A gas station safety operation intelligent management and control system based on multi-source data perception

Abstract

The application discloses a kind of based on multi-source data perception's gas field station safety operation intelligent management and control system, it is related to digital information transmission technical field, the present application includes edge computing gateway, with multiple perception nodes communication connection, for receiving and buffering multi-source heterogeneous perception data;Storage unit is used to store the preset feature threshold and dynamic window calculation parameter;Synchronization unit is used to extract the feature of key physical event from high-frequency perception data stream, according to feature and dynamic window calculation parameter dynamically generates time synchronization reference window, and utilizes time synchronization reference window to carry out time deviation compensation and data alignment to other perception data stream, forms space-time synchronization data unit.The present application constructs adaptive synchronization algorithm based on dynamic time window, and carries out time alignment to multi-source heterogeneous data with physical event feature as reference, can realize the accurate association of data in physical time dimension.

Description

Technical Field

[0001] This invention belongs to the field of digital information transmission technology, and in particular relates to an intelligent control system for safe operation of gas stations based on multi-source data perception. Background Technology

[0002] Gas station safety operations refer to a series of operating procedures and control measures implemented during gas storage, distribution, transportation, and equipment maintenance to ensure the safety of personnel, equipment, and the environment. In high-risk operations such as tank pre-cooling, valve switching, and pipeline replacement, transient changes during the operation often contain crucial risk information, such as pipeline vibrations caused by fluid impact, sudden temperature changes, and pressure fluctuations. Accurately capturing these transient characteristics and correlating them with operational behaviors is an important means of assessing operational risks.

[0003] Timing inaccuracies are a common problem in multi-source data fusion. Due to differences in hardware clocks, sampling triggering mechanisms, data transmission paths, and network protocols among different sensors, the timestamps carried by each data stream upon entering the analysis system exhibit inconsistent offsets from the actual physical events. This offset is not a fixed value but fluctuates dynamically with changes in network load, node processing capabilities, and transmission media. In multi-source data fusion analysis, if the data streams are not precisely aligned on the timeline, subsequent feature extraction and correlation judgments will be based on incorrect time correspondences.

[0004] To address the time synchronization problem of multi-source data, existing technologies mainly employ global clock synchronization methods based on Network Time Protocol (NTP) or Precision Time Protocol (PTP). These methods periodically synchronize the local clocks of each node at the network level, bringing them closer together. However, in industrial applications, this approach faces challenges. Clock synchronization accuracy is affected by network transmission jitter. In gas stations, where wireless transmission is prevalent and the electromagnetic environment is complex, the synchronization accuracy is insufficient to meet the analysis requirements of high-frequency transient events. Furthermore, this method only unifies the local clocks of nodes and does not address the issue of latency caused by path differences during data transmission; that is, data packets with the same physical timestamp still arrive at the analysis system at different times.

[0005] Other technologies employ a fixed-time-window alignment method, which sets a fixed time window length and treats multi-source data falling within the same window as occurring simultaneously. This method is effective when data sampling rates are similar and transmission delays are stable. However, in actual gas station operation scenarios, the sampling frequencies of different sensors can differ by several orders of magnitude, and data transmission paths involve a hybrid wired and wireless network, resulting in non-stationary delay distributions. A fixed window cannot simultaneously accommodate the temporal resolution of high-frequency data and the arrival delay fluctuations of low-frequency data, easily leading to alignment errors or the exclusion of valid data from the alignment window. Therefore, the following solutions are proposed to address these issues. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent management and control system for safe operation of gas stations based on multi-source data perception. By constructing an adaptive synchronization algorithm based on dynamic time windows, the system aligns multi-source heterogeneous data in time with physical event characteristics as a benchmark. This enables precise correlation of data in the physical time dimension and solves the problem of inaccurate time of multi-source data caused by differences in transmission paths in the prior art.

[0007] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: This invention relates to an intelligent control system for safe operation of gas stations based on multi-source data sensing, comprising: Multiple sensing nodes are used to collect multi-source heterogeneous sensing data within the target work area and assign a unique stream identifier to each data stream. An edge computing gateway, which is communicatively connected to the multiple sensing nodes, is used to receive and cache the multi-source heterogeneous sensing data; The storage unit is used to store preset feature thresholds and dynamic window calculation parameters; The synchronization unit is used to extract features of key physical events from high-frequency sensing data streams, dynamically generate a time synchronization reference window based on the features and the dynamic window calculation parameters, and use the time synchronization reference window to perform time deviation compensation and data alignment for other sensing data streams, forming a spatiotemporal synchronized data unit. The early warning unit is used to determine the operational safety status based on the spatiotemporal synchronization data unit and generate early warning instructions.

