Streaming data analysis method for smart city operation decision

Abstract

The invention relates to the technical field of traffic monitoring analysis, in particular to a streaming data analysis method for smart city operation decision, which comprises the following steps of: firstly, preliminarily determining a preliminary anomaly confidence coefficient for representing intersection node anomaly based on risk index values of various risk indexes; then, based on the diffusion influence of the accident in time and space, and based on the risk index value distribution and association condition of each neighborhood influence node, determining a moment anomaly confidence coefficient; determining an instability coefficient in combination with the characteristic that environmental factors may influence the occurrence of congestion events and in combination with environmental abnormal conditions; therefore, the traffic abnormity confidence coefficient representing the intersection node congestion influence is comprehensively determined and is more accurate. Furthermore, the traffic risk is quantified based on the time and space influence when the congestion event occurs, so that the effect of carrying out real-time adjustment on the operation decision according to the traffic risk scoring coefficient is better, and the congestion event processing efficiency is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic monitoring and analysis, and particularly to a method for analyzing streaming data for smart city operation decision-making. Background Art

[0002] Streaming data is a continuously generated, real-time, and dynamically changing data set, which mainly comes from sensors, social media messages, financial transactions, network logs, etc. Therefore, analyzing streaming data is not to process a fixed set of data at one time, but to continuously process the real-time generated data. In the smart city scenario, analyzing streaming data can obtain support data for smart city operation decision-making by monitoring real-time urban energy consumption data such as electricity, gas, transportation, and medical care, effectively improving the urban operation efficiency.

[0003] For traffic data, congestion events usually occur at intersections due to environmental impacts, traffic light failures, and traffic accidents. In order to respond to congestion events in a timely manner, the prior art usually performs similarity analysis on the real-time traffic flow data at intersections and the traffic flow data when congestion occurred in historical data of each intersection, and then determines whether a congestion event occurs at the intersection based on the calculated similarity.

[0004] However, in addition to traffic flow, the impact of congestion events will be reflected in various traffic congestion risk indicators such as real-time following distance, intersection lane occupancy rate, real-time vehicle speed, etc. Therefore, the method of analyzing congestion events only through traffic flow in the prior art has certain limitations, resulting in a relatively low accuracy of congestion event judgment; and the prior art does not quantify the severity of congestion events, resulting in relatively inefficient allocation of operation resources when dealing with congestion events, and unable to give appropriate treatment plans for congestion events of different severities. In summary, the method of analyzing congestion events by the similarity between real-time traffic flow and traffic flow in historical traffic accidents in the prior art has relatively low accuracy and relatively low efficiency in dealing with congestion events. Summary of the Invention

[0005] In order to solve the technical problems that the method of analyzing congestion events by the similarity between real-time traffic flow and traffic flow in historical traffic accidents in the prior art has relatively low accuracy and relatively low efficiency in dealing with congestion events, the purpose of this application is to provide a method for analyzing streaming data for smart city operation decision-making, and the specific technical solutions adopted are as follows: The first aspect of this application provides a method for analyzing streaming data for smart city operation decision-making, including: In a traffic data monitoring platform, collect the risk indicator values of each intersection node in the urban road network at each traffic congestion risk indicator at each moment; Determine the corresponding preliminary anomaly confidence according to the overall magnitudes of the risk index values of various traffic congestion risk indices corresponding to each intersection node at each moment; determine the neighborhood influence nodes of each intersection node at each moment according to the preliminary anomaly confidence. Determine the spatio-temporal anomaly confidence of each intersection node at each moment according to the risk index value distributions and numerical correlation situations of the corresponding various neighborhood influence nodes on various traffic congestion risk indices; determine the corresponding instability coefficient according to the environmental anomaly situation of each intersection node at each moment. Determine the traffic anomaly confidence of each intersection node at each moment according to the preliminary anomaly confidence, the spatio-temporal anomaly confidence, and the instability coefficient; determine the traffic risk scoring coefficient of each intersection node at each moment according to the temporal sequence change of the traffic anomaly confidence of each intersection node and the overall magnitude of the traffic anomaly confidence of the neighborhood influence nodes; perform real-time adjustment of operation decisions according to the traffic risk scoring coefficient.

[0006] Furthermore, the process of obtaining the preliminary anomaly confidence includes: In historical data, take each moment when a congestion event occurs at each intersection node as a historical congestion moment; determine the corresponding accident risk value according to the mean value of the risk index values of each traffic congestion risk index of each intersection node at all corresponding historical accident moments. At each moment, determine the corresponding anomaly intensity value as the difference between the risk index value of each traffic congestion risk index of each intersection node and the corresponding accident risk value; among all the traffic congestion risk indices of each intersection node at each moment, take the traffic congestion risk indices with corresponding anomaly intensity values less than the preset anomaly threshold as the anomaly basic indices of each intersection node. At each moment, determine the corresponding preliminary anomaly confidence according to the number of anomaly basic indices of each intersection node and the overall magnitude of the anomaly intensity values of each anomaly basic index.

[0007] Furthermore, the process of determining the corresponding preliminary anomaly confidence according to the number of anomaly basic indices of each intersection node and the overall magnitude of the anomaly intensity values of each anomaly basic index at each moment includes: At each moment, perform an accumulation operation on the normalized values of the anomaly intensity values of all the anomaly basic indices of each intersection node to determine the overall anomaly value of each intersection node; normalize the product between the number of anomaly basic indices of each intersection node and the overall anomaly value to determine the preliminary anomaly confidence of each intersection node at each moment.

[0008] Furthermore, the process of obtaining the neighborhood influence nodes includes: Take the product of the preliminary anomaly confidence level and the preset maximum neighborhood radius as the influence radius of each intersection node; take other intersection nodes whose chessboard distance from each intersection node is less than the influence radius as the neighborhood influence nodes of each intersection node.

