A vehicle-mounted terminal data acquisition and abnormality early warning method

Abstract

This invention relates to the field of traffic control and vehicle data monitoring technology, and discloses a method for data acquisition and anomaly warning from an on-board terminal. The method includes: receiving real-time data streams from the on-board terminal and parsing the protocol header; replacing the serial number and re-distributing the message when online status is confirmed; receiving parameter instructions to match abnormal behavior windows; switching the storage mode and resetting the cache area write address to reclaim cache space when duplicate messages occur; outputting spatiotemporal phase adjustment features to modify sampling window parameters; switching the sliding window to a gradient space window when the spatiotemporal phase adjustment features are less than the congestion threshold; sliding the window only when the displacement increment crosses the physical boundary of the grid or the state bit flips; and using an asynchronous network framework to push discrete warning command streams. This invention regulates the data window and cache cycle, eliminates redundant messages from competing for computing resources, and improves the stability of traffic warnings.

Description

Technical Field

[0001] This invention belongs to the field of traffic control and vehicle data monitoring technology, and in particular relates to a method for data acquisition and anomaly warning of vehicle-mounted terminals. Background Technology

[0002] Currently, spatiotemporal coordinate data is typically collected using vehicle-mounted devices, and trajectory messages are distributed to multiple platforms via wireless networks. After receiving the messages, the server parses the protocol header and message body, and uses a fixed sliding window to filter the spatiotemporal sequence in the time-series data stream to monitor vehicle travel paths. Under normal road traffic conditions, this ensures that the vehicle message upload rate is evenly distributed over time, keeping the underlying protocol parsing, data writing, and time-series feature comparison within normal loads and maintaining normal asynchronous message flow between the data pipeline and distributed storage modules. However, when a specific road segment experiences large-scale vehicle aggregation or stagnation, the limitations of the existing information architecture become fully apparent. As the physical displacement increment of vehicles approaches zero due to congestion, the vehicle-mounted devices trigger data retransmission mechanisms or frequently switch status messages, causing a burst of messages within the local grid in a short period of time. At this time, the traditional central server, heavily reliant on the ideal assumption of even distribution of data flow over time, still uses a fixed time-series window step size and static cache allocation rules, resulting in a massive accumulation of static messages containing duplicate information in the processing window and storage memory in a short period of time.

[0003] To cope with data surges caused by physical space congestion, industry strategies typically involve blindly expanding processor hardware cores or shortening cache lifecycles. However, increasing hardware cores leads to idle computing resources during non-congested periods, resulting in high equipment deployment costs. Shortening storage lifecycles, on the other hand, causes the loss of historical context within the time window, making it impossible for the system to fully trace the evolution of a vehicle's behavior when it resumes operation. Therefore, this linear improvement path, relying on hardware stacking or modification of a single time parameter, is prone to triggering the inherent conflict between operating costs and information integrity, and cannot solve the technical problem of reverse misalignment between vehicle physical displacement and computing power consumption. At the control logic level, existing technologies focus on communication link construction and physical channel partitioning, neglecting data... The communication logic is deeply coupled with the physical motion state. Chinese invention patent CN116233762B discloses a communication method, device, vehicle and storage medium between vehicles. It extracts feature information through image recognition and matches multiple access technology parameters to achieve precise allocation of communication channels and alleviate multi-vehicle broadcast conflicts at the link layer. The core logic of this solution anchors the deterministic construction of the communication link and does not penetrate to the adaptive control of message load. Under extremely low speed or static congestion conditions, even if the channel allocation is precise, the vehicle terminal still continuously reports redundant messages with almost zero physical displacement increment. It lacks consideration of the mapping relationship between physical displacement state and message reporting density. The complex event processing engine needs to process a massive amount of invalid and repetitive data streams within the defined channel, resulting in the ineffective occupation of computing resources.

[0004] Therefore, how to dynamically link the adjustment parameters of the data processing window and cache lifecycle with the physical displacement state of the vehicle, and adaptively allocate system computing resources under the condition of local vehicle gathering peak, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention aims to solve the problem of system computing power contention and resource squeeze caused by the disconnect between the processing window and cache cycle and the physical motion state of the vehicle.

[0006] In this technical solution, a method for data acquisition and anomaly warning of an in-vehicle terminal includes the following steps: Step S1: Receive the real-time data stream from the vehicle terminal, parse the 808 protocol header of the real-time data stream and determine the online status; when the online status is online, replace the packet header serial number, add header and footer identifiers and check bits, repackage the message and distribute the message; Step S2: Receive parameter instructions from multi-level platforms, match parameter instructions with abnormal behavior windows using discrete state control parameters, and when duplicate messages occur in the real-time data stream, switch the storage mode to compact release mode and reset the write address of the cache area to reclaim the data stream cache space occupied by duplicate messages. Step S3: Output the spatiotemporal phase adjustment feature quantity, and use the spatiotemporal phase adjustment feature quantity as the pre-judgment benchmark to modify the sampling window parameters of the complex event processing engine; when the spatiotemporal phase adjustment feature quantity is less than the congestion critical threshold, it is determined that the vehicle terminal is in a local clustering state, triggering the adaptive window width control flow, and switching the sliding window to a gradient space window based on the physical space displacement change rate. When the physical space displacement increment of the vehicle terminal does not cross the grid physical boundary and the state bit of the state register does not flip, the sliding window remains fixed. When the physical space displacement increment of the vehicle terminal crosses the grid physical boundary or the state bit of the state register flips, the sliding window slides forward and pushes the discrete early warning command flow to the control center bus through the asynchronous network framework.

