River and lake pollution tracing method and system and storage medium

Abstract

The invention provides a river and lake pollution tracing method and system and a storage medium, and the method comprises the steps: obtaining the hydrological basic data of a river and a lake, and converting the hydrological basic data into a directed graph structure to construct a hydrological topology network; original data of the monitoring points and the drain outlets are obtained, new nodes are generated based on the original data, and the new nodes are used for completing updating of the hydrological topology network; constructing a two-way path calculation framework based on the updated hydrological topology network, and obtaining a hydrological path distance from the sewage draining exit to the monitoring point based on the two-way path calculation framework; acquiring a pollutant attenuation coefficient, and constructing a pollution intensity attenuation matrix based on the hydrological path distance and the pollutant attenuation coefficient; acquiring the real-time pollution concentration of the monitoring point, and determining a pollution intensity estimation vector of the sewage draining exit based on the real-time pollution concentration and the pollution intensity attenuation matrix; and extracting a target sewage draining exit corresponding to the maximum component in the sewage draining exit pollution intensity estimation vector, and determining the target sewage draining exit as a main pollution source. The method can improve the tracing efficiency of the river and lake sewage.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydrological geographic information, in particular to a river and lake pollution tracing method and system and a storage medium. BACKGROUND

[0002] River and lake pollution has characteristics such as dynamic diffusion, concealment and complex tracing, which brings great challenges to river and lake pollution tracing work. The traditional river and lake pollution tracing work relies on manual operation, which is not only high in cost and low in efficiency, but also difficult to meet the demand of rapid tracing positioning. Therefore, if the river and lake pollution tracing work can be effectively coupled with geographic information spatial analysis technology, the intelligent and digital level of river and lake pollution tracing will be greatly improved, and efficiency optimization and operation safety upgrading will be realized at the same time.

[0003] In recent years, with the rapid development of sensing technology, geographic information technology and mathematical model evolution algorithm, the rapid tracing technology of river and lake pollution has become a research hotspot. The existing method obtains detection data of fixed observation points through high-precision water quality sensing technology, combines a probability distribution mathematical model, relies on geographic information technology analysis, and has achieved certain effect in the pollution tracing method. However, this method still faces high complexity in actual application, so the efficiency of river and lake sewage tracing is still somewhat low. SUMMARY

[0004] In order to improve the efficiency of river and lake sewage tracing, the embodiments of the present application provide a river and lake pollution tracing method, system and storage medium.

[0005] In a first aspect, the embodiments provide a river and lake pollution tracing method, which comprises: obtaining hydrological basic data of a river and lake, converting the hydrological basic data into a directed graph structure to construct a hydrological topology network; obtaining original data of a monitoring point and a sewage outlet, generating a new node based on the original data, and using the new node to complete the update of the hydrological topology network; constructing a bidirectional path calculation framework based on the updated hydrological topology network, and obtaining a hydrological path distance from the sewage outlet to the monitoring point based on the bidirectional path calculation framework; obtaining a pollutant attenuation coefficient, and constructing a pollution intensity attenuation matrix representing the influence of the pollution intensity of the sewage outlet on the monitoring point based on the hydrological path distance and the pollutant attenuation coefficient; obtaining a real-time pollution concentration of the monitoring point, and determining a sewage outlet pollution intensity estimation vector based on the real-time pollution concentration and the pollution intensity attenuation matrix; extracting a target sewage outlet corresponding to the maximum component in the sewage outlet pollution intensity estimation vector, and determining the target sewage outlet as a main pollution source.

[0006] In some embodiments, the hydrological base data includes river line elements, node spatial coordinates, and digital elevation model data, and the converting the hydrological base data into a directed graph structure to construct a hydrological topological network includes: extracting endpoint coordinates of all line segments in the river line elements to generate a node set of the directed graph structure; taking the start and end points of each river line element as the start point and end point of a directed edge respectively to form an edge set of the directed graph structure; obtaining an elevation value of each node in the node set based on the digital elevation model data, and determining the direction of the directed edge as the downstream direction of water flow if the elevation value of the start point of the directed edge is greater than the elevation value of the end point; if the elevation value of the start point of the directed edge is less than the elevation value of the end point, adjusting the start point and end point of the directed edge so that the direction of the adjusted directed edge is the downstream direction of water flow; constructing a hydrological topological network based on the node set, the edge set, and the direction of the directed edge.

[0007] In some embodiments, the original data includes monitoring spatial coordinates of monitoring points and pollution spatial coordinates of pollution points, and the generating new nodes based on the original data includes: calculating the vertical distances from the monitoring spatial coordinates and the pollution spatial coordinates to all edges in the hydrological topological network, and determining the edge corresponding to the minimum value among all the vertical distances as a target adjacent edge; obtaining target endpoint spatial coordinates of the target adjacent edge containing the endpoints, and obtaining projection coordinates of the monitoring points and the pollution outlets on the target adjacent edge by processing the target endpoint spatial coordinates, the monitoring spatial coordinates, and the pollution spatial coordinates through a vertical projection formula; dividing the target adjacent edge into two new edges at the projection coordinates, and taking the projection coordinates as new nodes.

[0008] In some embodiments, the bidirectional path calculation framework includes a forward path mode and a reverse path mode, and the constructing a bidirectional path calculation framework based on the updated hydrological topological network includes: obtaining the forward path mode and the reverse path mode in the updated hydrological topological network, wherein the enabling condition of the forward path mode is that the start point belongs to the node set of the pollution outlets, and the function is to simulate the propagation of pollutants along the downstream direction of the edge to output the downstream path of the pollution outlets to each monitoring point, and the enabling condition of the reverse path mode is that the start point belongs to the node set of the monitoring points, and the function is to simulate the backtracking of pollutants along the upstream direction of the edge to output the upstream path of the monitoring points to each pollution outlet; associating the downstream path of the pollution outlets to the monitoring points in the forward path mode and the upstream path of the monitoring points to the pollution outlets in the reverse path mode to form a bidirectional path calculation framework.

