Pipe network operation and maintenance management system based on data analysis

Abstract

The application discloses a pipe network operation and maintenance management system based on data analysis, and relates to the technical field of municipal infrastructure operation and maintenance management.The system comprises a device feature modeling module, a group anomaly identification module, a common mode risk traceability module, a dynamic risk modeling module, a regulation and control strategy optimization module, and a closed-loop adaptive updating module.The device feature modeling module is based on historical operation data and real-time acquisition parameters, and encodes the type, batch, position and time sequence of sensors to construct a device attribute mapping matrix.The application realizes intelligent adjustment of regulation and control parameters and avoidance of abnormal interference by constructing the device attribute mapping matrix and the group anomaly identification mechanism, combining external induced factor positioning and a dynamic risk weight graph, constructing a closed-loop optimization process, and improving the identification accuracy, control safety and operation resilience of the pipe network system in a multi-disturbance scenario.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of municipal infrastructure operation and maintenance management, and particularly relates to a pipe network operation and maintenance management system based on data analysis. BACKGROUND

[0002] The "pipe network operation and maintenance management system based on data analysis" refers to a digital management platform that utilizes multi-source data collection, fusion and intelligent analysis technology to comprehensively monitor, evaluate and decision support the operation state, fault risk and maintenance demand of urban or industrial pipe network systems (such as water supply and drainage, gas, heat, etc.). The system integrates real-time collection of key operation parameters such as pressure, flow, temperature, water quality through sensors, combines with historical operation and maintenance records, geographic information system (GIS) data and environmental factors, adopts data mining, anomaly detection, trend prediction and intelligent scheduling algorithms to realize precise management and optimization of pipe network operation efficiency, fault early warning, hidden danger positioning, maintenance plan formulation and other links, improve the intelligent level of pipe network operation and maintenance, reduce the cost of manual intervention, and ensure the safe, efficient and sustainable operation of public infrastructure.

[0003] The prior art has the following disadvantages:

[0004] In the existing pipe network operation and maintenance management system based on data analysis, a single-point anomaly tolerance strategy is usually adopted, that is, when the operation parameters collected by a certain sensor show sudden changes or deviate from the expected threshold, the system usually handles it as individual device failure, short-time interference or occasional fluctuation, thereby avoiding false action caused by misjudgment. However, in actual application, a large number of sensor devices deployed in the same subsystem or the same geographic area often have the same model, production batch or hardware architecture, and have high homogeneity. When such sensors encounter dramatic changes in electromagnetic environment (such as lightning, electrical equipment start-stop), firmware program defects (such as logic error, temperature drift failure) or are attacked by attackers through specific protocol interfaces to implement batch control operations, it is easy to cause group synchronous deviation or consistent distortion.

[0005] Due to the lack of an abnormal consistency recognition mechanism for the common characteristics of devices in the existing system, in the above-mentioned situation, the system still handles it as "multiple independent device single-point fluctuation", which cannot identify the potential systematic common mode risk source behind it, thereby leading to incorrect operation state judgment and scheduling decision. For example, in the case of multiple pressure sensors reporting false high data due to common mode interference, the system may mistakenly believe that the pipe section pressure is insufficient and enable pressure compensation measures, further exacerbating the overpressure risk, causing equipment damage, energy leakage and even pipe explosion accidents, with extremely serious consequences.

[0006] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0007] The purpose of the present application is to provide a pipe network operation and maintenance management system based on data analysis, which realizes intelligent adjustment of control parameters and avoidance of abnormal interference by constructing a device attribute mapping matrix and a group anomaly recognition mechanism, combining external induced factor positioning and a dynamic risk weight graph, constructing a closed-loop optimization process, and improving the identification accuracy, control safety and operation resilience of the pipe network system in a multi-disturbance scenario to solve the problems in the above background technology.

[0008] In order to achieve the above purpose, the present application provides the following technical scheme: a pipe network operation and maintenance management system based on data analysis, comprising a device feature modeling module, a group anomaly recognition module, a common mode risk tracing module, a dynamic risk modeling module, a control strategy optimization module and a closed-loop adaptive updating module:

[0009] The device feature modeling module encodes the type, batch, location and time sequence of the sensor based on historical operation data and real-time acquisition parameters to construct a device attribute mapping matrix.

[0010] The group anomaly recognition module uses the device attribute mapping matrix to identify homologous sensors, extracts their operation parameters within a unit time, calculates cross-point difference factors and time sequence synchronization factors, and generates a group anomaly preliminary judgment vector.

[0011] The common mode risk tracing module acquires electromagnetic interference, firmware upgrade and remote access records based on the group anomaly preliminary judgment vector, analyzes the time sequence relationship between them and abnormal deviation, identifies common mode risk inducing factors, and determines the corresponding inducing mechanism.

[0012] The dynamic risk modeling module reconstructs the parameter weights in the affected area according to the influence range of the common mode risk inducing factors, generates a dynamic risk weight graph in combination with the operation parameter trend, and represents the priority relationship between the control target and the resources.

[0013] The control strategy optimization module adjusts the valve opening, pump group start-stop sequence and alarm threshold in the target area based on the dynamic risk weight graph to avoid interference from common mode deviation sensors and improve control accuracy and safety.

[0014] The closed-loop adaptive updating module writes the control process data into the device attribute mapping matrix, updates the group anomaly preliminary judgment vector and the dynamic risk weight graph, and realizes closed-loop optimization of anomaly recognition and control parameters.

[0015] Preferably, the device attribute mapping matrix is constructed by the following steps:

[0016] Collect operation parameter records, operation and maintenance records, geographic information and environmental information covering the sensor operation period to form a unified data view.

[0017] The hardware information, manufacturing information, deployment environment information and runtime sequence information of the sensor are feature-encoded based on a data view;

[0018] The entire feature encoding is converted into a structured field and embedded into a device attribute mapping matrix in the form of a two-dimensional array;

[0019] The mapping matrix is subjected to missing value filling, logical conflict elimination and feature validity verification to complete the construction of the structured data base.

[0020] Preferably, the generation of the group anomaly preliminary judgment vector comprises the following steps:

[0021] The sensors with the same model number, production batch and deployment area are screened to form a homologous device set;

[0022] The running parameter data of the devices in the set within a unified time window are extracted, time-aligned and missing-completed;

[0023] The cross-point data difference factor and the time sequence synchronization factor between the devices in the set are calculated to obtain the behavior consistency analysis result;

[0024] A group anomaly preliminary judgment vector representing the synchronous offset abnormal state is generated based on a set threshold.