[0007] Preferably, in step S3, when the spatiotemporal phase adjustment feature is greater than or equal to the free flow threshold, the vehicle terminal is determined to be in normal driving state, and the adaptive window width control flow controls the complex event processing engine to maintain a sliding window with a standard time step to match the characteristics of the real-time data stream; wherein, the free flow threshold is 0.54m / packet, the congestion threshold is 0.03m / packet, and the grid physical boundary is 50m.

[0008] Preferably, step S1 includes the following sub-steps: step S11, parsing the 808 protocol header of the real-time data stream and extracting the terminal physical identifier and packet header sequence number from the real-time data stream; step S12, comparing the terminal physical identifier with the whitelist database, and determining that the online status of the vehicle terminal is online when the terminal physical identifier matches the whitelist database.

[0009] Preferably, step S2 includes the following sub-steps: step S21, receiving parameter instructions from the multi-level platform, and establishing a time-series mapping relationship between the parameter instructions of the multi-level platform using discrete state control parameters; step S22, monitoring the writing speed of distributed spatiotemporal data and the underlying information processing latency, and triggering the storage mode to switch to compact release mode when the writing speed and the underlying information processing latency are inversely misaligned.

[0010] Preferably, converting the storage mode to a compact release mode includes: identifying duplicate messages in the real-time data stream and discarding redundant fields in the duplicate messages, and shortening the lifespan of the corresponding trajectory logic file in the cache database from 300s to 45s, so as to release the data stream cache space occupied by duplicate messages.

[0011] Preferably, pushing discrete early warning command streams to the control center bus via an asynchronous network framework includes: capturing abnormal behavior features using an adaptive window width control flow, filling the alarm information data structure with spatiotemporal phase adjustment features and spatial grid topology encoding, generating a data stream containing the physical features of the target traffic node by retrieving the device alarm rule template, and pushing the data stream to the control center bus to establish a discrete early warning command stream.

[0012] Preferably, switching the sliding window to a gradient spatial window based on the rate of change of physical spatial displacement includes: dividing the geographic space into a grid network consisting of multiple grids, and defining the boundary of the grid network as the physical boundary of the grid.

[0013] Preferably, the state register state bit flipping includes: modifying the characteristic state bit of the state register when the vehicle corresponding to the vehicle terminal changes from a driving state to a stationary state, or from a stationary state to a driving state.

[0014] Preferably, after pushing the discrete early warning command stream to the control center bus through the asynchronous network framework, the method further includes: receiving the discrete early warning command stream by the control center bus, parsing the physical characteristics of the target traffic node, and adjusting the timing parameters of the traffic signal controller according to the physical characteristics of the target traffic node.

[0015] Compared with existing technologies, the vehicle-mounted terminal data acquisition and anomaly warning method of the present invention has the following advantages: 1. In vehicle terminal data acquisition and anomaly warning, the physical space displacement increment within the current grid is extracted by continuous spatiotemporal coordinate sequence and the total number of valid trajectory messages is counted. The spatiotemporal phase adjustment feature generated by the division operator of the two is used as the pre-judgment benchmark, so that the algorithm sampling window parameters of the complex event processing engine are deterministically correlated with the physical motion state of the vehicle. When local congestion causes a surge in data retransmission, the time sequence window is adaptively switched to the gradient space control window as assumed. The sliding window only slides forward when the physical space displacement crosses the grid boundary or the state register flips, thereby eliminating the parasitic interference of redundant messages on the computing resources of the computing node in the vehicle stationary state at the mechanism level.

[0016] 2. Based on the numerical range of the spatiotemporal phase adjustment feature quantity feedback, the lifecycle of the trajectory logic file in the cache database is dynamically controlled. When the high-frequency aggregation of the fleet causes a sudden surge in local data throughput, the file lifecycle is adjusted through a step-by-step shortening strategy. The long-term storage mode under the general stable distribution assumption is adaptively transformed into a compact release mode, and high-concurrency memory space is reclaimed and released in situ. This avoids the massive duplicate messages from causing a squeeze on the distributed storage module, and collaboratively ensures the stability of asynchronous writing of request parameters in cross-platform protocol instruction flow. It also solves the memory overload risk when the physical flow rate of data and the underlying information processing latency are reversed.