[0009] Further, the process of obtaining the spatio-temporal anomaly confidence level includes: In chronological order, divide all moments into at least two monitoring time periods, where the time lengths of all monitoring time periods are the same; for each traffic congestion risk indicator, arrange all risk indicator values corresponding to each neighborhood influence node in each monitoring time period in chronological order to determine the corresponding time-series risk indicator sequence. Take the time-series risk indicator sequence as a row vector, and determine the local spatio-temporal matrix corresponding to each monitoring time period for each traffic congestion risk indicator according to all time-series risk indicator sequences corresponding to all neighborhood influence nodes. Standardize the local spatio-temporal matrix and then decompose it using the SVD decomposition algorithm to obtain the corresponding sparse matrix; determine the event anomaly degree according to the mean value of the absolute values of all non-zero elements in the sparse matrix. Combine each row vector of the sparse matrix with any other row vector to determine the corresponding row vector pair; traverse all row vectors of the sparse matrix to determine all row vector pairs; take the normalized value of the Pearson correlation coefficient between the two row vectors in each row vector pair as the corresponding reference correlation; determine the event correlation according to the mean value of the reference correlations of all row vector pairs. Determine the reference confidence level of the intersection node corresponding to each traffic congestion risk indicator in each monitoring time period according to the normalized value of the product between the event anomaly degree and the event correlation; determine the overall confidence level of each intersection node in each monitoring time period according to the mean value of the reference confidence levels corresponding to all traffic congestion risk indicators. Under each intersection node, determine the corresponding spatio-temporal anomaly confidence level according to the overall confidence level of the monitoring time period in which each moment is located.

[0010] Further, the process of obtaining the instability coefficient includes: Determine the bad weather judgment coefficient according to the existence of bad weather; at each intersection node, determine the corresponding temperature standard deviation value according to the difference between the ambient temperature at each moment and the preset standard temperature; determine the corresponding humidity standard deviation value according to the difference between the ambient humidity at each moment and the preset standard humidity; determine the temperature and humidity anomaly value according to the product between the temperature standard deviation value and the humidity standard deviation value; normalize the product between the bad weather judgment coefficient and the temperature and humidity anomaly value to determine the instability coefficient of each intersection node at each moment.

[0011] Further, the process of obtaining the traffic anomaly confidence level includes: Determine the comparison confidence level of each intersection node at each moment according to the mean value between the instability coefficient and the spatio-temporal anomaly confidence level; normalize the product between the comparison confidence level and the preliminary anomaly confidence level to determine the traffic anomaly confidence level of each intersection node at each moment.

[0012] Further, the process of obtaining the traffic risk scoring coefficient includes: At each moment, determine the diffusion influence degree of each intersection node according to the mean value of the traffic anomaly confidence levels of all neighboring influence nodes corresponding to each intersection node; At each path node, regard the moment when the corresponding traffic anomaly confidence level is less than the preset confidence level threshold as the normal moment; regard the closest normal moment before each moment as the suspected event anomaly moment; arrange the traffic anomaly confidence levels of each intersection node at all moments in chronological order and perform curve fitting to determine the anomaly confidence level curve of each intersection node; determine the event evolution weight of each moment according to the change trend of the traffic anomaly confidence level between each moment on the anomaly confidence level curve and the suspected event anomaly moment; Normalize the product of the traffic anomaly confidence level, the event evolution weight, and the diffusion influence degree of each intersection node at each moment to determine the corresponding traffic risk scoring coefficient.

[0013] Further, the process of obtaining the event evolution weight includes: On the anomaly confidence level curve, when there is a maximum point between each moment and the corresponding suspected event anomaly moment, regard the mean value of the tangent slopes of all moments between the suspected event anomaly moment and the first maximum point after it as the event evolution weight of each moment; When there is no maximum point between each moment and the corresponding suspected event anomaly moment, regard the mean value of the tangent slopes of all moments between each moment and the corresponding suspected event anomaly moment as the event evolution weight of each moment.

[0014] Further, the process of adjusting the operation decision according to the traffic risk scoring coefficient includes: For each intersection node at the current moment: When the corresponding traffic risk scoring coefficient is greater than the preset first scoring threshold and less than or equal to the preset second scoring threshold, mark the corresponding intersection node as suspected of congestion on the navigation software; When the corresponding traffic risk score coefficient is greater than the preset second score threshold and less than or equal to the preset third score threshold, mark the corresponding intersection node as suspected of congestion on the navigation software and notify the traffic police to go to the corresponding intersection node for traffic command; When the corresponding traffic risk score coefficient is greater than the preset third score threshold, mark the corresponding intersection node as suspected of congestion on the navigation software, notify the traffic police to go to the corresponding intersection node for traffic command and set up accident warning signs on the road sections adjacent to the corresponding intersection node; wherein, the preset first score threshold is less than the preset second score threshold, and the preset second score threshold is less than the preset third score threshold.

[0015] In a second aspect, the present application provides a streaming data analysis system for smart city operation decision-making, the system includes: A data collection and preprocessing module, configured to collect the risk index values of each intersection node in the urban road network at each traffic congestion risk index at each moment in the traffic data monitoring platform; A first determination module, configured to determine the corresponding preliminary anomaly confidence according to the overall magnitude of the risk index values of various traffic congestion risk indexes corresponding to each intersection node at each moment; determine the neighborhood influence nodes of each intersection node at each moment according to the preliminary anomaly confidence; A second determination module, configured to determine the spatio-temporal anomaly confidence of each intersection node at each moment according to the risk index value distribution and numerical association of the corresponding neighborhood influence nodes on various traffic congestion risk indexes; determine the corresponding instability coefficient according to the environmental anomaly situation of each intersection node at each moment; An operation decision real-time adjustment module, configured to determine the traffic anomaly confidence of each intersection node at each moment according to the preliminary anomaly confidence, the spatio-temporal anomaly confidence and the instability coefficient; determine the traffic risk score coefficient of each intersection node at each moment according to the temporal and sequential changes of the traffic anomaly confidence of each intersection node and the overall magnitude of the traffic anomaly confidence of the neighborhood influence nodes; perform real-time adjustment of operation decisions according to the traffic risk score coefficient.

[0016] In a third aspect, the present application provides a computer device, including a memory and a processor. The memory is used to store computer program code, and the processor is used to call and run the computer program code from the memory to execute the method as described in the first aspect or any embodiment of the first aspect of the present application.

[0017] In a fourth aspect, the present application provides a computer program product, the computer program product includes computer program code, and when the computer program code is executed, it is used to execute the method as described in the first aspect or any embodiment of the first aspect of the present application.

[0018] In a fifth aspect, the present application provides a computer-readable storage medium storing computer program code which, when executed, performs the method according to the first aspect or any embodiment of the first aspect of the present application.

