A monitoring terminal and data processing method for a combined overflow pipeline system

Abstract

The application relates to the technical field of monitoring data processing, and discloses a monitoring terminal of a combined overflow pipeline system and a data processing method, which comprises the following steps: time alignment of liquid level data and gas concentration data based on timestamps, frequency domain filtering of time series data sets based on the set blocking bandwidth of astronomical tidal periods, segmentation and interception of target time series data sets based on a sliding time window in an edge computing terminal, determination of the correlation result of the liquid level and the gas concentration based on a correlation rule algorithm, matching of the change rate, the change gradient and the correlation result with the pipe network structure correlation, determination of the risk identification of the corresponding monitoring terminal based on the pipe section connection relationship and the flow direction information, and division of the event level based on the risk identification in the edge computing terminal. The monitoring data is processed through edge computing, so that the response timeliness, data efficiency and intelligent level of the combined overflow pipeline system monitoring are ensured.

Description

Technical Field

[0001] This invention relates to the field of monitoring data processing technology, and more specifically, to a monitoring terminal and data processing method for a combined overflow pipeline system. Background Technology

[0002] In coastal cities, combined sewer overflow (CSO) systems are a type of drainage system that collects and transports both domestic sewage and rainwater within the same pipe. Therefore, real-time data processing of CSO systems presents significant challenges. Firstly, the geographical characteristics of coastal areas mean that regular ocean tides can cause anomalies in monitoring data. Secondly, because the monitoring data is multi-source, all data must be transmitted to a central node for processing, heavily relying on network resources. This can easily lead to bandwidth pressure and processing bottlenecks in large-scale monitoring scenarios. Existing monitoring data processing methods struggle to meet the demands for high-frequency, low-latency monitoring, and lack operational feedback and edge execution mechanisms. A complete closed-loop processing mechanism for data acquisition, event identification, and early warning monitoring has not been established, thus limiting the response time, data efficiency, and intelligence level of combined sewer overflow system monitoring.

[0003] Therefore, it is necessary to design a monitoring terminal and data processing method for combined overflow pipeline systems to solve the problems existing in the current technology. Summary of the Invention

[0004] In view of this, the present invention proposes a monitoring terminal and data processing method for a combined sewer overflow pipeline system, aiming to solve the problems that the regular ocean tides cause anomalies in the monitoring data, and that all data needs to be transmitted to the central node for processing, which makes it difficult to meet the high-frequency and low-latency monitoring requirements of the monitoring data, thus limiting the response time, data efficiency and intelligence level of the combined sewer overflow pipeline system.

[0005] In one aspect, the present invention proposes a data processing method for a combined overflow piping system, comprising:

[0006] Several monitoring terminals are deployed in the combined overflow pipeline system to acquire liquid level data and at least one gas concentration data of the corresponding pipe section. The liquid level data and gas concentration data are time-aligned based on timestamps to determine a time series dataset.

[0007] The stopband width is set based on the astronomical tidal cycle, and the time series dataset is filtered in the frequency domain to determine the target time series dataset. The target time series dataset is segmented and truncated based on the sliding time window in the edge computing terminal. The data in each sliding time window is arranged according to the variable category and time order to determine the multivariate data matrix of each monitoring terminal.

[0008] Based on the multivariate data matrix, the rate of change of liquid level data and the gradient of change of gas concentration data corresponding to each monitoring terminal are determined, and the correlation result between liquid level and gas concentration is determined based on the association rule algorithm. The rate of change, gradient of change and correlation result are matched with the correlation relationship of pipeline network structure, and the risk identifier of the corresponding monitoring terminal is determined based on the pipeline segment connection relationship and flow direction information.

[0009] Within the edge computing terminal, events are classified into levels based on the risk identifier, and the time to upload to the central node is determined based on the severity of the event level.

[0010] Furthermore, in determining the target time series dataset, the process includes: performing a discrete Fourier transform on the time series dataset to determine the spectral data; determining a frequency range in the spectral data based on the center frequencies corresponding to the semi-diurnal and diurnal cycles; expanding the frequency range to both sides based on the center frequency to determine a stopband range; attenuating the amplitude of the spectral data within the stopband range; and performing an inverse Fourier transform on the processed spectral data to determine the target time series dataset.

[0011] Furthermore, in determining the multivariate data matrix, the following steps are taken: based on the segmented truncation results of the sliding time window, the data within each sliding time window are arranged in chronological order, and a multivariate data matrix for the corresponding monitoring terminal is constructed with variable category as the column dimension and time as the row dimension.

[0012] Furthermore, in determining the rate of change and the gradient of change, the method includes: performing a difference operation on the liquid level data and gas concentration data at adjacent time points in the multivariate data matrix, using the ratio of the difference result of the liquid level data to the corresponding time interval as the rate of change of liquid level, and using the ratio of the difference result of the gas concentration data to the corresponding time interval as the gradient of change of gas concentration.

[0013] Furthermore, in determining the correlation between liquid level and gas concentration, the process includes: statistically analyzing the sign consistency of the liquid level change rate and the gas concentration change gradient within the same time window; when the liquid level change rate and the gas concentration change gradient change directions are the same and remain consistent within at least two sliding time windows, they are determined to be synchronous changes; when the liquid level change rate and the gas concentration change gradient change directions are opposite and remain consistent within at least two time windows, they are determined to be inverse changes; based on the Eclat algorithm, the synchronous and inverse changes, several candidate itemsets are determined; frequent itemsets are determined based on the support of the candidate itemsets; and the correlation between liquid level and gas concentration is determined based on the frequent itemsets.

