A smart city traffic data processing system based on distributed sensors

Abstract

The application discloses a kind of intelligent city traffic data processing systems based on distributed sensor, it is related to urban traffic data processing technical field, by introducing the linkage mechanism of event trigger signal E (Tk) and dispatch state signal S (Tk), the dynamic switching of sensor sampling behavior is realized.Significantly reduce invalid data upload, save network bandwidth and storage resources;It improves the response speed of system to traffic mutation;Avoid the situation of insufficient sampling density at critical moment, improve the capture rate of abnormal event.System extracts the collected vehicle speed vS, traffic density pS and lane occupancy zS from standardized traffic data set, further constructs length traffic mutation rate HΓ and fluctuation intensity factor HΨ, and compares with the set threshold value.Improve the accuracy and stability of abnormal event identification;Support fine-grained, multi-level emergency judgment;Through sensitive parameter driving scheduling logic, improve the intelligent level of overall system.

Description

Technical Field

[0001] This invention relates to the field of urban traffic data processing technology, specifically to a smart city traffic data processing system based on distributed sensors. Background Technology

[0002] Intelligent transportation systems, as a core component of smart city construction, represent a crucial area of ​​convergence between urban informatization, automation, and intelligent control technologies. Within this broader field, traffic sensing technology serves as the primary link in the "collection-analysis-response" information loop, forming the foundation for constructing city-level traffic operation status. Especially in recent years, traffic sensing systems based on distributed sensor networks have gradually become the main technological means for urban road traffic monitoring and management due to their advantages such as wide spatial coverage, flexible node deployment, and abundant real-time data.

[0003] However, existing distributed traffic sensor systems generally employ a fixed-period sampling mechanism, where sensors perceive and upload data on traffic conditions at uniformly set time intervals (e.g., every 30 seconds or 60 seconds). While this mechanism is simple and easy to control, it reveals significant problems in practical applications: a large amount of data collected when traffic conditions are stable exhibits high redundancy and low value, with the system passively collecting a large amount of traffic information that does not show significant changes. This not only results in a significant waste of data storage resources but also places a higher load on transmission link bandwidth and the processing capabilities of edge computing nodes.

[0004] Traditional traffic status data collection is "non-selective and uniform across all time periods." This means that regardless of whether the traffic system is in a stable or fluctuating phase, sensors sample at a uniform frequency, lacking the ability to perceive and respond to sudden changes in traffic conditions. As a result, for the vast majority of the time, the system passively collects a large amount of "repetitive data under stable operation." When a genuine anomaly occurs, because the sampling frequency fails to automatically increase, the system cannot immediately identify the origin, development process, and wavefront propagation characteristics of the anomaly. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a smart city traffic data processing system based on distributed sensors, which solves the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a smart city traffic data processing system based on distributed sensors, comprising a data acquisition and preprocessing module, a traffic status perception module, an abnormal event identification module, a data scheduling and control module, a data aggregation and abnormal behavior analysis module, and a comprehensive evaluation and feedback module;

[0007] The data acquisition and preprocessing module collects traffic data through distributed sensors, and performs cleaning and normalization processing to obtain the traffic dataset TO;

[0008] The traffic condition perception module continuously senses the regional traffic condition at a low frequency of 1 minute and extracts sensitive parameters from the traffic dataset TO.

[0009] The abnormal event identification module performs mutation detection on the sensitive parameters it obtains, acquires event trigger signals, including the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, and compares them with the preset mutation rate threshold and fluctuation threshold to obtain the trigger condition E(Tk) and determine whether the event sampling mode is triggered.

[0010] The data scheduling control module receives the event trigger signal and trigger condition E(Tk), obtains the scheduling status signal S(Tk), and dynamically adjusts the sampling period TS and upload frequency fu of the sensor node to obtain the data scheduling strategy Ts(Tk).

[0011] The data aggregation and abnormal behavior analysis module aggregates the data triggered by the high-frequency sampling mode highTs in the data scheduling strategy Ts(Tk), identifies micro traffic behavior patterns, assesses the state of abnormal persistence, and obtains the speed disturbance attenuation index A(t).

[0012] The comprehensive evaluation and feedback module uses the combined length traffic flow mutation rate HΓ, fluctuation intensity factor HΨ, and speed disturbance attenuation index A(t) to calculate the traffic mutation complexity parameter TBx and evaluate the traffic system status.

[0013] Preferably, the data acquisition and preprocessing module includes a traffic data acquisition unit and a data preprocessing unit;

[0014] The traffic data acquisition unit collects vehicle speed vS through a laser speedometer, traffic flow density pS through a video stream and induction coil, and lane occupancy zS through a magnetoresistive induction coil.

[0015] Vehicle speed vS is obtained by the ratio of the distance of the speed measurement area to the time it takes for the vehicle to pass through it;

[0016] Traffic density pS is obtained by the ratio of the total number of vehicles to the number of road segments.

[0017] Lane occupancy rate zS is obtained by the ratio of the time a lane is occupied to the detection period.

[0018] The collected vehicle speed vS, traffic density pS, and lane occupancy zS are fitted to obtain the original dataset TW;

[0019] The raw dataset TW obtained by the data preprocessing unit is cleaned and normalized to obtain the traffic dataset TO;

[0020] Cleaning includes outlier detection and missing value imputation. Outlier detection uses the standard deviation method to detect outliers in the original dataset TW and removes them. Missing value imputation uses linear interpolation to fill in missing values ​​in the original dataset TW.

[0021] Normalization was performed on the original dataset TW using the Min-Max normalization method to obtain the traffic dataset TO.

[0022] The traffic dataset TO is obtained using the following formula:

[0023]

[0024] In the formula, TOb represents the b-th data item in the traffic dataset TO, TWb represents the b-th data item in the original dataset TW, minTWb represents the valley value of the b-th data item in the original dataset TW, and maxTWb represents the peak value of the b-th data item in the original dataset TW.

[0025] Preferably, the traffic state perception module includes a traffic basic state extraction unit and a sensitive dynamic feature calculation unit;

[0026] The traffic basic state extraction unit extracts the data points of the traffic dataset TO every minute, including the vehicle speed vS(Tk) of the k-th time slice, the traffic flow density pS(Tk) of the k-th time slice, and the lane occupancy rate zS(Tk) of the k-th time slice.

[0027] The sensitive dynamic feature calculation unit analyzes the vehicle speed vS(Tk), traffic density pS(Tk), and lane occupancy zS(Tk) of the k-th time slice to obtain sensitive parameters, including the traffic flow gradient change rate Γ(Tk) and the fluctuation intensity index Ψ(Tk); and judges the traffic flow status by the traffic flow gradient change rate Γ(Tk) and the traffic status by the fluctuation intensity index Ψ(Tk).

