Intelligent river governance monitoring method and system based on complex network and application of intelligent river governance monitoring method and system

Abstract

The invention discloses a river intelligent management monitoring method and system based on a complex network and an application thereof.According to the scheme, ground-air-sky monitoring linkage is ingeniously adopted, a ground sensor set is used as a normal state monitoring part, when monitoring feedback is abnormal, air / sky observation is triggered, and the real-time monitoring effect is achieved; in the aspect of data acquisition, a time-varying relation graph is constructed for de-exogenous sensor sequence data, and the overall abnormity of the network is effectively discovered based on the transition detection of correlation, partial correlation and mutual information matrixes and corresponding baselines; the abnormal monitoring nodes in the river network can be quickly positioned in combination with calculation of the node anomaly degree; according to the scheme, a self-adaptive monitoring closed loop of'commanding air / sky by the ground 'is realized, the advantages of high precision and high frequency of a ground sensor are exerted, planar coverage and quick response of a remote sensing technology are also considered, and the real-time performance, the accuracy and the resource utilization efficiency of river pollution monitoring are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to river monitoring technology, computational model application technical field, and more particularly to a river intelligent governance monitoring method and system based on complex networks and applications thereof. BACKGROUND

[0002] Traditional river pollution monitoring mainly relies on ground in-situ sensors or periodic manual sampling. Although this method has high quantitative accuracy, it has limited spatial coverage and is difficult to discover sudden pollution of tributaries or upstream sewage outlets in time. Moreover, its minute-hour level time resolution cannot fully capture the short-term peak value of pollution pulse emissions. In recent years, due to the advantages of large-scale and rapid acquisition of area information, unmanned aerial vehicles and satellite remote sensing technologies have been applied in water environment monitoring, but they are easily limited in cloudy, rainy, night, shaded or extremely narrow river conditions. Moreover, remote sensing products are mostly "optically active" indicators such as turbidity and chlorophyll, which are difficult to directly invert the accurate values of "optically inactive" pollutant concentrations such as COD and ammonia nitrogen.

[0003] However, due to the high cost of satellite remote sensing and unmanned aerial vehicle photography, the existing technology still lacks a real-time adaptive monitoring scheme that can organically couple ground high-precision continuous monitoring with air-space remote sensing wide-range coverage, so that both high-frequency data of sensor groups can be used to capture pollution pulses in time, and aerial photography / satellite can be accurately guided when necessary to quickly supplement area information with the least cruising task, thereby constructing a "point-surface-network" integrated monitoring closed loop. Without such a mechanism, it is often necessary to rely too much on ground deployment and ignore the overall watershed risk, or blindly launch aerial photography and satellite missions, wasting resources and missing the best opportunity to take evidence. SUMMARY

[0004] Therefore, the present application aims to provide a river intelligent governance monitoring method and system based on complex networks and applications thereof, which is reliable in implementation, flexible in application, and can initiate unmanned aerial vehicle photography and satellite remote sensing monitoring in a timely manner according to the monitoring situation to comprehensively optimize the scheduling of resources.

[0005] In order to achieve the above technical purposes, the technical scheme adopted by the present application is as follows: A river intelligent governance monitoring method based on complex networks, a sensor group for monitoring water quality, environmental conditions and / or pollutants is arranged on the main stream and tributaries in the preset river basin, which comprises: S01, acquiring multi-source monitoring data in the river basin by the sensor group, and generating a monitoring data sequence after collecting and preprocessing the multi-source monitoring data; S02, constructing a river network weighted directed graph GThe sensor group in-sensor relationship of the monitoring data sequence is measured by a sliding time window to obtain a correlation matrix, a partial correlation matrix and / or a mutual information matrix, and then a corresponding baseline matrix is estimated by using long-period data; S03, calculating a graph structure transition degree and a node anomaly degree based on the correlation matrix, the partial correlation matrix and the corresponding baseline matrix; S04, when one or more of the graph structure transition degree and the node anomaly degree exceed a corresponding preset threshold, determining that an abnormal event exists, and then triggering an air / space observation task generation; S05, according to the air / space observation task, scheduling a UAV and / or a satellite to collect data in a corresponding river basin area.

[0006] As a possible implementation, further, in the scheme S01, the multi-source monitoring data includes one or more of water quality data, hydrological data, environmental data and working condition data.

[0007] The water quality data includes one or more of COD, ammonia nitrogen, dissolved oxygen DO, pH, conductivity, turbidity and water temperature of the area where the sensor group is located.

[0008] The hydrological data includes one or more of water level, water flow velocity and flow of the area where the sensor group is located.

[0009] The environmental data includes one or more of rainfall, wind speed and wind direction of the area where the sensor group is located.

[0010] The working condition data includes pump station flow, water area gate operation and / or water intake of the area where the sensor group is located.

[0011] In order to facilitate tracking, the sensor groups deployed in the river according to the scheme all have unique IDs, and the data sequences generated by the sensor groups are also associated with the IDs of the sensor groups. In addition, the ID of each sensor group is also associated with the location information of its corresponding deployment.