[0014] Furthermore, in determining the risk identifier of the corresponding monitoring terminal, the process includes: determining the adjacency relationship between each monitoring terminal based on the connectivity and flow direction information of the pipe segment; calculating the difference between the change rate and change gradient of adjacent monitoring terminals based on the flow direction order; extracting monitoring terminals whose difference exceeds the difference threshold; comparing the correlation results of adjacent monitoring terminals; marking the corresponding monitoring terminal as an abnormal risk identifier when the extracted monitoring terminals have inconsistent correlation results; marking the corresponding monitoring terminal as a suspected risk identifier when the extracted monitoring terminals have consistent correlation results; and marking the corresponding monitoring terminal as a normal risk identifier when the difference does not exceed the difference threshold.

[0015] Furthermore, when classifying event levels, the process includes: counting the number of abnormal risk indicators and the number of suspected risk indicators, and recording the sum of the number of abnormal and suspected indicators as the risk quantity; pre-setting a first risk quantity and a second risk quantity, wherein the first risk quantity is greater than the second risk quantity; when the risk quantity is greater than or equal to the first risk quantity, it is determined as a level 1 event; when the risk quantity is less than the first risk quantity but greater than the second risk quantity, it is determined as a level 2 event; and when the risk quantity is less than or equal to the second risk quantity, it is determined as a level 3 event.

[0016] Furthermore, when determining the upload time to the central node, the process includes: performing a first-level upload operation on the first-level event and continuously sending new data packets; performing a second-level upload operation on the second-level event and making an interval judgment based on the time interval when new data packets are added; sending new data packets in batches when the time interval is reached; performing a third-level upload operation on the third-level event and caching new data packets to the edge computing terminal; and sending them centrally after the number of data packets reaches the data packet threshold.

[0017] Furthermore, when determining the upload time to the central node, the upload times for the first-level upload operation, the second-level upload operation, and the third-level upload operation increase sequentially.

[0018] Compared with existing technologies, the beneficial effects of this invention are as follows: By setting the stopband width based on the astronomical tidal cycle and performing frequency domain filtering on the time series dataset, abnormal interference in monitoring data caused by regular ocean tides in coastal areas is eliminated, ensuring that the target time series dataset accurately reflects the actual operating status of the pipeline system and solving data distortion caused by geographical characteristics. Relying on edge computing, the invention completes sliding time window interception, multivariate data matrix construction, change rate and gradient calculation, correlation result determination, risk identification, event level classification, and upload time determination, achieving local processing and intelligent judgment of multi-source monitoring data without transmitting all data to the central node. This reduces dependence on network resources, alleviates bandwidth pressure and central node processing bottlenecks in large-scale monitoring scenarios, and forms a closed-loop processing mechanism for data acquisition, event identification, and early warning monitoring. Simultaneously, based on the proximity processing characteristics of edge computing, it meets the high-frequency, low-latency monitoring requirements of combined sewer overflow pipeline systems. Combining pipeline network structure relationships, pipe segment connectivity relationships, and flow direction information, it achieves accurate risk identification, improving the response time, data processing efficiency, and intelligent management level of combined sewer overflow pipeline system monitoring.

[0019] On the other hand, this application also provides a monitoring terminal for a combined sewer overflow system, and a data processing method for the aforementioned combined sewer overflow system, including:

[0020] The cavity is made of 316L stainless steel or titanium alloy, and the cavity is provided with a transparent window of polytetrafluoroethylene facing the outside, and the transparent window is coated with a hydrophobic nano-coating.

[0021] A radar liquid level sensing module, wherein the radar liquid level sensing module is disposed on the inner wall of the cavity;

[0022] A wireless communication module is disposed on the inner wall of the cavity;

[0023] A gas sensing module, wherein the gas sensing module is fixedly connected to the wireless communication module;

[0024] A temperature and humidity sensing module is disposed at the bottom of the cavity;

[0025] A power management module is inserted into the side of the cavity;

[0026] The first control module is fixedly connected to the top of the temperature and humidity sensing module;

[0027] The second control module is disposed on the inner wall of the cavity.

[0028] It is understandable that the monitoring terminal and data processing method of the above-mentioned combined overflow pipeline system have the same beneficial effects, and will not be elaborated here. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in 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.

[0030] Figure 1 A flowchart illustrating a data processing method for a combined overflow piping system provided in an embodiment of the present invention;

[0031] Figure 2 An exploded view of a monitoring terminal for a combined overflow pipeline system provided in an embodiment of the present invention.

[0032] The module consists of: 101, First Control Module; 102, Radar Liquid Level Sensing Module; 103, Second Control Module; 104, Wireless Communication Module; 105, Gas Sensing Module; 106, Temperature and Humidity Sensing Module; 107, Cavity; and 108, Power Management Module. Detailed Implementation

[0033] 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.

[0034] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] See Figure 1 As shown in some embodiments of this application, a data processing method for a combined overflow piping system includes:

[0036] S100: Deploy several monitoring terminals within the combined overflow pipeline system and acquire liquid level data and at least one gas concentration data for the corresponding pipe section. Time-align the liquid level data and gas concentration data based on timestamps to determine the time series dataset.

[0037] S200: Based on the astronomical tidal cycle, the stopband width is set and the time series dataset is frequency-domain filtered to determine the target time series dataset. In the edge computing terminal, the target time series dataset is segmented and truncated based on the sliding time window. The data in each sliding time window is arranged according to the variable category and time order to determine the multivariate data matrix of each monitoring terminal.

[0038] S300: Based on the multivariate data matrix, determine the rate of change of liquid level data and the gradient of change of gas concentration data corresponding to each monitoring terminal, and determine the correlation result between liquid level and gas concentration based on the association rule algorithm. Match the rate of change, gradient of change and correlation result with the correlation relationship of pipeline network structure, and determine the risk label of the corresponding monitoring terminal based on the pipeline connection relationship and flow direction information.