[0028] The rate of change of traffic flow gradient Γ(Tk) is obtained by the following formula:

[0029]

[0030] In the formula, pS(Tk-1) represents the traffic density in the (k-1)th time slot, vS(Tk-1) represents the vehicle speed in the (k-1)th time slot, and Δt represents the time interval;

[0031] Traffic flow status is obtained through matching in the following ways:

[0032] When the rate of change of traffic flow gradient Γ(Tk) > 0, it indicates that the traffic volume is increasing and vehicles are queuing.

[0033] When the rate of change of traffic flow gradient Γ(Tk) ≤ 0, it indicates a decrease in traffic volume, resulting in signal disruption and traffic accidents.

[0034] The volatility index Ψ(Tk) is obtained using the following formula:

[0035]

[0036] In the formula, λ1 and λ2 represent adjustment coefficients, CS represents a constant, σvS(Tk) represents the standard deviation of vehicle speed in the k-th time slice, and PvS(Tk) represents the average vehicle speed in the k-th time slice. Indicates the rate of change in lane occupancy;

[0037] Traffic status is obtained through matching in the following ways:

[0038] When 0.5 > fluctuation intensity index Ψ(Tk) > 0, it indicates that the traffic condition is stable;

[0039] When 1.0 > fluctuation intensity index Ψ(Tk) ≥ 0.5, it indicates that the traffic conditions are unstable and there are phenomena such as speed changes, sudden braking and lane changes.

[0040] Preferably, the abnormal event identification module includes a mutation factor calculation unit and a threshold determination and triggering unit;

[0041] The mutation factor calculation unit performs mutation detection on the acquired traffic flow gradient change rate Γ(Tk) and fluctuation intensity index Ψ(Tk), and calculates the length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ.

[0042] The length-based traffic flow abrupt change rate HΓ is obtained using the following formula:

[0043]

[0044] In the formula, Γ(Tk-1) represents the rate of change of traffic flow gradient in the (k-1)th time slice, and CS represents a constant; to prevent division by zero, it is usually set to 0.001;

[0045] The volatility intensity factor HΨ is obtained using the following formula:

[0046]

[0047] In the formula, Ψ(Tk-1) represents the fluctuation intensity index of the (k-1)th time slice.

[0048] Preferably, the threshold determination and triggering unit compares the acquired length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ with the preset mutation rate threshold THΓ and fluctuation threshold THΨ to obtain the triggering condition E(Tk) and determine the event sampling mode triggering status:

[0049] The trigger condition E(Tk) is obtained using the following formula:

[0050]

[0051] The event sampling mode trigger state is obtained in the following way:

[0052] When the trigger condition E(Tk) = 0, it indicates that the traffic condition is stable, the event sampling mode is not triggered, and the normal low-frequency sampling is maintained.

[0053] When the trigger condition E(Tk) = 1, it means that the current traffic state meets the change condition, triggering the event sampling mode, performing high-frequency sampling, and calculating the change index Ω(Tk) to classify the sampling mode level;

[0054] The mutation index Ω(Tk) is obtained using the following formula:

[0055] Ω(Tk)=α1*HΓ+α2*HΨ;

[0056] In the formula, α1 and α2 represent the preset weight values ​​of the length-based traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively;

[0057] Sampling mode classification is obtained in the following way:

[0058] When 0 < mutation index Ω(Tk) < 0.1, it indicates the first-level mode, where fluctuations are detected with small amplitude;

[0059] When 0.1 ≤ mutation index Ω(Tk) < 0.3, it indicates a second-level mode, with changes in traffic flow and speed disturbances; start intermediate frequency sampling, such as a 20s cycle; upload event information to the central platform and annotate it; start a continuous observation window for 3 minutes;

[0060] When 0.3 ≤ mutation index Ω(Tk) < 0.6, it indicates the third level mode, which includes traffic accidents, temporary road closures, and traffic control failures; high-frequency sampling is initiated, such as a 5-second cycle; the signal timing strategy is updated by pushing the data to the traffic control platform; and it is recommended that police or patrol units be dispatched to the scene for verification.

[0061] When 0.6 ≤ mutation index Ω(Tk) < 1.0, it indicates the fourth level mode, major accidents, road closures, and extreme weather; all nodes start second-level sampling and reporting; notify the emergency platform and traffic police command center; activate the surrounding area collaborative response mechanism, adjust the green wave band at multiple intersections, and switch signal priorities; it is recommended to automatically report to the city traffic digital twin platform for regional analysis.

[0062] Preferably, the data scheduling control module includes an event status maintenance and scheduling judgment unit and a scheduling execution and node control unit;

[0063] The event status maintenance and scheduling judgment unit determines the current event status based on the acquired trigger condition E(Tk) and acquires the scheduling status signal S(Tk);

[0064] The formula for determining the current event status is as follows:

[0065]

[0066] In the formula, Co(Tk) represents the number of cycles that have been continuously in the "non-event state";

[0067] The scheduling status signal S(Tk) is obtained using the following formula:

[0068]

[0069] In the formula, High represents high frequency, low represents low frequency, and Δtr represents the hysteresis recovery period.

[0070] Preferably, the scheduling execution and node control unit dynamically adjusts the sampling period TS and upload frequency fu of the sensor node according to the acquired scheduling status signal S(Tk), and adopts different data scheduling strategies Ts(Tk) under different states.

[0071] The data scheduling policy Ts(Tk) is obtained through the following formula:

[0072]

[0073] In the formula, highTs represents high-frequency sampling mode with a sampling period TS of 5 seconds and an upload frequency fu of 0.2Hz; it activates edge caching, real-time upload channel, and reserved bandwidth resources. lowTs represents low-frequency sampling mode with a sampling period TS of 60 seconds and an upload frequency fu of 0.0167Hz; it enables local caching, batch upload, and saves network resources.

[0074] Preferably, the data aggregation and abnormal behavior analysis module includes a high-frequency data aggregation and behavior indicator extraction unit and a decay function analysis unit;

[0075] The high-frequency data aggregation and behavior index extraction unit fits the data collected by the high-frequency sampling mode highTs and extracts the micro traffic behavior index Bj(t), including the average speed PvSj of the j-th road segment at time t, the standard deviation of the vehicle speed σvSj of the j-th road segment at time t, and the intensity of driving behavior Rerj.

[0076] The intensity of driving behavior, Rerj, is obtained using the following formula:

[0077]

[0078] In the formula, Nj(t) represents the total number of vehicles in the j-th road segment at time t, and Π represents the indicator function. δv represents the rate of change of the vehicle's speed over time t, and δv represents the preset speed change threshold.

[0079] Preferably, the attenuation function analysis unit constructs the speed disturbance attenuation index A(t) based on the obtained micro traffic behavior index Bj(t);

[0080] The velocity disturbance decay exponent A(t) is obtained by the following formula:

[0081]

[0082] In the formula, to represents the initial time, and t represents the current time. Let represent the second derivative of the microscopic traffic behavior index, e represent a constant, L represent the attenuation coefficient, and τ represent the integral variable. t represents a specific point in time in history, between to and t.