[0012] In order to facilitate the tracking of past situations, in the scheme S01, when collecting data, the historical monitoring data of the area where the sensor group is located is also associated with it.

[0013] In S01, the preprocessing includes time alignment, outlier rejection, missing value filling processing of the multi-source monitoring data collected by the sensor group, and de-externalization processing of part of the data according to a preset condition; the de-externalization processing is to establish an external variable feature matrix and then remove the influence of the external variable on the corresponding data collected by each sensor group based on the external variable, which is defined as follows:

[0014] in, This refers to multi-source monitoring data after time alignment, outlier removal, and missing value imputation. This refers to the multi-source monitoring data after removing interference from external variables, i.e., the preprocessed monitoring data sequence. is the ridge regression coefficient.

[0015] As a preferred implementation method, preferably, in this scheme S02, the river network weighted directed graph... G Represented as ,in, It is a set of nodes, which contains nodes that represent the areas corresponding to the sensor groups in the watershed, i.e., monitoring points; It is a set of directed edges, which contain directed edges that represent the hydraulic flow direction between nodes, that is, the possible paths for upstream nodes to migrate to downstream nodes. is a weight function that corresponds to the directed edges in the directed edge set. It is used to quantify the connectivity strength or propagation probability between upstream nodes and downstream nodes.

[0016] As a preferred implementation method, preferably, in scheme S02, the weighted directed graph of the river network is constructed. G Subsequently, monitoring data sequences For the input item, where, Number the nodes. , Weighted directed graph of river network G The set of nodes, i.e., the total number of monitoring points; , This represents the sampling time of the sensor group.

[0017] The sliding window length is defined as follows: Step size is Then the first The data time period corresponding to the monitoring data sequence for each window is: ; data segments corresponding to the sliding time window The definition is as follows: ; Node set Correlation matrix of sensor group corresponding to middle node The definition is as follows:

[0018] in, , The first Nodes within a window , Data fragments, is a rank transform function, is a Pearson correlation function; is a rank transform function, is a rank transform function, , is a correlation matrix value between data segments of nodes ,

[0019] In the present scheme, the partial correlation matrix is estimated by using data segments under a sliding time window to estimate the covariance , which is defined as follows:

[0020] wherein, is a data segment of the th window, is a data matrix obtained by averaging the monitoring data sequence of the th window by column, is a sliding window length, is a data segment of the th window, is a covariance corresponding to the data segment of the th window.

[0021] The present scheme also inverts the covariance to obtain a precision matrix , which is defined as: and then calculates the partial correlation matrix , which is defined as follows:

[0022] wherein, , which represents the net linear dependence strength between node and node , i.e., the partial correlation value, is a correlation matrix value of the precision matrix at the th row and the th column, i.e., the inverse matrix of the linear covariance between node and node in the th window ; , are the diagonal elements of the precision matrix at the th and the th, respectively, which respectively represent the inverse of the conditional contrast of node and node after excluding the influence of other nodes; The value range of

[0023] In the present scheme, the mutual information matrix is a data segment The kNN proximity algorithm estimation is performed, which is defined as follows:

[0024] wherein, is the mutual information estimation value between the node and the node in the first window, is the derivative of the gamma function, i.e., the Digamma function, is the number of neighbors in the kNN proximity algorithm, is the length of the sliding window, i.e., the total number of nodes, is the neighbor number in the kNN proximity algorithm, , respectively represent the neighbor count in the one-dimensional projection space of the node , the node .

[0025] The scheme selects a time window with a length as a baseline window, which is longer than the sliding window , and performs an arithmetic average on the relationship matrix of each sliding window in the time period of the baseline window to obtain the baseline matrix corresponding to the correlation matrix, the partial correlation matrix and the mutual information matrix , , .

[0026] As a preferred selection implementation, preferably, the scheme S03 comprises: The correlation matrix , the partial correlation matrix and / or the mutual information matrix at time t are calculated with the baseline matrix , , estimated by the long-period data to obtain the corresponding global transition degree, i.e., the graph structure transition degree, and the calculation formula is as follows:

[0027]

[0028]

[0029] wherein, is the Frobenius norm, , , respectively are the correlation matrix , the partial correlation matrix and / or the mutual information matrix The global transition degree is used to measure the overall deviation of the linear synchronization of all node pairs in a weighted directed graph of a river network. G When several key edges in the middle suddenly rise, , , The corresponding increase serves as a notification of an anomaly in the overall network.

[0030] In this scheme, the calculation of node anomaly includes calculating the row norm and maximum deviation for each node, and the formula is as follows:

[0031] in, , Number the nodes. For about nodes row norm, For about nodes The maximum deviation, For node at time t , The correlation matrix values, For nodes , The baseline matrix values, This represents the total number of nodes, i.e., the total number of monitoring points.