[0017] 3. By using an asynchronous network framework to associate the abnormal event capture status with the data distribution channel, dynamic scheduling is triggered when the adaptive window width control flow captures micro-abnormal behavior characteristics. The spatiotemporal phase adjustment characteristics and spatial grid topology encoding produced in the previous steps are encapsulated in situ into the alarm information data structure. By calling the preset rule template of the matching device, a data stream containing the physical characteristics of specific traffic nodes is generated in real time. A discrete early warning command stream push path is established directly to the control center bus. A whole-machine collaborative closed loop is constructed, which integrates front-end protocol access, mid-end spatiotemporal phase calculation and back-end network asynchronous distribution, achieving decoupling of the underlying network communication program and business processing logic under high-throughput congestion conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart of the execution steps of a vehicle-mounted terminal data acquisition and anomaly warning method according to the present invention; Figure 2 This is a state transition diagram of a vehicle-mounted terminal data acquisition and anomaly early warning method according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A method for data acquisition and anomaly warning from an in-vehicle terminal includes the following steps: Step S1: Receive the real-time data stream from the vehicle terminal, parse the 808 protocol header of the real-time data stream and determine the online status; when the online status is online, replace the packet header serial number, add header and footer identifiers and check bits, repackage the message and distribute the message; Step S2: Receive parameter instructions from multi-level platforms, match parameter instructions with abnormal behavior windows using discrete state control parameters, and when duplicate messages occur in the real-time data stream, switch the storage mode to compact release mode and reset the write address of the cache area to reclaim the data stream cache space occupied by duplicate messages. Step S3: Output the spatiotemporal phase adjustment feature quantity, and use the spatiotemporal phase adjustment feature quantity as the pre-judgment benchmark to modify the sampling window parameters of the complex event processing engine; when the spatiotemporal phase adjustment feature quantity is less than the congestion critical threshold, it is determined that the vehicle terminal is in a local clustering state, triggering the adaptive window width control flow, and switching the sliding window to a gradient space window based on the physical space displacement change rate. When the physical space displacement increment of the vehicle terminal does not cross the grid physical boundary and the state bit of the state register does not flip, the sliding window remains fixed. When the physical space displacement increment of the vehicle terminal crosses the grid physical boundary or the state bit of the state register flips, the sliding window slides forward and pushes the discrete early warning command flow to the control center bus through the asynchronous network framework.

[0021] Preferably, in step S3, when the spatiotemporal phase adjustment feature is greater than or equal to the free flow threshold, the vehicle terminal is determined to be in normal driving state, and the adaptive window width control flow controls the complex event processing engine to maintain a sliding window with a standard time step to match the characteristics of the real-time data stream; wherein, the free flow threshold is 0.54m / packet, the congestion threshold is 0.03m / packet, and the grid physical boundary is 50m.

[0022] Preferably, step S1 includes the following sub-steps: step S11, parsing the 808 protocol header of the real-time data stream and extracting the terminal physical identifier and packet header sequence number from the real-time data stream; step S12, comparing the terminal physical identifier with the whitelist database, and determining that the online status of the vehicle terminal is online when the terminal physical identifier matches the whitelist database.

[0023] Preferably, step S2 includes the following sub-steps: step S21, receiving parameter instructions from the multi-level platform, and establishing a time-series mapping relationship between the parameter instructions of the multi-level platform using discrete state control parameters; step S22, monitoring the writing speed of distributed spatiotemporal data and the underlying information processing latency, and triggering the storage mode to switch to compact release mode when the writing speed and the underlying information processing latency are inversely misaligned.

[0024] Preferably, converting the storage mode to a compact release mode includes: identifying duplicate messages in the real-time data stream and discarding redundant fields in the duplicate messages, and shortening the lifespan of the corresponding trajectory logic file in the cache database from 300s to 45s, so as to release the data stream cache space occupied by duplicate messages.

[0025] Preferably, pushing discrete early warning command streams to the control center bus via an asynchronous network framework includes: capturing abnormal behavior features using an adaptive window width control flow, filling the alarm information data structure with spatiotemporal phase adjustment features and spatial grid topology encoding, generating a data stream containing the physical features of the target traffic node by retrieving the device alarm rule template, and pushing the data stream to the control center bus to establish a discrete early warning command stream.

[0026] Preferably, switching the sliding window to a gradient spatial window based on the rate of change of physical spatial displacement includes: dividing the geographic space into a grid network consisting of multiple grids, and defining the boundary of the grid network as the physical boundary of the grid.

[0027] Preferably, the state register state bit flipping includes: modifying the characteristic state bit of the state register when the vehicle corresponding to the vehicle terminal changes from a driving state to a stationary state, or from a stationary state to a driving state.

[0028] Preferably, after pushing the discrete early warning command stream to the control center bus through the asynchronous network framework, the method further includes: receiving the discrete early warning command stream by the control center bus, parsing the physical characteristics of the target traffic node, and adjusting the timing parameters of the traffic signal controller according to the physical characteristics of the target traffic node.