[0083] Preferably, the comprehensive evaluation and feedback module constructs the traffic change complexity parameter TBx based on the obtained length traffic flow change rate HΓ, fluctuation intensity factor HΨ, and speed disturbance attenuation index A(t), and evaluates the traffic system state through the traffic change complexity parameter TBx to obtain the traffic state level;

[0084] The traffic mutation complexity parameter TBx is obtained using the following formula:

[0085]

[0086] In the formula, α1 and α2 represent the preset weight values ​​of the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively, and β represents the preset weight value of the speed disturbance attenuation index A(t).

[0087] Traffic status levels are obtained through the following matching method:

[0088] When 0 < traffic mutation complexity parameter TBx < 0.3, it indicates normal operation and normal fluctuations; maintain low-frequency sampling and record basic data;

[0089] When the traffic change complexity parameter TBx is less than 0.5 and 0.3, it indicates that the local fluctuations are enhanced and observation is required; the sampling frequency should be appropriately increased and the observation buffer should be activated.

[0090] When the traffic mutation complexity parameter TBx is between 0.5 and 0.7, it indicates a moderate abnormal state; medium-frequency sampling is initiated, sensor coverage is enhanced, and the data may be reported to the platform.

[0091] When the traffic change complexity parameter TBx is between 0.7 and 1.0, it indicates a sudden change in traffic conditions and a high risk. Immediately switch to high-frequency sampling mode, feed back to the intervention platform, and issue a dispatch signal system or linkage command.

[0092] This invention provides a smart city traffic data processing system based on distributed sensors, which has the following advantages:

[0093] (1) During system operation, a linkage mechanism between the event trigger signal E(Tk) and the scheduling status signal S(Tk) is introduced to achieve dynamic switching of sensor sampling behavior. The system only enters high-frequency sampling mode when it identifies sudden or non-stable traffic events, such as traffic changes or accident fluctuations. This significantly reduces the upload of invalid data, saves network bandwidth and storage resources, improves the system's response speed to traffic changes, avoids insufficient sampling density at critical moments, and improves the abnormal event capture rate.

[0094] The system extracts vehicle speed (vS), traffic density (pS), and lane occupancy (zS) from a standardized traffic dataset. It further constructs the length-based traffic flow mutation rate (HΓ) and fluctuation intensity factor (HΨ), comparing them with set thresholds to form a structured and quantitative mutation triggering mechanism. This improves the accuracy and stability of abnormal event identification; supports fine-grained, multi-level emergency event judgment; and enhances the overall system intelligence level by driving scheduling logic through sensitive parameters.

[0095] (2) By combining different types of sensors, such as laser speedometers, video streaming systems, and magnetoresistive induction coils, into the same data acquisition module, vehicle speed (vS), traffic density (pS), and lane occupancy (zS) are collected respectively. This technically achieves targeted and precise acquisition of different physical quantities, improving the coverage dimension and sampling quality of the data. Simultaneously, all collected data are uniformly processed using a defined formula and fused into a raw dataset (TW), forming a spatiotemporal integrated representation of traffic elements in terms of data structure. This ensures the accuracy of each traffic characteristic parameter, avoiding data distortion caused by the accuracy error of a single sensor; enhances the acquisition system's responsiveness to complex road conditions and traffic state transitions; and provides high-dimensional, highly consistent basic data support for subsequent behavior analysis and dynamic recognition.

[0096] (3) A mutation factor calculation unit is introduced, and two mutation sensitivity factors are constructed by comprehensively using the traffic flow gradient change rate Γ(Tk) and the fluctuation intensity index Ψ(Tk): the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ. Compared with the traditional judgment method that relies on a single flow threshold, this method more accurately identifies the mutation trend and evolution direction through time series difference + exponential scaling analysis. It improves the accuracy of emergency event identification, especially in the stage of minor disturbances and early accident evolution; supports the capture of dynamic changes in the traffic system rather than static threshold exceedances; and establishes a "source-expansion-stability" continuity judgment basis for traffic disturbances, enhancing the coherence and interpretability of identification.

[0097] (4) By setting up an event state maintenance and scheduling judgment unit, a continuous "non-event" counter Co(Tk) and a delayed recovery period Δtr are introduced to dynamically monitor the triggering condition E(Tk), and the sampling state is switched to low-frequency mode only after the continuous stable state is met. This "event response → delayed recovery" mechanism solves the problem of frequent changes in sampling mode caused by frequent fluctuations in traffic state in a short period of time after a sudden event.

[0098] To avoid frequent switching of sampling frequency in the system at critical states, maintain the continuity of data acquisition and the stability of system operation; to have fault-tolerant characteristics, improve the robustness of identification when dealing with "tailing disturbances" or slight echoes after sudden events; and to provide a judgment mechanism with memory function for sensor scheduling, improve the intelligence of system judgment. Attached Figure Description

[0099] Figure 1 This is a schematic diagram of the block flow of a smart city traffic data processing system based on distributed sensors according to the present invention.

[0100] Figure 2 This is a schematic diagram of the steps of the traffic flow status determination system of the present invention;

[0101] Figure 3 This is a bar chart showing the variation of the mutation index of this invention with the mutation rate of the traffic flow length and the fluctuation intensity factor.

[0102] Figure 4 This is a line graph showing the change in the mutation index of this invention. Detailed Implementation

[0103] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0104] Example 1

[0105] This invention provides a smart city traffic data processing system based on distributed sensors. Please refer to [link / reference]. Figure 1 - Figure 4 It includes a data acquisition and preprocessing module, a traffic condition perception module, an abnormal event identification module, a data scheduling and control module, a data aggregation and abnormal behavior analysis module, and a comprehensive evaluation and feedback module;

[0106] The data acquisition and preprocessing module collects traffic data through distributed sensors, and performs cleaning and normalization processing to obtain the traffic dataset TO;

[0107] The traffic condition perception module continuously senses the regional traffic condition at a low frequency of 1 minute and extracts sensitive parameters from the traffic dataset TO.

[0108] The abnormal event identification module performs mutation detection on the sensitive parameters it obtains, acquires event trigger signals, including the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, and compares them with the preset mutation rate threshold and fluctuation threshold to obtain the trigger condition E(Tk) and determine whether the event sampling mode is triggered.

[0109] The data scheduling control module receives the event trigger signal and trigger condition E(Tk), obtains the scheduling status signal S(Tk), and dynamically adjusts the sampling period TS and upload frequency fu of the sensor node to obtain the data scheduling strategy Ts(Tk).

[0110] The data aggregation and abnormal behavior analysis module aggregates the data triggered by the high-frequency sampling mode highTs in the data scheduling strategy Ts(Tk), identifies micro traffic behavior patterns, assesses the state of abnormal persistence, and obtains the speed disturbance attenuation index A(t).

[0111] The comprehensive evaluation and feedback module uses the combined length traffic flow mutation rate HΓ, fluctuation intensity factor HΨ, and speed disturbance attenuation index A(t) to calculate the traffic mutation complexity parameter TBx and evaluate the traffic system status.