[0008] Furthermore, the sensing nodes include vibration sensing nodes, sound wave sensing nodes, temperature sensing nodes, pressure sensing nodes, and visual sensing nodes; when the edge computing gateway receives the multi-source heterogeneous sensing data, each data packet carries a local timestamp, and the edge computing gateway classifies and caches the data streams according to the stream identifier.

[0009] Furthermore, the feature extraction of key physical events by the synchronization unit specifically includes: performing sliding window calculation on the high-frequency vibration or sound wave data stream to obtain the short-time energy and spectral centroid within the window; when the short-time energy exceeds a preset energy threshold and the spectral centroid falls within a preset feature frequency band, it is determined that a key physical event has been captured.

[0010] Furthermore, the dynamic generation of the time synchronization reference window by the synchronization unit specifically includes: calculating the width of the time synchronization reference window based on the characteristic intensity of the captured key physical event, wherein the characteristic intensity includes the ratio of the event peak energy to the environmental noise energy, and the degree of deviation between the event spectrum centroid and the baseline spectrum centroid; the time synchronization reference window extends forward and backward by half the width, with the time when the key physical event is observed in the vibration data stream as the center.

[0011] Furthermore, the width of the time synchronization reference window is positively correlated with the ratio of the event peak energy to the environmental noise energy, and is also positively correlated with the degree of deviation between the event spectrum centroid and the baseline spectrum centroid.

[0012] Furthermore, the synchronization unit uses the time synchronization reference window to align the visual perception data stream, specifically including: within the time synchronization reference window, analyzing the average temperature difference of the thermal imaging area between adjacent frames in the visual perception data stream, determining the timestamp of the video frame in which the average temperature difference first exceeds a set threshold as the visual alignment point, and using the visual alignment point as the precise physical event time reference.

[0013] Furthermore, the synchronization unit uses the time synchronization reference window to align the temperature sensing data stream or pressure sensing data stream, specifically including: within the time synchronization reference window, using the visual alignment point as the time reference, interpolating and resampling the temperature sensing data stream or pressure sensing data stream to obtain compensated synchronization data.

[0014] Furthermore, the early warning unit's judgment of operational safety status based on the spatiotemporal synchronization data unit specifically includes: extracting vibration data and temperature data within the same spatiotemporal synchronization data unit, and calculating the thermal vibration coupling risk index; extracting video keyframes and pressure data within the same spatiotemporal synchronization data unit, and identifying the correlation between worker posture and pressure fluctuations through an image recognition model.

[0015] Furthermore, the warning instruction generated by the warning unit encapsulates a risk level, risk type, synchronization data unit identifier, and execution action code; the system also includes a communication module, used to select different network transmission slices according to the risk level and distribute the warning instruction to the on-site execution terminal.

[0016] Furthermore, the edge computing gateway is also used to annotate the spatiotemporal synchronization data unit and its corresponding physical event reference time, and store them in the storage unit for subsequent risk retrospective analysis.

[0017] The present invention has the following beneficial effects: 1. This invention constructs a dynamic time window and uses the physical event characteristics in high-frequency vibration data as a synchronization benchmark to achieve time alignment of data streams under different transmission paths. This method does not rely on a single fixed clock or global synchronization protocol. Instead, under the condition that there is an uncertain delay in data transmission, it establishes an alignment benchmark based on the signal characteristics corresponding to the event itself at the physical level. This improves the consistency and traceability of multi-source data in the time dimension and reduces the impact of synchronization deviations caused by network fluctuations or node clock drift on subsequent analysis.

[0018] 2. The time synchronization window width in this invention is adaptively adjusted according to the characteristic intensity of the sensed event, so that the system can expand or shrink the time alignment range accordingly when key physical events with different intensities and frequency domain characteristics occur. This method enables the system to cope with factors such as changes in environmental noise and fluctuations in the quality of sensor signals during operation, maintains the stability of the alignment capability of key events under changing working conditions, and improves the applicability and robustness of the system in complex working scenarios.