[0029] Example 1: In a continuously operating traffic and logistics monitoring system, the operation mode for data interaction and real-time anomaly warning between the vehicle-mounted terminal and the upper-level platform is as follows: The system receives real-time data streams from the vehicle-mounted terminal. The internal data parsing module parses the 808 protocol header of the data stream and extracts the terminal's physical identifier and packet header sequence number. The online status of the vehicle-mounted terminal is determined by comparing it with a preset whitelist database. If the terminal is online, the packet header sequence number is replaced, and specific header and footer identifiers and check bits are added to complete the message repackaging process. The system distributes the repackaged data stream to the corresponding business processing queue. When performing message repackaging, the system strips the start and end characters of the 808 protocol header from the original real-time data stream and extracts the original packet header sequence number. The system first assigns a unique serial number to the header serial number field. Then, it calls the local serial number counter to generate a globally unique new serial number, which is then overwritten in place by the original header serial number field. The reconstructed message data is then appended with standard two-byte CRC16 checksums, and hexadecimal fixed identifiers are encapsulated at the beginning and end of the message. After repackaging, the system uses a first-in-first-out (FIFO) queue mechanism to directly push the reconstructed complete message into the distribution channel of the high-concurrency message queue, thereby achieving real-time smooth distribution of the business processing queue. Upon receiving request commands from the upper-level platform based on the 809 protocol, the system asynchronously writes the parsed request parameters to the Redis cache database and obtains the network address and port information of the data service through the cluster management node. Based on the request parameters, it then directs the data to the specified vehicle-mounted... The terminal issues control commands and establishes a data transmission channel to retrieve real-time data streams. Specifically, when the system matches parameter commands and abnormal behavior windows from multiple platforms, the parameter commands issued by the upper-level platform contain control level codes for specific spatiotemporal regions. The system parses these control level codes into discrete window correction coefficients. Simultaneously, it performs matrix mapping between the aforementioned spatiotemporal phase adjustment feature, which serves as a discrete state control parameter, and these window correction coefficients. This dynamically adjusts the upper limit of the sliding window time used to capture specific abnormal behaviors. When the discrete state control parameters reflect that the vehicle is in a low-speed, clustered state, and the control level code indicates a high-load road segment, the system adaptively stretches the judgment delay threshold of the abnormal behavior window from the baseline of 30 seconds to 120 seconds to match the vehicle... The system analyzes the long-term temporal characteristics of vehicles under congested conditions to achieve a strong causal mapping between parameter commands and abnormal behavior windows. Based on the unique identifier of the vehicle terminal, the system extracts a continuous spatiotemporal coordinate sequence within a preset historical sampling period. By calculating the Euclidean distance between the current sampling point and the previous sampling point, the system obtains the physical spatial displacement increment of the vehicle terminal within the current spatial grid area. It also counts the total number of valid trajectory messages received by the network interface within this sampling period. The physical spatial displacement increment is divided by the total number of valid trajectory messages to generate a spatiotemporal phase adjustment feature. To eliminate the time reference loss caused by non-fixed-period reporting, this invention introduces a fixed sliding time observation window when counting the total number of valid trajectory messages. Within this fixed time observation window...The system not only counts the absolute total number of received valid messages, but also uses the difference between the timestamps of the latest and earliest received messages to perform frequency correction on the total number of messages. If the actual time span is shortened due to channel congestion, the total number of messages is multiplied by the ratio of the baseline time span to the actual time span to obtain the corrected total number of messages for division. This ensures that even under conditions of terminal frequency conversion reporting or network packet loss, the generated characteristic quantity still possesses standard time-frequency constraints and physical consistency. This characteristic quantity, as a discrete state control parameter characterizing the nonlinear correlation between the physical displacement increment and the message reporting density, is bound to the distributed spatiotemporal data processing module of the traffic control server.