[0112] In this embodiment, by introducing a linkage mechanism between the event trigger signal E(Tk) and the scheduling status signal S(Tk), dynamic switching of sensor sampling behavior is achieved. The system only enters high-frequency sampling mode when it identifies sudden or non-stable traffic events, such as traffic anomalies or accident fluctuations. This significantly reduces invalid data uploads, saving network bandwidth and storage resources; improves the system's response speed to traffic anomalies; avoids insufficient sampling density at critical moments; and enhances the abnormal event detection rate.

[0113] The system extracts vehicle speed (vS), traffic density (pS), and lane occupancy (zS) from a standardized traffic dataset. It further constructs the length-based traffic flow mutation rate (HΓ) and fluctuation intensity factor (HΨ), comparing them with set thresholds to form a structured and quantitative mutation triggering mechanism. This improves the accuracy and stability of abnormal event identification; supports fine-grained, multi-level emergency event judgment; and enhances the overall system intelligence level by driving scheduling logic through sensitive parameters.

[0114] To address the evolutionary state following abrupt events, the system constructs a disturbance response function based on vehicle speed time series, thereby deriving the speed disturbance decay index A(t), which characterizes whether the disturbance is spreading, weakening, or persisting. It captures traffic wavefront propagation trends and initial accident signs; supports dynamic traffic state modeling and trend prediction; provides a basis for subsequent regulation and guidance, and possesses proactive intervention capabilities. Multiple traffic abrupt event indicators are integrated to construct a unified traffic abrupt event complexity parameter TBx, and a grading strategy is used to assess the system's current operational status. A risk grading model is established to provide scientific support for signal control and police dispatch; achieving a shift from passive response to proactive prediction and tiered response.

[0115] The entire process, from data acquisition, status recognition, event triggering, sampling scheduling, behavior analysis to risk assessment, forms a complete workflow. Modules are logically nested, and data flow and control flow are highly integrated. The closed-loop processing of the entire process forms a unified system architecture; each module is decoupled and its functions are focused, facilitating subsequent expansion and deployment; it supports deployment on edge computing platforms or cloud-based distributed platforms, improving system portability and deployment flexibility.

[0116] Example 2

[0117] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 and Figure 2 Specifically: the data acquisition and preprocessing module includes a traffic data acquisition unit and a data preprocessing unit;

[0118] The traffic data acquisition unit collects vehicle speed vS through a laser speedometer, traffic flow density pS through a video stream and induction coil, and lane occupancy zS through a magnetoresistive induction coil.

[0119] Vehicle speed vS is obtained by the ratio of the distance of the speed measurement area to the time it takes for the vehicle to pass through it;

[0120] Traffic density pS is obtained by the ratio of the total number of vehicles to the number of road segments.

[0121] Lane occupancy rate zS is obtained by the ratio of the time a lane is occupied to the detection period.

[0122] The collected vehicle speed vS, traffic density pS, and lane occupancy zS are fitted to obtain the original dataset TW;

[0123] The raw dataset TW obtained by the data preprocessing unit is cleaned and normalized to obtain the traffic dataset TO;

[0124] Cleaning includes outlier detection and missing value imputation. Outlier detection uses the standard deviation method to detect outliers in the original dataset TW and removes them. Missing value imputation uses linear interpolation to fill in missing values ​​in the original dataset TW.

[0125] Normalization was performed on the original dataset TW using the Min-Max normalization method to obtain the traffic dataset TO.

[0126] The traffic dataset TO is obtained using the following formula:

[0127]

[0128] In the formula, TOb represents the b-th data item in the traffic dataset TO, TWb represents the b-th data item in the original dataset TW, minTWb represents the valley value of the b-th data item in the original dataset TW, and maxTWb represents the peak value of the b-th data item in the original dataset TW.

[0129] The traffic condition perception module includes a basic traffic condition extraction unit and a sensitive dynamic feature calculation unit;

[0130] The traffic basic state extraction unit extracts the data points of the traffic dataset TO every minute, including the vehicle speed vS(Tk) of the k-th time slice, the traffic flow density pS(Tk) of the k-th time slice, and the lane occupancy rate zS(Tk) of the k-th time slice.

[0131] The sensitive dynamic feature calculation unit analyzes the vehicle speed vS(Tk), traffic density pS(Tk), and lane occupancy zS(Tk) of the k-th time slice to obtain sensitive parameters, including the traffic flow gradient change rate Γ(Tk) and the fluctuation intensity index Ψ(Tk); and judges the traffic flow status by the traffic flow gradient change rate Γ(Tk) and the traffic status by the fluctuation intensity index Ψ(Tk).

[0132] The rate of change of traffic flow gradient Γ(Tk) is obtained by the following formula:

[0133]

[0134] In the formula, pS(Tk-1) represents the traffic density in the (k-1)th time slot, vS(Tk-1) represents the vehicle speed in the (k-1)th time slot, and Δt represents the time interval;

[0135] Traffic flow status is obtained through matching in the following ways:

[0136] When the rate of change of traffic flow gradient Γ(Tk) > 0, it indicates that the traffic volume is increasing and vehicles are queuing.

[0137] When the rate of change of traffic flow gradient Γ(Tk) ≤ 0, it indicates a decrease in traffic volume, resulting in signal disruption and traffic accidents.

[0138] The volatility index Ψ(Tk) is obtained using the following formula:

[0139]

[0140] In the formula, λ1 and λ2 represent adjustment coefficients, CS represents a constant, σvS(Tk) represents the standard deviation of vehicle speed in the k-th time slice, and PvS(Tk) represents the average vehicle speed in the k-th time slice. Indicates the rate of change in lane occupancy;

[0141] Traffic status is obtained through matching in the following ways:

[0142] When 0.5 > fluctuation intensity index Ψ(Tk) > 0, it indicates that the traffic condition is stable;

[0143] When 1.0 > fluctuation intensity index Ψ(Tk) ≥ 0.5, it indicates that the traffic conditions are unstable and there are phenomena such as speed changes, sudden braking and lane changes.

[0144] In this embodiment, by combining different types of sensors, such as laser speedometers, video streaming systems, and magnetoresistive induction coils, into the same data acquisition module, vehicle speed (vS), traffic density (pS), and lane occupancy (zS) are collected respectively. This technically achieves targeted and precise acquisition of different physical quantities, improving the data coverage and sampling quality. Simultaneously, all collected data are uniformly processed using a defined formula and fused into a raw dataset TW, forming a spatiotemporal integrated representation of traffic elements in terms of data structure. This ensures the accuracy of each traffic characteristic parameter, avoiding data distortion caused by the accuracy error of a single sensor; enhances the acquisition system's responsiveness to complex road conditions and traffic state transitions; and provides high-dimensional, highly consistent basic data support for subsequent behavior analysis and dynamic recognition.