[0019] The present application has the following beneficial effects: The present application first preliminarily determines a preliminary anomaly confidence level characterizing the anomaly of the intersection node based on the risk index values of various risk indicators; then, based on the diffusion impact of accidents in time and space, and based on the risk index value distribution and association of each neighboring influence node, determines the moment anomaly confidence level; and combines the instability coefficient by combining the environmental anomaly situation considering the characteristics that environmental factors may affect the occurrence of congestion events; thereby comprehensively determining a traffic anomaly confidence level that more accurately characterizes the congestion impact of the intersection node according to the preliminary anomaly confidence level, the spatio-temporal anomaly confidence level, and the instability coefficient; further quantifies the traffic risk based on the time and space impacts when the congestion event occurs, so that the obtained traffic risk scoring coefficient can not only more accurately determine whether a congestion event occurs, but also quantify the impact degree of the congestion event, making the effect of real-time adjustment of operation decisions based on the traffic risk scoring coefficient better and the processing efficiency of congestion events higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 Flowchart of a method for streaming data analysis of smart city operation decision-making provided by an embodiment of the present invention; Figure 2 Schematic diagram of an urban road network provided by an embodiment of the present invention; Figure 3 Structural diagram of a system for streaming data analysis of smart city operation decision-making provided by an embodiment of the present invention; Figure 4 Schematic diagram of the structure of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, in conjunction with the accompanying drawings and preferred embodiments, a method for streaming data analysis for smart city operation decision-making proposed according to the present invention, including its specific implementation manner, structure, features and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment, and specific features, structures or characteristics in one or more embodiments can be combined in any suitable form. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0024] The following specifically describes the specific solution of a method for streaming data analysis for smart city operation decision-making provided by the present invention in conjunction with the accompanying drawings.

[0025] An embodiment of the present application provides a method for streaming data analysis for smart city operation decision-making. Please refer to Figure 1 , which shows a flowchart of a method for streaming data analysis for smart city operation decision-making provided by an embodiment of the present invention. The method includes: Step S101: In the traffic data monitoring platform, collect the risk index values of each intersection node in the urban road network at each traffic congestion risk index at each moment.

[0026] All road intersections within the monitoring range in the traffic data monitoring platform are taken as intersection nodes, and the road between two road intersections with connected roads is taken as a connection line to form an urban road network; please refer to Figure 2 , which shows a schematic diagram of an urban road network provided by an embodiment of the present invention, where each point represents an intersection node, that is, the node corresponding to each road intersection, and the connection line between the nodes is the road between the road intersections.

[0027] In a specific implementation manner of the embodiment of the present invention, the sampling frequency is set to be collected once per minute, and the sampling time period is set to 30 minutes before the current moment, which can be adjusted according to the specific implementation environment.

[0028] In a specific implementation manner of the embodiment of the present invention, the traffic congestion risk indicators include the number of vehicles at intersections, the real-time following distance, the lane occupancy rate at intersections, and the real-time vehicle speed; the traffic data monitoring platform is used to call the data values of each intersection node in the urban road network at each moment for each traffic congestion risk indicator; cameras, radars, and geomagnetic coils are installed at each road intersection. At each moment, the real-time vehicle speed and following distance of each vehicle at the road intersection are monitored by the radar; road intersection images are collected by the camera, and each vehicle is tracked in real time through the object detection method, and the traffic flow and lane occupancy rate at the intersection are determined according to the number of vehicles; for the number of vehicles at the intersection, the normalized value of the number of vehicles at the intersection is used as the corresponding risk indicator value, that is, the greater the number of vehicles at the corresponding intersection, the more likely a congestion event occurs. For the real-time following distance, the opposite number of the difference between the normalized value of the average following distance of all vehicles in the road intersection corresponding to each intersection node at each moment and the real number 1 is used as the corresponding risk indicator value; that is, the smaller the overall following distance of each vehicle in the corresponding intersection, the closer the distance between vehicles, and the more likely a congestion event occurs. For the real-time vehicle speed, the opposite number of the difference between the normalized value of the average real-time vehicle speed of all vehicles in the road intersection corresponding to each intersection node at each moment and the real number 1 is used as the corresponding risk indicator value; that is, the smaller the overall real-time vehicle speed of each vehicle in the corresponding intersection, the more likely the vehicle is affected by congestion and the slower the vehicle travels, that is, the higher the possibility of a congestion event. For the lane occupancy rate at the intersection, the normalized value of the lane occupancy rate at the intersection is used as the corresponding risk indicator value; that is, the greater the lane occupancy rate at the corresponding intersection, the more in line with the situation where vehicles are more concentrated on the lane during a congestion event, and the greater the probability of a corresponding congestion event occurring. It should be noted that the implementer can select other traffic congestion risk indicators according to the specific implementation environment, and when calculating the risk indicator value, it is necessary to make the risk indicator value of the corresponding traffic congestion risk indicator positively correlated with the possibility of a congestion event occurring. And in this application, through normalization processing, the influence of the dimensions of different traffic congestion risk indicators on the subsequent calculation process is reduced, making the subsequent processing process more robust.

[0029] Step S102: Determine the corresponding preliminary abnormal confidence according to the overall magnitude of the risk indicator values of various traffic congestion risk indicators corresponding to each intersection node at each moment; determine the neighborhood influence nodes of each intersection node at each moment according to the preliminary abnormal confidence.

[0030] Since the greater the risk indicator value corresponding to the traffic congestion risk indicator, the more likely the corresponding intersection node corresponds to a congestion event, therefore, the possibility of a congestion event occurring is initially characterized based on the numerical characteristics of the risk indicator value. Preferably, in some possible implementation manners of the embodiment of the present invention, the process of obtaining the preliminary abnormal confidence includes: In historical data, each moment when a congestion event occurs at each intersection node is used as a historical congestion moment; according to the mean value of the risk index values of each traffic congestion risk index at all corresponding historical accident moments for each intersection node, the corresponding accident risk value is determined; at each moment, the difference between the risk index value of each traffic congestion risk index of each intersection node and the corresponding accident risk value is determined as the corresponding abnormal intensity value; among all traffic congestion risk indexes of each intersection node at each moment, the traffic congestion risk indexes with corresponding abnormal intensity values less than the preset abnormal threshold are used as the abnormal basic indexes of each intersection node. In a specific implementation manner of the embodiment of the present invention, the preset abnormal threshold is set to 0 and can be adjusted according to the specific implementation environment.