[0032] As a preferred implementation method, solution S04 preferably includes: The graph structure transition degree and node anomaly degree of the nodes corresponding to the sensor group are compared and judged according to the preset threshold. When one or more of the graph structure transition degree and node anomaly degree exceed the corresponding preset threshold, it is determined that there is an abnormal event, the sensor group ID of the node corresponding to the abnormal event is recorded, and at the same time, the generation of the space / sky observation task is triggered.

[0033] As an example, the correlation matrix Partial correlation matrix Mutual information matrix Corresponding global transition degree , , The system compares the data with the corresponding preset thresholds. If one of the thresholds exceeds the preset threshold, an abnormal event is identified. The sensor group ID of the node corresponding to the abnormal event is recorded, and an air / sky observation task is triggered.

[0034] As a preferred implementation method, solution S05 preferably includes: The number of nodes recording abnormal events is counted, and when the number of nodes exceeds a preset number value, the unmanned aerial vehicle and the satellite are dispatched to jointly collect data in the corresponding river basin area; when the number of nodes is less than or equal to the preset number value, the unmanned aerial vehicle is dispatched to collect data in the corresponding river basin area.

[0035] As a preferred selection implementation, preferably, in the scheme S05, when the unmanned aerial vehicle is selected to collect data, the position of the sensor group ID is obtained, then the flight trajectory of the unmanned aerial vehicle is planned, and the flight route information is generated, and the unmanned aerial vehicle collects data according to the flight route information.

[0036] As a preferred selection implementation, preferably, in the scheme, when the flight trajectory of the unmanned aerial vehicle is planned, the flight path is minimized, which includes the following: The node information of the recorded abnormal event is collected, the two-dimensional position information of the corresponding sensor group in the river is obtained according to the node information, and then the Euclidean distance between different nodes is calculated, which is defined as follows:

[0037] Among them, The distance from node to node , , The coordinate value of node in two-dimensional space, , The coordinate value of node in two-dimensional space.

[0038] As a preferred selection implementation, preferably, the objective function of the scheme is defined as follows when the flight path is minimized:

[0039] Among them, The number of nodes recording abnormal events, The path length, , The node number, The distance from node to node , It is a binary degree constraint variable, which is used to ensure that the distance between node and the remaining nodes is calculated once.

[0040] Based on the above, the scheme also proposes a river intelligent governance monitoring system based on complex network, which includes: A monitoring terminal management unit is in communication connection with a plurality of sensor groups, the plurality of sensor groups are respectively arranged on the main stream and tributaries in the preset river basin, and are used for monitoring water quality, environmental conditions and / or pollutants; the monitoring terminal management unit obtains multi-source monitoring data in the river basin through the sensor groups, collects and pre-processes the multi-source monitoring data, and generates a monitoring data sequence; A complex network unit is configured to construct a river network weighted directed graph according to the monitoring data sequence G The sensor group intra-group relationship of the monitoring data sequence is measured by using a sliding time window, to obtain a correlation matrix, a partial correlation matrix and / or a mutual information matrix, and then a corresponding baseline matrix is estimated by using long-period data; A data judgment unit is configured to calculate a graph structure transition degree and a node anomaly degree based on the correlation matrix, the partial correlation matrix and the corresponding baseline matrix; A task generation unit is configured to determine that an abnormal event exists when one or more of the graph structure transition degree and the node anomaly degree exceeds a corresponding preset threshold value, and then trigger generation of a space / sky observation task; A work scheduling module is configured to schedule a drone and / or a satellite to collect data in a corresponding basin area of the river according to the space / sky observation task.

[0041] Based on the above, the present scheme further proposes a river remote monitoring management method, which applies the above-mentioned river intelligent governance monitoring method based on a complex network; the method comprises the following steps: A01, receiving a monitoring instruction about pollution of the river, and performing S01-S05; A02, obtaining the data collected in S05, and generating a monitoring management report.

[0042] Compared with the prior art, the application has the beneficial effects that: the scheme ingeniously provides monitoring guarantee for the monitoring node area in the river basin through the monitoring linkage of ground-air-space, uses the ground sensor group as the normal monitoring part, triggers air / space observation when the monitoring feedback occurs abnormally, provides on-site monitoring basis for river intelligent monitoring, and effectively discovers the overall network anomaly through the construction of the time-varying relationship graph based on the de-externalized sensor sequence data and the transition detection based on the correlation, partial correlation and mutual information matrix and the corresponding baseline; in combination with the calculation of the node anomaly degree, the abnormal monitoring node in the river network can be quickly located; by counting the number of recorded abnormal events, the scheme optimizes the monitoring cost, improves the pertinence and minimizes the flight path on the basis, plans the unmanned aerial vehicle flight route and timely schedules satellite remote sensing for monitoring data collection, compared with the prior art, the scheme realizes the consideration of the timeliness and accuracy of the river intelligent governance monitoring, triggers aerial / satellite only when necessary, ensures the spatial and temporal alignment of the areal information and "point" data, greatly improves the resource utilization efficiency, at the same time, the scheme is positioned based on the global transition degree and the node level anomaly degree, can significantly reduce the interested area and avoid invalid cruising; and the flight route is planned based on the Euclidean shortest path model, which can reduce the flight mileage and the number of tasks. The scheme realizes the adaptive monitoring closed loop of "letting the ground command the air / space", which not only gives play to the high precision and high frequency advantages of the ground sensor, but also takes into account the areal coverage and rapid response of the remote sensing technology, and significantly improves the real-time performance, accuracy and resource utilization efficiency of the river pollution monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, a brief introduction will be given to the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creating labor.