[0145] This embodiment designs a systematic data preprocessing mechanism, introducing outlier detection and removal, as well as missing value imputation, into the original dataset TW to solve problems such as jumps and omissions in the original collected data. Simultaneously, Min-Max normalization is introduced at the data structure layer to standardize the data into a standard traffic dataset TO. This significantly improves the completeness and usability of the collected data; ensures that subsequent model processing stages are not affected by dirty data or scaling bias; and enhances the reliability and stability of traffic situation analysis results.

[0146] This system quantifies and abstracts the road conditions of each segment spatially, and constructs a continuous state flow at the "minute level" in time, achieving near real-time tracking of the basic state. This ensures the system has "continuous perception capability" for any sudden trends; avoids information gaps or misjudgments due to state jumps; and constructs a temporal resolution that naturally matches the traffic change process, facilitating modeling and recognition.

[0147] In this embodiment, two dynamic features with significant identification value are designed in the state perception module:

[0148] Traffic flow gradient change rate Γ(Tk): used to determine the spatial and temporal dynamic evolution of traffic flow;

[0149] The fluctuation intensity index Ψ(Tk) is used to determine velocity consistency, behavioral disturbances, and state stability.

[0150] Both functions construct derivative-type functions based on speed, density, and occupancy, respectively, exhibiting characteristics such as directionality, abrupt change, and continuity, making them particularly suitable for dynamic recognition scenarios. They can accurately characterize common traffic abrupt change patterns such as "queue accumulation" or "fluctuation diffusion"; support the establishment of a highly sensitive triggering mechanism in the system to achieve behavior-level event perception; and have the potential for threshold control and level classification, supporting adaptive response design.

[0151] By defining threshold judgment rules for the traffic flow gradient change rate Γ(Tk) and fluctuation intensity index Ψ(Tk), a two-way judgment logic for "traffic flow status" and "traffic stability status" is established. This transforms traffic status from "numerical features" to "semantic categories," enabling a structural understanding and classification of traffic mutations; providing precise trigger signals for subsequent scheduling and control; reducing the system's reliance on expert experience or fuzzy logic; and improving its level of intelligence.

[0152] Example 3

[0153] This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 2 - Figure 4 Specifically: the abnormal event identification module includes a mutation factor calculation unit and a threshold determination and triggering unit;

[0154] The mutation factor calculation unit performs mutation detection on the acquired traffic flow gradient change rate Γ(Tk) and fluctuation intensity index Ψ(Tk), and calculates the length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ.

[0155] The length-based traffic flow abrupt change rate HΓ is obtained using the following formula:

[0156]

[0157] In the formula, Γ(Tk-1) represents the rate of change of traffic flow gradient in the (k-1)th time slice;

[0158] The volatility intensity factor HΨ is obtained using the following formula:

[0159]

[0160] In the formula, Ψ(Tk-1) represents the fluctuation intensity index of the (k-1)th time slice.

[0161] The threshold determination and triggering unit compares the acquired length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ with the preset mutation rate threshold THΓ and fluctuation threshold THΨ to obtain the triggering condition E(Tk) and determine the event sampling mode triggering status:

[0162] The trigger condition E(Tk) is obtained using the following formula:

[0163]

[0164] The event sampling mode trigger state is obtained in the following way:

[0165] When the trigger condition E(Tk) = 0, it indicates that the traffic condition is stable, the event sampling mode is not triggered, and the normal low-frequency sampling is maintained.

[0166] When the trigger condition E(Tk) = 1, it means that the current traffic state meets the change condition, triggering the event sampling mode, performing high-frequency sampling, and calculating the change index Ω(Tk) to classify the sampling mode level;

[0167] The mutation index Ω(Tk) is obtained using the following formula:

[0168] Ω(Tk)=α1*HΓ+α2*HΨ;

[0169] In the formula, α1 and α2 represent the preset weight values ​​of the length-based traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively;

[0170] Sampling mode classification is obtained in the following way:

[0171] When 0 < mutation index Ω(Tk) < 0.1, it indicates the first-level mode, where fluctuations are detected with small amplitude;

[0172] When 0.1 ≤ mutation index Ω(Tk) < 0.3, it indicates the second level mode, with traffic flow changes and speed disturbances.

[0173] When 0.3 ≤ mutation index Ω(Tk) < 0.6, it indicates the third level mode, which includes traffic accidents, temporary road closures, and traffic control failures.

[0174] When 0.6 ≤ mutation index Ω(Tk) < 1.0, it indicates a level 4 pattern, including major accidents, road closures, and extreme weather.

[0175] Specific examples:

[0176] Set α1 and α2 to 0.6 and 0.4 respectively;

[0177] The length-based traffic flow abrupt change rate HΓ = 0.05 was obtained.

[0178] The volatility intensity factor HΨ = 0.08;

[0179] Calculate and obtain the mutation index Ω(Tk):

[0180] Ω(Tk)=α1*HΓ+α2*HΨ

[0181] = 0.6 × 0.05 + 0.4 × 0.08;

[0182] =0.062

[0183] The mutation index Ω(Tk) = 0.062, indicating a first-level pattern;

[0184] Table 1. Mutation Index Grading Assessment Table:

[0185] Group number Length of traffic flow change rate HΓ Fluctuation intensity factor HΨ Mutation index Ω(Tk) Sampling mode level Group 1 0.60 0.70 0.640 Fourth level Group 2 0.40 0.45 0.420 Level 3 Group 3 0.25 0.30 0.270 Second level Group 4 0.12 0.15 0.132 Second level Group 5 0.05 0.08 0.062 First level

[0186] In this embodiment, a mutation factor calculation unit is introduced, which comprehensively uses the traffic flow gradient change rate Γ(Tk) and the fluctuation intensity index Ψ(Tk) to construct two mutation sensitivity factors: the length-flow mutation rate HΓ and the fluctuation intensity factor HΨ. Compared with the traditional judgment method that relies on a single flow threshold, this method more accurately identifies the mutation trend and evolution direction through time series difference + exponential scaling analysis. This improves the accuracy of sudden event identification, especially in the stage of minor disturbances and early accident evolution; supports the capture of dynamic changes in the traffic system rather than static threshold exceedances; and establishes a "source-expansion-stability" continuity judgment basis for traffic disturbances, enhancing the coherence and interpretability of identification.

[0187] In this embodiment, the threshold determination and triggering unit forms a triggering condition function E(Tk) by comparing two mutation factors with a system preset threshold in real time. This mechanism achieves quantitative triggering and a clearly structured event response signal output. Transforming subjective judgment into system logic enables automatic access to the intelligent control link; provides a reliable condition for the scheduling control module to determine whether to enter high-frequency mode; avoids frequent state switching, and improves system stability and the rationality of the sampling strategy.

[0188] By introducing the traffic mutation index formula Ω(Tk), this system linearly weights two disturbance factors to derive a single numerical index, thereby quantifying the intensity of the current mutation. Subsequently, a four-level segmented matching rule is established to map different disturbance intensities to different response levels. The system response levels corresponding to different traffic states are clearly defined, supporting tiered intervention from "minor disturbance – moderate fluctuation – strong mutation – system anomaly"; a differentiated sampling mechanism is implemented; this helps improve the efficiency of sampling resource utilization and reduce unnecessary high-frequency global operations; and event risk level labels are established to provide decision-making basis for upper-level platform scheduling, alarm push notifications, and resource guidance.