[0039] S400: Within the edge computing terminal, events are classified based on risk identifiers, and the time to upload to the central node is determined based on the severity of the event.

[0040] Specifically, several monitoring terminals are deployed within the combined sewer overflow pipeline system. The number of monitoring terminals can be dynamically adjusted according to the length of the combined sewer overflow pipeline. These monitoring terminals collect data at various pipe sections within the network, meaning they are front-end devices that directly contact the internal environment of the pipeline network and acquire operational data. Edge computing terminals are devices deployed at the pipeline site to perform local computation with the monitoring terminals. All monitoring terminals establish real-time communication connections with the edge computing terminals, which uniformly manage and schedule the data collection work of each monitoring terminal. Each monitoring terminal synchronously acquires liquid level data and at least one gas concentration data for its corresponding pipe section. The liquid level data represents the real-time height of the mixed sewage and rainwater within the pipe section where the monitoring terminal is located, directly reflecting the water accumulation and drainage smoothness within the pipe section. The gas concentration data represents the real-time content of various gases produced by the fermentation and reaction of the mixed sewage within the pipe section. Specifically, at least one or more gases from hydrogen sulfide, methane, carbon monoxide, and oxygen can be selected. When the monitoring terminal collects data, it generates a unique timestamp for each data point. The timestamp is a unique identifier that records the precise moment of data collection, which can accurately pinpoint the collection time of each data point. Based on the timestamp, the liquid level data and gas concentration data are time-aligned. The liquid level data and gas concentration data generated by the same monitoring terminal and at the same collection time are matched one by one to eliminate data with mismatched timestamps or misaligned collection times. This ensures that there are corresponding and synchronously collected liquid level data and gas concentration data at each time point. Finally, all the matching data are integrated to determine the time series dataset, which is a collection of data arranged in chronological order of collection time.The stopband width is set based on the astronomical tidal cycle, and frequency domain filtering is performed on the time series dataset. The astronomical tidal cycle represents the inherent periodicity of the ebb and flow of tides in coastal areas. In particular, external interference factors that cause false abnormal fluctuations in pipeline monitoring data are identified. These tidal interferences are not due to abnormalities in the pipeline's own operation, so they need to be specifically eliminated. The edge computing end sets the stopband width based on the astronomical tidal cycle and performs frequency domain filtering on the locally stored time series dataset. This eliminates the need to transmit massive amounts of time series data to the central node, avoiding the consumption of network bandwidth. The stopband width is a specifically defined frequency range that needs to be filtered for the interference frequencies corresponding to the astronomical tidal cycle, and this range perfectly matches the tides. The frequency characteristics of the tidal cycle are utilized without affecting the real-time data of the pipeline network. Frequency domain filtering converts the time series dataset presented in the conventional time dimension into a frequency dimension, accurately identifying the frequency signals belonging to tidal interference. By using the stopband width, these interference signals are eliminated and weakened, retaining the data that can truly reflect the operating status of the pipeline network. Finally, the target time series dataset is determined. The target time series dataset eliminates external tidal interference and only retains the time series data of the actual operating status of the combined overflow pipeline system itself. The entire process is completed locally at the edge computing end, which not only improves the filtering efficiency but also avoids network waste caused by interference data transmission and solves the problem of data anomalies caused by tides. The edge computing end uses a sliding time window to segment the target time series dataset. Then, it arranges the data within each sliding time window according to variable category and time order. Variable category indicates the different data types that distinguish liquid level data and gas concentration data. Time order indicates that the data within the window is arranged in the order of collection time from morning to evening. By using two dimensions to organize the data, a multivariate data matrix for each monitoring terminal is finally determined. The multivariate data matrix is ​​a well-organized structured data corresponding to a single monitoring terminal, which can clearly show the changing patterns of various types of data over time. For example, the row dimension has 5 rows, corresponding to 5 consecutive sliding time windows, each window lasting 10 minutes. The time column in the row records the end time of each sliding window (uniformly marked by the edge computing end and consistent with the timestamp of the data collected by the monitoring terminal) to fully cover the monitoring period. The number of column dimensions is set according to the gas type. Liquid level data and gas concentration data are two variable categories, corresponding to the data collected by the monitoring terminal respectively. The multivariate data matrix clearly presents the changing patterns of the two types of data over time, providing a structured and intuitive data foundation for edge computing to calculate the rate of liquid level change, the gradient of gas concentration change, and to explore the relationship between the two.

[0041] Understandably, the edge computing terminal calculates the rate of change of liquid level data and the gradient of change of gas concentration data for each monitoring terminal based on a multivariate data matrix. The rate of change represents the magnitude of the rise and fall of liquid level data in the pipe section per unit time, directly reflecting the rate of increase or decrease of water volume within the pipe section. The gradient of change represents the magnitude of the rise and fall of gas concentration data in the pipe section per unit time, directly reflecting the rate of accumulation or dissipation of hazardous gases within the pipe section. All calculations are completed at the edge computing terminal. The association rule algorithm is used to discover the inherent linkage between liquid level changes and gas concentration changes. The association results represent the linkage relationship between the two changes and are also used to judge the pipe network. A crucial indicator of operational anomalies is the matching of the calculated rate of change, gradient of change, and correlation results between the edge computing end and the local pipe network structure. The pipe network structure correlation represents the overall layout and linkage logic between various pipe segments of the combined overflow pipeline system, and is stored in advance by the edge computing end. The pipe network structure correlation can be obtained through a combination of data integration and on-site surveys, or by extracting information such as pipe segment layout and node connections from existing drainage facility GIS databases and pipe network as-built drawings. Alternatively, a pipe network spatial model can be constructed using GIS to sort out the linkage relationship and EPANET can be used to simulate hydraulic characteristics to determine the correlation. By combining local pipe segment connectivity and flow direction information, the risk identifier of the corresponding monitoring terminal is determined. Pipe segment connectivity indicates the physical connection status between each pipe segment, flow direction information indicates the flow direction of the mixed water in the pipe, and risk identifier represents the risk label marked by the edge computing terminal, indicating the operational status of the corresponding pipe segment of the monitoring terminal. This intuitively distinguishes between the normal operation and the risk status of the pipe segment. The edge computing terminal automatically classifies the event level based on the risk identifiers of all monitoring terminals and autonomously decides the time to upload to the central node based on the degree of event level. The central node represents the remote global data aggregation platform. Through a hierarchical upload strategy, the edge computing terminal does not need to transmit the entire data to the central node, which not only ensures real-time response to risk events but also reduces network bandwidth consumption and avoids the bandwidth pressure and processing delay risks of the original monitoring mode. At the same time, the edge computing terminal realizes closed-loop processing of data acquisition, local processing, and hierarchical upload, improving the response time and intelligence level of data management and monitoring of combined sewer overflow pipeline systems.