[0031] The accident risk value characterizes the numerical characteristics of the risk index values of each traffic congestion risk index when a congestion event occurs at the corresponding intersection node. If the risk index value detected in real time is closer to the corresponding accident risk value, or even larger than the accident risk value, it indicates that the corresponding intersection node is more in line with the characteristics of the congestion event in the corresponding traffic congestion risk index, and the greater the possibility of the corresponding congestion event occurring. Correspondingly, the larger the risk index value, the larger the calculated abnormal intensity value; when the preset abnormal threshold is set to 0, it means that if the risk index value of the corresponding traffic congestion risk index is greater than the accident risk value, the corresponding traffic congestion risk index conforms to the risk index characteristics during the traffic congestion event. Therefore, it is defined as an abnormal basic index. The more the number of corresponding abnormal basic indexes, the more likely a traffic congestion event will occur; therefore, further combining the numerical characteristics of the abnormal intensity value and the number characteristics of the abnormal basic index, the preliminary abnormal confidence degree representing the possibility of the traffic congestion event is comprehensively determined; at each moment, according to the number of abnormal basic indexes of each intersection node and the overall magnitude of the abnormal intensity values of each abnormal basic index, the corresponding preliminary abnormal confidence degree is determined.

[0032] Preferably, in some possible implementation manners of the embodiment of the present invention, the process of determining the corresponding preliminary abnormal confidence degree according to the number of abnormal basic indexes of each intersection node and the overall magnitude of the abnormal intensity values of each abnormal basic index at each moment includes: At each moment, the normalized values of the abnormal intensity values of all abnormal basic indexes of each intersection node are accumulated to determine the overall abnormal value of each intersection node; the product of the number of abnormal basic indexes and the overall abnormal value of each intersection node is normalized to determine the preliminary abnormal confidence degree of each intersection node at each moment. Only analyzing the abnormal intensity values of the abnormal basic indexes can enhance the influence of the traffic congestion risk indexes that conform to the congestion event, making the representation of the possibility of the congestion event more accurate; and through the normalization process, the value of the preliminary abnormal confidence degree is limited to within 0 to 1, avoiding the influence of the dimension.

[0033] In a specific implementation manner of the embodiment of the present invention, the process of obtaining the preliminary anomaly confidence is expressed by the formula: ; where is the preliminary anomaly confidence of the th intersection node at the th moment; is the number of anomaly basic indicators of the th intersection node at the th moment; is the risk indicator value of the th anomaly basic indicator of the th intersection node at the th moment; is the overall anomaly of the th intersection node at the th moment; is a linear normalization function. It should be noted that, unless otherwise specified, all normalization methods in the embodiments of the present invention adopt linear normalization, which can be adjusted according to the specific implementation environment.

[0034] For each intersection node, when a traffic congestion event occurs, the congestion usually affects other adjacent intersection nodes. Therefore, the present application further determines other intersection nodes that may be affected by each intersection node. And the greater the preliminary anomaly confidence of the intersection node, the more likely the corresponding congestion event is real. Therefore, a greater influence range is given for more accurate analysis.

[0035] Preferably, in some possible implementation manners of the embodiment of the present invention, the process of obtaining the neighborhood influence nodes includes: Taking the product of the preliminary anomaly confidence and the preset maximum neighborhood radius as the influence radius of each intersection node; taking other intersection nodes whose chessboard distance from each intersection node is less than the influence radius as the neighborhood influence nodes of each intersection node. In a specific implementation manner of the embodiment of the present invention, the preset maximum neighborhood radius is set to 1.5 km, which can be adjusted according to the specific implementation environment. Among them, the chessboard distance is more in line with the path planning of the city, and the implementer can also replace the chessboard distance with the Euclidean distance or the road path extension length according to the specific implementation environment, which will not be further elaborated here.

[0036] Step S103: Determine the spatio-temporal anomaly confidence of each intersection node at each moment according to the risk indicator value distribution and numerical association of the corresponding neighborhood influence nodes on various traffic congestion risk indicators; determine the corresponding instability coefficient according to the environmental anomaly situation of each intersection node at each moment.

[0037] Furthermore, based on the analysis of the impact of adjacent influencing nodes on time and space when traffic congestion occurs, while more accurately characterizing the likelihood of congestion events at each intersection node, a more accurate quantitative analysis of the impact degree after the occurrence of congestion events is carried out, so that when the spatio-temporal anomaly confidence level is higher, the likelihood of traffic congestion at the corresponding intersection node is higher, and the impact on other roads when traffic congestion occurs is greater, that is, the congestion is more serious.

[0038] Preferably, in some possible implementation manners of the embodiments of the present invention, the process of obtaining the spatio-temporal anomaly confidence level includes: In the time sequence, all moments are divided into at least two monitoring time periods, where the time lengths of all monitoring time periods are the same; under each traffic congestion risk index, all risk index values corresponding to each neighborhood influencing node in each monitoring time period are arranged in time sequence to determine the corresponding time series risk index sequence; using the time series risk index sequence as a row vector, according to all time series risk index sequences corresponding to all neighborhood influencing nodes, a local spatio-temporal matrix corresponding to each monitoring time period under each traffic congestion risk index is determined. In the local spatio-temporal matrix, each column represents all risk index values corresponding to all neighborhood influencing nodes at the same moment, and each row corresponds to the time series risk index sequence. In a specific implementation manner of the embodiments of the present invention, the number of monitoring time periods is set to 5, which can be adjusted according to the specific implementation environment. By dividing the monitoring time periods, not only can the complexity of matrix analysis be reduced to improve the analysis efficiency, but also the time series impact of congestion events can be captured more sensitively by analyzing local time series characteristics.

[0039] The local spatio-temporal matrix is standardized and then decomposed by the SVD decomposition algorithm to obtain the corresponding sparse matrix; according to the mean value of the absolute values of all non-zero elements in the sparse matrix, the event anomaly degree is determined; when decomposing the matrix by the SVD decomposition algorithm, a low-rank matrix and a sparse matrix will be obtained; the low-rank matrix reflects the main traffic patterns in the corresponding urban road network, which is the main part of the matrix and reflects the operation rules of traffic under normal conditions; while the sparse matrix represents the sudden anomalies in traffic information, that is, low-frequency anomaly data. The larger the absolute value of its non-zero element, the more likely it is that the corresponding traffic information machine risk index value will have a mutation anomaly, and the more it conforms to the abnormal change characteristics of the risk index value of the congestion event; therefore, when the event anomaly degree representing the overall absolute value of all non-zero elements in the sparse matrix is larger, the corresponding neighborhood influencing nodes are more likely to be affected by the congestion event, and the credibility of the occurrence of the congestion event at the corresponding intersection node is higher; therefore, when the time anomaly degree is larger, the spatio-temporal anomaly confidence level is larger.