[0009] In some embodiments, constructing the pollution intensity attenuation matrix representing the influence of the pollution outlets on the pollution intensity of the monitoring points based on the hydrological path distance and the pollution attenuation coefficient comprises: Suppose the set of pollution outlets is J = {j1, j2, …, jn} and the set of monitoring points is I = {i1, i2, …, im}, where n is the number of pollution outlets and m is the number of monitoring points; Define the pollution intensity attenuation matrix A ∈ R m×n , where the matrix element Aij represents the pollution intensity attenuation coefficient of the jth pollution outlet on the ith monitoring point. Based on the pollution attenuation coefficient, calculate the matrix element Aij = exp(-a × Lij), where a is the pollution attenuation coefficient and Lij is the hydrological path distance from the jth pollution outlet to the ith monitoring point. If there is no feasible downstream path from the jth pollution outlet to the ith monitoring point in the updated hydrological topology network, set Aij = 0.

[0010] In some embodiments, determining the pollution intensity estimation vector of the pollution outlets based on the real-time pollution concentration and the pollution intensity attenuation matrix comprises: Arrange the real-time pollution concentration of the monitoring points in the order of monitoring number to obtain the pollution intensity observation vector b, where b is an m*1 vector. Use the Bayesian ridge regression algorithm to minimize the residual of the product of the pollution intensity attenuation matrix A and the pollution intensity estimation vector X and the pollution intensity observation vector b as the optimization objective to construct the objective function , where λ is the regularization parameter and ‖.‖2 is the L2 norm. Optimize the regularization parameter λ through k-fold cross-validation to select the λ that minimizes the cross-validation error opt . Substitute λ opt into the closed-form solution of the Bayesian ridge regression algorithm X = (A T A+λ opt I) -1 A T b, where I is an m-order identity matrix, to obtain the pollution intensity estimation vector of the pollution outlets.

[0011] In some embodiments, the Bayesian ridge regression algorithm is also used to solve the ill-conditioned equation problem caused by the number of monitoring points being less than the number of pollution outlets, where Modify the matrix A T A by the regularization parameter λ opt , so that the modified A T A+λ opt I is a positive definite matrix.

[0012] In some embodiments, the determining the target pollution outlet as the main pollution source further comprises: Based on the edge direction of the updated hydrological topology network, starting from the target pollution outlet, traversing all downstream edges to generate a pollution expansion path of the main pollution source; Generating a spatial element file of the pollution expansion path, the position of the main pollution source, and the maximum component in the pollution intensity estimation vector of the pollution outlet.

[0013] In a second aspect, the embodiment provides a river and lake pollution tracing system, which comprises a network construction module, a processing module, and a tracing module, wherein, The network construction module is configured to obtain hydrological basic data of a river and lake, convert the hydrological basic data into a directed graph structure to construct a hydrological topology network, obtain original data of monitoring points and pollution outlets, generate new nodes based on the original data, and update the hydrological topology network using the new nodes; The processing module is configured to construct a bidirectional path calculation framework based on the updated hydrological topology network, obtain hydrological path distances from pollution outlets to monitoring points based on the bidirectional path calculation framework, obtain a pollutant attenuation coefficient, and construct a pollution intensity attenuation matrix representing the influence of pollution outlets on the pollution intensity of monitoring points based on the hydrological path distances and the pollutant attenuation coefficient; The tracing module is configured to obtain real-time pollution concentrations of monitoring points, determine a pollution intensity estimation vector of a pollution outlet based on the real-time pollution concentrations and the pollution intensity attenuation matrix, extract a target pollution outlet corresponding to the maximum component in the pollution intensity estimation vector of the pollution outlet, and determine the target pollution outlet as the main pollution source.

[0014] In a third aspect, the embodiment provides a computer-readable storage medium having a computer program stored thereon, the computer program being capable of running on a processor, and the computer program being executed by the processor to implement the river and lake pollution tracing method of the first aspect.

[0015] By adopting the above method, the application obtains hydrological basic data of rivers and lakes, converts the hydrological basic data into a directed graph structure to construct a hydrological topology network; obtains original data of monitoring points and sewage outlets, generates new nodes based on the original data, and uses the new nodes to complete the update of the hydrological topology network; constructs a bidirectional path calculation framework based on the updated hydrological topology network, obtains hydrological path distances from the sewage outlets to the monitoring points based on the bidirectional path calculation framework; obtains a pollutant attenuation coefficient, constructs a pollution intensity attenuation matrix representing the influence of the pollution intensity of the sewage outlets on the monitoring points based on the hydrological path distances and the pollutant attenuation coefficient; obtains real-time pollution concentrations of the monitoring points, determines a pollution intensity estimation vector of the sewage outlets based on the real-time pollution concentrations and the pollution intensity attenuation matrix; extracts a target sewage outlet corresponding to a maximum component in the pollution intensity estimation vector of the sewage outlets, and determines the target sewage outlet as a main pollution source. In this way, the application can reduce the demand for manual work of personnel directly participating in pollution tracing by combining the spatial analysis function of the geographic information platform, thereby effectively improving work efficiency and reducing the incidence of accidents and injuries, and improving safety. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a river and lake pollution tracing method block diagram provided by the application.

[0017] Figure 2 is a method block diagram for converting hydrological basic data into a directed graph structure to construct a hydrological topology network provided by the application.

[0018] Figure 3 is a method block diagram for generating new nodes based on original data provided by the application.

[0019] Figure 4 is a flowchart of a river and lake pollution tracing method provided by the application.

[0020] Figure 5 is a connection diagram of a river and lake pollution tracing system provided by the application. DETAILED DESCRIPTION

[0021] To more clearly understand the purpose, technical solutions and advantages of the application, the application is described and explained in detail below in conjunction with the drawings and embodiments. However, those of ordinary skill in the art should understand that the application can be implemented without these details. It is obvious to those of ordinary skill in the art that various changes can be made to the embodiments disclosed in the application, and the general principles defined in the application can be applied to other embodiments and application scenarios without deviating from the principles and scope of the application. Therefore, the application is not limited to the embodiments shown, but conforms to the broadest scope claimed in the application.

[0022] The embodiments of the application are described in further detail below in conjunction with the drawings of the specification.

[0023] Figure 1 is a river and lake pollution tracing method block diagram provided by the present application. As shown in Figure 1 , a river and lake pollution tracing method includes the following steps: Step S100, obtaining hydrological basic data of the river and lake, and converting the hydrological basic data into a directed graph structure to construct a hydrological topology network.