[0042] In some embodiments of this application, determining the target time series dataset includes: performing a discrete Fourier transform on the time series dataset to determine the spectral data; determining a frequency range in the spectral data based on the center frequencies corresponding to the semi-diurnal and diurnal cycles; expanding the frequency range to both sides based on the center frequency to determine the stopband range; attenuating the amplitude of the spectral data within the stopband range; and performing an inverse Fourier transform on the processed spectral data to determine the target time series dataset.

[0043] Specifically, the edge computing end performs a Discrete Fourier Transform (DFT) on the time-aligned time-series dataset. The DFT transforms the time-series dataset, originally arranged in a time-dimension format, from the time domain to the frequency domain. This allows for the separation of different frequency signal components mixed within the time-series data, distinguishing tidal interference signals from the actual operating signals of the pipeline network in the frequency domain. After the transformation, spectral data is determined. This spectral data, after the DFT, represents the amplitude of different frequency signals in the frequency domain, clearly reflecting the time... The distribution of various frequency signals in the inter-sequence data is shown. The astronomical tidal cycles are divided into semi-diurnal and diurnal cycles. A semi-diurnal cycle indicates the periodic pattern of two high tides and two low tides per day in coastal areas, while a diurnal cycle indicates the periodic pattern of one high tide and one low tide per day. The center frequency represents the frequency of the concentrated tidal interference signals corresponding to the semi-diurnal and diurnal cycles, respectively. The center frequency of the semi-diurnal cycle is pre-set to 1.932 cycles / day, and the center frequency of the diurnal cycle is set to 0.997 cycles / day. (Edge computing root...) The frequency range is determined from the spectral data based on the center frequencies corresponding to the semi-diurnal and diurnal tidal cycles. This frequency range, defined by the center frequency, represents the main frequency distribution area of ​​tidal interference signals. Extending the frequency range to both sides of the center frequency, a stopband is determined. This stopband, defined in the spectral data, covers ±5% of the center frequency offset caused by local geographical factors. The stopband is a frequency range in the frequency domain where signal amplitude attenuation is required. Subsequently, the amplitude of the spectral data is attenuated within the defined stopband, effectively setting the signal amplitude within the stopband. With an attenuation amplitude greater than 20 dB, while preserving the non-periodic signal reflecting the actual working conditions, it accurately eliminates the regular fluctuations in liquid level and gas concentration data caused by tides. The edge computing end performs an inverse Fourier transform on the spectrum data after amplitude attenuation processing. The inverse Fourier transform converts the processed frequency domain spectrum data back to the time domain to ensure the temporal continuity and integrity of the data. Moreover, the inverse Fourier transform does not produce data distortion, information loss, or timing disorder, and can accurately output time-series data that can be directly used for subsequent analysis, thus improving the data processing efficiency of the edge computing end.

[0044] In some embodiments of this application, determining the multivariate data matrix includes: based on the segmented truncation results of the sliding time window, arranging the data in each sliding time window in chronological order, and constructing a multivariate data matrix for the corresponding monitoring terminal with variable category as the column dimension and time as the row dimension.

[0045] In some embodiments of this application, determining the rate of change and the gradient of change includes: performing a difference operation on the liquid level data and gas concentration data at adjacent time points in the multivariate data matrix, using the ratio of the difference result of the liquid level data to the corresponding time interval as the rate of change of liquid level, and using the ratio of the difference result of the gas concentration data to the corresponding time interval as the gradient of change of gas concentration.

[0046] Specifically, when determining the multivariate data matrix, the edge computing end uses the results of previous sliding time window segmentation of the target time series dataset. For all data within each sliding time window, it arranges them according to the chronological order of data collection to ensure clear and uninterrupted data sequence within the same window. Using variable categories as columns (representing different types of monitoring data such as liquid level and gas concentration) and time as rows, a multivariate data matrix corresponding to a single monitoring terminal is constructed. When determining the rate of change and gradient, the edge computing end uses the multivariate data matrix as a basis to perform differential operations on the liquid level and gas concentration data corresponding to two adjacent time points. The calculations reflect the increase or decrease of data at adjacent time points. Then, the difference calculation result of the liquid level data is compared with the time interval corresponding to these two adjacent time points to determine the liquid level change rate. The liquid level change rate represents the magnitude of the rise and fall of the liquid level data in the pipe section per unit time. At the same time, the difference calculation result of the gas concentration data is compared with the time interval corresponding to the adjacent time points to determine the gas concentration change gradient. The gas concentration change gradient represents the magnitude of the rise and fall of the gas concentration data in the pipe section per unit time. By determining the liquid level change rate and the liquid level change rate, the degree of change of the two types of data can be accurately obtained, providing a data foundation for subsequent exploration of the correlation between liquid level and gas concentration and the determination of risk indicators.