[0044] Figure 1 is a brief implementation flowchart of the monitoring method of the scheme; Figure 2 is a brief schematic diagram of the scheme monitoring method, in which the sensor groups are deployed on the main stream and tributaries of the river to form monitoring nodes (schematically shown as dots in the figure); Figure 3 is a flight trajectory (schematically shown as a dashed line in the figure) of the unmanned aerial vehicle for aerial data collection above the river in the monitoring method of the scheme; Figure 4 is a brief connection diagram of the unit modules of the monitoring system of the scheme. DETAILED DESCRIPTION

[0045] The application will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly pointed out that the following examples are only for illustration of the application, but do not limit the scope of the application. Similarly, the following examples are only part of the embodiments of the application, not all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of the application.

[0046] As shown in the embodiment, the scheme is a river intelligent governance monitoring method based on a complex network. The main stream and tributaries in the river basin are each deployed with a sensor group for monitoring water quality, environmental conditions and / or pollutants (see FIG. 2). Figure 1 The sensor group includes: Figure 2 S01, obtaining multi-source monitoring data in the river basin through the sensor group, and generating a monitoring data sequence after collecting and preprocessing the multi-source monitoring data; S02, constructing a river network weighted directed graph S03, measuring the intra-group relationship of the sensor group in the monitoring data sequence by a sliding time window to obtain a correlation matrix, a partial correlation matrix and / or a mutual information matrix, and then estimating a corresponding baseline matrix using long-period data; G S04, calculating the graph structure transition degree and the node anomaly degree based on the correlation matrix, the partial correlation matrix and the corresponding baseline matrix; S05, when one or more of the graph structure transition degree and the node anomaly degree exceed the corresponding preset threshold, determining that an abnormal event exists, and then triggering the generation of an air / space observation task; S06, scheduling a drone and / or a satellite to collect data in the corresponding river basin area according to the air / space observation task. The scheme obtains multi-source monitoring data in the river basin through the sensor group, which serves as daily monitoring information to provide on-site data support for the water quality of each node monitoring position in the river for the background management personnel. The constructed river network weighted directed graph helps to associate the nodes where the sensor groups are deployed in the river network, providing correlation judgment for subsequent integrated analysis.

[0047] For the scenario of abnormality occurring in the river, the monitoring nodes in the corresponding area will have abnormal monitoring data. By calculating the graph structure transition degree and the node anomaly degree of the correlation matrix, the partial correlation matrix and the corresponding baseline matrix formed by the monitoring data sequence obtained by the node, quantitative analysis can be realized on the data. When one or more of the graph structure transition degree and the node anomaly degree exceed the corresponding preset threshold, it is determined that an abnormal event exists, and then the air / space observation task is triggered to generate, providing linkage measure guarantee for more detailed and global river condition monitoring in the next step.

[0048] For the scenario of abnormality occurring in the river, the monitoring nodes in the corresponding area will have abnormal monitoring data. By calculating the graph structure transition degree and the node anomaly degree of the correlation matrix, the partial correlation matrix and the corresponding baseline matrix formed by the monitoring data sequence obtained by the node, quantitative analysis can be realized on the data. When one or more of the graph structure transition degree and the node anomaly degree exceed the corresponding preset threshold, it is determined that an abnormal event exists, and then the air / space observation task is triggered to generate, providing linkage measure guarantee for more detailed and global river condition monitoring in the next step.

[0049] In terms of data collection types, as a possible implementation, further, in the present scheme S01, the multi-source monitoring data includes one or more of water quality data, hydrological data, environmental data, and working condition data.

[0050] The water quality data includes one or more of COD, ammonia nitrogen, dissolved oxygen DO, pH, conductivity, turbidity, and water temperature of the region corresponding to the sensor group.

[0051] The hydrological data includes one or more of water level, water flow speed, and flow of the region corresponding to the sensor group.

[0052] The environmental data includes one or more of rainfall, wind speed, and wind direction of the region corresponding to the sensor group.

[0053] The working condition data includes pump station flow, water area gate operation, and / or water intake of the region corresponding to the sensor group.

[0054] For convenience of tracking, the sensor group deployed in the river in the present scheme has a unique ID, and the data sequence generated by the sensor group is also associated with the ID of the sensor group. In addition, the ID of each sensor group is also associated with the location information of its corresponding deployment.

[0055] For convenience of tracing past situations, in the present scheme S01, when collecting data, the historical monitoring data of the region corresponding to the sensor group is also associated.