[0025] Preferably, the identification of the common mode risk inducing factor and the determination of the corresponding inducing mechanism comprise the following steps:

[0026] Based on the time period marked by the group anomaly preliminary judgment vector, the corresponding electromagnetic interference record, firmware upgrade information and remote access log are extracted;

[0027] The occurrence time of the electromagnetic interference record, firmware upgrade information and remote access log is respectively compared with the time sequence of the group offset start time;

[0028] The time axes of the three types of events are unified to construct a time sequence correlation graph, and the influence range and credibility of each event on the abnormal offset are analyzed;

[0029] According to the event coincidence degree and the influence degree, the common mode risk inducing factor causing the group offset is determined.

[0030] Preferably, the generation of the dynamic risk weight map comprises the following steps:

[0031] Based on the spatial influence range of the common mode risk inducing factor, the controlled device set in the risk intervention area is calibrated;

[0032] The pressure, flow and temperature parameters of the devices in the set within a fixed period before and after the common mode event are extracted, and the running response intensity is calculated;

[0033] The comprehensive risk weight value of each device is calculated in combination with the control criticality of the device in the pipe network.

[0034] According to the comprehensive risk weight values of all devices and the deployment positions, a dynamic risk weight map is constructed, and a risk gradient channel is marked with a weight difference value.

[0035] Preferably, adjusting the valve opening degree in the target area, the pump group start-stop sequence and the alarm threshold value comprises the following steps:

[0036] Extracting nodes with a risk weight score higher than 75 in the dynamic risk weight map, and marking a control intervention device list;

[0037] According to the risk level of the control point, adjusting the opening set value of the electric valve, and gradually executing according to a preset amplitude;

[0038] Adjusting the start-stop timing and running rate parameters of the pump group according to the risk level;

[0039] Temporarily correcting the threshold parameters of the alarm model, and adding a homologous device comparison rule to control the alarm triggering condition.

[0040] Preferably, after writing the running data in the control execution process into the device attribute mapping matrix, the following steps are included:

[0041] Writing the pressure value, flow rate, temperature and control device response state data into the corresponding fields;

[0042] Extracting the running parameter data of the homologous sensor, calculating the cross-point data difference factor and the time sequence synchronization factor;

[0043] Generating an updated group anomaly preliminary judgment vector based on the cross-point data difference factor and the time sequence synchronization factor;

[0044] According to the updated preliminary judgment vector, refreshing the dynamic risk weight map, and adjusting the risk level and response priority of each node.

[0045] In the above technical solution, the technical effects and advantages provided by the present application are as follows:

[0046] The application realizes accurate identification of the same source device group by constructing the device attribute mapping matrix, systematically integrating the multi-dimensional attributes of the sensor model, batch, deployment location and running time sequence; further, the preliminary judgment mechanism of group abnormal deviation is first established through the analysis of difference factors and synchronization factors, and combined with external induced events such as electromagnetic interference, firmware update and remote access behavior, the positioning and mechanism identification of the common mode risk source are realized. On this basis, the quantitative mapping from the running state to the control strategy is realized through the dynamic risk weight diagram, so that the system can dynamically adjust the valve, pump group and alarm threshold, avoid false data interference, and ensure the accuracy and safety of regulation and control. Finally, through the continuous write-back of running data and model iterative update, a data-driven closed-loop optimization mechanism is constructed, realizing the continuous adaptation and self-evolution of the system to complex abnormal situations. The scheme not only improves the accuracy and efficiency of abnormal response, but also significantly enhances the resilience and stability of the pipe network system in a multi-disturbance environment, and has wide practical application value. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only represent some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0048] Figure 1 A module schematic diagram of the pipe network operation and maintenance management system based on data analysis of the present application. DETAILED DESCRIPTION

[0049] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the inventive aspects to those skilled in the art.

[0050] The present application provides a pipe network operation and maintenance management system based on data analysis as shown in Figure 1 The pipe network operation and maintenance management system based on data analysis comprises a device feature modeling module, a group anomaly identification module, a common mode risk traceability module, a dynamic risk modeling module, a regulation and control strategy optimization module and a closed-loop adaptive updating module.

[0051] The device feature modeling module performs multi-dimensional feature coding on the model, production batch, installation location and running time sequence of the sensor device based on full historical running data and real-time acquisition parameters, generates a device attribute mapping matrix, and provides structured basic data for abnormal consistency analysis;

[0052] Based on the deep processing of the full historical operation data and real-time collected parameters of the sensor devices deployed in the pipe network system, a data infrastructure for supporting abnormal consistency identification is constructed, specifically a device attribute mapping matrix containing multiple dimension fields. The process includes the following steps:

[0053] Collect raw data and associated information covering the entire cycle of pipe network operation to form a raw data set that can be used for device attribute coding. The data set includes: the operation parameter records of various sensors from installation to the current time period, specifically including the pressure value change curve of the pressure sensor, the flow rate change record of the flow sensor, the temperature response log of the temperature sensor, and the dissolved oxygen content, conductivity, and turbidity values reported by the water quality monitoring device; at the same time, it also includes all operation and maintenance records of the device in its life cycle, such as repair time, replacement parts details, downtime, debugging period, and the specific time of each fault occurrence and the handling method of the operation and maintenance personnel. In addition, geographic information data needs to be extracted, including the exact coordinates of the sensor device in the city pipe network, the pipe segment number it belongs to, the upstream and downstream pipe diameter, the water supply or heating area identification it belongs to, and the environmental factors such as terrain elevation difference, surrounding building density, and manhole cover closure. After cleaning, timestamp alignment, and exception removal of the above types of data, a complete structure data index set is formed, and a unified multi-source data view is constructed with the sensor unique identification code as the primary key.

[0054] Based on the above data view, each sensor device is encoded from four aspects: hardware information, manufacturing information, deployment environment information, and runtime sequence information. In terms of hardware information, record its product model, rated range, sampling accuracy, signal output method (such as 4-20mA analog, Modbus digital communication, NB-IoT wireless transmission), working temperature range, and power input specification; in terms of manufacturing information, specify the device manufacturer name, production factory number, product batch number, factory date, and product certificate number; in terms of deployment environment, extract the pipe material type (such as ductile iron, polyethylene, steel pipe), burial depth, well site number, valve or pump group number connected, pipe installation year, historical water hammer impact frequency, and surrounding high pressure equipment distribution; in terms of runtime sequence, count the number of device start and stop, daily effective data upload rate, frequency of abnormal data segment, response delay of data change with adjacent devices, and maximum daily data variation amplitude in the past month. All the above features are converted into structured fields, numbered and classified according to the unified field coding specification, and a field dictionary is established to ensure consistency between different devices.