[0030] The system uses the spatiotemporal phase adjustment feature as a pre-judgment benchmark for the complex event processing engine, activating the adaptive window width control flow. When this feature is greater than or equal to 0.54 m / packet, the system determines that the vehicle terminal is in a normal driving state, and the complex event processing engine maintains a sliding window with a standard time step to perform feature matching on the time-series data stream. When this feature is less than 0.03 m / packet, the system determines that the vehicle terminal is in a physically localized clustered state, triggering the window update mechanism to back off, switching the sliding window to a gradient space window based on the rate of change of physical spatial displacement. The system sets the physical boundary of the grid to 50 m. When the physical spatial displacement increment of the vehicle terminal crosses the physical boundary of the grid, or when the corresponding state register flips its state bit, the sliding window slides forward. The system uses this mechanism to effectively cut off the redundant trajectory messages that affect the computing power of the computing nodes when the vehicle is stationary. Source occupancy: When the vehicle terminal changes from a driving state to a stationary state, the characteristic state bit of the status register is modified to trigger a window update. When the complex event processing engine captures an abnormal event under the adaptive window width control flow, an early warning mechanism is triggered. Utilizing the t-io network framework based on asynchronous input / output technology, the device alarm rule template is retrieved, and an alarm information format containing the current spatial grid topology encoding and spatiotemporal phase adjustment characteristics is dynamically generated. This alarm information is then distributed and pushed to the distribution bus of the traffic control center in real time, ultimately outputting a discrete early warning command stream. This command stream is received by the traffic signal controller, which adjusts its timing parameters according to the physical characteristics of the target traffic node. The traffic signal timing adjustment establishes a nonlinear dynamic mapping between the node vehicle congestion density and the green light duration based on the traffic flow rate conservation principle. The green light extension duration for a specific phase of the traffic signal controller is determined according to the following formula: ,in, This determines the number of seconds to extend the green light for the current phase to be adjusted in the traffic signal controller. The value is an integer ranging from 0 to 30. This is a dimensionless gain increment control constant, with a value of 25. The dimensionless spatiotemporal phase state adjustment characteristic output from the preliminary step is taken, with a value range of 0 to 0.54. The spatiotemporal phase state adjustment characteristic is affected by the local clustering state of the vehicle terminal. When the flow is reduced, the traffic signal controller extends the green light time for the corresponding convergence direction to improve the spatial flow guidance rate. In practical engineering derivation, the formula is established based on the nonlinear dynamic mapping between the vehicle congestion density at the node and the green light duration. The dimensionless gain control constant K, with a value of 25, is derived and calibrated based on the maximum permissible extension of the green light in a single instance by the traffic signal controller, combined with the maximum free flow characteristic of the system, through boundary condition derivation. When vehicles are congested and completely stationary, the spatiotemporal phase adjustment characteristic is... Approaching zero, the theoretical extension time calculated by the formula is 25 seconds, which is accurately nested within the 0 to 30 seconds safety control range set by the system. This can maximize the spatial flow guidance and avoid secondary computational and traffic resource squeeze on intersecting roads due to the extreme extension of one-way green lights.

[0031] Under this condition, when the physical space displacement increment of the vehicle terminal does not cross the physical boundary of the grid and the status bit of the status register does not flip, the system keeps the sliding window fixed, thereby eliminating the preemption of computing node computing resources by redundant trajectory messages when the vehicle is stationary. At the same time, the system shortens the life cycle of the trajectory logic file in the Redis cache database from 300s to 45s, and releases the cache memory in place to offset the storage squeeze caused by high-frequency messages, thereby decoupling the business processing logic from the underlying network push program.

[0032] Example 2: This example objectively demonstrates the engineering effectiveness and creativity of its technical solution through multi-dimensional engineering verification with a control group. Specific experimental steps and data are as follows: Experimental Platform and Parameter Settings: This experiment uses a traffic logistics monitoring simulation test bench, which integrates a high-precision clock synchronization module, a multi-channel signal generator, and a data acquisition and analysis server. The experiment is designed to simulate typical urban congestion conditions and stress test the dynamic correlation between traffic flow density and terminal data reporting frequency. To ensure the objectivity and authenticity of the experimental data, during the experiment, the system superimposed random Gaussian white noise with an amplitude of 0.5V and harmonic interference with a frequency of 50Hz onto the original data stream through the signal generator to simulate channel noise and electromagnetic interference in a real vehicle communication environment. The experimental design parameters are as follows: the sampling period is set to achieve an optimal balance between the real-time performance of data acquisition and the system's data processing load and energy consumption. When the environmental noise variance exceeds 0.15, the system automatically increases the time step of the sliding window to filter out noise jitter. The control experiment and data analysis include a control group and an experimental sample group. Control group one uses the existing technology with a conventional fixed-period polling acquisition method, while control group two is an improved control group that removes the spatiotemporal phase adjustment feature feedback mechanism. The experimental sample group uses a gradient spatial window control strategy. The data flow stress test is based on the principle of discrete event queuing theory. A packet drop probability matrix and transmission delay variance are dynamically constructed in the software stream processing transmission link to simulate high-concurrency network conditions. The system establishes the network packet loss rate. With data bus throughput The linear mapping formula: ,in, The simulated packet loss rate for the application layer transmission link ranges from 0 to 0.30. This is the congestion decay factor constant, with a value of 0.0005. This represents the total number of packets received by the network bus per unit of time, ranging from 0 to 600. The total number of received packets can be adjusted... The input scale is used to generate random digital packet loss perturbations of 0% to 30% at the stream processing front end to detect the stability coefficient of the complex event processing engine under non-stationary data stream input conditions.