[0024] The present application takes a certain river and lake in a city as an application object, and describes from the tracing end of tracing the pollution of the river and lake. The above hydrological basic data includes river channel line elements, node spatial coordinates and digital elevation model data. Among them, the river channel line elements can be obtained from the vector topographic map provided by the natural resources bureau of the city, and the river channel line elements include a plurality of continuous line segments. Then the starting point and the ending point of the continuous line segments are extracted, and the coordinates of each starting point and each ending point in the topographic map are labeled to obtain the node spatial coordinates, such as the coordinates of a point (113.52°E, 34.78°N). In addition, 30M resolution advanced satellite thermal emission and reflection radiometer global digital elevation model data can be used to extract the digital elevation model data of each endpoint by using ArcGIS software, such as the digital elevation model data is 125.6m.

[0025] Then the obtained hydrological basic data is converted into a directed graph structure to construct a hydrological topology network. Figure 2 is a method block diagram for converting hydrological basic data into a directed graph structure to construct a hydrological topology network provided by the present application. As shown in Figure 2 , converting the hydrological basic data into a directed graph structure to construct a hydrological topology network includes the following steps: Step S101, extracting the endpoint coordinates of all line segments in the river channel line elements to generate a node set of the directed graph structure.

[0026] Step S102, taking the starting and ending points of each river channel line element as the starting point and the ending point of the directed edge respectively to form an edge set of the directed graph structure.

[0027] Step S103, obtaining the elevation value of each node in the node set based on the digital elevation model data, and determining the direction of the directed edge as the downstream direction of water flow if the elevation value of the starting point of the directed edge is greater than the elevation value of the ending point.

[0028] Step S104, if the elevation value of the starting point of the directed edge is less than the elevation value of the ending point, adjusting the starting point and the ending point of the directed edge so that the direction of the adjusted directed edge is the downstream direction of water flow.

[0029] Step S105, constructing a hydrological topology network based on the node set, the edge set and the direction of the directed edge.

[0030] Specifically, the river line element reflects the spatial distribution form of the actual river channel of rivers and lakes, and is the physical basis for generating the node set and the edge set. The node spatial coordinates, as the core constituent elements of the node set, accurately locate the spatial position of each node and ensure the spatial authenticity of the network. The digital elevation model data provide elevation information of the nodes, which is the key basis for judging the downstream direction of water flow and avoiding the contradiction between the network direction and the actual water flow.

[0031] For the node set V, the two endpoint coordinates of each river line element are extracted one by one, and all the extracted endpoint coordinates are summarized to form the node set V of the directed graph structure. Each coordinate corresponds to an independent node, representing a key spatial point in the river channel, such as a river inflection point, a tributary intersection, etc.

[0032] For the edge set E, first, the starting endpoint of a single river line element is taken as the starting point of a directed edge, and the terminal endpoint is taken as the terminal point of a directed edge, thereby forming a directed edge. Then, all the directed edges corresponding to the river line elements are summarized to form the edge set E of the directed graph structure. The edge set E directly reflects the connectivity of the river channel, such as the connection of the main river channel and the straight flow, and the connection of the upstream river channel and the downstream river channel.

[0033] Next, through the digital elevation model data, the elevation value, such as the altitude, of each node in the node set V is queried. For each directed edge in the edge set E, the starting point elevation value corresponding to the starting endpoint is compared with the terminal point elevation value corresponding to the terminal endpoint. If the starting point elevation value is greater than the terminal point elevation value, it conforms to the natural law that water flows from high to low, and the direction of the directed edge is directly determined as the downstream direction of water flow. If the starting point elevation value is less than the terminal point elevation value, it means that the initially set direction of the starting endpoint to the terminal endpoint is opposite to the actual water flow direction, which needs to be further adjusted.

[0034] Among them, for the directed edge whose initial direction is opposite to the actual water flow, the direction correction can be realized by exchanging the starting endpoint and the terminal endpoint, i.e. for the directed edge whose starting point elevation value is less than the terminal point elevation value, the previous starting endpoint is transformed into the terminal endpoint, and the previous terminal endpoint is transformed into the starting endpoint. The new starting elevation value after exchange is greater than the new terminal elevation value, ensuring that the direction of the adjusted directed edge is still the downstream direction of water flow. In this way, the directions of all directed edges are consistent with the actual water flow direction, providing a correct direction benchmark for subsequent pollution diffusion simulation and pollution source tracing.

[0035] Finally, using node set V as spatial points and the corrected edge set E as connecting channels, with the downstream direction of water flow as the directional attribute of the edges, a complete hydrological topological network is formed, encompassing spatial location information, connectivity relationships, and water flow direction attributes. By clearly defining the composition of basic hydrological data, standardizing node and edge generation logic, and relying on digital elevation model data to ensure consistency in water flow direction, an accurate hydrological topological network is constructed. This network provides reliable support at the data structure level for subsequent mapping of monitoring points and sewage outlets, bidirectional path calculation, and pollution source identification, laying the foundation for efficient and accurate tracing of river and lake pollution sources.

[0036] Step S200: Obtain the raw data of monitoring points and sewage outlets, generate new nodes based on the raw data, and use the new nodes to complete the update of the hydrological topology network.

[0037] The raw data includes the monitoring spatial coordinates of the monitoring points and the discharge spatial coordinates of the discharge points. The monitoring spatial coordinates represent the actual installation location of the monitoring equipment, and the discharge spatial coordinates represent the actual location of the pollutant discharge outlet. The raw data of the monitoring points and discharge outlets can be obtained by reading the location of the monitoring equipment and the location of the pollutant discharge outlet. Figure 3 This is a flowchart illustrating the method for generating new nodes based on original data provided in this application. Figure 3 As shown, generating a new node based on the original data includes the following steps: Step S201: Calculate the vertical distances from the monitoring spatial coordinates and the sewage discharge spatial coordinates to all edges in the hydrological topology network, and determine the edge corresponding to the minimum value among all vertical distances as the target neighbor edge.

[0038] Step S202: Obtain the spatial coordinates of the target endpoints, including the target adjacent side. Process the target endpoint spatial coordinates, monitoring spatial coordinates, and sewage discharge spatial coordinates using the vertical projection formula to obtain the projection coordinates of the monitoring point and sewage discharge outlet on the target adjacent side.

[0039] Step S203: The target adjacent edge is split into two new edges at the projection coordinates, and the projection coordinates are used as new nodes.