[0056] In S01, the preprocessing includes time alignment, outlier rejection, and missing value filling of the multi-source monitoring data collected by the sensor group, and de-externalization of part of the data according to the preset conditions; the de-externalization processing is to establish an external variable feature matrix and then remove the influence of external variables on the corresponding data collected by each sensor group, which is defined as follows:

[0057] wherein, is the multi-source monitoring data after time alignment, outlier rejection, and missing value filling, is the multi-source monitoring data after removing the interference of external variables, i.e., the monitoring data sequence after preprocessing; is the ridge regression coefficient.

[0058] In terms of complex network construction, as a relatively optimal selection implementation, preferably, in the present scheme S02, the river network weighted directed graph G is represented as wherein, is a node set, which contains nodes representing the areas corresponding to the sensor groups in the river basin, i.e. monitoring points; is a directed edge set, which contains directed edges representing the hydraulic flow direction between nodes, i.e. the possible paths of migration from the upstream node to the downstream node; is a weight function corresponding to the directed edges of the directed edge set, used to quantify the connectivity strength or propagation possibility between the upstream node and the downstream node.

[0059] On the basis of the above, as a preferred implementation, preferably, in the scheme S02, the river network weighted directed graph is constructed by G monitoring data sequence as input items, wherein, is the node number, , is the node set of the river network weighted directed graph, i.e. the total number of monitoring points; G , , is the sampling time of the sensor group.

[0060] By establishing the river network weighted directed graph, it is helpful to associate the monitoring nodes corresponding to the sensor groups dispersed in the river network, providing guarantee for subsequent data analysis.

[0061] wherein the sliding window length is defined as , the step is , the data period of the monitoring data sequence corresponding to the th window is ; the data segment under the corresponding sliding time window is defined as follows: ; The correlation matrix of the sensor group corresponding to the node pair in the node set is defined as follows:

[0062] wherein, , are the data segments of the nodes , in the th window, respectively, is a rank transformation function, is a Pearson correlation function; is the correlation matrix value between the data segments of the nodes , in the th window.

[0063] ​This scheme uses data segments within a sliding time window for calculating the partial correlation matrix. Estimate covariance Its definition is as follows:

[0064] in, For the first Data fragments of a window, For the first Monitoring data sequence of each window The data matrix after column averaging The length of the sliding window. For the first The covariance corresponding to the data fragments in each window.

[0065] Based on the above, this scheme also inverses the covariance matrix to obtain the accuracy matrix. Its definition is: Then calculate the partial correlation matrix. Its definition is as follows:

[0066] in, , which represents a node ,node The net linear dependence strength, i.e., the partial correlation value, For the accuracy matrix in the th Okay, number The correlation matrix values ​​of the columns, i.e., the nodes ,node In the The inverse matrix of the linear covariance of each window ; , The precision matrix is ​​located on the diagonal. , No. Each element represents a node. ,node The reciprocal of its own conditional contrast after removing the influence of other nodes; The value range is [-1, 1].

[0067] In this scheme, the mutual information matrix To analyze data segments within a sliding time window The kNN neighbor algorithm is used for estimation, and its definition is as follows:

[0068] in, In the first Within each window, nodes the mutual information estimation value between nodes , is the derivative of the gamma function, i.e. the Digamma function, is the number of neighbors in the kNN proximity algorithm, is the length of the sliding window, i.e. the total number of nodes, is the neighbor number in the kNN proximity algorithm, , respectively represent the neighbor count in the one-dimensional projection space of the node , the node .

[0069] The scheme selects a time window of length as a baseline window, which is longer than the sliding window , and makes an arithmetic average of the relationship matrix of each sliding window in the time period of the baseline window to obtain the baseline matrix , , .

[0070] In the scheme, the graph structure transition degree is used to express the overall abnormal situation in the river network, which helps the background management personnel to know the global situation in the river network. As a preferred implementation manner, the scheme S03 comprises: The correlation matrix , the partial correlation matrix and / or the mutual information matrix at time t are calculated with the baseline matrix , , obtained by long-period data estimation to obtain the corresponding global transition degree, i.e. the graph structure transition degree, and the calculation formula is as follows:

[0071]

[0072]

[0073] wherein, is the Frobenius norm, , , are respectively the global transition degrees of the correlation matrix , the partial correlation matrix and / or the mutual information matrix , which are used to measure the overall deviation degree of linear synchronization of all node pairs. When some key edges in the river network weighted directed graph G rise sharply, 、 、 Corresponding rise, to prompt the network global emergence of abnormal events.

[0074] In the scheme, the node anomaly degree is used to express the monitoring situation of the node area where the sensor group is located. When the background management personnel learns that the global abnormal event occurs, the node anomaly degree can be used to quickly locate the abnormal source and the trajectory of its development and evolution. The calculation of the node anomaly degree includes the calculation of the row norm and the maximum deviation of each node, and the formula is as follows:

[0075] Wherein, 、 is the node number, is the row norm of the node , is the maximum deviation of the node , is the correlation matrix value of the node 、 at time t, is the baseline matrix value of the node 、 , is the total number of nodes, that is, the total number of monitoring points.