[0047] In some embodiments of this application, determining the correlation between liquid level and gas concentration includes: statistically analyzing the sign consistency of the liquid level change rate and the gas concentration change gradient within the same time window; when the liquid level change rate and the gas concentration change gradient change direction are the same and remain consistent within at least two sliding time windows, they are determined to be synchronous changes; when the liquid level change rate and the gas concentration change gradient change direction are opposite and remain consistent within at least two time windows, they are determined to be inverse changes; several candidate itemsets are determined based on the Eclat algorithm, synchronous changes, and inverse changes; frequent itemsets are determined based on the support of the candidate itemsets; and the correlation between liquid level and gas concentration is determined based on the frequent itemsets.

[0048] Specifically, the sign consistency of the liquid level change rate and the gas concentration change gradient within the same sliding time window is statistically analyzed. Sign consistency indicates whether the positive and negative signs corresponding to the liquid level change rate and the gas concentration change gradient are the same, directly reflecting whether their change directions are consistent. When the change directions of the liquid level change rate and the gas concentration change gradient are the same, and this change direction remains stable within at least two consecutive sliding time windows, their change relationship is determined to be a synchronous change relationship. A synchronous change relationship can be defined as the liquid level change rate being in an upward state and the gas concentration change gradient also being in an upward state simultaneously, and this common upward state is maintained for multiple consecutive sliding time windows, or the liquid level change rate... When the liquid level change rate and the gas concentration change gradient are both in a decreasing state, and this common decreasing state is maintained for multiple consecutive sliding time windows, and the change direction of the liquid level change rate is opposite to that of the gas concentration change gradient, and this change direction remains stable for at least two consecutive sliding time windows, then the relationship between the two is determined to be an inverse relationship. An inverse relationship can be that the liquid level change rate is in an increasing state while the gas concentration change gradient is in a decreasing state, and this increasing-decreasing state is maintained for multiple consecutive sliding time windows, or that the liquid level change rate is in a decreasing state while the gas concentration change gradient is in an increasing state, and this decreasing-increasing state is maintained for multiple consecutive sliding time windows. Based on the Eclat algorithm and the established synchronous and inverse change relationships, several candidate item sets are determined. The number of candidate item sets is determined by the amount of data. Eclat is used to mine the correlation patterns between data items. The candidate item sets are the data sets to be screened, which are formed by combining the relationship between the rate of change of liquid level and the gradient of change of gas concentration. Frequent item sets are determined based on the support of the candidate item sets. Support is the frequency of the candidate item set appearing in the entire sliding time window. Frequent item sets are item sets whose frequency of occurrence exceeds a certain minimum support threshold. It is the basis for generating association rules. Based on the frequent item sets obtained by screening, the association results between liquid level and gas concentration that can reflect the actual operating status of the pipeline network are finally determined. By statistically analyzing the sign consistency between the rate of change of liquid level and the gradient of change of gas concentration within the same sliding time window, and limiting the consistency to at least two sliding time windows before determining synchronous or reverse change relationships, misjudgments caused by single data fluctuations and instantaneous interference can be effectively eliminated, improving the stability and reliability of change relationship identification. The Eclat algorithm is used to construct candidate itemsets in conjunction with change relationships, and frequent itemsets are selected based on support. This allows for the lightweight discovery of stable intrinsic linkage patterns between liquid level and gas concentration at the edge computing end, avoiding false associations that interfere with judgments. This further enhances the intelligent capabilities of edge computing, thereby reducing dependence on central nodes and improving the response accuracy and anti-interference capability of monitoring combined overflow pipeline systems.

[0049] In some embodiments of this application, determining the risk identifier of a corresponding monitoring terminal includes: determining the adjacency relationship between each monitoring terminal based on the connectivity and flow direction information of the pipe segment; calculating the difference between the change rate and change gradient of adjacent monitoring terminals based on the flow direction order; extracting monitoring terminals whose difference exceeds the difference threshold; comparing the association results of adjacent monitoring terminals; marking the corresponding monitoring terminal as an abnormal risk identifier when the extracted monitoring terminals have inconsistent association results; marking the corresponding monitoring terminal as a suspected risk identifier when the extracted monitoring terminals have consistent association results; and marking the corresponding monitoring terminal as a normal risk identifier when the difference does not exceed the difference threshold.