[0019] 3. This invention integrates multi-dimensional information such as temperature, vibration, pressure and vision based on the aligned synchronous data units, enabling joint analysis of various types of sensing data on the same physical time reference. This method ensures that the semantic correlation between different physical quantities is preserved, avoids feature mismatch or misjudgment due to time misalignment, and provides a more reliable fusion data foundation for subsequent operation risk assessment, which is conducive to improving the overall grasp of the risk situation in complex operation processes.

[0020] 4. This invention uses the spatiotemporally synchronized data units as input for early warning judgment and encapsulates the generated early warning instructions into standard data packets for distribution through a digital information transmission network. This method enables early warning information to enter the transmission stage in a structured manner after completing multi-source data fusion and risk judgment, reducing processing delays and judgment biases caused by inconsistent data times, and helping to improve the efficiency of information flow from perception to response. Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an intelligent control system for safe operation of gas stations based on multi-source data perception, according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 As shown, this invention is an intelligent control system for safe operation of gas stations based on multi-source data perception. The control system includes: Multiple sensing nodes are used to collect multi-source heterogeneous sensing data within the target work area and assign a unique stream identifier to each data stream. An edge computing gateway communicates with multiple sensing nodes to receive and cache multi-source heterogeneous sensing data. The storage unit is used to store preset feature thresholds and dynamic window calculation parameters; The synchronization unit is used to extract features of key physical events from high-frequency sensing data streams, dynamically generate a time synchronization reference window based on the features and dynamic window calculation parameters, and use the time synchronization reference window to compensate for time deviations and align data with other sensing data streams, thus forming a spatiotemporal synchronized data unit. The early warning unit is used to determine the operational safety status based on the spatiotemporal synchronized data unit and generate early warning instructions.

[0025] The sensing nodes include vibration sensing nodes, sound wave sensing nodes, temperature sensing nodes, pressure sensing nodes, and visual sensing nodes. When the edge computing gateway receives multi-source heterogeneous sensing data, each data packet carries a local timestamp, and the edge computing gateway classifies and caches the data streams according to the stream identifier.

[0026] The features extracted by the synchronization unit for key physical events specifically include: performing sliding window calculations on high-frequency vibration or acoustic data streams to obtain short-time energy and spectral centroid within the window; when the short-time energy exceeds a preset energy threshold and the spectral centroid falls within a preset characteristic frequency band, it is determined that a key physical event has been captured.

[0027] The synchronization unit dynamically generates a time synchronization reference window, specifically by: calculating the width of the time synchronization reference window based on the characteristic intensity of the captured key physical events, where the characteristic intensity includes the ratio of the event peak energy to the environmental noise energy, and the degree of deviation between the event spectrum centroid and the baseline spectrum centroid; the time synchronization reference window extends forward and backward by half the width of the key physical event in the vibration data stream, with the time of observation of the key physical event in the vibration data stream as the center.

[0028] The width of the time synchronization reference window is positively correlated with the ratio of event peak energy to ambient noise energy, and also positively correlated with the degree of deviation between the event spectrum centroid and the baseline spectrum centroid.

[0029] The synchronization unit uses a time synchronization reference window to align the visual perception data stream. Specifically, it analyzes the average temperature difference of the thermal imaging area between adjacent frames in the visual perception data stream within the time synchronization reference window, determines the timestamp of the video frame whose average temperature difference first exceeds a set threshold as the visual alignment point, and uses the visual alignment point as the accurate physical event time reference.

[0030] The synchronization unit uses a time synchronization reference window to align the temperature sensing data stream or pressure sensing data stream. Specifically, within the time synchronization reference window, the temperature sensing data stream or pressure sensing data stream is interpolated and resampled using the visual alignment point as the time reference to obtain compensated synchronization data.

[0031] The early warning unit's assessment of operational safety status based on spatiotemporal synchronized data units specifically includes: extracting vibration and temperature data within the same spatiotemporal synchronized data unit and calculating the thermal vibration coupling risk index; extracting video keyframes and pressure data within the same spatiotemporal synchronized data unit and identifying the correlation between worker posture and pressure fluctuations through an image recognition model.

[0032] The warning instructions generated by the warning unit are encapsulated with risk level, risk type, synchronous data unit identifier and execution action code; the system also includes a communication module, which is used to select different network transmission slices according to the risk level and distribute the warning instructions to the on-site execution terminal.