[0040] Combine each row vector of the sparse matrix with any other row vector to determine the corresponding row vector pair; traverse all row vectors of the sparse matrix to determine all row vector pairs; use the normalized value of the Pearson correlation coefficient between the two row vectors in each row vector pair as the corresponding reference relevance; determine the event relevance according to the mean value of the reference relevance of all row vector pairs. The non-zero element values in the sparse matrix reflect the anomalies in traffic data. Therefore, the higher the data correlation between different rows in the sparse matrix, the more similar the anomalies in the traffic data of each neighborhood influence node, the more likely they are affected by the same congestion event, and the greater the likelihood of a congestion event occurring at the corresponding intersection node and the more serious the congestion impact; therefore, when the event relevance is greater, the corresponding spatio-temporal anomaly confidence level is greater.

[0041] Furthermore, combine the event relevance and event anomaly degree under all traffic congestion risk indicators to comprehensively characterize the spatio-temporal anomaly confidence level. According to the normalized value of the product between the event anomaly degree and the event relevance, determine the reference confidence level of the corresponding intersection node under each traffic congestion risk indicator in each monitoring time period; according to the mean value of the reference confidence levels corresponding to all traffic congestion risk indicators, determine the overall confidence level of each intersection node in each monitoring time period; under each intersection node, determine the corresponding spatio-temporal anomaly confidence level according to the overall confidence level of the monitoring time period at each moment; so that the obtained moment anomaly confidence level can characterize the likelihood of a congestion event in terms of time and the impact degree when a congestion event occurs in terms of space.

[0042] In a specific implementation manner of the embodiment of the present invention, the process of obtaining the overall confidence level is expressed by the formula: ; where is the overall confidence level of the th intersection node in the th monitoring time period; is the total number of traffic congestion risk indicators; is the mean value of the reference relevance of all row vector pairs corresponding to the sparse matrix of the th traffic congestion risk indicator, the th intersection node in the th monitoring time period, that is, the event relevance; is the mean value of the absolute values of all non-zero elements corresponding to the sparse matrix of the th traffic congestion risk indicator, the th intersection node in the th monitoring time period, that is, the event anomaly degree; is the th traffic congestion risk indicator, the th intersection node in the The reference confidence level for each monitoring time period.

[0043] Furthermore, analyze each intersection node at each moment from the environmental aspect. Considering that in bad weather, such as rain, snow, heavy fog, hail and other weather, the possibility of accidents will increase due to slippery roads or limited visibility; abnormal temperature and humidity may affect driving safety, and abnormal temperature and humidity usually correspond to bad weather. Therefore, further determine the corresponding instability coefficient according to the environmental abnormality of each intersection node at each moment, so that the greater the instability coefficient, the higher the possibility of traffic accidents, that is, congestion events, occurring at the corresponding intersection node.

[0044] Preferably, in some possible implementation manners of the embodiments of the present invention, the process of obtaining the instability coefficient includes: Determine the bad weather judgment coefficient according to whether there is bad weather; at each intersection node, determine the corresponding temperature standard deviation value according to the difference between the environmental temperature at each moment and the preset standard temperature; determine the corresponding humidity standard deviation value according to the difference between the environmental humidity at each moment and the preset standard humidity; determine the temperature and humidity abnormality value according to the product of the temperature standard deviation value and the humidity standard deviation value; normalize the product of the bad weather judgment coefficient and the temperature and humidity abnormality value to determine the instability coefficient of each intersection node at each moment. In a specific implementation manner of the embodiments of the present invention, the preset standard temperature is set to 20 degrees Celsius, and the preset standard humidity is set to 40%, and the implementer adaptively adjusts according to the season or month in which the current moment is located. It should be noted that the humidity in the embodiments of the present invention is the standard humidity, and the environmental temperature is collected in real time through a temperature sensor installed at the intersection, and the environmental humidity is collected in real time through a humidity sensor.

[0045] Among them, in a specific implementation manner of the embodiments of the present invention, when there is bad weather at the corresponding intersection node at the corresponding moment, set the corresponding bad weather judgment coefficient to 1; when there is no bad weather at the corresponding intersection node at the corresponding moment, set the corresponding bad weather judgment coefficient to 0; by setting 0 and 1, it can be ensured that the obtained instability coefficient characterizes the severity of bad weather, so as to more accurately characterize the possibility of congestion events occurring. When there is bad weather, the greater the temperature standard deviation value and the greater the humidity standard deviation value, the more abnormal the environment caused by the corresponding bad weather, and the greater the instability coefficient of the corresponding environmental impact; that is, the greater the temperature and humidity abnormality value, the higher the credibility of traffic accidents caused by bad weather interference leading to congestion events.

[0046] In some possible implementation manners of the embodiments of the present invention, the process of obtaining the instability coefficient is expressed by the formula: ; where is the instability coefficient of the th intersection node at the th moment; is the ambient temperature of the th intersection node at the th moment; is the preset standard temperature; is the ambient humidity of the th intersection node at the th moment; is the preset standard humidity; is the absolute value symbol; is the temperature standard deviation value of the th intersection node at the th moment; is the humidity standard deviation value of the th intersection node at the th moment; is the temperature and humidity anomaly value of the th intersection node at the th moment; is the bad weather judgment coefficient of the th intersection node at the th moment.

[0047] Step S104: Determine the traffic anomaly confidence of each intersection node at each moment according to the preliminary anomaly confidence, spatio-temporal anomaly confidence, and instability coefficient; determine the traffic risk scoring coefficient of each intersection node at each moment according to the temporal variation of the traffic anomaly confidence of each intersection node and the overall magnitude of the traffic anomaly confidence of the neighboring influence nodes; perform real-time adjustment of the operation decision according to the traffic risk scoring coefficient.