[0076] In the air-space cooperation aspect, as a relatively optimal selection implementation manner, preferably, the scheme S04 comprises: The node corresponding to the sensor group is compared and judged according to the preset threshold value of the graph structure transition degree and the node anomaly degree. When one or more of the graph structure transition degree and the node anomaly degree exceeds the corresponding preset threshold value, it is judged that there is an abnormal event, the sensor group ID of the node corresponding to the abnormal event is recorded, and at the same time, the air / space observation task is triggered to generate.

[0077] As a relatively optimal selection implementation manner, preferably, the scheme S05 comprises: The number of nodes of the recorded abnormal event is counted. When the number of nodes exceeds the preset number value, the unmanned aerial vehicle and the satellite are dispatched to jointly collect data on the corresponding river basin area of the river; when the number of nodes is less than or equal to the preset number value, the unmanned aerial vehicle is dispatched to collect data on the corresponding river basin area of the river.

[0078] As an example, for example, when all the nodes are 100, the number of nodes of the recorded abnormal event is 21, which exceeds the preset first threshold value of 20, then the unmanned aerial vehicle and the satellite are dispatched to jointly collect data on the corresponding river basin area of the river. When the number of nodes of the recorded abnormal event is 19, the unmanned aerial vehicle is dispatched to collect data on the corresponding river basin area of the river.

[0079] The scheme can improve the reliability of the river air cruise monitoring by flexibly selecting the unmanned aerial vehicle to collect data in the corresponding river basin area according to the recorded abnormal events. When a large area of abnormality occurs, satellite remote sensing is used to collect data, which can provide timely, reliable and diversified data reference for the background supervisor, and provide a data reference basis for formulating intervention measures, thereby improving the efficiency and pertinence of monitoring response.

[0080] In the scheme, since the river network generally has a large span, in order to improve the efficiency and flexibility of the unmanned aerial vehicle aerial photography and to improve the amount of effective data collected by single cruise, as a preferred selection implementation manner, in the scheme S05, when the unmanned aerial vehicle is selected to collect data, the position of the sensor group ID is obtained, and then the unmanned aerial vehicle navigation track is planned to generate navigation route information, and the unmanned aerial vehicle collects data according to the navigation route information.

[0081] For path planning, the unmanned aerial vehicle navigation track is planned to minimize the navigation path, which includes the following: The node information of the recorded abnormal event is collected, the two-dimensional position information of the corresponding sensor group in the river is obtained according to the node information, and then the Euclidean distance between different nodes is calculated, which is defined as follows:

[0082] Among them, is the distance from node to node , , is the coordinate value of node in two-dimensional space, , is the coordinate value of node in two-dimensional space.

[0083] As a preferred selection implementation manner, the objective function of the scheme is defined as follows for the purpose of minimizing the navigation path:

[0084] Among them, is the number of nodes recorded abnormal events, is the path length, , is the node number, is the distance from node to node , is a binary degree constraint variable, which is used to ensure that the distance between node and the remaining nodes is calculated once.

[0085] wherein, Figure 3 An example of path planning when taking aerial photography by a UAV is shown.

[0086] Based on the above, the present scheme also proposes a river remote monitoring management method, which applies the river intelligent governance monitoring method based on complex network as described above; it includes: A01, receiving monitoring instructions about pollution occurring in the river, executing S01-S05; A02, obtaining the data collected in S05, and generating a monitoring management report.

[0087] In combination Figure 4 As shown, based on the above, the present scheme also proposes a river intelligent governance monitoring system based on complex network, which includes: A monitoring terminal management unit is in communication with a plurality of sensor groups, which are respectively deployed on the main stream and tributaries in the river's pre-set drainage basin, and are used to monitor water quality, environmental conditions and / or pollutants; the monitoring terminal management unit obtains multi-source monitoring data in the river basin through the sensor groups, collects and pre-processes them to generate a monitoring data sequence; A complex network unit is used to construct a river network weighted directed graph according to the monitoring data sequence G The sensor group's intra-group relationship of the monitoring data sequence is measured by a sliding time window to obtain a correlation matrix, a partial correlation matrix and / or a mutual information matrix, and then a corresponding baseline matrix is estimated using long-period data; A data judgment unit is used to calculate the graph structure transition degree and node anomaly degree based on the correlation matrix, the partial correlation matrix and its corresponding baseline matrix; A task generation unit is used to determine the existence of an abnormal event when one or more of the graph structure transition degree and the node anomaly degree exceeds the corresponding preset threshold, and then trigger the generation of space / sky observation tasks; A work scheduling module is used to schedule the data collection of the corresponding drainage area of the river by the UAV and / or satellite according to the space / sky observation task.

[0088] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0089] If the integrated unit is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0090] The above only describes some embodiments of the present application, and does not limit the protection scope of the present application. Any equivalent device or equivalent process transformation, or direct or indirect application in other related technical fields based on the content of the present application specification and drawings are also included in the patent protection scope of the present application.