[0040] First, the vertical distance is calculated to filter target neighbors. This involves using the monitoring spatial coordinates and the sewage discharge spatial coordinates as target points, traversing all directed edges in the obtained hydrological topology network, and calculating the vertical distance from the target point to each directed edge. The vertical distance can be calculated using plane analytical geometry formulas, such as the formula for the distance from a point to a line. Among all calculated vertical distances, the directed edge corresponding to the minimum value is selected as the target neighbor. This minimum vertical distance filtering ensures a perfect match between the target neighbor and the actual river segment, preventing mismatches between monitoring points and tributaries, and between sewage discharge outlets and main canals.

[0041] Then, the projected coordinates of the new node are calculated using the vertical projection formula. The core of this process is to convert the spatial coordinates of the target point into the coordinates of a point on the target's adjacent side. First, the spatial coordinates of the target endpoints corresponding to the target's adjacent side are obtained, that is, the coordinates of the two endpoints of the target's adjacent side in the hydrological topology network, denoted as endpoint P1(x1,y1) and endpoint P2(x2,y2). Simultaneously, the coordinates of the target points at the monitoring point and the sewage outlet are retrieved, denoted as P0(x0,y0). Applying the vertical projection formula, such as the analytical geometry formula based on vector projection, the target point P0 is projected onto the target's adjacent side line segment P1P2, obtaining the projected coordinates Pq(xq,yq). The core logic of the projection formula is to ensure that line segment P0Pq is perpendicular to the target's adjacent side P1P2, and that Pq lies on line segment P1P2, not on its extension, to avoid the projected point exceeding the actual river channel range. If the projected coordinates Pq are on the extension of line segment P1P2, they need to be adjusted to the coordinates of the nearest endpoint. If Pq is outside P1, then the coordinates of P1 are taken to ensure that the new node always falls on the target adjacent side corresponding to the actual river segment.

[0042] Specifically, calculate vectors with vector Calculate projection coefficients . represents the dot product of vectors. Let t represent the vector magnitude. If 0 ≤ t ≤ 1, meaning the projection point lies on line segment P1P2, then the projected coordinates are xq = x1 + t(x2 - x1) and yq = y1 + t(y2 - y1). If t < 0, meaning the projection point is outside P1, then Pq = P1. If t > 1, meaning the projection point is outside P2, then Pq = P2.

[0043] Next, the target neighboring edge is segmented to generate new nodes. By adding nodes through edge splitting, the physical update of the topology network is completed, formally incorporating the monitoring point and the sewage outlet into the network. Specifically, the target neighboring edge P1P2 is first segmented from the projected coordinates Pq into two new edges: one consisting of P1 and Pq, and the other consisting of Pq and P2. Then, the projected coordinates Pq are defined as a new node, and attribute information is associated with it: if the new node corresponds to a monitoring point, then the monitoring point number, real-time and historical pollution concentration data, and monitoring equipment model are associated with it. If the new node corresponds to a sewage outlet, then the sewage outlet number, sewage type, and designed discharge volume are associated with it.

[0044] Finally, the two new edges replace the original target adjacent edges, and the new nodes and attributes are added to the node set of the hydrological topology network, ultimately forming an updated hydrological topology network that includes monitoring points and sewage outlets. This logic, through selecting target adjacent edges, calculating projected coordinates, and splitting edges to generate new nodes, accurately integrates actual monitoring points and sewage outlets into the obtained hydrological topology network, solving the problem of the disconnect between the basic network and the actual pollution scenario. This not only improves spatial positioning accuracy, network functionality, and data storage integration, but also directly provides a reliable and accurate data foundation for subsequent bidirectional path modeling and pollution source intensity calculation, while reducing labor costs and safety risks.

[0045] Step S300: Construct a bidirectional path calculation framework based on the updated hydrological topology network, and obtain the hydrological path distance from the sewage outlet to the monitoring point based on the bidirectional path calculation framework.

[0046] The bidirectional path computation framework includes forward path mode and reverse path mode. The construction of the bidirectional path computation framework based on the updated hydrological topology network includes the following steps: Step S301: Obtain the forward path mode and the reverse path mode in the updated hydrological topology network. The forward path mode is enabled when the starting point belongs to the set of sewage discharge nodes. Its function is to simulate the downstream propagation of pollutants along the edge and output the downstream path from the sewage discharge outlet to each monitoring point. The reverse path mode is enabled when the starting point belongs to the set of monitoring point nodes. Its function is to simulate the upstream backflow of pollutants along the edge and output the reverse path from the monitoring point to each sewage discharge outlet.

[0047] Step S302: Associate the downstream path from the discharge outlet to the monitoring point in the forward path mode and the upstream path from the monitoring point to the discharge outlet in the reverse path mode to form a bidirectional path calculation framework.

[0048] First, two modes of bidirectional path simulation are defined: forward path mode and reverse path mode. The core is to design dedicated path calculation rules for different source tracing needs, clarifying the triggering conditions, calculation logic, and output results of the two modes to ensure non-overlapping functions and coverage of all source tracing scenarios. The forward path mode, a dedicated mode for pollution diffusion simulation, identifies the starting point type through node attributes. If the starting point is marked as a discharge outlet, the forward path mode is automatically activated. It simulates the natural diffusion process of pollutants from the discharge outlet along the water flow direction, locating the monitoring points that the discharge outlet may affect. The calculation logic of the forward path mode is as follows: taking the target discharge outlet node as the starting point S, following the downstream direction of the directed edges, traversing all downstream directed edges originating from the starting point S, forming a continuous sequence of nodes to edges. When traversing to a node marked as a monitoring point, the branch traversal stops, and the complete path from the starting point S to that monitoring point is recorded. Finally, for each individual discharge outlet, the set of downstream paths to all potentially affecting monitoring points is output. Each path includes a path node sequence, a path edge sequence, and the total path distance.

[0049] The reverse path mode, a dedicated mode for pollution source tracing, identifies the starting point type through node attributes. If the starting point is marked as a monitoring point, the reverse path mode is automatically activated. It traces back from the monitoring point along the reverse direction of the water flow to locate pollution outlets that may be discharging pollutants to that monitoring point. The calculation logic of the reverse path mode is as follows: taking the target monitoring point node as the starting point T, it traverses all upstream directed edges that can connect to the starting point T along the upstream direction of the water flow, forming a continuous backtracking sequence from node to edge. When traversing to a node marked as a discharge outlet, the traversal of that branch stops, and the complete path from the starting point T to that discharge outlet is recorded. Finally, for each individual monitoring point, it outputs a set of reverse paths from it to all discharge outlets that may contribute pollutants. Each path includes a path node sequence, a path edge sequence, and a total path distance.