[0055] The above-mentioned features are embedded in the device attribute mapping matrix in a structured form. The mapping matrix is constructed in the form of a two-dimensional array, where each row corresponds to a unique sensor device and each column corresponds to a specific encoding field. The matrix has a total of 48 fields, covering 12 hardware specification fields, 6 manufacturing information fields, 18 deployment environment fields, and 12 runtime sequence fields. Each field has a clear unit, data type, and value range. For example, the "well site number" in the deployment environment is a string field with a value of city number plus geographic coordinate encoding; "signal output mode" is a classification field with a value range limited to analog, voltage digital, serial communication, or wireless communication; "daily data upload rate" is a floating-point field with a value range of 0 to 1. All numerical fields are standardized with units, classification fields are encoded with one-hot encoding, and Boolean fields are encoded with binary encoding. The generated matrix is exported to the data repository in CSV format and used through the data interface for subsequent query, feature selection, and device grouping processing.

[0056] The constructed device attribute mapping matrix is subjected to multi-layer verification to ensure data consistency and integrity. First, field-by-field check for missing values, zero padding for null values in Boolean fields, and historical mean interpolation for missing values in numerical fields; second, screen for logical conflicts between fields, such as a sensor being in the "running state" field with a stop state while having sampling records within the past week; third, statistical analysis of field values to identify outliers, such as devices with "daily abnormal segment number" exceeding the reasonable threshold marked as "feature deviation anomaly"; finally, verify the availability of the mapping matrix by selecting three representative deployment areas, extracting the attribute vectors of all sensor devices, comparing their feature similarity distribution curves, and confirming the matrix's ability to distinguish between homologous and non-homologous devices in real data. The device attribute mapping matrix allows for field-level incremental updates when device status changes (such as model replacement, re-deployment, or communication method change) after construction, ensuring the mapping matrix remains effective throughout the entire pipeline life cycle and providing a high-precision, strongly interpretable structured data foundation for subsequent group anomaly identification.

[0057] The core role of this step is to provide a unified, complete, and comparable structured data basis for subsequent anomaly consistency identification. By encoding the key attributes of sensor devices such as model, production batch, installation location, and runtime sequence in multiple dimensions, the dispersed, heterogeneous, and unstructured data can be transformed into a device attribute matrix with semantic association and logical structure. This matrix not only encodes and abstracts the static attributes and dynamic behavior of each sensor device, but also provides technical support for comparing the group characteristics between homologous devices. In practical applications, sensors of the same model or same production batch often have similar response characteristics and potential common defects, while the running environment and time sequence information determine their sensitivity to external disturbances or systemic risks. Therefore, by standardizing the modeling of the above information, the system can identify the structural homology relationship between devices when faced with large-scale sensor data, and conduct synchronous analysis of group behavior accordingly. Once the system detects multiple devices with abnormal deviation at the same time, it can quickly query their belonging in the device attribute mapping matrix to determine whether they have common mode attributes or deployment relevance, thereby avoiding misjudgment of group system failure as individual fluctuations of multiple independent devices. Therefore, the construction of this structured matrix is not only the basis of data management, but also the key prerequisite for ensuring the logical rigor of the identification algorithm and the reliability of the judgment result.

[0058] The group anomaly identification module, with the support of the device attribute mapping matrix, identifies homologous sensors of the same model, production batch, and deployment area, obtains their running parameter data within a unit time, calculates cross-point data difference factors and time sequence synchronization factors, generates a group anomaly preliminary judgment vector, and identifies the highly consistent abnormal deviation behavior of multiple homologous sensors within the same time period.

[0059] Relying on the device attribute mapping matrix that has been constructed, the identification of sensor group abnormal deviation behavior is carried out. This identification process revolves around the comparison of homologous device behavior and specifically includes device screening, parameter extraction, behavior consistency analysis, and abnormal preliminary judgment vector generation, which clearly realizes the judgment of whether multiple sensors have group synchronous deviation within the same time period. The specific steps are as follows:

[0060] The registered sensor devices are conditionally filtered by the model field, production batch field and deployment area field recorded in the device attribute mapping matrix. The specific filtering criteria are: the model field is completely consistent, for example, all are "PS-2020B" type pressure sensors; the production batch field is the same production serial number, such as "B0421"; the deployment area field corresponds to a continuous geographical location, specifically, the city pipe network number is continuous or the geographical coordinates fall within the same subnet range, for example, all devices are located in the pressure pipeline numbered "Z001" to "Z008". Through the joint conditions of the three types of fields, the devices that meet the conditions are divided into a homologous device set with structural consistency and similar operating environment. The devices within this set are highly consistent in structure, manufacturing and deployment, providing a basis for comparing group operating behavior.

[0061] From the corresponding operating data of all devices in the homologous device set, the complete data segment within a certain unit time window is extracted. The time window can be a fixed length, for example, 30 consecutive minutes, and the specific value is set according to the frequency of the target parameter. The extracted data parameters include but are not limited to: the output pressure value of the pressure sensor (unit: megapascal), the measured temperature value of the temperature sensor (unit: Celsius), and the instantaneous flow rate of the flow sensor (unit: cubic meters per hour). During the extraction process, the data sequences of each device are aligned by timestamp to ensure that the start and end times of each device's data sequence are consistent, and the interval spacing is uniform at one record per minute. If some data is missing at certain time points, for example, the sensor does not upload records at a certain minute, then the valid data of the device in the previous and next five minutes is referred to, and the gap is filled by linear interpolation, thereby ensuring the data integrity in the subsequent processing link.

[0062] After the data preparation is completed, the operating parameters between each device in the homologous device set are analyzed for synchronization and difference. For difference analysis, the numerical difference between the parameter values of two devices at the same time point is calculated in units of one minute, the absolute value is recorded, and the average difference value in the entire time period is accumulated as a measure of the degree of device offset. For synchronization analysis, the direction of parameter fluctuation of each device at consecutive time points is observed. For example, if four out of five devices show a consistent upward trend at three consecutive time points, it is recorded as a "synchronous upward behavior". This statistical process is repeated multiple times within the entire time window, and the proportion of synchronous change behavior is finally calculated. The results of the above two analyses together form a response behavior description quantity for the device set within a specified time window, reflecting whether the overall behavior presents a coordinated and synchronized change state.