[0189] The embodiment effectively constructs a closed-loop processing framework from data-driven to strategy-linked by setting a clear logical judgment chain: "Calculate the sensitivity factor → compare the preset threshold → output the trigger signal → generate the mutation index → ​​match the level mode".

[0190] It can not only determine "whether a mutation has occurred", but also identify the "degree of mutation" and "response methods";

[0191] It forms a system linkage interface from the data layer to the control layer, which is convenient for embedding into edge devices or traffic management platforms;

[0192] Enhance the autonomous decision-making level and dynamic adjustment capability of the entire intelligent transportation system.

[0193] Example 4

[0194] This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 Specifically: the data scheduling and control module includes an event status maintenance and scheduling judgment unit and a scheduling execution and node control unit;

[0195] The event status maintenance and scheduling judgment unit determines the current event status based on the acquired trigger condition E(Tk) and acquires the scheduling status signal S(Tk);

[0196] The formula for determining the current event status is as follows:

[0197]

[0198] In the formula, Co(Tk) represents the number of cycles that have been continuously in the "non-event state";

[0199] The scheduling status signal S(Tk) is obtained using the following formula:

[0200]

[0201] In the formula, High represents high frequency, low represents low frequency, and Δtr represents the hysteresis recovery period.

[0202] The scheduling execution and node control unit dynamically adjusts the sampling period TS and upload frequency fu of the sensor node based on the acquired scheduling status signal S(Tk), and adopts different data scheduling strategies Ts(Tk) under different states.

[0203] The data scheduling policy Ts(Tk) is obtained through the following formula:

[0204]

[0205] In the formula, highTs represents high-frequency sampling mode with a sampling period TS of 5 seconds and an upload frequency fu of 0.2Hz; lowTs represents low-frequency sampling mode with a sampling period TS of 60 seconds and an upload frequency fu of 0.0167Hz.

[0206] In this embodiment, by setting up an event state maintenance and scheduling judgment unit, a continuous "non-event" counter Co(Tk) and a delayed recovery period Δtr are introduced to dynamically monitor the triggering condition E(Tk), and the sampling state is switched to low-frequency mode only after a continuous stable state is met. This "event response → delayed recovery" mechanism solves the problem of frequent changes in sampling mode caused by frequent fluctuations in traffic conditions in a short period of time after a sudden event.

[0207] To avoid frequent switching of sampling frequency in the system at critical states, maintain the continuity of data acquisition and the stability of system operation; to have fault-tolerant characteristics, improve the robustness of identification when dealing with "tailing disturbances" or slight echoes after sudden events; and to provide a judgment mechanism with memory function for sensor scheduling, improve the intelligence of system judgment.

[0208] This embodiment decouples the judgment process from the execution process in terms of logical structure. First, the state judgment module generates a scheduling state signal S(Tk) based on the event triggering conditions and lag period. Then, the execution unit drives the adjustment of specific sampling behaviors according to S(Tk). This hierarchical control architecture improves the logical interpretability of the system response and the structural clarity of operation. It increases the transparency of the system's scheduling decisions, facilitating system development, verification, and functional maintenance; it achieves clear separation of responsibilities between modules, enhancing scalability and deployment flexibility; and it supports future replacement or training optimization of the state judgment function.

[0209] In this embodiment, the sampling period TS and upload frequency fu are mapped to the sampling state S(Tk) through scheduling execution and node control unit, which is then used to define a set of explicit scheduling policies Ts(Tk).

[0210] High-frequency mode: TS = 5s, fu = 0.2Hz;

[0211] Low frequency mode: TS = 60s, fu = 0.0167Hz;

[0212] This parameterized scheduling mechanism transforms abstract "high-frequency / low-frequency" operations into quantitative index control, facilitating automated operation by embedding it into control programs, edge platforms, and remote terminals. It improves the standardization of sampling strategies, aiding system maintenance and platform-level deployment; enables precise control of network traffic and computing resources, avoiding resource waste; and ensures sufficient data density during periods of high volatility while reducing system workload during periods of stability.

[0213] This embodiment truly establishes a closed-loop sensing and sampling linkage mechanism of "event perception → scheduling state switching → node control response," enabling sensor nodes to no longer work passively but to have the ability to schedule and adjust autonomously on demand. It effectively reduces unnecessary high-frequency data uploads under normal traffic conditions, improving the overall data utilization rate of the sensing system; in abnormal situations, it quickly switches to high-frequency sampling mode to ensure that the data density meets the needs of identification and analysis; and it strengthens the system's adaptability in resource-constrained environments, such as scenarios with limited power and communication bandwidth at edge nodes.

[0214] Example 5

[0215] This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 1 Specifically: the data aggregation and abnormal behavior analysis module includes a high-frequency data aggregation and behavior indicator extraction unit and a decay function analysis unit;

[0216] The high-frequency data aggregation and behavior index extraction unit fits the data collected by the high-frequency sampling mode highTs and extracts the micro traffic behavior index Bj(t), including the average speed PvSj of the j-th road segment at time t, the standard deviation of the vehicle speed σvSj of the j-th road segment at time t, and the intensity of driving behavior Rerj.

[0217] The intensity of driving behavior, Rerj, is obtained using the following formula:

[0218]

[0219] In the formula, Nj(t) represents the total number of vehicles in the j-th road segment at time t, and Π represents the indicator function. δv represents the rate of change of the vehicle's speed over time t, and δv represents the preset speed change threshold.

[0220] The attenuation function analysis unit constructs the speed disturbance attenuation index A(t) based on the obtained micro traffic behavior index Bj(t);

[0221] The velocity disturbance decay exponent A(t) is obtained by the following formula:

[0222]

[0223] In the formula, to represents the initial time, and t represents the current time. Let represent the second derivative of the micro-level traffic behavior index, e represent a constant, L represent the attenuation coefficient, and τ represent the integral variable.

[0224] The comprehensive assessment and feedback module constructs the traffic change complexity parameter TBx based on the obtained length traffic flow change rate HΓ, fluctuation intensity factor HΨ, and speed disturbance attenuation index A(t), and evaluates the traffic system status through the traffic change complexity parameter TBx to obtain the traffic status level.

[0225] The traffic mutation complexity parameter TBx is obtained using the following formula:

[0226]

[0227] In the formula, α1 and α2 represent the preset weight values ​​of the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively, and β represents the preset weight value of the speed disturbance attenuation index A(t).

[0228] Traffic status levels are obtained through the following matching method:

[0229] When 0 < traffic mutation complexity parameter TBx < 0.3, it indicates normal operation and normal fluctuations;

[0230] When 0.3 ≤ traffic change complexity parameter TBx < 0.5, it indicates increased local fluctuations, which need to be observed.