[0033] Under localized traffic density of 120 vehicles per kilometer, in Control Group 1, when vehicles were stationary, the continuous sending of repetitive heartbeat messages led to a peak computing load of 88.5% on the computing nodes. Due to network bus congestion, 25.3% of trajectory anomaly warning messages were lost. Although Control Group 2 introduced some anomaly warning logic, the lack of feedback correction for spatiotemporal phase adjustment features resulted in a lag in sliding window switching during vehicle deceleration, causing the warning logic trigger point to deviate from the preset boundary by approximately 15.2m, with a warning response delay of 4.8s. Under the same conditions, the experimental group dynamically triggered the gradient space window using spatiotemporal phase adjustment features, keeping the computing load on the computing nodes stably below 32.6%. By eliminating redundant messages, the warning message loss rate was reduced to 0.3%. Furthermore, because the triggering conditions had clear physical displacement crossing boundary characteristics, the deviation between the trajectory coordinates triggered by the warning and the actual physical value was controlled within 0.8m, and the logic response time was shortened to 0.4s.

[0034] To verify the rationality of setting the free-flow threshold and congestion threshold, gradient experiments were conducted on the feature quantities. Experimental data revealed that when the free-flow threshold is set to 0.54 m / packet, the system achieves an optimal balance between data throughput and computational resource consumption. In actual calibration, the above thresholds and boundary values ​​were determined based on simulation experiments of urban traffic flow dynamics. The free-flow threshold of 0.54 m / packet corresponds to the ratio of the baseline speed of vehicles under smooth urban conditions to the standard reporting frequency. Setting it too high could lead to normal vehicles being misjudged as clustered. The congestion threshold of 0.03 m / packet corresponds to the lower limit of the creeping displacement when vehicles experience extreme congestion or creeping. Setting it below this value would prevent the system from effectively sensing small clustering data spikes. The grid physical boundary is set to 50 m, an engineering spatial scale determined based on half the average queue length at urban road intersections. If the threshold is too large, it will be impossible to accurately capture the microscopic evolution of local clusters. If it is too small, it will lead to frequent triggering of the sliding window, resulting in redundant consumption of computing resources. If the threshold is increased to 1.25 m / packet, normally driving vehicles will be frequently identified as clustered, increasing the number of sliding window switching by 140% and causing unnecessary computing power fluctuations. If the congestion threshold is lowered to 0.002 m / packet, the system will be too sensitive to small displacements, triggering false warnings under conditions such as road vibration, with the false alarm rate rising to 18.4%. Experimental results show that this invention effectively avoids the nonlinear collapse of system computing power resources under congestion peaks by physically anchoring the sliding window parameters with spatiotemporal phase adjustment features. The features serve as the judgment benchmark, performing a deterministic physical quantization mapping between the microscopic motion state of vehicles and the macroscopic system load, ensuring the stability of the abnormal warning logic and the determinism of computing resource allocation under extreme traffic conditions.

[0035] Example 3: This example combines Figures 1 to 2 This paper describes a method for data acquisition and anomaly warning from an in-vehicle terminal, such as... Figure 1 As shown, step S1, parsing the protocol header and repackaging and distributing, is connected to step S2, matching instructions and switching to mode when encountering duplicates, via a solid arrow marked with a real-time data stream. Step S2, matching instructions and switching to mode when encountering duplicates, is connected to the cache database to reclaim duplicate message space via a dashed arrow marked with a compact release mode. At the same time, step S2, matching instructions and switching to mode when encountering duplicates, is connected to step S3, pushing early warning based on feature quantities, via a solid arrow marked with a logical flow. Step S3, pushing early warning based on feature quantities, pushing early warning based on window changes, is connected to the complex event processing engine's adaptive switching sampling window via a dashed arrow marked with a pre-determination of modified parameters.

[0036] like Figure 2 As shown, the normal driving state includes maintaining a standard time step sliding window, the trajectory logic file being in long-term storage mode, and the trajectory message upload rate being uniformly distributed along the time axis. When the spatiotemporal phase adjustment characteristic is less than the congestion critical threshold, it switches to the local clustering state via solid arrows. The local clustering state includes switching the sliding window to a gradient spatial window based on the rate of change of physical space displacement, converting repeated messages to a compact release mode, shortening the life cycle, and pushing discrete warning command streams. When the spatiotemporal phase adjustment characteristic is greater than or equal to the free flow threshold, it switches back to normal driving via solid arrows. When the link stability coefficient is below the preset stability threshold range, the system switches from the normal driving state to the edge state via a solid arrow. When the communication environment triggers the edge state, the system switches from the local clustering state to the edge state via a solid arrow. The edge state includes injecting implicit negative event records containing network jitter parameters and automatically stretching the window time step to filter out false trajectories. After the abnormal link recovers, it triggers automatic re-verification and secondary smoothing. When the link stability coefficient rises back to above the stability threshold range, the system switches back from the edge state to the normal driving state via a solid broken line arrow.