[0063] The analysis result is converted into a preliminary judgment vector for identifying group abnormal deviation. A judgment threshold is set, for example, when more than 80% of the devices in the device set have a synchronization behavior ratio higher than 70% in the time window, and the average parameter difference value of each device is significantly deviated from the normal value range of the historical data of the set (such as more than ±15% range), the time window is marked as a "group deviation suspicious time period". Correspondingly, a preliminary judgment vector is generated for the time period, and the vector value of "1" represents that there is a synchronization deviation suspicion, and "0" represents no abnormality. The preliminary judgment vector is indexed by time period and bound with the device set identifier, which is used for subsequent process calling for identifying potential common mode induced mechanism, and can be used as a preliminary signal for triggering the control response mechanism.

[0064] Through the above steps, the consistency of the running behavior of the sensors deployed in the same area, with the same model and production source, in the same time period can be effectively analyzed, so as to determine whether there is a synchronization deviation abnormal phenomenon with non-individuality but systematic background. This method effectively breaks through the traditional way of processing sensor single-point abnormality in isolation, improves the accuracy of multi-device joint behavior identification and the foresight of response, and provides a technical precondition for subsequent processing of common mode interference, electromagnetic influence or human manipulation risk.

[0065] The role of this step is to analyze the running behavior of a group of sensor devices with the same model, the same production batch, and deployed in adjacent areas, to identify whether there is a highly consistent abnormal deviation phenomenon in a specific time period, so as to effectively find potential systematic risk or common mode interference problem. The traditional pipe network operation and maintenance system usually adopts single-point abnormality judgment logic, that is, when the data fluctuation of a certain sensor exceeds the threshold, the system regards it as an isolated device fault or incidental abnormality, and does not have the ability to analyze the consistency of the behavior of multiple devices. However, in actual application, a large number of sensors often come from the same production line, have the same hardware architecture, communication protocol and internal circuit design, and are concentrated in a certain area in deployment. When subjected to external electromagnetic interference, batch configuration error or remote manipulation behavior, it is easy to show group synchronization distortion. Therefore, this step filters the key attributes recorded in the device attribute mapping matrix, locks the device set with structural homology relationship, and further extracts the running data of each device in unit time to calculate the parameter difference value and change trend consistency index between devices with minute or second level granularity, to determine whether there is a similar deviation behavior mode occurring simultaneously in a large area and multiple devices. The finally generated group abnormality preliminary judgment vector can not only be used as an input basis for subsequent common mode risk analysis, but also as a warning signal in real-time monitoring system to prompt the dispatcher that the current abnormality may not be an isolated event, but a potential wide-area coordinated distortion phenomenon, thereby avoiding false control decisions and resource misallocation, and improving the sensing ability and operation response accuracy of the system.

[0066] The common mode risk traceability module collects electromagnetic interference records, firmware upgrade information, and remote access behavior logs based on the time period corresponding to the group anomaly preliminary judgment vector, analyzes the time sequence correlation between the abnormal offset starting point and the above external events, judges whether there is a common mode risk inducing factor, and determines the corresponding inducing mechanism;

[0067] To accurately identify potential common mode inducing factors of multiple sensor devices that have highly consistent abnormal offsets within the same time period, it is necessary to retrieve relevant external environmental data, device software state change information, and network access behavior records based on the time period marked by the aforementioned generated group anomaly preliminary judgment vector, and comprehensively analyze the time sequence relationship between the abnormal starting point and external factors to determine whether there is a traceable common mode risk event and further clarify its inducing mechanism. This process includes the following steps:

[0068] According to the time period marked as "synchronous anomaly" in the group anomaly preliminary judgment vector, the starting time, ending time, and device set identification of the time period are extracted. Specifically, in the time window with a preliminary judgment vector value of "1", the start minute and end minute are determined, and the corresponding device attribute information is located, including sensor number, deployment area number, product model, and batch number. Subsequently, according to the deployment area number, all electrical interference record logs in the area are retrieved, which come from interference monitoring devices deployed in substations, distribution cabinets, and key power control points. The data fields include interference occurrence time, interference type (such as surge, voltage dip, electromagnetic pulse), interference intensity level (units: volts / meter or amperes / meter), and interference range. Each event in the record is marked with an accurate timestamp with an accuracy of not less than one second, and is accompanied by interference source location information and influence radius. By comparing the time sequence relationship between the interference event time and the group offset starting time, it is determined whether there is a time overlap or a high-intensity electromagnetic interference occurring in advance.

[0069] Within the same time period, all firmware upgrades or configuration modification events are retrieved from the sensor device maintenance record platform. This type of information is regularly archived by the device management platform, and each record contains device unique number, upgrade operator identification, operation time, pre-upgrade version number, post-upgrade version number, upgrade content summary, and upgrade completion status. Records involving measurement accuracy parameter adjustment, signal processing logic optimization, or communication protocol version change in the upgrade content are particularly noteworthy. Compare the upgrade operation time with the abnormal offset time. If both are within the same time window, or the upgrade operation occurs within five minutes before the anomaly, it is considered a potential internal configuration change inducer. In particular, when multiple devices complete firmware upgrades within a short period of time, and data synchronization offset occurs immediately after the upgrade, it is necessary to further associate the upgrade content with the change in device response characteristics to determine whether the anomaly is caused by program logic mismatch or parameter initialization error.

[0070] The remote access behavior log involving the above-mentioned sensor device in the same time period is obtained, and the legitimacy of the access behavior is analyzed. The log includes remote access initiation IP address, access timestamp, operation type (such as remote data reading, parameter issuing, remote restart), access account ID, device response state, and log check value. The following two situations are focused on during screening: first, whether there are a large number of concentrated access requests initiated from abnormal IP addresses; second, whether there are operations of batch configuration change through privilege escalation. If such behavior is found to occur before the group abnormal offset time point, and the access object set is concentratedly directed to the abnormal device set, it can be preliminarily determined that there may be abnormal manipulation or configuration abuse behavior. In addition, for the records of operation type "remote parameter batch issuing" in the access log, it is necessary to further check whether the operation execution is successful and whether the issued content involves sensor parameter change items, such as threshold setting, filter period adjustment, and other key control fields.