[0050] Then, the two path modes are associated to form a bidirectional path calculation framework. The core is to establish a mapping relationship between forward and reverse paths to achieve collaborative verification and data interoperability, avoiding path deviations caused by the independent operation of the two modes. Specifically, taking the pairing of sewage outlet and monitoring point as the core, the forward path mode output of the sewage outlet to the monitoring point is bound to the reverse path mode output of the monitoring point to the sewage outlet, forming a pair of bidirectional paths. If the forward path from the sewage outlet to the monitoring point and the reverse path from the monitoring point to the sewage outlet are the opposite directions of the same physical path, they are marked as a bidirectional path pair from sewage outlet to monitoring point after association.

[0051] For each pair of associated paths, the total distance of the downstream path is verified to be equal to the total distance of the upstream path. If they are equal, the path is considered valid, and the path pair is retained; its total distance is the hydrological path distance from the discharge outlet to the monitoring point. If they are not equal, the path calculation is considered to have an error, and the process is repeated until the distances are consistent. All verified discharge outlet-monitoring point bidirectional path pairs are summarized to form a complete bidirectional path calculation framework. This framework includes all valid discharge outlet-monitoring point pairs, the bidirectional path for each pair, a unified hydrological path distance, and a path status marker indicating validity. By defining bidirectional path patterns and forming the framework through association verification, a bidirectional path calculation system that is collaborative in diffusion source tracing, verifiable in paths, and reusable in data is constructed. This ensures the accuracy of the hydrological path distance and provides efficient and accurate data support for subsequent pollution quantification calculations, while reducing the system's computational load, thereby indirectly improving the efficiency of river and lake wastewater source tracing.

[0052] Step S400: Obtain the pollutant attenuation coefficient, and construct a pollution intensity attenuation matrix based on the hydrological path distance and the pollutant attenuation coefficient to characterize the impact of the discharge outlet on the pollution intensity of the monitoring point.

[0053] The aforementioned pollutant attenuation coefficient characterizes the rate of attenuation of pollutant concentration per unit path distance, i.e., the exponential attenuation law of pollutants along the hydrological path. This pollutant attenuation coefficient can be obtained through one of the following methods: experimental measurement, model inversion, or literature correction. The construction of a pollution intensity attenuation matrix characterizing the impact of the discharge outlet on the pollution intensity at the monitoring point, based on the hydrological path distance and the pollutant attenuation coefficient, includes the following steps: Step S401: Let the set of sewage outlets be J = {j1,j2,…,jn} and the set of monitoring points be I = {i1,i2,…,im}, where n is the number of sewage outlets and m is the number of monitoring points.

[0054] Step S402, define the pollution intensity attenuation matrix A∈R m×n The matrix element Aij represents the pollution intensity attenuation coefficient of the j-th discharge outlet to the i-th monitoring point.

[0055] Step S403: Based on the pollutant attenuation coefficient, calculate the matrix element Aij = exp(-a × Lij), where a is the pollutant attenuation coefficient and Lij is the hydrological path distance from the j-th discharge outlet to the monitoring point.

[0056] Step S404: If there is no feasible downstream path from the j-th sewage outlet to the i-th monitoring point in the updated hydrological topology network, then set Aij = 0.

[0057] First, define the dimensions and physical meaning of the attenuation matrix, that is, clarify the basic structure of the matrix and establish a one-to-one correspondence between matrix elements and sewage outlets and monitoring points. Specifically, construct an m x n real matrix A ∈ R. m×n The rows of the matrix correspond to the set of monitoring points I, with the first row corresponding to i1, the second row to i2, ..., the m-th row to im; the columns of the matrix correspond to the set of sewage outlets J, with the first column to j1, the second column to j2, ..., the n-th column to jn. The element Aij in the i-th row and j-th column represents the attenuation coefficient of the pollutant emitted from the j-th sewage outlet after it travels along the hydrological path to the i-th monitoring point. A value closer to 1 indicates less attenuation and a stronger pollution impact on the monitoring point; a value closer to 0 indicates more attenuation and a weaker pollution impact on the monitoring point.

[0058] Then, based on the exponential decay model, the effective pairing Aij is calculated. That is, the exponential decay model quantifies the decay law of pollutants with path distance, ensuring that Aij conforms to the actual transport characteristics of pollutants in rivers and lakes. Specifically, in the river and lake hydrological scenario, the decay of pollutants exhibits an exponential decay trend with transport distance, i.e., Aij = exp(-a × Lij), where exp() is the natural exponential function, and a is the pollutant decay coefficient, such as a = 0.001m. -1 Lij represents a 0.1% decrease in pollutant concentration per 1m of transmission, where Lij is the hydrological path distance from the j-th discharge outlet to the i-th monitoring point.

[0059] Next, for combinations of discharge outlets to monitoring points with no feasible downstream path, their pollution correlation is clarified by setting them to zero, avoiding invalid data from interfering with subsequent calculations. Specifically, within the aforementioned bidirectional path framework, if there is no feasible downstream path from discharge outlet j to monitoring point i (e.g., discharge outlet j is located in a closed tributary, i is located in an external main stream, and there is no connecting channel between them, or the path is blocked by a dam), then the combination of discharge outlet j and monitoring point i is determined to be an invalid pairing. For invalid pairs, pollutants cannot be transferred from discharge outlet j to monitoring point i; therefore, the pollution contribution of discharge outlet j to monitoring point i is zero. In this case, Aij is set to 0, which facilitates reducing the computational load during subsequent river and lake pollution source tracing, thereby indirectly improving the efficiency of river and lake wastewater source tracing.

[0060] Finally, all calculated Aij values ​​are arranged according to the monitoring points and the order of discharge outlets, forming a complete m*n matrix A. Matrix A contains the calculated exponential values ​​of valid pairings and the 0 value of invalid pairings. By defining the matrix dimensions, quantifying Aij, handling invalid terms, and integrating the matrix logic, the physical laws of pollutant transport attenuation are transformed into a standardized mathematical matrix. On the one hand, the accurate data from the above steps ensures the accuracy of the matrix, providing reliable input for subsequent pollution source intensity calculations; on the other hand, quantitative correlation and invalid filtering improve the accuracy and efficiency of source tracing.