[0050] Specifically, the adjacency relationships between monitoring terminals are determined based on the connectivity and flow direction information of the pipe segments. The connectivity of the pipe segments represents the physical layout of direct or indirect connections between segments in a combined sewer overflow system. The flow direction information indicates the flow direction of the mixed water within the pipe. The adjacency relationship represents the correspondence between monitoring terminals directly connected according to the water flow direction and where water can flow directly through the pipe network. That is, monitoring terminals deployed in upstream pipe segments are adjacent to monitoring terminals in directly connected downstream pipe segments. Based on the flow direction sequence, the liquid level change rate and gas concentration change gradient corresponding to adjacent monitoring terminals are respectively... The difference calculation reflects the degree of difference in parameter changes between adjacent monitoring terminals. The calculated difference is compared with a pre-set difference threshold, which is a critical value pre-set by the edge computing end to determine whether there are abnormalities in parameter changes between adjacent monitoring terminals. The difference threshold can be dynamically adjusted according to the combined sewer overflow pipeline system in different coastal areas. Monitoring terminals whose calculated differences exceed the difference threshold are extracted. Simultaneously, the correlation results between liquid level and gas concentration between adjacent monitoring terminals are compared. Due to the physical characteristics of connected pipe sections and continuous water flow in the combined sewer overflow pipeline system, under normal operating conditions, adjacent pipe sections... The operating parameters and variation patterns of the pipe section will remain in a continuous, gradual, and relatively consistent state. Under normal operating conditions with the pipe section connected and water flow smooth, the rate of change of liquid level and the gradient of change of gas concentration between adjacent monitoring terminals distributed along the flow direction will not suddenly jump. The difference calculated by the two will be within the difference threshold range. At the same time, the correlation results between liquid level and gas concentration will also maintain a certain degree of synchronous or reverse consistency. In other words, under normal operating conditions without blockage, siltation, leakage, or other faults, environmental factors such as water composition, siltation degree, water flow velocity, and ventilation conditions in the pipe will not experience local abrupt changes. Liquid level changes directly determine the residence time and spatial volume of water in the pipe. The pressure state and gas concentration are directly determined by the liquid level, the degree of sewage fermentation, and the gas diffusion conditions. The linkage between the two is determined by the stable environment inside the pipeline. Therefore, when the liquid level and gas concentration in the upstream pipe section change synchronously or in opposite directions, this inherent linkage determined by the environment will be smoothly transmitted to the downstream adjacent pipe section with the water flow. There will be no situation where one pipe section changes synchronously and the adjacent pipe section suddenly changes in opposite directions. Only when the pipeline has faults such as blockage, siltation, leakage, or abnormal local accumulation of gas, which disrupts the continuity of water flow and environment between pipe sections, will the correlation between liquid level and gas concentration change abruptly and become inconsistent.Based on the extracted monitoring terminals and the comparison of correlation results with adjacent monitoring terminals, the corresponding monitoring terminals are marked as normal risk indicators. This means that the water flow in the pipe section where the corresponding monitoring terminal is located is smooth and the operating status is stable, with no safety risks. When the difference between the rate of change and the gradient of change of adjacent monitoring terminals exceeds the difference threshold, it indicates that the parameter has abruptly changed, representing abnormal fluctuations in the operating status of the pipe section. However, a single parameter abrupt change cannot distinguish whether it is a local fault or a fluctuation in the overall environment. Therefore, further verification is needed by combining the correlation results. If the correlation results of adjacent monitoring terminals are inconsistent at this time, it means that this parameter abrupt change is not a normal fluctuation in the overall pipe network, but a substantial fault such as blockage, siltation, leakage, or local gas accumulation in a local pipe section. The fault will break the regular transmission of adjacent pipe sections, so it is marked as an abnormal risk indicator, indicating the existence of safety risks. If the parameter changes abruptly but the correlation results of adjacent monitoring terminals are consistent, it means that the flow and water level of the entire pipe network fluctuate synchronously, which may not represent a substantial safety hazard. Therefore, it is marked as a suspected risk indicator, indicating that the risk status is pending. By combining the pipeline topology and the parameter correlation between adjacent terminals, a comprehensive judgment is made to ensure the lightweight and low-latency reliability of the edge computing terminal.

[0051] In some embodiments of this application, the process of classifying event levels includes: counting the number of abnormal risk markers and the number of suspected risk markers, and recording the sum of the number of abnormal and suspected risk markers as the risk quantity; pre-setting a first risk quantity and a second risk quantity; the first risk quantity being greater than the second risk quantity; when the risk quantity is greater than or equal to the first risk quantity, it is determined as a level 1 event; when the risk quantity is less than the first risk quantity but greater than the second risk quantity, it is determined as a level 2 event; and when the risk quantity is less than or equal to the second risk quantity, it is determined as a level 3 event.

[0052] In some embodiments of this application, determining the time to upload to the central node includes: performing a first-level upload operation for first-level events and continuously sending new data packets; performing a second-level upload operation for second-level events and making interval judgments based on time intervals when new data packets are added, sending new data packets in batches when the time interval is reached; performing a third-level upload operation for third-level events, caching new data packets at the edge computing end, and sending them centrally after the number of data packets reaches the data packet threshold.

[0053] In some embodiments of this application, when determining the upload time to the central node, the upload time for the first-level upload operation, the second-level upload operation, and the third-level upload operation increases sequentially.

[0054] Specifically, the number of abnormal risk markers on all monitoring terminals is counted, with the abnormal number representing the total number of monitoring terminals marked as abnormal risk markers. The number of suspected risk markers on all monitoring terminals is also counted, with the suspected number representing the total number of monitoring terminals marked as suspected risk markers. The risk count directly reflects the overall risk scale of the pipeline network. The edge computing terminal pre-sets a first risk count and a second risk count. When the calculated risk count is greater than or equal to the first risk count, it indicates that there are a large number of abnormal or suspected abnormal monitoring terminals in the pipeline network, and the current pipeline network operating status is then determined. A Level 1 event represents the highest-risk pipeline event, such as serious malfunctions like blockages in multiple pipe sections or large-scale gas accumulation. If the number of events is less than the number of Level 1 events but greater than the number of Level 2 events, it indicates the presence of a certain number of abnormal or suspected abnormal monitoring terminals within the pipeline network, thus classifying it as a Level 2 event. A Level 2 event represents a pipeline event with a medium-risk level. If the number of events is less than or equal to the number of Level 2 events, it indicates the presence of only a very small number of abnormal or suspected abnormal monitoring terminals within the pipeline network, with an overall risk level that is extremely low or essentially nonexistent, thus classifying it as a Level 3 event. A Level 3 event represents the lowest-risk pipeline event. When determining the upload time to the central node, the central node refers to a remote platform that performs global control, data aggregation, and unified management of the combined sewer overflow pipeline system. For Level 1 events, a Level 1 upload operation is performed. This Level 1 upload operation represents the highest priority upload method for the highest-risk events, meaning that new data packets are sent continuously as soon as they are generated at the edge computing end. New data packets represent data newly generated at the edge computing end after local data processing and risk assessment, containing monitoring data and analysis results, ensuring that high-risk event information is transmitted as quickly as possible. For Level 2 events, a Level 2 upload operation is performed. This Level 2 upload operation represents the normal priority upload method for medium-risk events. When new data packets are generated, a preset time interval is used for interval judgment, and when the time interval is reached, the generated new data packets are sent in batches. For Level 3 events, a Level 3 upload operation is performed. The operation describes a low-priority upload method for low-risk events. New data packets are first cached in the local storage space of the edge computing terminal. They are only sent in a centralized manner after the number of cached data packets reaches a data packet threshold. The upload time for level 1, level 2, and level 3 upload operations increases sequentially, meaning that level 1 events are uploaded faster, level 2 events are next, and level 3 events are uploaded slower. By classifying events into levels based on the number of risks, the overall risk level of the pipeline network can be objectively quantified, avoiding misjudgments caused by single terminal anomalies. At the same time, the three-level differentiated upload strategy matched with the event level leverages the advantages of local processing at the edge computing terminal, reducing network bandwidth consumption and the processing pressure on the central node. The sequentially increasing upload time further optimizes the allocation of transmission resources, achieving a balance between the timeliness of risk response and data transmission, and improving the intelligence level of data processing for combined overflow pipeline systems.