[0048] So far, the present application has determined the preliminary anomaly confidence representing the numerical characteristics of the risk index value of the congestion event, the spatio-temporal anomaly confidence representing the time and space influence characteristics of the congestion event, and the instability coefficient representing the environmental anomaly influence. The three parameters together represent the occurrence possibility and influence of the congestion event. Therefore, further determine the traffic anomaly confidence of each intersection node at each moment according to the preliminary anomaly confidence, spatio-temporal anomaly confidence, and instability coefficient; the higher the traffic anomaly confidence, the higher the possibility of the corresponding path node having a congestion event and the greater the congestion influence.

[0049] Preferably, in some possible implementation manners of the embodiments of the present invention, the process of obtaining the traffic anomaly confidence includes: Determine the comparison confidence of each intersection node at each moment according to the mean between the instability coefficient and the spatio-temporal anomaly confidence; normalize the product between the comparison confidence and the preliminary anomaly confidence to determine the traffic anomaly confidence of each intersection node at each moment. Since the instability coefficient is likely to take the value of 0, in order to avoid a large impact of 0 on the calculation results, after first obtaining the comparison confidence by taking the mean of the instability coefficient and the spatio-temporal anomaly confidence, then weight the initially obtained preliminary anomaly confidence, so that the obtained traffic anomaly confidence is more accurate.

[0050] In a specific implementation manner of the embodiment of the present invention, the process of obtaining the traffic anomaly confidence is expressed by the formula: ; where is the traffic anomaly confidence of the th intersection node at the th moment; is the instability coefficient of the th intersection node at the th moment; is the spatio-temporal anomaly confidence of the th intersection node at the th moment; is the preliminary anomaly confidence of the th intersection node at the th moment.

[0051] Furthermore, it is necessary to analyze from the impact of congestion events. When the impact of a congestion accident is large, this impact usually radiates to adjacent intersection nodes, making the adjacent intersection nodes also have a relatively high traffic anomaly confidence; and it is necessary to consider that the occurrence of congestion events is usually relatively rapid, that is, as long as a congestion event occurs, the calculated traffic anomaly confidence will show a relatively rapid numerical change. Therefore, further comprehensively characterize the impact and occurrence probability of congestion events according to the temporal sequence change of the traffic anomaly confidence of each intersection node and the overall magnitude of the traffic anomaly confidence of neighboring influence nodes, so that the greater the obtained traffic risk scoring coefficient, the greater the probability of the occurrence of a congestion event and the greater the corresponding impact when the congestion event occurs.

[0052] Preferably, in some possible implementation manners of the embodiment of the present invention, the process of obtaining the traffic risk scoring coefficient includes: At each moment, according to the mean value of the traffic anomaly confidence levels of all neighboring influence nodes corresponding to each intersection node, determine the diffusion influence degree of each intersection node. The neighboring influence nodes are the nodes that may be affected by the congestion of the corresponding intersection node. The greater the mean value of the traffic anomaly confidence levels of the corresponding neighboring influence nodes, the more likely it is that congestion events may occur in the neighboring influence nodes under the influence of the corresponding intersection node, and then the higher the probability of a congestion event occurring at the intersection node.

[0053] At each path node, take the moments when the corresponding traffic anomaly confidence level is less than the preset confidence threshold as normal moments; take the closest normal moment before each moment as the suspected event anomaly moment; arrange the traffic anomaly confidence levels of each intersection node at all moments in chronological order and then perform curve fitting to determine the anomaly confidence curve of each intersection node; according to the change trend of the traffic anomaly confidence level between each moment and the suspected event anomaly moment on the anomaly confidence curve, determine the event evolution weight of each moment. In a specific implementation manner of the embodiment of the present invention, the preset confidence threshold is set to 0.4 and can be adjusted according to the specific implementation environment.

[0054] Preferably, in some possible implementation manners of the embodiment of the present invention, the process of obtaining the event evolution weight includes: On the anomaly confidence curve, when there is a maximum point between each moment and the corresponding suspected event anomaly moment, take the mean value of the tangent slopes of all moments between the suspected event anomaly moment and the first maximum point after it as the event evolution weight of each moment; when there is no maximum point between each moment and the corresponding suspected event anomaly moment, take the mean value of the tangent slopes of all moments between each moment and the corresponding suspected event anomaly moment as the event evolution weight of each moment.

[0055] When the congestion event starts to spread, the traffic anomaly confidence level will show a sudden increase and then remain stable; therefore, first analyze by taking the closest normal moment before each moment as the suspected event anomaly moment. The sudden increase in the traffic anomaly confidence level is manifested as a relatively high slope in the upward trend of the traffic anomaly confidence level. Therefore, the greater the mean value of the tangent slopes of all moments between the suspected event anomaly moment and the first maximum point after it, the more it conforms to the characteristics of the sudden increase in the traffic anomaly confidence level, and the greater the probability of a congestion event occurring.

[0056] It should be noted that when the corresponding moment is the normal moment, the corresponding time evolution weight is set to 0; that is, when the traffic anomaly confidence at the corresponding moment is less than the preset confidence threshold, it is considered that there is no congestion event at the intersection node at the corresponding moment. Therefore, setting the event evolution weight to 0 is more in line with the objective facts, and the implementer can adjust the numerical value according to the specific implementation environment.

[0057] In addition, the traffic anomaly confidence itself represents the possibility of a congestion event. Therefore, further normalize the product of the traffic anomaly confidence, event evolution weight, and diffusion influence degree of each intersection node at each moment to determine the corresponding traffic risk scoring coefficient; the greater the traffic risk scoring coefficient, the higher the possibility and the greater the impact of the corresponding intersection node having a congestion event. Finally, make more accurate and efficient operation decision adjustments according to the traffic risk scoring coefficient.

[0058] In a specific implementation manner of the embodiment of the present invention, the process of obtaining the traffic risk scoring coefficient is expressed by the formula: ; where is the traffic risk scoring coefficient of the th intersection node at the th moment; is the traffic anomaly confidence of the th intersection node at the th moment; is the mean value of the traffic anomaly confidences of all neighboring influence nodes corresponding to the th intersection node at the th moment, that is, the corresponding diffusion influence degree; is the event evolution weight of the th intersection node at the th moment.