[0231] When 0.5 ≤ traffic mutation complexity parameter TBx < 0.7, it indicates a moderate abnormal state;

[0232] When 0.7 ≤ traffic change complexity parameter TBx < 1.0, it indicates a sudden change in traffic status and a high risk.

[0233] In this embodiment, under the high-frequency sampling mode after the event is triggered, a high-frequency data aggregation and behavior index extraction unit is set up to process the data collected by the sensor nodes from a fine-scale temporal-spatial dimension, extracting a micro-level traffic behavior index set Bj(t), including the average speed PvSj of the j-th road segment at time t, the standard deviation of the vehicle speed σvSj of the j-th road segment at time t, and the intensity of driving behavior Rerj. The intensity is extracted by comparing the rate of change of vehicle speed with a threshold, realistically reflecting the fluctuations in individual behavior within the traffic flow. This enables a micro-level characterization of abnormal traffic behaviors, such as sudden braking, lane changing, and acceleration; expands the system's perception depth from the "road segment level" to the "driving behavior level"; and enhances the system's early identification capability for latent anomalies such as minor disturbances, accident precursors, and traffic wave fronts.

[0234] This embodiment uses a decay function analysis unit to perform second-order derivative analysis on the micro-behavioral index Bj(t) and embeds an exponential decay kernel function to construct the disturbance response index A(t) of the traffic system at time t, thereby reflecting the system's "sensitivity, memory, and evolutionary trend" to historical disturbances. This function attenuates and superimposes the acceleration changes of early disturbances onto the current moment, forming a trend curve of the disturbance over time. It can determine whether traffic anomalies are weakening, maintaining, or spreading; it has a strong ability to capture dynamic trends and is suitable for complex scenarios such as traffic waves and accident chains; it establishes a "disturbance absorption capacity" assessment mechanism for the traffic system as a basis for subsequent intervention decisions.

[0235] The comprehensive assessment and feedback module integrates three key factors: the traffic flow mutation rate HΓ, the fluctuation intensity factor HΨ, and the speed disturbance attenuation index A(t), and constructs the traffic mutation complexity parameter TBx through weighted fusion. This index sets four traffic state levels (from normal operation to systemic mutation) based on segmented intervals, enabling the system to quantify the current operating state based on the index value and output clear state labels. This establishes a clearly structured and logically continuous traffic system state level system, facilitating the platform to implement functions such as automatic labeling, response strategy linkage, and risk warning; and enabling a leap from "whether it is abnormal" to "how severe the abnormality is," supporting the deployment of a hierarchical dispatch response mechanism.

[0236] This embodiment transforms the static judgment of the original indicator values ​​into a "derivative trend + exponential weight superposition" model through formula construction, forming a structural identification framework for behavioral disturbances in the system, further enhancing the adaptive perception capability of the traffic system. It strengthens the system's understanding of the evolution process of abnormal behavior; reduces misjudgments of short-term fluctuations; enhances the robustness of the system's judgment under continuous time-series data; and supports higher-level data-driven prediction and intervention models.

[0237] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart city traffic data processing system based on distributed sensors, characterized in that: It includes a data acquisition and preprocessing module, a traffic condition perception module, an abnormal event identification module, a data dispatch and control module, a data aggregation and abnormal behavior analysis module, and a comprehensive evaluation and feedback module; The data acquisition and preprocessing module collects traffic data through distributed sensors, and performs cleaning and normalization processing to obtain the traffic dataset TO; The traffic condition perception module continuously senses the regional traffic condition at a low frequency of 1 minute and extracts sensitive parameters from the traffic dataset TO. The abnormal event identification module performs mutation detection on the sensitive parameters it obtains, acquires event trigger signals, including the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, and compares them with the preset mutation rate threshold and fluctuation threshold to obtain the trigger condition E(Tk) and determine whether the event sampling mode is triggered. The data scheduling control module receives the event trigger signal and trigger condition E(Tk), obtains the scheduling status signal S(Tk), and dynamically adjusts the sampling period TS and upload frequency fu of the sensor node to obtain the data scheduling strategy Ts(Tk). The data aggregation and abnormal behavior analysis module aggregates the data when the high-frequency sampling mode highTs is triggered in the data scheduling strategy Ts(Tk), identifies micro traffic behavior patterns, assesses the state of abnormal persistence, and obtains the speed disturbance attenuation index A(t). The comprehensive assessment and feedback module integrates the traffic flow mutation rate HΓ, the fluctuation intensity factor HΨ, and the speed disturbance attenuation index A(t) to calculate the traffic mutation complexity parameter TBx and assess the traffic system status.

2. The smart city traffic data processing system based on distributed sensors according to claim 1, characterized in that: The data acquisition and preprocessing module includes a traffic data acquisition unit and a data preprocessing unit; The traffic data acquisition unit collects vehicle speed vS through a laser speedometer, traffic flow density pS through a video stream and induction coil, and lane occupancy zS through a magnetoresistive induction coil. Vehicle speed vS is obtained by the ratio of the distance of the speed measurement area to the time it takes for the vehicle to pass through it; Traffic density pS is obtained by the ratio of the total number of vehicles to the number of road segments. Lane occupancy rate zS is obtained by the ratio of the time a lane is occupied to the detection period. The collected vehicle speed vS, traffic density pS, and lane occupancy zS are fitted to obtain the original dataset TW; The raw dataset TW obtained by the data preprocessing unit is cleaned and normalized to obtain the traffic dataset TO; Cleaning includes outlier detection and missing value imputation. Outlier detection uses the standard deviation method to detect outliers in the original dataset TW and removes them. Missing value imputation fills in the missing values ​​in the original dataset TW using linear interpolation; Normalization was performed on the original dataset TW using the Min-Max normalization method to obtain the traffic dataset TO. The traffic dataset TO is obtained using the following formula: ； In the formula, TOb represents the b-th data item in the traffic dataset TO, TWb represents the b-th data item in the original dataset TW, minTWb represents the valley value of the b-th data item in the original dataset TW, and maxTWb represents the peak value of the b-th data item in the original dataset TW.