[0037] Example 4: In handling data transmission fluctuations caused by transient interruptions in the communication link of an in-vehicle terminal, the stability of the communication link is quantified in real time, and the judgment criteria for abnormal behavior windows are dynamically adjusted based on the stability. This achieves engineering stability in trajectory data acquisition. In application scenarios located in high-density mobile communication base station switching areas, when the in-vehicle terminal faces high-frequency network base station switching, link signal attenuation and packet loss cause the raw data stream to exhibit a non-linear jitter state. At this time, the in-vehicle terminal cannot maintain a stable data reporting sequence. During data reception, the system's communication protocol processing unit monitors the signal strength and packet loss rate of the wireless communication link in real time, extracts the signal-to-noise ratio and transmission delay difference from the real-time data stream, and constructs a link stability coefficient characterizing the link communication quality. When the link stability coefficient is below the preset link stability threshold range, the system determines that the current communication is interrupted. When the communication environment triggers an edge-state condition, implicit negative event records are injected into the cache database storing trajectory data. These records contain the current communication interruption timestamp and the corresponding network jitter parameter. In practice, the link stability coefficient is obtained by weighted summation of the normalized signal-to-noise ratio and the normalized transmission delay difference, with weights of 0.6 and 0.4 respectively. The network jitter parameter is defined as the standard deviation of the time interval between consecutive received packets. When the link stability coefficient is lower than the preset threshold of 0.35, the system triggers an edge-state condition and calculates the stretching of the sliding window time step. This stretching is equal to the base time step multiplied by the ratio of the network jitter parameter to the system's inherent communication cycle. Through this deterministic mathematical mapping relationship, the complex event processing engine can adaptively extend the waiting time of the sliding window, thereby providing accurate displacement offset compensation.

[0038] When performing sliding window feature matching, the complex event processing engine reads and parses implicit negative event records in the cache database, using them as compensation factors to calculate the displacement offset of the sliding window. Based on the compensation factors, the complex event processing engine automatically stretches the time step, making the sliding window's judgment of abnormal behavior lag behind the communication link recovery delay time, thereby filtering out false trajectory anomalies caused by network jitter. When the system determines that the link stability coefficient of the vehicle terminal has risen above the link stability threshold range, the complex event processing engine triggers automatic re-verification logic, reads all data samples marked as associated with implicit negative events within the sliding window, and performs secondary smoothing processing to verify the continuity of trajectory coordinates.

[0039] Under high-frequency handover conditions where the vehicle terminal experiences an average of 5 base station handovers per minute, test data shows that by introducing implicit negative event recording injection logic, the system successfully identified and filtered false alarms of trajectory anomalies caused by communication jitter. The warning accuracy improved from 78.2% before the repair to 96.5%. Furthermore, the system dynamically adjusts the compensation weight of the sliding window according to the fluctuation range of the link stability coefficient, keeping the physical time difference between trajectory determination delay and communication base station handover time within 0.1s. This verifies that the data acquisition and warning logic has deterministic engineering implementation under edge conditions of unstable communication links.

[0040] Example 5: In an environment oriented towards multi-terminal cluster deployment, by executing a pre-boot calibration procedure and offline parameter benchmark calibration, data acquisition anomalies caused by environmental noise interference and differences in hardware manufacturing processes are resolved. Before system deployment, through offline calibration tests, raw messages and positioning coordinates of different vehicle terminals under no-load, full-load, high vibration, and high and low temperature extreme conditions are collected. The correlation curvature between the reporting cycle and physical displacement increment of the vehicle terminal acquisition module under standard environment is fitted, and a standard state feature baseline library is constructed. By comparing the above multi-dimensional baseline data offline, the effective sampling frequency fluctuation range of the vehicle terminal acquisition module under standard environment is calibrated and locked. This serves as a unified logical anchor point for distributed processing, eliminating parameter benchmark offsets caused by individual hardware differences.

[0041] When the vehicle-mounted terminal enters the field deployment condition, the system automatically executes the pre-power-on calibration procedure. After powering on, the vehicle-mounted terminal performs a 10-minute silent calibration mode, recording in real time the satellite positioning coordinate drift and speed sensor noise deviation when the vehicle is stationary. The system writes the measured deviation values ​​into the vehicle-mounted terminal's non-volatile memory as an environmental compensation factor. During system operation, the traffic control server performs instantaneous offset correction on the received real-time trajectory message data based on the environmental compensation factor, compensating for mechanical vibration displacement artifacts generated by the trajectory messages under high-frequency vibration conditions. Specifically, this is done when constructing standard state characteristics. During baseline analysis, the system uses the least squares method to perform quadratic polynomial fitting on the physical displacement increment and reporting period under extreme conditions such as no-load and full-load, thereby obtaining the correlation curvature equation characterizing the standard state. In order to eliminate mechanical vibration displacement artifacts, the system constructs a high-pass filter using the acquired speed sensor noise deviation during operation. From the received real-time latitude and longitude coordinate sequence, it subtracts the satellite positioning coordinate drift represented by the environmental compensation factor and the virtual displacement generated by the time integration of the noise deviation. Through this direct data difference subtraction step, the random coordinate jitter caused by the high-frequency mechanical vibration of the vehicle is filtered out.