[0059] Preferably, in some possible implementation manners of the embodiment of the present invention, the process of making operation decision adjustments according to the traffic risk scoring coefficient includes: For each intersection node at the current moment: When the corresponding traffic risk scoring coefficient is greater than the preset first scoring threshold and less than or equal to the preset second scoring threshold, mark the corresponding intersection node as suspected of congestion on the navigation software; when the corresponding traffic risk scoring coefficient is greater than the preset second scoring threshold and less than or equal to the preset third scoring threshold, mark the corresponding intersection node as suspected of congestion on the navigation software and notify the traffic police to go to the corresponding intersection node for traffic command; when the corresponding traffic risk scoring coefficient is greater than the preset third scoring threshold, mark the corresponding intersection node as suspected of congestion on the navigation software, notify the traffic police to go to the corresponding intersection node for traffic command and set up accident warning signs on the road sections adjacent to the corresponding intersection node. In some possible implementation manners of the embodiments of the present invention, the preset first scoring threshold is set to 0.5, the preset second scoring threshold is set to 0.7, and the preset third scoring threshold is set to 0.9. As the traffic risk scoring coefficient gradually increases, the possibility of a congestion event occurring becomes higher and higher. When the traffic risk scoring coefficient is greater than the preset first scoring threshold, it is considered that a congestion event has occurred; during the continuous increase process, the greater the value of the traffic risk scoring coefficient, the greater the impact. It should be noted that the specific operation decision can be adjusted according to the specific implementation environment and will not be further elaborated here.

[0060] In summary, a streaming data analysis method for smart city operation decision-making first initially determines a preliminary anomaly confidence level representing the anomaly of intersection nodes based on the risk indicator values of various risk indicators; then, based on the diffusion impact of accidents in time and space, determines the moment anomaly confidence level based on the risk indicator value distribution and association of each neighboring influence node; and combines the characteristics that environmental factors may affect the occurrence of congestion events and combines environmental anomaly situations to determine the instability coefficient; thereby, based on the preliminary anomaly confidence level, spatio-temporal anomaly confidence level, and instability coefficient, comprehensively determines a traffic anomaly confidence level that more accurately represents the congestion impact of intersection nodes; further quantifies the traffic risk based on the time and space impacts when a congestion event occurs, so that the obtained traffic risk scoring coefficient can not only more accurately determine whether a congestion event occurs, but also quantify the impact degree of the congestion event, making the effect of real-time adjustment of operation decision-making according to the traffic risk scoring coefficient better and the processing efficiency of congestion events higher.

[0061] This application also provides a streaming data analysis system for smart city operation decision-making. Please refer to Figure 3 , which shows the structure diagram of a streaming data analysis system for smart city operation decision-making provided by an embodiment of the present invention. The system includes: a data acquisition and preprocessing module 301, a first determination module 302, a second determination module 303, and an operation decision real-time adjustment module 304.

[0062] The data acquisition and preprocessing module 301 is used to collect the risk index values of each intersection node in the urban road network at each moment for each traffic congestion risk index in the traffic data monitoring platform. The first determination module 302 is used to determine the corresponding preliminary anomaly confidence level according to the overall magnitude of the risk index values of various traffic congestion risk indexes corresponding to each intersection node at each moment; determine the neighborhood influence nodes of each intersection node at each moment according to the preliminary anomaly confidence level. The second determination module 303 is used to determine the spatio-temporal anomaly confidence level of each intersection node at each moment according to the risk index value distribution and numerical association of the corresponding neighborhood influence nodes for various traffic congestion risk indexes; determine the corresponding instability coefficient according to the environmental anomaly situation of each intersection node at each moment. The operation decision real-time adjustment module 304 is used to determine the traffic anomaly confidence level of each intersection node at each moment according to the preliminary anomaly confidence level, spatio-temporal anomaly confidence level and instability coefficient; determine the traffic risk score coefficient of each intersection node at each moment according to the temporal sequence change of the traffic anomaly confidence level of each intersection node and the overall magnitude of the traffic anomaly confidence level of the neighborhood influence nodes; perform real-time adjustment of operation decisions according to the traffic risk score coefficient.

[0063] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, a streaming data analysis system for smart city operation decision-making and a method embodiment of a streaming data analysis method for smart city operation decision-making provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0064] An embodiment of the present application also provides a computer device. Please refer to Figure 4 , which shows a schematic structural diagram of a computer device provided by an embodiment of the present invention. The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can execute any of the above-described streaming data analysis methods for smart city operation decision-making.

[0065] An embodiment of the present application also provides a computer program product. When the computer program product runs on a computer device, the computer device can execute any of the above-described streaming data analysis methods for smart city operation decision-making.

[0066] The embodiment of the present application also provides a computer-readable storage medium, in which computer program code is stored. When the computer program code runs on a computer device, the computer device can execute any of the above-introduced streaming data analysis methods for smart city operation decision-making.

[0067] In the embodiments provided in the present application, it should be understood that the provided computer device, computer program product, and computer-readable storage medium are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can refer to the beneficial effects in the methods provided above, which will not be elaborated here.

[0068] It should be noted that the above sequence of the embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A streaming data analysis method for smart city operation decision-making, characterized in that: The method comprises: In the traffic data monitoring platform, the risk index value of each intersection node in the urban road network at each moment on each traffic congestion risk index is collected; Determine the corresponding preliminary abnormality confidence according to the overall size of the risk index values ​​of various traffic congestion risk indicators corresponding to each intersection node at each moment; determine the neighborhood influencing node of each intersection node at each moment according to the preliminary abnormality confidence; According to the distribution of risk index values ​​and numerical correlation of the corresponding neighborhood influencing nodes on various traffic congestion risk indicators, the spatiotemporal anomaly confidence of each intersection node at each moment is determined; according to the environmental anomaly of each intersection node at each moment, the corresponding instability coefficient is determined; According to the preliminary anomaly confidence, the spatiotemporal anomaly confidence and the instability coefficient, the traffic anomaly confidence of each intersection node at each moment is determined; according to the temporal changes of the traffic anomaly confidence of each intersection node and the overall size of the traffic anomaly confidence of the neighborhood influencing nodes, the traffic risk scoring coefficient of each intersection node at each moment is determined; and the operational decision is adjusted in real time according to the traffic risk scoring coefficient.