[0033] Edge computing gateways are also used to label spatiotemporal synchronization data units and their corresponding physical event reference times, and store them in storage units for subsequent risk retrospective analysis.

[0034] The specific application of this embodiment is as follows: Step S1: Deployment of multi-source sensing nodes and identification of data flow features At the LNG tank pre-cooling site, multiple sensing nodes are deployed, forming a heterogeneous sensor network. All nodes are connected via industrial switches that support Time-Sensitive Networking Protocol (TSP), but non-deterministic delays are allowed during data packet transmission.

[0035] Step S11, Vibration / Acoustic Sensing Nodes: Six triaxial accelerometers and four acoustic emission sensors are installed at the precooling pipe inlet, elbow, and base of the storage tank. These nodes collect data at the highest sampling rate to capture the structural transient response induced by LNG liquid flow.

[0036] Step S12, Temperature / Pressure Sensing Nodes: Install 12 platinum resistance thermometers and 4 pressure transmitters along the pipeline and tank wall. These nodes collect data at a low frequency to monitor the cooling rate and pressure fluctuations.

[0037] Step S13, Visual Perception Node: Deploy two explosion-proof infrared thermal imaging and visible light binocular cameras in the work area to collect video streams at a rate of 25 frames per second, which are used to identify the position of the workers, their operating posture, and the surface temperature field of the equipment.

[0038] Step S14, Data Stream Identification: At the data link layer, a unique stream identifier (Stream ID) is assigned to each data stream, and its physical attributes are encapsulated. For example, vibration data stream. The identifier is {StreamID: VIB_01, Type: Vibration, SampleRate: 20k, Unit:} SensLoc: Pipe_Inlet}; Temperature data stream The identifier is {StreamID: TMP_05, Type: Temperature, SampleRate: 1, Unit: ℃, SensLoc: Tank_Wall_Mid}.

[0039] Step S2: Coarse-grained time synchronization and acquisition of multi-source data streams Upon system startup, an initial coarse synchronization is performed on all sensing nodes using the Precise Time Protocol (PTP, IEEE 1588), controlling the time deviation of all devices within ±1ms. Subsequently, each node begins collecting data and transmitting it with its local timestamp. The data packets are transmitted to the edge computing gateway through their respective network paths (wired / wireless hybrid).

[0040] Step S21: During data packet transmission, uncertain transmission delays occur due to factors such as network congestion and retransmission. Therefore, the timestamp of the data packets received by the edge gateway With respect to the actual time of the data There is an unknown bias: In the formula, This represents the clock drift amount after initial synchronization.

[0041] Step S22: The edge gateway caches all received raw data streams into a circular buffer and stores them according to their data stream identifiers. At this point, the data streams are out of alignment on the timeline.

[0042] Step S3: Constructing a dynamic baseline window based on job status feature matching This step aims to dynamically identify key physical events in the pre-cooling process from high-frequency vibration / acoustic data streams, and to construct a dynamic time synchronization window based on the occurrence time of these events.

[0043] Step S31, Key Event Feature Extraction: At the start of the precooling operation, the LNG liquid enters the ambient temperature pipeline for the first time, instantly vaporizing and triggering a violent gas-liquid two-phase flow. This flow generates a transient vibration wave packet with distinct characteristics on the pipeline wall. The edge computing gateway continuously monitors the data stream from the vibration sensing node (VIB_01). Perform sliding window processing. Calculate the short-time energy within the window. : ; In the formula, The length of the sliding window. For the sample index within the window, For a moment The amplitude of vibration acceleration.

[0044] At the same time, calculate the spectral centroid within the calculation window. : ; In the formula, For the first Frequency values ​​at each frequency point For the signal within the window at frequency Amplitude spectrum at that location, The number of points in the Fourier transform.

[0045] when Exceeding the preset threshold and Located in the LNG vaporization characteristic frequency band At that time, the system determined that a pre-cool start event had been captured, and the physical occurrence time of this event was recorded as . Due to data stream latency, the event was observed in the vibration data stream at the following time. .