[0071] The data from the above three sources are combined for time sequence matching and causal analysis. The electromagnetic interference record, firmware upgrade information, and remote access log are unified on the time axis to establish a time sequence correlation graph with "group offset time" as the anchor point. In the graph, the electromagnetic event, upgrade behavior, and remote access behavior are respectively taken as event nodes, and their occurrence time, object, and influence level are marked. They are arranged in chronological order, and the time difference between each event and the abnormal starting point is marked. By analyzing the time sequence relationship, the degree of overlapping of the action area, and the number of affected devices of these events, it is determined whether there is an obvious inducing event. For example, when an electromagnetic interference record appears within two minutes before the abnormality, the interference influence radius covers the deployment positions of all abnormal devices, the interference intensity reaches the warning level or above, and the device parameter set appears a synchronous offset phenomenon after that, it can be determined that the interference event is a high-possibility common mode risk source. At the same time, if there is a firmware upgrade behavior in the same period but only involves a small number of devices, or the remote access log has no suspicious operation, it is determined that the electromagnetic interference is the main inducing factor. After the analysis, the type, time, action range, and credibility of the common mode inducing factors are recorded to provide accurate basis for subsequent risk response and control measures.

[0072] The role of this step is to further analyze whether there is an external inducing factor causing the group anomaly after identifying that multiple homologous sensors have highly consistent abnormal offsets in the same time period, thereby avoiding misjudgment of systematic risks as individual device failures, and improving the accuracy of anomaly identification and the reliability of operation decisions. Specifically, this step locates the time period marked in the group anomaly preliminary judgment vector, extracts three types of key data covering this period: first, electromagnetic interference records, such as strong electromagnetic field changes caused by lightning strikes, instantaneous start-stop of electrical equipment, transformer operation, etc.; second, sensor firmware upgrade information, including version updates, parameter configuration changes, program logic reloads, and other operation records that may affect device measurement or communication behavior; third, remote access behavior logs, involving remote reading and writing, parameter issuance, start-stop control, and other operations performed by external personnel or systems through network interfaces. By aligning these data on the time axis and accurately comparing them with the starting point of the group anomaly offset, it is determined whether the three types of events occurred shortly before the offset, whether they cover all devices that appeared abnormal, and whether they have the potential to affect data measurement, signal transmission, or logic execution, thereby determining whether there is a common mode risk inducing factor. If a certain electromagnetic interference event or a firmware upgrade operation is found to be highly coincident with a wide range of device offsets and is adjacent in time to the offset occurrence point, it can be inferred that the event is one of the inducing mechanisms. This analysis process plays a key role in converting from "result identification" to "cause tracing", not only improving the control ability of the source of the anomaly, but also providing a basis for subsequent regulation and optimization, preventing ineffective or excessive responses due to false identification, and ensuring the stability and safety of the pipeline system.

[0073] A dynamic risk modeling module reconstructs the weights of the operating parameters in the affected area according to the influence range of the common mode risk inducing factor, generates a dynamic risk weight map based on the variation trends of multi-point pressure, flow and temperature, and represents the priority of the regulation and control target and the distribution relationship of the control resources between nodes.

[0074] In response to the group anomaly of sensors induced by the common mode risk inducing factor, real-time evaluation of key operating parameters in the affected area is required, and a set of dynamic risk weight maps with structural logic is constructed from the perspective of risk propagation, thereby providing a basis for regulation strategy development. This process involves risk impact positioning, operating data evaluation, risk weight quantification, and map construction, with the following specific steps:

[0075] According to the identified common mode risk inducing factors, the influence area and object range in the spatial dimension are determined. The specific operation includes extracting the influence radius of the interference event, the electromagnetic source position, the action time period, and the influence intensity level. If it is an electromagnetic interference event, the coordinate point where the interference occurs and its radius of influence are obtained, usually in meters, combined with the GIS geographic information system to determine whether it covers a certain specific water supply trunk road or heat return main line. For firmware upgrade type risk inducers, according to the device number listed in the upgrade log, combined with the device attribute mapping matrix, the corresponding physical deployment point and device channel are marked. Remote access operation is located relying on the target device number listed in the access control record. In this way, a clear set of controlled devices and physical deployment range are formed, marked as "risk intervention area".

[0076] For all sensor devices in the above "risk intervention area", fixed sampling periods (for example, 30 minutes before and after the event) are set before and after the common mode event, and the instantaneous pressure value (unit: megapascal) of the pressure sensor, the flow rate (unit: cubic meters per hour) of the flow sensor, and the medium temperature (unit: degree Celsius) of the temperature sensor are extracted. The time series of these parameters are organized in units of minutes, and the average value, maximum value, and change amplitude of each device in the two time periods before and after the event are quantitatively analyzed and compared with the historical statistical value of the device in the past 72 hours. For example, if the pressure of a certain pressure sensor increases by more than 1.5 times the 72-hour average fluctuation amplitude within 5 minutes after the event, it is marked as "severe change device". This change flag is used as a risk response factor of the current state of the device to form a "controlled area operation response list", which lists each device number, parameter type, change amplitude, response intensity level (scored from 0 to 100), and a Boolean flag indicating whether it has crossed the set safety threshold.

[0077] According to the state performance of each device in the "operation response list" and the location role of the device in the pipe network, a comprehensive risk weight evaluation is performed. Specifically, the following two contents are included: one is the operation response intensity, that is, the actual offset degree of each parameter fluctuation, for example, a pressure surge of 1.2 megapascal, a flow rate reduction of 40 cubic meters / hour, and a temperature drop of 10 degrees Celsius, which corresponds to different levels of risk score values; the other is the device control criticality, that is, whether the device is located in the main trunk of the water supply pipe network, whether it is directly connected to a primary pump station or a main control valve, and whether it supplies services to multiple end users in the downstream area. These two types of information are combined and processed to form a set of comprehensive weight values containing "fluctuation intensity score + control importance score". For example, the operation response intensity score of a certain node is 85 points, and the control criticality score is 90 points, so the final node risk weight value is (85+90) / 2=87.5. This weight value is used to represent the risk degree and control intervention priority of the device under the influence of this common mode risk.

[0078] Based on the node information with all weight values being determined, a complete dynamic risk weight graph is constructed. The graph takes physical deployment points as node units, and each node is attached with equipment number, geographic coordinates, running parameter type, current parameter value, comprehensive risk weight value and state label. The edges in the graph are physical connection relationships of pipelines, such as the pipeline segment path connected from the main water supply valve to the downstream distribution valve, and the direction mark represents the fluid flow direction. If the risk weight difference between two nodes is greater than a set threshold (such as 20 points), the edge is marked as a “risk gradient channel” for judging the risk propagation path. The high weight nodes in the graph structure are preferentially concerned by the control strategy, and the valve opening control, pump group start sequence and alarm threshold adjustment in the area where the nodes are located will trigger operation instructions based on the graph structure.