2. A streaming data analysis method for smart city operation decision-making according to claim 1, characterized in that: The process of obtaining the preliminary abnormality confidence level includes: In the historical data, each moment when a congestion event occurs at each intersection node is taken as the historical congestion moment; the corresponding accident risk value is determined according to the average of the risk index values ​​of each traffic congestion risk index of each intersection node at all corresponding historical accident moments; At each moment, the difference between the risk index value of each traffic congestion risk index of each intersection node and the corresponding accident risk value is used to determine the corresponding abnormal intensity value; among all the traffic congestion risk indicators of each intersection node at each moment, the traffic congestion risk indicator with a corresponding abnormal intensity value less than a preset abnormal threshold is used as the abnormal basic indicator of each intersection node; At each moment, the corresponding preliminary anomaly confidence is determined according to the number of abnormal basic indicators of each intersection node and the overall size of the abnormal intensity values ​​of each abnormal basic indicator.

3. A streaming data analysis method for smart city operation decision-making according to claim 2, characterized in that: The process of determining the corresponding preliminary abnormality confidence at each moment according to the number of abnormal basic indicators of each intersection node and the overall magnitude of the abnormal intensity values ​​of each abnormal basic indicator includes: At each moment, the normalized values ​​of the abnormal intensity values ​​of all abnormal basic indicators of each intersection node are accumulated to determine the overall abnormal value of each intersection node; the product between the number of abnormal basic indicators of each intersection node and the overall abnormal value is normalized to determine the preliminary abnormal confidence of each intersection node at each moment.

4. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of obtaining the neighborhood influence node includes: The product of the preliminary anomaly confidence and the preset maximum neighborhood radius is used as the influence radius of each intersection node; other intersection nodes whose chessboard distance from each intersection node is less than the influence radius are used as neighborhood influencing nodes of each intersection node.

5. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of obtaining the spatiotemporal anomaly confidence level includes: In terms of time sequence, all moments are divided into at least two monitoring time periods, where all monitoring time periods have the same length; under each traffic congestion risk index, all risk index values ​​corresponding to each neighborhood influencing node in each monitoring time period are arranged in time sequence to determine the corresponding time series risk index sequence; Taking the time series risk indicator sequence as a row vector, according to all time series risk indicator sequences corresponding to all neighborhood influence nodes, determine the local space-time matrix corresponding to each traffic congestion risk indicator in each monitoring time period; The local spatiotemporal matrix is ​​normalized and then decomposed by an SVD decomposition algorithm to obtain a corresponding sparse matrix; the abnormality degree of the event is determined according to the mean of the absolute values ​​of all non-zero elements in the sparse matrix; Combine each row vector of the sparse matrix with any other row vector to determine the corresponding row vector tuple; traverse all row vectors of the sparse matrix to determine all row vector tuples; use the normalized value of the Pearson correlation coefficient between two row vectors in each row vector tuple as the corresponding reference correlation; determine the event correlation according to the mean of the reference correlations of all row vector tuples; Determine the reference confidence of the intersection node corresponding to each traffic congestion risk index in each monitoring time period according to the normalized value of the product of the abnormality degree of the event and the correlation of the event; determine the overall confidence of each intersection node in each monitoring time period according to the average of the reference confidences corresponding to all traffic congestion risk indicators; At each intersection node, the corresponding spatiotemporal anomaly confidence is determined based on the overall confidence of the monitoring time period at each moment.

6. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of obtaining the instability coefficient includes: Determine the bad weather judgment coefficient according to whether there is bad weather; at each intersection node, determine the corresponding temperature standard deviation value according to the difference between the ambient temperature at each moment and the preset standard temperature; determine the corresponding humidity standard deviation value according to the difference between the ambient humidity at each moment and the preset standard humidity; determine the temperature and humidity abnormal value according to the product between the temperature standard deviation value and the humidity standard deviation value; normalize the product between the bad weather judgment coefficient and the temperature and humidity abnormal value to determine the instability coefficient of each intersection node at each moment.

7. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of obtaining the traffic anomaly confidence level includes: According to the mean value between the instability coefficient and the spatiotemporal anomaly confidence, the comparative confidence of each intersection node at each moment is determined; the product between the comparative confidence and the preliminary anomaly confidence is normalized to determine the traffic anomaly confidence of each intersection node at each moment.

8. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of obtaining the traffic risk scoring coefficient includes: At each moment, the diffusion influence degree of each intersection node is determined according to the mean of the traffic anomaly confidence of all neighboring influencing nodes corresponding to each intersection node; At each path node, the moment when the corresponding traffic anomaly confidence is less than the preset confidence threshold is regarded as a normal moment; the closest normal moment before each moment is regarded as the abnormal moment of the suspected event; the traffic anomaly confidence of each intersection node at all moments is arranged in chronological order and then curve fitting is performed to determine the abnormal confidence curve of each intersection node; according to the traffic anomaly confidence change trend between each moment on the abnormal confidence curve and the abnormal moment of the suspected event, the event evolution weight of each moment is determined; The product of the traffic anomaly confidence level of each intersection node at each moment, the event evolution weight and the diffusion impact degree is normalized to determine the corresponding traffic risk score coefficient.

9. The streaming data analysis method for smart city operation decision-making according to claim 8 is characterized in that: The process of obtaining the event evolution weight includes: On the anomaly confidence curve, when there is a maximum point between each moment and the corresponding suspected event abnormal moment, the average of the tangent slopes of all moments between the suspected event abnormal moment and the first maximum point thereafter is used as the event evolution weight of each moment; When there is no maximum point between each moment and the corresponding suspected event abnormal moment, the mean of the tangent slopes of all moments between each moment and the corresponding suspected event abnormal moment is used as the event evolution weight of each moment.

10. The streaming data analysis method for smart city operation decision-making according to claim 1 is characterized in that: The process of adjusting the operation decision according to the traffic risk scoring coefficient includes: For each intersection node at the current moment: When the corresponding traffic risk score coefficient is greater than a preset first score threshold and less than or equal to a preset second score threshold, the corresponding intersection node is marked as suspected of congestion on the navigation software; When the corresponding traffic risk score coefficient is greater than the preset second score threshold and less than or equal to the preset third score threshold, the corresponding intersection node is marked as suspected congestion on the navigation software and the traffic police are notified to go to the corresponding intersection node for traffic control; When the corresponding traffic risk score coefficient is greater than the preset third score threshold, the corresponding intersection node is marked as suspected congestion on the navigation software, and the traffic police is notified to go to the corresponding intersection node to conduct traffic control and set up accident warning signs on the road sections adjacent to the corresponding intersection node; wherein the preset first score threshold is less than the preset second score threshold, and the preset second score threshold is less than the preset third score threshold.

Images