[0046] Step S32, Dynamic Reference Window Parameter Adaptation: Based on the extracted key event feature intensity, dynamically calculate the width of the time synchronization reference window. : ; In the formula, The peak energy of the detected vibration event, The average energy of environmental vibration and noise; The centroid of the detected event spectrum, The baseline spectral centroid; The preset base window width; and For the weighting coefficients, satisfying This formula is used to balance the influence of energy and spectral characteristics on the window width. The physical meaning of this formula is: the more intense the event (higher energy ratio) and the greater the deviation of the characteristic frequency from the baseline, the larger the dynamic window... The wider the window, the better, to accommodate potentially greater data transmission latency jitter. Extending forward and backward from the center This forms a search interval on the time axis.

[0047] Step S4: Adaptive Spatiotemporal Synchronization of Heterogeneous Data Streams The dynamic reference window constructed using step S3 It performs precise time deviation compensation and alignment for all other data streams (temperature, pressure, video) in the circular buffer.

[0048] Step S41, Frame matching of video data stream: The video stream is transmitted at 25fps, and each frame has its local timestamp. The system matches the frame within a dynamic reference window. Within the search, video frames are physically semantically associated with vibration events. By analyzing infrared thermal imaging changes on the pipe surface within the video frames, frames showing abrupt temperature field changes (caused by LNG injection) are identified. Let the video frame sequence be... Calculate the average temperature difference of the thermal imaging region between adjacent frames. Within the reference window, First time exceeding the set threshold video frame timestamps Defined as the visual alignment point for this event.

[0049] Step S42, Synchronization of Temperature / Pressure Data Streams: Temperature and pressure data have low sampling rates and change slowly. The system uses linear interpolation within the reference window, based on... This precise physical event time reference is used for temperature flow. and pressure flow Perform resampling. Calculate the synchronization compensation amount for each temperature data point. : ; In the formula, This is the local timestamp included with the temperature data packet. The compensated temperature value. Obtained through interpolation function: ; This interpolation method is cubic spline interpolation, and its functional form is... In the interval The above satisfies: ; In the formula, The time point for the original temperature data. , , , The spline coefficients are determined through continuity and boundary conditions. This process ensures that all data are available at the time the physical event occurs. Semantic alignment was achieved at precise time points.

[0050] Step S43, Data Fusion and Annotation: The aligned vibration, sound wave, temperature, pressure, and video keyframe data are encapsulated into a unified synchronized data unit, and its absolute physical time reference is annotated. This unit forms the minimum data foundation for all subsequent security analyses.

[0051] Step S5: Spatiotemporal synchronization data fusion and operational safety status assessment By utilizing the synchronous data units generated in step S4, multi-dimensional fusion analysis is performed to achieve accurate identification of risks in pre-cooling operations.

[0052] Step S51, Temperature Field-Vibration Fusion Risk Assessment: Extract temperature data (tank wall temperature) and vibration data (pipeline vibration amplitude) from the synchronous data unit. Calculate the thermal-vibration coupling risk index. : ; In the formula, The effective value of the vibration signal within the synchronization unit. The vibration baseline is the minimum allowable level for safe operation; Given the current cooling rate of the storage tank, This is the maximum permissible cooling rate (e.g., 5°C / min). This index... This intuitively reflects the coupling risk between vibration intensity and cooling rate.

[0053] Step S52, Visual-Pressure Fusion Pose Determination: Extract keyframes from the synchronized data unit and use a pre-trained convolutional neural network (CNN) model to identify the relative position of the operator and the valve operating rod. Simultaneously, obtain the pipeline pressure within the unit. If the CNN recognizes the posture of "the operator's hands are on the valve control panel," and If abnormal fluctuations occur in a short period of time (such as a sudden drop in pressure), the system will determine that there is a risk of non-standard operation.

[0054] Step S6: Generation and distribution of early warning instructions based on information transmission Based on the judgment result of step S5, the edge computing gateway generates warning instructions of different levels. These instructions are distributed to the execution terminals on site in a specific data packet format via the transmission protocol defined by H04L.

[0055] Step S61, Warning Instruction Encapsulation: The warning instruction contains {Timestamp, RiskLevel, RiskType, SyncUnitID, ActionCode}. RiskLevel represents the risk level (1-urgent, 2-warning, 3-notice), and SyncUnitID points to the synchronization data unit that triggered the warning, facilitating post-event review.