3. The smart city traffic data processing system based on distributed sensors according to claim 1, characterized in that: The traffic condition perception module includes a basic traffic condition extraction unit and a sensitive dynamic feature calculation unit; The traffic basic state extraction unit extracts the data points of the traffic dataset TO every minute, including the vehicle speed vS(Tk) of the k-th time slice, the traffic flow density pS(Tk) of the k-th time slice, and the lane occupancy rate zS(Tk) of the k-th time slice. The sensitive dynamic feature calculation unit analyzes the vehicle speed vS(Tk), traffic density pS(Tk), and lane occupancy zS(Tk) of the k-th time slice to obtain sensitive parameters, including the traffic flow gradient change rate Γ(Tk) and the fluctuation intensity index Ψ(Tk). The unit then uses the traffic flow gradient change rate Γ(Tk) to determine the traffic flow status and the fluctuation intensity index Ψ(Tk) to determine the traffic status. The rate of change of traffic flow gradient Γ(Tk) is obtained by the following formula: ； In the formula, pS(Tk-1) represents the traffic density in the (k-1)th time slot, vS(Tk-1) represents the vehicle speed in the (k-1)th time slot, and Δt represents the time interval; Traffic flow status is obtained through matching in the following ways: When the rate of change of traffic flow gradient Γ(Tk) > 0, it indicates that the traffic volume is increasing and vehicles are queuing. When the rate of change of traffic flow gradient Γ(Tk) ≤ 0, it indicates a decrease in traffic volume, resulting in signal disruption and traffic accidents. The volatility index Ψ(Tk) is obtained using the following formula: ； In the formula, This represents the adjustment coefficient, and CS represents a constant. Let PvS(Tk) represent the standard deviation of vehicle speed in the k-th time slice, and let PvS(Tk) represent the average vehicle speed in the k-th time slice. d represents the rate of change of lane occupancy, and d represents the integral function. Traffic status is obtained through matching in the following ways: When 0.5 > fluctuation intensity index Ψ(Tk) > 0, it indicates that the traffic condition is stable; When 1.0 > fluctuation intensity index Ψ(Tk) ≥ 0.5, it indicates that the traffic condition is unstable and there are phenomena such as speed change, sudden braking and lane change.

4. The smart city traffic data processing system based on distributed sensors according to claim 3, characterized in that: The abnormal event identification module includes a mutation factor calculation unit and a threshold determination and triggering unit; The mutation factor calculation unit performs mutation detection on the obtained traffic flow gradient change rate Γ(Tk) and fluctuation intensity index Ψ(Tk), and calculates the length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ. The length-based traffic flow abrupt change rate HΓ is obtained using the following formula: ； In the formula, Γ(Tk-1) represents the rate of change of traffic flow gradient in the (k-1)th time slice; The volatility intensity factor HΨ is obtained using the following formula: ； In the formula, Ψ(Tk-1) represents the fluctuation intensity index of the (k-1)th time slice.

5. A smart city traffic data processing system based on distributed sensors according to claim 4, characterized in that: The threshold determination and triggering unit compares the acquired length traffic flow mutation rate HΓ and fluctuation intensity factor HΨ with the preset mutation rate threshold THΓ and fluctuation threshold THΨ to obtain the triggering condition E(Tk) and determine the event sampling mode triggering status: The trigger condition E(Tk) is obtained using the following formula: ； The event sampling mode trigger state is obtained in the following way: When the trigger condition E(Tk) = 0, it indicates that the traffic condition is stable, the event sampling mode is not triggered, and the normal low-frequency sampling is maintained. When the trigger condition E(Tk) = 1, it means that the current traffic state meets the change condition, triggering the event sampling mode, performing high-frequency sampling, and calculating the change index Ω(Tk) to classify the sampling mode level; The mutation index Ω (Tk) is obtained using the following formula: ； In the formula, These represent the preset weight values ​​for the length-based traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively. Sampling mode classification is obtained in the following way: When 0 < mutation index Ω (Tk) < 0.1, it indicates the first-level mode, and fluctuations are detected; When 1 ≤ mutation index Ω (Tk) < 0.3, it indicates the second level mode, with traffic flow changes and speed disturbances. When 3 ≤ mutation index Ω (Tk) < 0.6, it indicates the third level mode, which includes traffic accidents, temporary road closures, and traffic control failures. When 6 ≤ mutation index Ω (Tk) < 1.0, it indicates a level 4 pattern, road closures, and extreme weather.

6. A smart city traffic data processing system based on distributed sensors according to claim 5, characterized in that: The data scheduling and control module includes an event status maintenance and scheduling judgment unit and a scheduling execution and node control unit; The event status maintenance and scheduling judgment unit determines the current event status based on the acquired trigger condition E(Tk) and acquires the scheduling status signal S(Tk). The formula for determining the current event status is as follows: ； In the formula, Co(Tk) represents the number of consecutive cycles in the "non-event state"; The scheduling status signal S(Tk) is obtained using the following formula: ； In the formula, High represents high frequency, low represents low frequency, and Δtr represents the hysteresis recovery period.

7. A smart city traffic data processing system based on distributed sensors according to claim 6, characterized in that: The scheduling execution and node control unit dynamically adjusts the sampling period TS and upload frequency fu of the sensor node based on the acquired scheduling status signal S(Tk), and adopts different data scheduling strategies Ts(Tk) under different states. The data scheduling policy Ts(Tk) is obtained using the following formula: ； In the formula, highTs represents high-frequency sampling mode; lowTs represents low-frequency sampling mode.

8. A smart city traffic data processing system based on distributed sensors according to claim 7, characterized in that: The data aggregation and abnormal behavior analysis module includes a high-frequency data aggregation and behavior indicator extraction unit and a decay function analysis unit; The high-frequency data aggregation and behavior indicator extraction unit fits the data collected in the high-frequency sampling mode highTs and extracts the micro-traffic behavior indicator Bj(t), including the average speed PvSj of the j-th road segment at time t and the standard deviation of the vehicle speed of the j-th road segment at time t. And the intensity of driving behavior (Rerj); The intensity of driving behavior, Rerj, is obtained using the following formula: ； In the formula, Nj(t) represents the total number of vehicles on the j-th road segment at time t. Indicates an indicator function, This represents the rate of change of the vehicle's speed over time t. The preset speed change threshold is represented by vSj(t), which represents the speed of the j-th road segment at time t.

9. A smart city traffic data processing system based on distributed sensors according to claim 8, characterized in that: The attenuation function analysis unit constructs the speed disturbance attenuation index A(t) based on the obtained micro traffic behavior index Bj(t). The velocity disturbance decay exponent A(t) is obtained by the following formula: ； In the formula, to represents the initial time, and t represents the current time. The second derivative of the microscopic traffic behavior index is represented by e, which is a constant, and L is the attenuation coefficient. This represents the integral variable.

10. A smart city traffic data processing system based on distributed sensors according to claim 9, characterized in that: The comprehensive assessment and feedback module constructs the traffic change complexity parameter TBx based on the obtained length traffic flow change rate HΓ, fluctuation intensity factor HΨ, and speed disturbance attenuation index A(t), and evaluates the traffic system status through the traffic change complexity parameter TBx to obtain the traffic status level. The traffic mutation complexity parameter TBx is obtained using the following formula: ； In the formula, β represents the preset weight values ​​of the length traffic flow mutation rate HΓ and the fluctuation intensity factor HΨ, respectively, and β represents the preset weight value of the speed disturbance attenuation index A(t); Traffic status levels are obtained through the following matching method: When 0 < traffic mutation complexity parameter TBx < 0.3, it indicates normal operation and normal fluctuations; When 3 ≤ traffic change complexity parameter TBx < 0.5, it indicates local fluctuations that need to be observed. When 5 ≤ traffic mutation complexity parameter TBx < 0.7, it indicates an abnormal state; When 7 ≤ the traffic change complexity parameter TBx < 1.0, it indicates a sudden change in traffic status.

Images