Claims

1. A smart river management and monitoring method based on complex networks, wherein sensor arrays for monitoring water quality, environmental conditions, and / or pollutants are deployed on the main stream and tributaries of a pre-defined river basin, characterized in that, It includes: S01. Acquire multi-source monitoring data within the river basin through sensor arrays, collect and preprocess the data to generate a monitoring data sequence; S02, Construct a weighted directed graph of the river network. G The intra-group relationship of the monitoring data sequence is measured by a sliding time window to obtain the correlation matrix, partial correlation matrix and / or mutual information matrix, and then the corresponding baseline matrix is ​​estimated using long-period data. S03. Calculate the graph structure transition degree and node anomaly degree based on the correlation matrix, partial correlation matrix and their corresponding baseline matrix; S04. When one or more of the graph structure transition degree and node anomaly degree exceed the corresponding preset threshold, it is determined that there is an abnormal event, and then the generation of the space / sky observation task is triggered. S05. Based on the air / space observation mission, schedule drones and / or satellites to collect data on the corresponding watershed area of ​​the river.

2. The intelligent river management and monitoring method based on complex networks as described in claim 1, characterized in that, In S01, the multi-source monitoring data includes one or more of the following: water quality data, hydrological data, environmental data, and operating condition data. The water quality data includes one or more of the following: COD, ammonia nitrogen, dissolved oxygen (DO), pH, conductivity, turbidity, and water temperature in the area corresponding to the sensor group. The hydrological data includes one or more of the following: water level, water flow velocity, and flow rate in the area corresponding to the sensor group. The environmental data includes one or more of the following: rainfall, wind speed, and wind direction in the area corresponding to the sensor group. The operating data includes the pumping station flow rate, water gate operation, and / or water intake in the area corresponding to the sensor group; Each sensor group deployed in the river has a unique ID, and the corresponding data sequence it generates is also associated with the sensor group's ID. In S01, when collecting data, the historical monitoring data of the area corresponding to the sensor group is also associated with it; In S01, the preprocessing includes time alignment, outlier removal, missing value filling, and de-exogenization of some data according to preset conditions for the multi-source monitoring data collected by the sensor group. The de-exogenization process involves establishing a feature matrix of exogenous variables. Then, based on the external variables, the data collected by each sensor group are processed to eliminate the influence of the external variables, as defined below: in, This refers to multi-source monitoring data after time alignment, outlier removal, and missing value imputation. This refers to the multi-source monitoring data after removing interference from external variables, i.e., the preprocessed monitoring data sequence. is the ridge regression coefficient.

3. The intelligent river management and monitoring method based on complex networks as described in claim 1 or 2, characterized in that, In S02, the weighted directed graph of the river network G Represented as ,in, It is a set of nodes, which contains nodes that represent the areas corresponding to the sensor groups in the watershed, i.e., monitoring points; It is a set of directed edges, which contain directed edges that represent the hydraulic flow direction between nodes, that is, the possible paths for upstream nodes to migrate to downstream nodes. is the weight function, which corresponds to the directed edges of the directed edge set, and is used to quantify the connectivity strength or propagation probability between upstream nodes and downstream nodes. In S02, the weighted directed graph of the river network is constructed. G Subsequently, monitoring data sequences For the input item, where, Number the nodes. , Weighted directed graph of river network G The set of nodes, i.e., the total number of monitoring points; , The sampling time of the sensor group; Define the sliding window length as Step size is Then the first The data time period corresponding to the monitoring data sequence for each window is: ; data segments corresponding to the sliding time window The definition is as follows: ； Node set Correlation matrix of sensor group corresponding to middle node The definition is as follows: in, , The first Nodes within a window , Data fragments, It is a rank transformation function. The Pearson correlation function; For the first Nodes within a window , The correlation matrix values ​​between data fragments; When calculating the partial correlation matrix, data segments under a sliding time window are used. Estimate covariance Its definition is as follows: in, For the first Data fragments of a window, For the first Monitoring data sequence of each window The data matrix after column averaging The length of the sliding window. For the first The covariance of each window's data segment; Inverse the covariance matrix to obtain the precision matrix. Its definition is: Then calculate the partial correlation matrix. Its definition is as follows: in, , which represents a node ,node The net linear dependence strength, i.e., the partial correlation value, For the accuracy matrix in the th Okay, number The correlation matrix values ​​of the columns, i.e., the nodes ,node In the The inverse matrix of the linear covariance of each window ; , The precision matrix is ​​located on the diagonal. , No. Each element represents a node. ,node The reciprocal of its own conditional contrast after removing the influence of other nodes; The value range is [-1, 1]; The mutual information matrix To analyze data segments within a sliding time window The kNN neighbor algorithm is used for estimation, and its definition is as follows: in, In the first Within each window, nodes With nodes Mutual information estimates between them This is the derivative of the gamma function, i.e., the Digamma function. The number of neighbors in the kNN nearest neighbor algorithm. The length of the sliding window, i.e., the total number of nodes. The neighbor numbers in the kNN nearest neighbor algorithm. , These represent the nodes respectively. ,node Counting neighbors in a one-dimensional projected space; By selecting length The time window is used as the baseline window, and its length is greater than that of the sliding window. Within the time period of the baseline window, the arithmetic mean of the relation matrices for each sliding window is calculated to obtain the baseline matrices corresponding to the correlation matrix, partial correlation matrix, and mutual information matrix. , , .