[0056] Step S62, Adaptive Transmission Strategy: For emergency commands with RiskLevel=1, the system immediately transmits them with the highest priority to on-site audible and visual alarms, workers' smart wristbands, and the central control room via 5G URLLC (Ultra-Reliable Low-Latency Communication) slices, and forcibly stops the pre-cooling pump. For warning commands with RiskLevel=2, they are transmitted to the central control room display screen and on-site tablet terminals via standard 5G eMBB slices, accompanied by voice prompts. The entire process of generating, encapsulating, and distributing warning information is completed within the framework of digital information transmission (HO4L), ensuring reliable and timely delivery of commands.

[0057] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0058] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A smart control system for safe operation of gas stations based on multi-source data sensing, characterized in that, The control system includes: Multiple sensing nodes are used to collect multi-source heterogeneous sensing data within the target operating area and assign a unique stream identifier to each data stream; An edge computing gateway, which is communicatively connected to the multiple sensing nodes, is used to receive and cache the multi-source heterogeneous sensing data; The storage unit is used to store preset feature thresholds and dynamic window calculation parameters; The synchronization unit is used to extract features of key physical events from high-frequency sensing data streams, dynamically generate a time synchronization reference window based on the features and the dynamic window calculation parameters, and use the time synchronization reference window to perform time deviation compensation and data alignment for other sensing data streams, forming a spatiotemporal synchronized data unit. The early warning unit is used to determine the operational safety status based on the spatiotemporal synchronization data unit and generate early warning instructions.

2. The intelligent control system for safe operation of gas stations based on multi-source data perception as described in claim 1, characterized in that, The sensing nodes include vibration sensing nodes, sound wave sensing nodes, temperature sensing nodes, pressure sensing nodes, and visual sensing nodes; when the edge computing gateway receives the multi-source heterogeneous sensing data, each data packet carries a local timestamp, and the edge computing gateway classifies and caches the data streams according to the stream identifier.

3. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 1, characterized in that, The specific features extracted by the synchronization unit for key physical events include: performing sliding window calculations on high-frequency vibration or acoustic data streams to obtain short-time energy and spectral centroid within the window; when the short-time energy exceeds a preset energy threshold and the spectral centroid falls within a preset characteristic frequency band, it is determined that a key physical event has been captured.

4. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 3, characterized in that, The synchronization unit dynamically generates a time synchronization reference window, specifically by: calculating the width of the time synchronization reference window based on the characteristic intensity of the captured key physical events, wherein the characteristic intensity includes the ratio of the event peak energy to the environmental noise energy, and the degree of deviation between the event spectrum centroid and the baseline spectrum centroid; the time synchronization reference window extends forward and backward by half the width, with the time when the key physical event is observed in the vibration data stream as the center.

5. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 4, characterized in that, The width of the time synchronization reference window is positively correlated with the ratio of the event peak energy to the environmental noise energy, and is also positively correlated with the degree of deviation between the event spectrum centroid and the baseline spectrum centroid.

6. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 4, characterized in that, The synchronization unit uses the time synchronization reference window to align the visual perception data stream, specifically by: analyzing the average temperature difference of the thermal imaging area between adjacent frames in the visual perception data stream within the time synchronization reference window, determining the timestamp of the video frame in which the average temperature difference first exceeds a set threshold as the visual alignment point, and using the visual alignment point as the precise physical event time reference.

7. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 6, characterized in that, The synchronization unit uses the time synchronization reference window to align the temperature sensing data stream or pressure sensing data stream. Specifically, within the time synchronization reference window, the temperature sensing data stream or pressure sensing data stream is interpolated and resampled using the visual alignment point as the time reference to obtain compensated synchronization data.

8. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 1, characterized in that, The early warning unit performs operational safety status assessment based on the spatiotemporal synchronization data unit, specifically including: extracting vibration and temperature data within the same spatiotemporal synchronization data unit and calculating the thermal vibration coupling risk index; extracting video keyframes and pressure data within the same spatiotemporal synchronization data unit and identifying the correlation between worker posture and pressure fluctuations through an image recognition model.

9. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 1, characterized in that, The warning command generated by the warning unit encapsulates the risk level, risk type, synchronization data unit identifier, and execution action code; the system also includes a communication module, which is used to select different network transmission slices according to the risk level and distribute the warning command to the on-site execution terminal.

10. The intelligent control system for safe operation of gas stations based on multi-source data perception according to claim 1, characterized in that, The edge computing gateway is also used to annotate the spatiotemporal synchronization data unit and its corresponding physical event reference time, and store them in the storage unit for subsequent risk retrospective analysis.

Images