[0061] Step S500: Obtain the real-time pollution concentration at the monitoring point, and determine the pollution intensity estimation vector of the discharge outlet based on the real-time pollution concentration and the pollution intensity attenuation matrix.

[0062] By combining the pollution intensity attenuation matrix A constructed above with the real-time pollution concentration at the monitoring point, the pollution intensity estimation vector x of the discharge outlet is obtained using the Bayesian ridge regression algorithm. The pollution intensity estimation vector X of the discharge outlet directly reflects the pollution emission intensity of each discharge outlet. The actual pollution concentration at the monitoring point can be obtained by reading the data detected by the monitoring equipment. Determining the pollution intensity estimation vector of the discharge outlet based on the real-time pollution concentration and the pollution intensity attenuation matrix includes the following steps: Step S501: Arrange the real-time pollution concentrations of the monitoring points in order of monitoring number to obtain the pollution intensity observation vector b, where b is an m*1 vector.

[0063] Step S502: Using the Bayesian ridge regression algorithm, the objective function is constructed with the goal of minimizing the product of the pollution intensity attenuation matrix A and the pollution intensity estimation vector X at the discharge outlet, as well as the residual between the product and the pollution intensity observation vector b. Where λ is the regularization parameter and ||.|2 is the L2 norm.

[0064] Step S503: Optimize the regularization parameter λ using k-fold cross-validation to select the λ that minimizes the cross-validation error. opt .

[0065] Step S504, λ opt Substituting the closed-form solution X = (A) into the Bayesian Ridge Regression algorithm T A+λ opt I) -1 A T b, where I is an m-order identity matrix, yields the pollution intensity estimation vector of the sewage outlet.

[0066] The pollution intensity estimation vector x of the sewage outlet is determined by following the logic of data adaptation, objective function construction, parameter optimization, and closed-loop solution. Based on the pollution intensity attenuation matrix obtained above, the solution of the pollution intensity estimation vector x of the sewage outlet is completed in four steps.

[0067] First, a pollution intensity observation vector b is constructed, converting the monitoring data into a vector form suitable for the algorithm. Specifically, following the row order of the pollution intensity attenuation matrix A (i.e., the order of monitoring points i1, i2, ..., im), the real-time pollution concentration of each monitoring point is sorted sequentially, forming an m*1 column vector b. The i-th element bi in the vector corresponds to the real-time pollution concentration of monitoring point i in the i-th row of the pollution intensity attenuation matrix A. Here, the pollution intensity observation vector b is the target output data of the algorithm, i.e., the monitored pollution result. Subsequently, the pollution outlet intensity vector X is adjusted to make AX as close as possible to b.

[0068] Then, the objective function for Bayesian ridge regression is constructed, which is based on the least squares principle plus L2 regularization to balance fitting accuracy and solution stability, avoiding the overfitting problem of traditional least squares. Specifically, the objective function takes the form of... The sum of squared residuals and the square of the L2 norm represent the deviation between the theoretical pollution concentration AX calculated from the pollution intensity attenuation matrix A by the discharge outlet intensity vector X and the actual concentration b at the monitoring point. The smaller this value, the higher the fitting accuracy. Here, λ is the L2 regularization term, and λ is the regularization parameter. The squared L2 norm of X is used to constrain the absolute value of X, preventing individual elements in X from becoming excessively large, i.e., overfitting, such as forcibly amplifying the intensity of a sewage outlet to match a certain abnormal monitoring value. It also addresses the ill-conditioned equation problem caused by the number of monitoring points being less than the number of sewage discharges, i.e., A... T A. This is an irreversible problem. λ is a regularization parameter that controls the strength of regularization. The larger λ is, the stronger the constraint on X, and the more stable the solution, but it may be underfitting. The smaller λ is, the weaker the constraint, and the higher the fitting accuracy, but it may be overfitting, requiring optimization through subsequent steps. The objective function is to find the optimal balance between minimizing the deviation between theoretical and actual concentrations and preventing excessively extreme discharge intensity. In other words, it is to find the optimal balance between ensuring the accuracy of source tracing and ensuring the rationality of the solution, ensuring that the solved X can both match the monitoring data and reflect the actual discharge situation.

[0069] Next, the regularization parameter λ is optimized using k-fold cross-validation, and the optimal λ is selected. opt Specifically, a candidate range for λ can be defined through experience or literature, generating multiple candidate λ values. For each candidate λ, k-fold cross-validation is performed. This involves taking k-1 subsets of the training set and constructing a decay matrix A_train and an observation vector b_train for training. The decay matrix A_train and the observation vector b_train are then substituted into the objective function to solve for the discharge outlet intensity X_train of the training set. The validation error is calculated using the remaining one subset of the validation set (A_val, b_val), as shown below. Repeat the process k times to calculate the average verification error for λ. Among all candidate λ, select the λ with the smallest average verification error, denoted as λmin. opt Determining λ through a data-driven approach avoids overfitting or underfitting caused by traditional empirical λ setting, ensuring that the subsequently solved X is the optimal solution with strong generalization ability.

[0070] In addition, the Bayesian ridge regression algorithm is also used to handle ill-conditioned equation problems caused by the number of monitored items being less than the number of pollutants discharged, where the regularization parameter λ is used. opt For matrix A T A is corrected so that the corrected A T A+λ opt I is a positive definite matrix.

[0071] Finally, λ opt Substituting the values ​​into the objective function and storing them in the analytical closed-form solution, we solve for the pollution intensity estimation vector X of the discharge outlet. The objective function of Bayesian ridge regression has an analytical closed-form solution, requiring no iteration and exhibiting high computational efficiency. Let λ... opt Substituting the values ​​allows for direct solution, thus improving the efficiency of tracing the source of sewage in rivers and lakes. Specifically, the closed-form solution formula is: X = (A T A+λ opt I) - 1 A T b, A T For the transpose of the pollution intensity attenuation matrix A, A T Let A be an n*n information matrix, if m <n,A T A is usually irreversible, constituting an ill-conditioned equation. Adding λ... opt After I, A T A+λ opt If I becomes a positive definite matrix, it is necessarily invertible, thus resolving the ill-conditioned problem. I is an n*n identity matrix, ensuring that the regularization term uniformly corrects the information matrix. (A) T A+λ opt I) -1 A is the inverse of the corrected information matrix, with dimensions n*n. T Let b be an n*1 vector. Project the m-dimensional observation vector B onto the n-dimensional sewage outlet space. Substitute the relevant information into the above X = (A T A+λ opt I) -1 A TIn equation b, the pollution intensity estimation vector X of the discharge outlet is obtained. The j-th element xj in X represents the pollution emission intensity of the j-th discharge outlet (the unit is consistent with b, such as the equivalent emission intensity corresponding to mg / L). The larger xj is, the greater the contribution of the discharge outlet to the overall pollution. In this way, by standardizing data input, regularizing the objective function, data-driven optimization of λ, and efficient closed-form solution, the core defects of traditional regression algorithms, such as no solution, instability, and low efficiency, can be alleviated. The obtained pollution intensity estimation vector X of the discharge outlet is the direct basis for subsequent identification of the main pollution source. At the same time, through deep collaboration with the previous steps, data consistency and computational efficiency are ensured, thereby improving the efficiency of tracing the source of sewage in rivers and lakes.