[0079] The role of this step is to further quantitatively evaluate each running node within the influence range of the common mode risk inducing factor after identifying it, and construct a “dynamic risk weight graph” in a structured way to determine the control priority and resource allocation relationship of each node under the current abnormal situation. Since the pipeline network system is composed of numerous sensors, valves, pump stations and control devices, there are significant differences in structural position, functional attributes, running state and risk exposure degree of each node, so the response degree of each node to the abnormality will also be different after electromagnetic interference, firmware upgrade anomaly or remote control anomaly. If these response differences are not dynamically identified and hierarchically classified, the system may have problems such as misissuance of control instructions, imbalance of resource allocation or error of risk transfer path. Through this step, for each node in the affected area, the response intensity can be calculated in combination with its pressure change value before and after the abnormality, flow velocity fluctuation amplitude, temperature deviation trend and other running parameters, and then different weight scores are given according to its key degree in the water supply or heating system, such as whether it is a main node, whether it is connected to multiple user terminals, etc. Further, all nodes are constructed into a network structure graph with flow direction relationship and weight distribution, in which the nodes with higher weight are preferentially controlled and intervened, and resource allocation should also be more inclined to such high-risk nodes. The finally generated dynamic risk weight graph not only improves the accuracy of the system response to the abnormality, but also has the ability of real-time updating and trend tracking, which can be used to guide the valve opening adjustment, pump group start-stop sequence optimization and alarm mechanism sensitivity setting, and is an important foundation support for realizing intelligent scheduling and risk closed-loop management.

[0080] The control strategy optimization module adjusts the valve opening, pump group start-stop sequence and alarm threshold of the analysis model in the target area according to the results of the dynamic risk weight graph, to avoid data interference from the common mode offset sensor and ensure the accuracy and safety of the system control behavior.

[0081] Based on the high-risk nodes and propagation paths identified in the dynamic risk weight map, parameter adjustment operations are carried out around the operating control points of the target area, specifically for the optimization of valve opening control, pump group start-stop timing and operation rate management, and threshold setting strategies in the system alarm model. Through the above adjustments, the interference of the common mode offset sensor on the operation and regulation of the pipe network is eliminated, ensuring the stability of the control logic and the accuracy of the response behavior. The entire process includes the following steps:

[0082] Extract the node information of all nodes in the dynamic risk weight map with a risk weight score higher than 75 points, identify the corresponding control device number, the pipe section it is in, the upstream and downstream connection relationship, and the device type (such as regulating valve, booster pump, pressure measurement point). The system determines the specific device list that needs to be controlled according to the graph, and sets the action priority for each node. Nodes with a score of 90 points or more are marked as first-level control points and are given priority for adjustment; nodes with a score between 75 and 89 points are considered as second-level control points and are adjusted in coordination with the first-level nodes. For example, the main water supply pipe section node P-006 is located in the center of the common mode offset influence, so its corresponding electric butterfly valve and outlet variable frequency pump are determined as first-level control points and enter the adjustment state.

[0083] For electric valve devices in first-level and second-level control points, according to the position of the node and the water supply stability requirements of the downstream area, the opening degree parameters are reconfigured. If the current opening degree is full opening (100%), the system will automatically adjust the opening degree set value to 80% after confirming that the downstream flow is sufficient and the pressure value is in a high trend, and gradually execute it in five minutes with an increment of 5%, to avoid instantaneous hydraulic fluctuation. For example, valve V-102 is located at the transition position of the main line to the branch pipe, and the dynamic map shows that the downstream pressure value is above 0.85 MPa (the design upper limit is 0.8 MPa) for 15 consecutive minutes, so the system will gradually reduce its opening degree from 100% to 80%, to suppress excessive flow release. For control points in the same area that are downstream and directly connected to high-risk nodes, such as V-104 and V-105, they are reduced to 90% and 85% respectively, forming a local flow dispersion pressure control strategy.

[0084] At the node involving the water pump equipment, the start-stop sequence and operation mode of the pump group are adjusted. If the current scheduling strategy originally plans to start the main water supply pump P-201 and the backwater pump P-202 after 5 minutes, the system will determine whether the start operation needs to be delayed or cancelled according to the risk weight evaluation result of the backwater area in the atlas. For example, when the branch pressure connected to P-202 is continuously at a high level due to sensor false alarm, the system determines that the current pressure compensation behavior will cause an overpressure risk, and delays the start time of P-202 by 15 minutes and keeps it in a low-speed preheating state. The main water supply pump P-201 is operated in a slow start mode, and the starting speed is set to be reduced from 1450 rpm to 1150 rpm, and a real-time pressure data check is performed before starting the pump to ensure that all monitoring points do not exceed the upper limit of operation. The entire pump group operation state is controlled by the atlas feedback cycle, and each control cycle is 10 minutes. The system will dynamically adjust the pump group operation parameters after updating the risk evaluation in the next round.

[0085] The alarm model currently deployed at each node is temporarily reset to prevent frequent triggering of invalid alarms due to distortion of sensor output data caused by common mode offset. Taking the pressure monitoring point as an example, the original alarm upper limit is set to 0.85 MPa and the lower limit is set to 0.25 MPa. If the sampling value of a node exceeds the upper limit for three consecutive times, and the corresponding sensor has been identified as a common mode offset affected object, the system will temporarily increase the upper limit to 0.95 MPa, and add a "same source device comparison rule" to the alarm condition. That is, if more than 70% of the same type of sensors show consistent offset, the alarm information will not be immediately triggered for execution, but will be transferred to the manual review process. This mechanism can minimize the risk of control errors caused by group false alarms without affecting the speed of abnormal response. The adjusted alarm threshold will be maintained for 30 minutes, and if the data returns to normal within this period, the original setting will be restored.