4. The intelligent river management and monitoring method based on complex networks as described in claim 3, characterized in that, S03 includes: The correlation matrix at time t Partial correlation matrix and / or mutual information matrix Baseline matrix obtained from long-period data estimation , , The calculation is performed to obtain the corresponding global transition degree, i.e., the graph structure transition degree, and the calculation formula is as follows: in, It is the Frobenius norm. , , These are related to the correlation matrix. Partial correlation matrix and / or mutual information matrix The global transition degree is used to measure the overall deviation of the linear synchronization of all node pairs in a weighted directed graph of a river network. G When several key edges in the middle suddenly rise, , , The corresponding increase serves as a notification of an anomaly in the overall network. The calculation of node anomaly includes calculating the row norm and maximum deviation for each node, using the following formula: in, , Number the nodes. For about nodes row norm, For about nodes The maximum deviation, For node at time t , The correlation matrix values, For nodes , The baseline matrix values, This represents the total number of nodes, i.e., the total number of monitoring points.

5. The intelligent river management and monitoring method based on complex networks as described in claim 4, characterized in that, S04 includes: The graph structure transition degree and node anomaly degree of the nodes corresponding to the sensor group are compared and judged according to the preset threshold. When one or more of the graph structure transition degree and node anomaly degree exceed the corresponding preset threshold, it is determined that there is an abnormal event, the sensor group ID of the node corresponding to the abnormal event is recorded, and at the same time, the generation of the space / sky observation task is triggered.

6. The intelligent river management and monitoring method based on complex networks as described in claim 5, characterized in that, S05 includes: The system counts the number of nodes that record abnormal events. When the number of nodes exceeds a preset value, it coordinates drones and satellites to jointly collect data on the corresponding watershed area of ​​the river. When the number of nodes is less than or equal to the preset value, it coordinates drones to collect data on the corresponding watershed area of ​​the river.

7. The intelligent river management and monitoring method based on complex networks as described in claim 6, characterized in that, In S05, when selecting a drone for data collection, the deployment location is obtained based on the sensor group ID. Then, the drone's flight path is planned to generate flight route information, which the drone then uses to collect data.

8. The intelligent river management and monitoring method based on complex networks as described in claim 7, characterized in that, When planning the flight path of a drone, the goal is to minimize the flight path, which includes the following: The node information of recorded abnormal events is collected, and the two-dimensional position information of the corresponding sensor group in the river is obtained based on the node information. Then, the Euclidean distance between different nodes is calculated, which is defined as follows: in, For nodes To node distance, , For nodes Coordinate values ​​in two-dimensional space, , For nodes Coordinate values ​​in two-dimensional space; The objective function for planning with the goal of minimizing the navigation path is defined as follows: in, The number of nodes that recorded abnormal events. For path length, , Number the nodes. For nodes To node distance, It is a binary degree constraint variable, used to ensure node... The distance to all other nodes has been calculated once.

9. A river intelligent governance and monitoring system based on complex networks, characterized in that, It includes: The monitoring terminal management unit is connected to multiple sensor groups, which are deployed on the main stream and tributaries of a river in a predetermined basin and are used to monitor water quality, environmental conditions and / or pollutants. The monitoring terminal management unit acquires multi-source monitoring data in the river basin through the sensor groups, collects and preprocesses the data, and generates a monitoring data sequence. Complex network units are used to construct weighted directed graphs of river networks based on monitoring data sequences. G The intra-group relationship of the monitoring data sequence is measured by a sliding time window to obtain the correlation matrix, partial correlation matrix and / or mutual information matrix, and then the corresponding baseline matrix is ​​estimated using long-period data. The data judgment unit is used to calculate the graph structure transition degree and node anomaly degree based on the correlation matrix, partial correlation matrix and their corresponding baseline matrix; The task generation unit is used to determine the existence of an abnormal event when one or more of the graph structure transition degree and node anomaly degree exceed the corresponding preset threshold, and then trigger the generation of air / sky observation tasks. The task scheduling module is used to schedule UAVs and / or satellites to collect data on the corresponding watershed area of ​​the river according to the air / space observation task.

10. A method for remote monitoring and management of rivers, characterized in that, Its application includes the intelligent river management and monitoring method based on complex networks as described in any one of claims 1 to 8; It includes: A01. Receive monitoring instructions regarding river pollution and execute S01-S05; A02: Obtain the data collected in S05 and generate a monitoring and management report.

Images