[0072] Step S600: Extract the target sewage outlet corresponding to the largest component in the pollution intensity estimation vector of the sewage outlet, and determine the target sewage outlet as the main pollution source.

[0073] The pollution intensity estimation vector X of the sewage outlet includes several elements x. By comparing the magnitudes of all elements x, the sewage outlet corresponding to the element with the largest value is determined as the target sewage outlet, which is the main pollution source. This application, combined with the spatial analysis function of a geographic information platform, can reduce the need for manual labor in pollution source tracing, thereby effectively improving work efficiency, reducing the incidence of accidents and injuries, and enhancing safety.

[0074] Furthermore, after identifying the target discharge outlet as the primary pollution source, the process also includes: based on the edge directions of the updated hydrological topology network, traversing all downstream edges from the target discharge outlet to generate the pollution propagation path of the primary pollution source. The pollution propagation path, the location of the primary pollution source, and the largest component in the pollution intensity estimation vector of the discharge outlet are then used to generate a spatial feature file. In other words, this application supports pollution source tracing and pollution diffusion path simulation with limited monitoring data, and generates corresponding geospatial feature files that can be directly loaded and visualized by mainstream geographic information platforms, effectively contributing to the analysis and processing of downstream tasks.

[0075] Figure 4 This is a flowchart illustrating the river and lake pollution source tracing method provided in this application. For example... Figure 4 As shown, the first step is to construct the hydrological topology network and map the nodes. This includes building a graph network, determining the flow direction, and mapping the nodes. Specifically, the graph network is constructed first, then the flow direction is determined based on the graph network, and finally, the nodes are mapped.

[0076] Next, pollution diffusion and source tracing path modeling is performed, which includes downstream path modeling, upstream path modeling, and a two-way computational framework. Specifically, downstream path modeling is performed first, followed by upstream path modeling, and finally, these two models are integrated to obtain the two-way computational framework.

[0077] Finally, pollution source identification and path generation are performed. This includes creating a pollution intensity attenuation matrix, identifying the main pollution source, and generating pollution paths. Specifically, the pollution intensity attenuation matrix is ​​first determined, then the main pollution source is identified based on this matrix, and finally, the pollution path is generated. This application not only combines the spatial analysis functions of a geographic information platform, reducing the need for manual labor in pollution source tracing, thereby effectively improving work efficiency and reducing the incidence of accidents and injuries, and enhancing safety, but also supports the simulation of pollution source tracing and pollution diffusion paths with limited monitoring data, generating corresponding geospatial element files that can be directly loaded and visualized by mainstream geographic information platforms, effectively contributing to the analysis and processing of downstream tasks.

[0078] Figure 5 This is a schematic diagram of a river and lake pollution source tracing system provided in this application. (For example...) Figure 5 As shown, a river and lake pollution tracing system includes: a network construction module, a processing module, and a tracing module.

[0079] The system comprises several modules: a network construction module, a processing module, and a source tracing module. The network construction module acquires basic hydrological data of rivers and lakes, converts this data into a directed graph structure to construct a hydrological topology network, acquires raw data from monitoring points and sewage outlets, generates new nodes based on this data, and updates the hydrological topology network using these new nodes. The processing module constructs a bidirectional path calculation framework based on the updated hydrological topology network, obtains the hydrological path distance from the sewage outlet to the monitoring point, acquires the pollutant attenuation coefficient, and constructs a pollution intensity attenuation matrix representing the impact of the sewage outlet on the pollution intensity of the monitoring point based on the hydrological path distance and the pollutant attenuation coefficient. The source tracing module acquires the real-time pollution concentration at the monitoring point, determines the pollution intensity estimation vector of the sewage outlet based on the real-time pollution concentration and the pollution intensity attenuation matrix, extracts the target sewage outlet corresponding to the largest component in the pollution intensity estimation vector, and identifies the target sewage outlet as the main pollution source.

[0080] The other functions performed by the aforementioned network construction module, processing module, and source tracing module, as well as the technical details of each function, are the same as or similar to the corresponding features in the previously described method for tracing the source of river and lake pollution, and therefore will not be repeated here.

[0081] This application also provides a computer storage medium storing a computer program that, when run on a computer, enables the computer to execute the steps in the previously described method for tracing the source of river and lake pollution.

[0082] It should be understood that although the steps in the flowcharts in the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order requirement for the execution of these steps, and they can be performed in other orders.

[0083] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for tracing pollution of rivers and lakes, characterized by, The method comprises: obtaining hydrological basic data of rivers and lakes, converting the hydrological basic data into a directed graph structure to construct a hydrological topology network; obtaining original data of monitoring points and sewage outlets, generating new nodes based on the original data, and using the new nodes to complete the update of the hydrological topology network; based on the updated hydrological topology network, a bidirectional path calculation framework is constructed, and the hydrological path distance from the sewage outlet to the monitoring point is obtained based on the bidirectional path calculation framework; obtaining the pollutant attenuation coefficient, and based on the hydrological path distance and the pollutant attenuation coefficient, constructing a pollution intensity attenuation matrix representing the influence of the sewage outlet on the pollution intensity of the monitoring point; obtaining the real-time pollution concentration of the monitoring point, and based on the real-time pollution concentration and the pollution intensity attenuation matrix, determining the pollution intensity estimation vector of the sewage outlet; extracting the target sewage outlet corresponding to the maximum component in the pollution intensity estimation vector of the sewage outlet, and determining the target sewage outlet as the main pollution source.