[0055] In summary, the beneficial effects of this invention are as follows: By setting the stopband width based on the astronomical tidal cycle and performing frequency domain filtering on the time series dataset, abnormal interference in monitoring data caused by regular ocean tides in coastal areas is eliminated, ensuring that the target time series dataset accurately reflects the actual operating status of the pipeline system and solving data distortion caused by geographical characteristics. Relying on edge computing, the invention completes sliding time window interception, multivariate data matrix construction, change rate and gradient calculation, correlation result determination, risk identification, event level classification, and upload time determination, achieving local processing and intelligent judgment of multi-source monitoring data without transmitting all data to the central node. This reduces dependence on network resources, alleviates bandwidth pressure and central node processing bottlenecks in large-scale monitoring scenarios, and forms a closed-loop processing mechanism for data acquisition, event identification, and early warning monitoring. Simultaneously, based on the proximity processing characteristics of edge computing, it meets the high-frequency, low-latency monitoring requirements of combined sewer overflow pipeline systems. Combining pipeline network structure relationships, pipe segment connectivity relationships, and flow direction information, it achieves accurate risk identification, improving the response time, data processing efficiency, and intelligent management level of combined sewer overflow pipeline system monitoring.

[0056] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides a monitoring terminal for a combined sewer overflow system and a data processing method for the combined sewer overflow system, including:

[0057] The cavity 107 is made of 316L stainless steel or titanium alloy. The cavity 107 has a transparent window made of polytetrafluoroethylene facing outward, and the transparent window is coated with a hydrophobic nano-coating.

[0058] Radar liquid level sensing module 102 is disposed on the inner wall of cavity 107.

[0059] Wireless communication module 104 is disposed on the inner wall of cavity 107;

[0060] Gas sensing module 105, gas sensing module 105 is fixedly connected to wireless communication module 104;

[0061] Temperature and humidity sensing module 106 is located at the bottom of cavity 107;

[0062] The power management module 108 is inserted into the side of the cavity 107;

[0063] The first control module 101 is fixedly connected to the top of the temperature and humidity sensing module 106;

[0064] The second control module 103 is disposed on the inner wall of the cavity 107.

[0065] Specifically, cavity 107 provides the mounting base for each module. Cavity 107 is made of 316L stainless steel or titanium alloy, and the outer wall thickness of the fully sealed cavity 107 is no less than 3 mm to ensure long-term structural stability and corrosion resistance in high-salt and high-humidity environments. The outer surface of the transparent window, made of polytetrafluoroethylene, is coated with a hydrophobic nano-coating with a thickness of 50-200 nanometers to prevent condensation droplets from affecting radar signal transmission and reception. Radar level sensing module 102 is a non-contact frequency-modulated continuous wave (FMCW) radar module operating in the 24 GHz or 77 GHz frequency band. Gas sensing module 105 communicates with the external pipeline environment through a waterproof and breathable membrane integrated on the top of cavity 107, allowing external gas molecules to enter and be measured while maintaining an IP68 sealing rating. The temperature and humidity sensing module 106 is located at the bottom of the cavity 107 and is used to collect temperature and humidity data inside the combined overflow pipe. The gas sensing module 105 is an electrochemical sensing module that integrates hydrogen sulfide, methane, and carbon monoxide. The wireless communication module 104 is a low-power wide-area network (LPWAN) communication module, specifically a narrowband Internet of Things (NB-IoT) module or a LoRaWAN module. The power management module 108 is inserted into the side of the cavity 107. The power management module 108 is woken up by the clock (RTC) of the first control module 101 and the second control module 103 at preset time intervals (e.g., 60 minutes). After being woken up, the power management module 108 sequentially powers on the gas sensing module 105, the temperature and humidity sensing module 106, and the radar liquid level sensing module 102, and completes the acquisition of liquid level and multi-channel gas concentration data.

