning explanation chain from the current to the historical.

[0047] Example 2, an embodiment of the present invention, provides a system for detecting abnormal points in time-series marine ecological environment warning monitoring data, including a data collection and standardization processing module, an ecological factor-feedback mapping atlas construction module, and an anomaly recognition module based on counterfactual generation and atlas perturbation. The data collection and standardization processing module is used to construct a unified data input basis. The ecological factor-feedback mapping atlas construction module is used to depict the coupled dynamic relationships between different monitoring indicators in the ecosystem. The anomaly recognition based on counterfactual generation and atlas perturbation is used to identify the perturbation points in the monitoring data, judge the impact degree on the ecological feedback system, and form the basis for early warning decision-making.

[0048] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0049] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, which can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0050] More specific examples (a non-exhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer diskettes (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then storing it in a computer memory.

[0051] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

[0052] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for detecting abnormal points in time-series marine ecological environment early warning monitoring data, characterized in that, Including: Continuously collect the background interference factors of the marine ecological environment to form a multi-source and multi-dimensional time-series monitoring dataset of the marine ecological environment, preprocess the dataset, and unify it into a standard input format; Based on the causal relationships and co-evolution characteristics among the monitoring factors in the multi-source and multi-dimensional time-series monitoring dataset of the marine ecological environment, each monitoring factor is regarded as a node, and the feedback path among the monitoring factors is regarded as an edge. Combining temporal causality, mutual information intensity, feedback propagation, and graph attention mechanism, the edge weights are comprehensively determined, and an EFMG graph is constructed by depicting the coupled propagation relationship among the factors; Through comparative analysis of the propagation path and structural influence of the EFMG graph, a sliding window local time-series model is constructed. For each time point, an excluded counterfactual evolution path is generated. The feedback propagation differences between the original path and the counterfactual path in the EFMG graph structure are compared. The perturbation moment is identified through the counterfactual evolution path, and combined with the historical event similarity and spatial consistency information, the early warning credibility is evaluated, and a graded early warning response is carried out.

2. The method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to claim 1, wherein: The preprocessing of the dataset includes using the K-nearest neighbor algorithm or Bayesian inference method to fill in the missing values when the monitoring factors are missing at time points, screening the initial outliers for the time series of each monitoring factor using the Z-score or interquartile range method, completing the time alignment of the monitoring factors using a unified timestamp mechanism, and performing unified numerical conversion using the minimum-maximum normalization or standard deviation normalization method.

3. The method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to claim 2, characterized in that: The construction of the EFMG graph includes that during the construction process, the nodes represent various monitoring factors, the edges represent the feedback relationship, and the weights of the edges are determined by the fusion of causal analysis, information correlation, time lag, and graph attention mechanism. The EFMG graph dynamically updates and changes with the monitoring time window.

4. The method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to claim 3, wherein: The EFMG graph is used to represent the feedback path among the factors. During the construction process, the evolution process of the indicators is modeled based on a sliding window, and whether there is a feedback connection among the monitoring factors is determined by combining the lagged causal relationship and co-information intensity.

5. The abnormal point detection method for time - series ocean ecological environment early warning monitoring data according to claim 4, wherein: Generating an excluded counterfactual evolution path for each time point includes using a sliding window local time-series model to infer and generate an alternative path in the case of a non-occurring point under the front and back information of a given time series window.

6. The method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to claim 5, characterized in that: Generating an excluded counterfactual evolution path for each time point includes identifying the time point by comparing the propagation results of the original sequence and the counterfactual path in the feedback structure of the EFMG graph, constructing a candidate anomaly set to evaluate the early warning credibility, and carrying out a graded early warning response.

7. The method for detecting abnormal points in time-series marine ecological environment early warning monitoring data according to claim 6, characterized in that: Evaluating the early warning credibility and carrying out a graded early warning response includes outputting the early warning credibility score of each candidate anomaly point based on four factors: abnormal perturbation score, EFMG graph structure change, historical event similarity, and spatial consistency, and dividing it into multiple early warning levels according to the scoring results and triggering the corresponding early warning response strategy.

8. A system adopting the method for detecting abnormal points in time-series ocean ecological environment early warning monitoring data as described in any one of claims 1 to 7, characterized in that: Including a data collection and standardization processing module, an ecological factor - feedback mapping graph construction module, and an anomaly identification module based on counterfactual generation and graph perturbation; The data collection and standardization processing module is used to construct a unified data input basis; The ecological factor - feedback mapping graph construction module is used to depict the coupled dynamic relationship among different monitoring indicators in the ecosystem; The anomaly recognition based on counterfactual generation and graph perturbation is used to identify the perturbation points in the monitoring data, judge the impact degree on the ecological feedback system, and form the basis for early warning decision-making.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method for detecting abnormal points in the time-series marine ecological environment early warning monitoring data according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method for detecting abnormal points in the time-series marine ecological environment early warning monitoring data according to any one of claims 1 to 7 are implemented.

Images