[0042] When the system executes the anomaly judgment logic, it calls the environmental compensation factor and the standard state feature baseline library, and performs Euclidean distance matching calculation between the real-time trajectory data and the normal data distribution in the baseline library. When the matching error exceeds the preset anomaly judgment threshold, the system triggers the early warning mechanism. The judgment threshold is set based on the normal data fluctuation range of vehicles traveling in different urban road grids obtained by the system based on the offline calibration phase test. Three times the standard deviation of the data fluctuation in the grid is set as the physical critical point for anomaly judgment. By deploying pre-calibration and operation period compensation mechanisms, false early warning triggers caused by equipment operation deviations or environmental differences are eliminated, ensuring that the data stream received by the traffic control center has engineered spatiotemporal consistency.

[0043] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A method for data acquisition and anomaly warning of an on-board terminal, characterized in that, Includes the following steps: Step S1: Receive the real-time data stream from the vehicle terminal, parse the 808 protocol header of the real-time data stream and determine the online status; when the online status is online, replace the packet header serial number, add header and footer identifiers and check bits, repackage the message and distribute the message; Step S2: Receive parameter instructions from multi-level platforms, match parameter instructions with abnormal behavior windows using discrete state control parameters, and when duplicate messages occur in the real-time data stream, switch the storage mode to compact release mode and reset the write address of the cache area to reclaim the data stream cache space occupied by duplicate messages. Step S3: Output the spatiotemporal phase adjustment feature quantity, and use the spatiotemporal phase adjustment feature quantity as the pre-judgment benchmark to modify the sampling window parameters of the complex event processing engine; When the spatiotemporal phase adjustment characteristic is less than the congestion threshold, the vehicle terminal is determined to be in a local clustering state, triggering an adaptive window width control flow. The sliding window is switched to a gradient space window based on the rate of change of physical space displacement. The sliding window remains fixed as long as the increment of physical space displacement of the vehicle terminal does not cross the physical boundary of the grid and the state bit of the state register does not flip. When the increment of physical space displacement of the vehicle terminal crosses the physical boundary of the grid or the state bit of the state register flips, the sliding window slides forward and pushes a discrete early warning command flow to the control center bus through the asynchronous network framework.

2. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, In step S3, when the spatiotemporal phase adjustment feature is greater than or equal to the free flow threshold, the vehicle terminal is determined to be in normal driving state. The adaptive window width control flow controls the complex event processing engine to maintain a sliding window with a standard time step to match the characteristics of the real-time data stream. The free flow threshold is 0.54m / packet, the congestion threshold is 0.03m / packet, and the grid physical boundary is 50m.

3. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, Step S1 includes the following sub-steps: Step S11, parse the 808 protocol header of the real-time data stream and extract the terminal physical identifier and packet header sequence number from the real-time data stream; Step S12, compare the terminal physical identifier with the whitelist database. When the terminal physical identifier matches the whitelist database, determine that the online status of the vehicle terminal is online.

4. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, Step S2 includes the following sub-steps: Step S21, receive parameter instructions from the multi-level platform, and establish a time-series mapping relationship between the parameter instructions of the multi-level platform using discrete state control parameters; Step S22, monitor the writing speed of distributed spatiotemporal data and the underlying information processing latency, and trigger the storage mode to switch to compact release mode when the writing speed and the underlying information processing latency are inversely misaligned.

5. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, Converting the storage mode to a compact release mode includes: identifying duplicate messages in the real-time data stream and discarding redundant fields in the duplicate messages, and shortening the lifespan of the corresponding trajectory logic file in the cache database from 300s to 45s, in order to release the data stream cache space occupied by duplicate messages.

6. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, Pushing discrete early warning command streams to the control center bus via an asynchronous network framework includes: capturing abnormal behavior features using adaptive window width control flow, filling the alarm information data structure with spatiotemporal phase adjustment features and spatial grid topology encoding, generating a data stream containing the physical features of the target traffic node by retrieving the device alarm rule template, and pushing the data stream to the control center bus to establish discrete early warning command streams.

7. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, Switching the sliding window to a gradient spatial window based on the rate of change of physical spatial displacement includes: dividing the geographic space into a grid network consisting of multiple grids, and defining the boundaries of the grid network as the physical boundaries of the grid.

8. The method for data acquisition and anomaly warning of an on-board terminal according to claim 1, characterized in that, The status register status bit flipping includes: when the vehicle corresponding to the vehicle terminal changes from a driving state to a stationary state, or from a stationary state to a driving state, the characteristic status bit of the status register is modified.

9. The method for data acquisition and anomaly warning of an in-vehicle terminal according to claim 1, characterized in that, After pushing discrete early warning command streams to the control center bus via an asynchronous network framework, the process also includes: receiving discrete early warning command streams from the control center bus, parsing the physical characteristics of the target traffic nodes, and adjusting the timing parameters of the traffic signal controller based on the physical characteristics of the target traffic nodes.

Images