[0086] The role of this step is to adjust the key regulating parameters in the pipe network control system after identifying and quantifying the common mode risk impact area, according to the node risk level and operation state feedback results provided by the dynamic risk weight map, to avoid control misjudgment and response failure caused by the group offset of sensors. Common mode offset sensors often output a large range of consistent distorted data in a short time, such as synchronous pressure rising or synchronous flow rate falling. If the system performs control operations based on such data without distinction, it is easy to cause valve error opening, pump group misoperation, frequent system alarms, and even pipe explosion, heat fluctuation and other engineering risks. Therefore, this step reads the control device state of the high-risk node in the dynamic risk weight map to accurately adjust the valve opening, such as reducing the opening from 100% to 80% to weaken the false high pressure caused by false pressure compensation behavior; staggered or soft start strategy configuration is configured for the pump group start-stop sequence, such as delaying the start of two groups of pumps for 5 minutes and segmented speed-up to prevent water hammer impact caused by instantaneous flow rate surge; at the same time, for the alarm logic driven by sensor data in the analysis model, the upper and lower limits of the alarm threshold are dynamically improved, or a time lag and group consistency discrimination mechanism is introduced to avoid frequent invalid alarms caused by false abnormalities. Through these adjustment means, the system no longer completely relies on the current collected abnormal data, but makes comprehensive judgments based on risk distribution, structural characteristics and running trend to achieve more robust, safe and intelligent control response. This step not only improves the system's ability to adapt to data anomalies, but also provides key control support for ensuring the continuity, safety and stability of pipe network systems such as water supply and heat supply.

[0087] The closed-loop adaptive updating module writes the running data during the execution of the control operation into the device attribute mapping matrix after completing the control operation, reextracts the cross-point data difference factor and time sequence synchronization factor of the homologous sensors, continuously analyzes the offset behavior in unit time, and updates the group abnormality preliminary judgment vector and the dynamic risk weight map, realizing the continuous closed-loop optimization of abnormal consistency identification and control parameter setting.

[0088] To realize the continuous closed-loop optimization of abnormal consistency identification and control parameter setting, after completing the control operation guided based on the dynamic risk weight map, all kinds of real-time running data during the execution of this control operation need to be written into the device attribute mapping matrix, and based on the updated matrix, the response behavior of the homologous sensors in the current time period is reanalyzed, so as to further update the group abnormality preliminary judgment vector and the dynamic risk weight map, ensuring that the system has the ability of continuous perception, feedback correction and adaptive adjustment. This process includes the following steps:

[0089] All running data during the current round of regulation execution is written into the device attribute mapping matrix. The running data includes real-time pressure values of pressure sensors (unit: megapascal), flow rate changes of flow sensors (unit: cubic meters per hour), medium temperatures of temperature sensors (unit: degrees Celsius), and response state parameters of control-type devices (such as electric valves and pump sets), for example, actual opening percentage of valves, pump set startup delay time, frequency of frequency converter start and stop, and the like. Each type of data is marked with a unified timestamp, with a time accuracy of no less than minute level, to ensure complete and comparable time sequence of the data, and is classified into a corresponding mapping matrix row item according to a unique device identification number. At the same time, to ensure data quality, the system performs format checking and outlier screening on all data before writing, to ensure that there is no repeated sampling, format loss, or severely offset values.

[0090] Based on the updated device attribute mapping matrix, the same type, same production batch, and adjacent deployment location of the same source sensor set in the current time period are re-screened, and the corresponding running parameter data is extracted. In the data preparation stage, it is ensured that the extracted time period covers the whole process of control execution, for example, from 5 minutes before the regulation starts to 15 minutes after the regulation ends, to form a complete behavior observation window with a minimum length of no less than 20 minutes. The running data of all same source devices in the time period is aligned, and a cross-point data difference factor is calculated to measure the numerical offset amplitude of different devices at the same time point. At the same time, a time sequence synchronization factor is extracted to evaluate the consistency between devices in the change trend, for example, whether they simultaneously show a pressure rise or flow decrease trend in a certain time period. These factors together form the basis of the group behavior characteristics in the current period.

[0091] According to the difference factor and synchronization factor extracted above, a group anomaly preliminary judgment vector in the current time period is regenerated. If more than 70% of the same source sensors in the current period show a highly consistent offset direction (such as synchronous pressure rise and amplitude exceeding twice the historical average) within 10 minutes, the period is marked as a potential abnormal area, and the result is converted into a new preliminary judgment vector, which is used as the input of the next round of common mode risk identification. At the same time, the new behavior characteristics are compared and analyzed with the previous round of behavior characteristics to judge whether the device response changes before and after the regulation operation significantly converge, for example, whether the average difference factor decreases by more than 30% or the standard deviation of the synchronization factor is reduced to less than 60% of the previous value. If the above indicators change significantly, the current regulation strategy is recorded as an effective control measure; otherwise, it is suggested that the strategy model parameters need to be adjusted.

[0092] The dynamic risk weight graph is refreshed synchronously according to the updated group anomaly preliminary judgment vector. The specific operation is: the risk level of each node is re-evaluated, if the response parameter tends to be stable due to the regulation operation, the risk score value is reduced; if some nodes still have abnormal fluctuations or even an expanding trend after regulation, the score value is increased. The edge weight between the nodes in the graph is also adjusted, which is used to reflect the change of the risk propagation path. For example, the risk score of a certain upstream node is reduced from 88 to 72 after regulation, and the system adjusts it from the first response area to the second response area, and the corresponding pump group control authority and response priority are also adjusted. The whole graph refreshing process is automatically executed after each regulation operation is completed, forming a risk dynamic reconstruction mechanism based on data feedback, ensuring that each subsequent control strategy is based on the current latest running state, realizing the fusion of identification update, regulation correction and behavior prediction ability under data driving.

[0093] The role of this step is to realize the data-identification-regulation closed-loop feedback mechanism in the intelligent pipe network operation and management process, and through the write-back, analysis and update of the operation data generated in the regulation execution stage, the system has the ability of self-learning, self-correction and dynamic optimization. In the traditional pipe network control system, the regulation behavior is mainly one-time execution, that is, the control instruction is triggered according to the current state, but there is no real-time evaluation and mechanism adjustment of the execution result, which easily leads to lag of regulation effect or slow reaction to subsequent state changes. In this step, by writing the operation parameter data such as pressure, flow and temperature in the regulation execution process into the device attribute mapping matrix, the system obtains the complete correspondence between control behavior and result. On this basis, the homologous sensors are re-analyzed for synchronization and difference, cross-point data difference factors and time sequence synchronization factors are extracted, and it is judged whether the sensor group shows convergence or continues to have abnormal deviation under the current regulation influence. If it is found that the fluctuation of most device parameters tends to decrease and the deviation trend weakens, it means that the current regulation measure is effective; otherwise, it suggests that the control logic needs to be further optimized or the risk source needs to be re-evaluated. At the same time, by updating these analysis results to the group anomaly preliminary judgment vector and the dynamic risk weight graph, the system can reconstruct the current risk distribution pattern and node response priority, so as to adopt more accurate and reasonable scheduling decisions in the next round of control. This process realizes the continuous tracking of common mode abnormal response, the rolling identification of risk state and the adaptive adjustment of control scheme, and is a key supporting link to build high stability and high intelligence of urban or industrial pipe network operation and management ability.