[0066] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A data processing method for a combined overflow piping system, characterized in that, include: Several monitoring terminals are deployed in the combined overflow pipeline system to acquire liquid level data and at least one gas concentration data of the corresponding pipe section. The liquid level data and gas concentration data are time-aligned based on timestamps to determine a time series dataset. The stopband width is set based on the astronomical tidal cycle, and the time series dataset is filtered in the frequency domain to determine the target time series dataset. The target time series dataset is segmented and truncated based on the sliding time window in the edge computing terminal. The data in each sliding time window is arranged according to the variable category and time order to determine the multivariate data matrix of each monitoring terminal. Based on the multivariate data matrix, the rate of change of liquid level data and the gradient of change of gas concentration data corresponding to each monitoring terminal are determined, and the correlation result between liquid level and gas concentration is determined based on the association rule algorithm. The rate of change, gradient of change and correlation result are matched with the correlation relationship of pipeline network structure, and the risk identifier of the corresponding monitoring terminal is determined based on the pipeline segment connection relationship and flow direction information. Within the edge computing terminal, events are classified into levels based on the risk identifier, and the time to upload to the central node is determined based on the severity of the event level.

2. The data processing method for a combined overflow piping system according to claim 1, characterized in that, Determining the target time series dataset includes: performing a discrete Fourier transform on the time series dataset to determine the spectral data; determining a frequency range in the spectral data based on the center frequencies corresponding to the semi-diurnal and diurnal cycles; expanding the frequency range to both sides based on the center frequency to determine a stopband range; attenuating the amplitude of the spectral data within the stopband range; and performing an inverse Fourier transform on the processed spectral data to determine the target time series dataset.

3. The data processing method for a combined overflow piping system according to claim 2, characterized in that, When determining the multivariate data matrix, the following steps are taken: based on the segmented truncation results of the sliding time window, the data within each sliding time window are arranged in chronological order, and a multivariate data matrix for the corresponding monitoring terminal is constructed with variable category as the column dimension and time as the row dimension.

4. The data processing method for a combined overflow piping system according to claim 3, characterized in that, When determining the rate of change and the gradient of change, the method includes: performing a difference operation on the liquid level data and gas concentration data at adjacent time points in the multivariate data matrix, using the ratio of the difference result of the liquid level data to the corresponding time interval as the rate of change of liquid level, and using the ratio of the difference result of the gas concentration data to the corresponding time interval as the gradient of change of gas concentration.

5. The data processing method for a combined overflow piping system according to claim 4, characterized in that, When determining the correlation between liquid level and gas concentration, the process includes: statistically analyzing the sign consistency of the liquid level change rate and the gas concentration change gradient within the same time window; when the liquid level change rate and the gas concentration change gradient change direction are the same and remain consistent within at least two sliding time windows, they are determined to be synchronous changes; when the liquid level change rate and the gas concentration change gradient change direction are opposite and remain consistent within at least two time windows, they are determined to be inverse changes; based on the Eclat algorithm, the synchronous and inverse changes, several candidate itemsets are determined; frequent itemsets are determined based on the support of the candidate itemsets; and the correlation between liquid level and gas concentration is determined based on the frequent itemsets.

6. The data processing method for a combined overflow piping system according to claim 5, characterized in that, When determining the risk identifier of the corresponding monitoring terminal, the process includes: determining the adjacency relationship between each monitoring terminal based on the connectivity and flow direction information of the pipeline segment; calculating the difference between the change rate and change gradient of adjacent monitoring terminals based on the flow direction order; extracting monitoring terminals whose difference exceeds the difference threshold; comparing the correlation results of adjacent monitoring terminals; marking the corresponding monitoring terminal as an abnormal risk identifier when the correlation results of the extracted monitoring terminals are inconsistent; marking the corresponding monitoring terminal as a suspected risk identifier when the correlation results of the extracted monitoring terminals are consistent; and marking the corresponding monitoring terminal as a normal risk identifier when the difference does not exceed the difference threshold.

7. The data processing method for a combined overflow piping system according to claim 6, characterized in that, When classifying events into levels, the process includes: counting the number of abnormal risk indicators and the number of suspected risk indicators, and recording the sum of the number of abnormal and suspected indicators as the risk quantity. A first risk quantity and a second risk quantity are preset, wherein the first risk quantity is greater than the second risk quantity. When the risk quantity is greater than or equal to the first risk quantity, it is determined as a level 1 event. When the risk quantity is less than the first risk quantity but greater than the second risk quantity, it is determined as a level 2 event. When the risk quantity is less than or equal to the second risk quantity, it is determined as a level 3 event.

8. The data processing method for a combined overflow piping system according to claim 7, characterized in that, When determining the upload time to the central node, the process includes: performing a first-level upload operation for the first-level event and continuously sending new data packets; performing a second-level upload operation for the second-level event and making an interval judgment based on the time interval when new data packets are added; sending new data packets in batches when the time interval is reached; performing a third-level upload operation for the third-level event and caching new data packets to the edge computing terminal; and sending them centrally after the number of data packets reaches the data packet threshold.

9. The data processing method for a combined overflow piping system according to claim 8, characterized in that, When determining the upload time to the central node, the upload times for the first-level upload operation, the second-level upload operation, and the third-level upload operation increase sequentially.

10. A monitoring terminal for a combined sewer overflow system, applied to the data processing method for the combined sewer overflow system as described in any one of claims 1-9, characterized in that, include: The cavity is made of 316L stainless steel or titanium alloy, and the cavity is provided with a transparent window of polytetrafluoroethylene facing the outside, and the transparent window is coated with a hydrophobic nano-coating. A radar liquid level sensing module, wherein the radar liquid level sensing module is disposed on the inner wall of the cavity; A wireless communication module is disposed on the inner wall of the cavity; A gas sensing module, wherein the gas sensing module is fixedly connected to the wireless communication module; A temperature and humidity sensing module is disposed at the bottom of the cavity; A power management module is inserted into the side of the cavity; The first control module is fixedly connected to the top of the temperature and humidity sensing module; The second control module is disposed on the inner wall of the cavity.

Images