2. The method of claim 1, wherein, The hydrological basic data includes river line elements, node spatial coordinates and digital elevation model data, and the conversion of the hydrological basic data into a directed graph structure to construct a hydrological topology network comprises: extracting the end point coordinates of all line segments in the river line elements to generate a node set of the directed graph structure; the start and end points of each river line element are respectively taken as the start point and end point of a directed edge to form an edge set of the directed graph structure; based on the digital elevation model data, the elevation value of each node in the node set is obtained, and if the elevation value of the start point of the directed edge is greater than the elevation value of the end point, the direction of the directed edge is determined as the downstream direction of the water flow; if the elevation value of the start point of the directed edge is less than the elevation value of the end point, the start and end points of the directed edge are adjusted so that the direction of the adjusted directed edge is the downstream direction of the water flow; based on the node set, the edge set and the direction of the directed edge, a hydrological topology network is constructed.

3. The method of claim 2, wherein, The original data includes monitoring spatial coordinates of monitoring points and sewage spatial coordinates of sewage points, and the generation of new nodes based on the original data comprises: calculating the vertical distance from the monitoring spatial coordinates and the sewage spatial coordinates to all edges in the hydrological topology network, and determining the edge corresponding to the minimum value in all vertical distances as the target adjacent edge; obtaining the target end point spatial coordinates contained in the target adjacent edge, and obtaining the projection coordinates of the monitoring point and the sewage outlet on the target adjacent edge by processing the target end point spatial coordinates, the monitoring spatial coordinates and the sewage spatial coordinates through the vertical projection formula; the target adjacent edge is divided into two new edges at the projection coordinates, and the projection coordinates are taken as new nodes.

4. The method of claim 2, wherein, The bidirectional path calculation framework includes a forward path mode and a reverse path mode, and the construction of the bidirectional path calculation framework based on the updated hydrological topology network comprises: obtaining the forward path mode and the reverse path mode in the updated hydrological topology network, wherein the enabling condition of the forward path mode is that the start point belongs to the node set of the sewage outlet, and the function is to simulate the propagation of pollutants along the downstream direction of the edge, and output the downstream path from the sewage outlet to each monitoring point, and the enabling condition of the reverse path mode is that the start point belongs to the node set of the monitoring point, and the function is to simulate the backtracking of pollutants along the upstream direction of the edge, and output the upstream path from the monitoring point to each sewage outlet; The downstream path from the pollution outlet to the monitoring point in the forward path mode and the upstream path from the monitoring point to the pollution outlet in the upstream path mode are associated to form a bidirectional path calculation framework.

5. The method of claim 4, wherein, The pollution intensity attenuation matrix representing the influence of the pollution outlet on the pollution intensity of the monitoring point is constructed based on the hydrological path distance and the pollutant attenuation coefficient, and includes: Suppose that the pollution outlet set is J = {j1, j2, …, jn} and the monitoring point set is I = {i1, i2, …, im}, where n is the number of pollution outlets and m is the number of monitoring points; A∈R m×n , the matrix element Aij represents the pollution intensity attenuation coefficient of the jth pollution outlet to the ith monitoring point; Based on the pollutant attenuation coefficient, the matrix element Aij = exp(-a×Lij) is calculated, where a is the pollutant attenuation coefficient and Lij is the hydrological path distance from the jth pollution outlet to the ith monitoring point. If there is no feasible downstream path from the jth pollution outlet to the ith monitoring point in the updated hydrological topology network, Aij = 0 is set.

6. The method of claim 5, wherein, The pollution intensity estimation vector of the pollution outlet is determined based on the real-time pollution concentration and the pollution intensity attenuation matrix, and includes: The real-time pollution concentration of the monitoring point is arranged in the order of the monitoring number to obtain a pollution intensity observation vector b, where b is an m*1 vector. The Bayesian ridge regression algorithm is adopted, a product of a pollution intensity attenuation matrix A and a pollution intensity estimation vector X of a pollution outlet and a residual error of a pollution intensity observation vector b are taken as optimization targets, and a target function is constructed Wherein, λ is a regularization parameter, and ‖.‖2 is an L2 norm. The regularization parameter λ is optimized by k-fold cross-validation, and the λ that minimizes the cross-validation error is selected opt ; Substitute λ opt into the closed-form solution of Bayesian ridge regression algorithm X = (A T A+λ opt I) -1 A T b, where I is an m-order identity matrix, to obtain the estimated pollution intensity vector of the pollution outlet.

7. The method of claim 6, wherein, The Bayesian ridge regression algorithm is also used to solve the ill-conditioned equation problem caused by the number of monitoring points being less than the number of pollution outlets, and includes: By the regularization parameter λ opt To the matrix A T A is modified so that the modified A T A + λ opt I is a positive definite matrix.

8. The method of claim 1, wherein, After the target pollution outlet is determined as the main pollution source, the pollution expansion path of the main pollution source is generated by starting from the target pollution outlet and traversing all downstream edges. A space element file is generated based on the pollution expansion path, the location of the main pollution source, and the maximum component in the pollution intensity estimation vector of the pollution outlet. The system includes a network construction module, a processing module, and a tracing module, wherein 9. A river and lake pollution traceability system characterized by, The network construction module is used to obtain hydrological basic data of rivers and lakes, convert the hydrological basic data into a directed graph structure to construct a hydrological topology network, obtain original data of monitoring points and pollution outlets, generate new nodes based on the original data, and update the hydrological topology network using the new nodes; The processing module is used to construct a bidirectional path calculation framework based on the updated hydrological topology network, obtain the hydrological path distance from the pollution outlet to the monitoring point based on the bidirectional path calculation framework, obtain the pollutant attenuation coefficient, and construct a pollution intensity attenuation matrix representing the influence of the pollution outlet on the pollution intensity of the monitoring point based on the hydrological path distance and the pollutant attenuation coefficient; The tracing module is used to obtain the real-time pollution concentration of the monitoring point, determine the pollution intensity estimation vector of the pollution outlet based on the real-time pollution concentration and the pollution intensity attenuation matrix, extract the target pollution outlet corresponding to the maximum component in the pollution intensity estimation vector of the pollution outlet, and determine the target pollution outlet as the main pollution source. The computer program is executed by the processor to implement a river and lake pollution tracing method according to any one of claims 1 to 8.

10. A computer readable storage medium having stored thereon a computer program, capable of running on a processor, characterized in that, ​

Images