[0094] Through the above pipe network operation and maintenance management system based on data analysis, the intelligent optimization of precise identification and control response of common mode abnormality is realized, and the structural short board of the existing system in dealing with the deviation problem of the homologous sensor group is effectively made up. The application constructs the equipment attribute mapping matrix, systematically integrates the multi-dimensional attributes such as the model, batch, deployment position and running time sequence of the sensor, realizes the accurate identification of the homologous equipment group, and further realizes the preliminary judgment mechanism of the group abnormal deviation through the analysis of the difference factor and the synchronization factor, and realizes the positioning and mechanism identification of the common mode risk source combined with the external induced events such as electromagnetic interference, firmware update and remote access behavior. On this basis, the quantitative mapping from the running state to the control strategy is realized through the dynamic risk weight diagram, so that the system can dynamically adjust the valve, pump group and alarm threshold, avoid false data interference, and guarantee the accuracy and safety of control. Finally, through the continuous write-back of running data and the iterative update of the model, a closed-loop optimization mechanism driven by data is constructed, and the continuous adaptation ability and self-evolution ability of the system to complex abnormal situations are realized. The scheme not only improves the precision and efficiency of abnormal response, but also significantly enhances the resilience and stability of the pipe network system in the multi-disturbance environment, and has wide practical application value.

[0095] The above only describes some exemplary embodiments of the application by way of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the application. Therefore, the above drawings and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the application.

Claims

1. A data analysis-based pipe network operation and maintenance management system, characterized in that, The device feature modeling module, the group anomaly recognition module, the common mode risk traceability module, the dynamic risk modeling module, the control strategy optimization module, and the closed-loop adaptive updating module are included. The device feature modeling module encodes the type, batch, location, and time sequence of the sensor based on historical operation data and real-time collected parameters to construct a device attribute mapping matrix. The group anomaly recognition module uses the device attribute mapping matrix to identify homologous sensors, extracts their operation parameters within a unit time, calculates cross-point data difference factors and time sequence synchronization factors, and generates a group anomaly preliminary judgment vector. The common mode risk traceability module collects electromagnetic interference, firmware upgrade, and remote access records based on the group anomaly preliminary judgment vector, analyzes their time sequence relationship with abnormal deviation, identifies common mode risk inducing factors, and determines their corresponding inducing mechanisms. The dynamic risk modeling module reconstructs the parameter weights in the affected area according to the influence range of the common mode risk inducing factors, generates a dynamic risk weight map combining the operation parameter trend, and represents the priority relationship between the control target and the resources. The control strategy optimization module adjusts the valve opening, pump group start-stop sequence, and alarm threshold in the target area based on the dynamic risk weight map to avoid interference from common mode deviation sensors, improving control accuracy and safety. The closed-loop adaptive updating module writes the control process data into the device attribute mapping matrix, updates the group anomaly preliminary judgment vector and the dynamic risk weight map, and realizes closed-loop optimization of anomaly recognition and control parameters. Generating the group anomaly preliminary judgment vector includes the following steps: Selecting sensors with the same type, production batch, and deployment area to form a homologous device set; Extracting the operation parameter data of the device set within a unified time window, aligning the time, and completing the missing data; Calculating the cross-point data difference factors and time sequence synchronization factors between the devices in the set to obtain the behavior consistency analysis result; Generating a group anomaly preliminary judgment vector representing the synchronous deviation abnormal state based on the set threshold; Generating the dynamic risk weight map includes the following steps: Based on the spatial influence range of the common mode risk inducing factors, the controlled device set in the risk intervention area is labeled; Extracting the pressure, flow, and temperature parameters of the devices in the set within a fixed period before and after the common mode event, and calculating the operation response intensity; Combining the control criticality of the devices in the pipe network, calculating the comprehensive risk weight value of each device; According to the comprehensive risk weight value of all devices and the deployment location, a dynamic risk weight map is constructed, and the risk gradient channel is marked with weight difference value.

2. The data analysis based pipe network operation and management system according to claim 1, characterized in that, Constructing the device attribute mapping matrix includes the following steps: Collecting operation parameter records covering the sensor operation period, operation and maintenance records, geographic information, and environmental information to form a unified data view; Encoding the hardware information, manufacturing information, deployment environment information, and operation time sequence information of the sensor based on the data view; Convert all feature encodings into structured field embedded two-dimensional array form in the device attribute mapping matrix; Performing missing value filling, logical conflict elimination, and feature validity verification on the mapping matrix to complete the structured data foundation construction.

3. The data analysis based pipe network operation and management system according to claim 1, characterized in that, Identifying common mode risk inducing factors and determining their corresponding inducing mechanisms includes the following steps: Based on the time period marked by the group anomaly preliminary judgment vector, the corresponding electromagnetic interference records, firmware upgrade information and remote access logs are extracted; The occurrence time of electromagnetic interference records, firmware upgrade information and remote access logs is compared with the time of group offset start time respectively; The time axis of the three types of events is unified, the time sequence correlation graph is constructed, and the influence range and credibility of each event on abnormal offset are analyzed; According to the event coincidence degree and influence degree, the common mode risk inducing factor causing group offset is determined.

4. The data analysis based pipe network operation and management system according to claim 1, characterized in that, The steps of adjusting the valve opening degree, pump group start-stop sequence and alarm threshold in the target area include: Extract the nodes with risk weight score higher than 75 in the dynamic risk weight graph, and label the control intervention equipment list; According to the risk level of control point, adjust the opening set value of electric valve, and execute gradually according to the preset amplitude; According to the risk level, adjust the start-stop time sequence and running speed parameters of pump group; Temporarily correct the threshold parameters of alarm model, and add homologous equipment comparison rules to control alarm triggering condition.

5. The data analysis based pipe network operation and management system according to claim 1, characterized in that, After writing the running data in the regulation and control execution process into the equipment attribute mapping matrix, the following steps are included: Write pressure value, flow rate, temperature and control equipment response state data to the corresponding field; Extract the running parameter data of homologous sensors, calculate the cross-point data difference factor and time sequence synchronization factor; Based on the cross-point data difference factor and time sequence synchronization factor, the updated group anomaly preliminary judgment vector is generated; According to the updated preliminary judgment vector, refresh the dynamic risk weight graph, adjust the risk level and response priority of each node.

Images