A method for monitoring urban lifeline underground pipe network data

Abstract

The present invention discloses a method for monitoring urban lifeline underground pipeline network data, including: S1, obtaining historical pipeline network data; S2, respectively calibrating data tags for the historical pipeline network data; S3, constructing an abnormal / fault data curve according to the historical data; S4, associating the abnormal / fault data curve with a processing solution; S5, inputting the historical pipeline network data with calibrated data tags into a deep learning model for training of classification tasks; S6, obtaining real-time pipeline network data; S7, inputting the real-time pipeline network data into a comprehensive deep learning model to obtain labels of these real-time data, constructing data curves of these real-time data, and calling an abnormal data curve or a fault data curve stored in a database for comparison based on the data labels of these real-time data, and outputting an abnormal or fault warning message and pushing a processing solution associated with the abnormal data curve or the fault data curve when the curve distance prediction value is met.

Description

Technical Field

[0001] The present invention relates to the field of data monitoring, and in particular to a method for monitoring urban lifeline underground pipeline network data. Background Art

[0002] The overall goal of the construction of the urban underground pipeline network geographic information system and the safety operation monitoring system is to continuously promote under the guidance of new technologies and new concepts, in accordance with the development idea of "risk management and moving the checkpoint forward". By establishing a scientific, standardized, systematic, and dynamic long-term mechanism for underground pipeline network safety management and guarantee, using advanced and reliable safety management concepts, public safety science and technology, Internet of Things technology, and mobile information technology, to carry out all-round Internet of Things monitoring, evaluation, and refined governance of the risk hazards of the city's underground pipeline network. With digital, networked, intelligent, and interactive construction, on the basis of the existing underground pipeline network comprehensive information management system and data in the urban construction archives, by integrating the existing infrastructure (including important bridges) safety operation and management information system resources of gas companies, water supply companies, and municipal management offices, coordinating the resources of each unit, and constructing an all-round and three-dimensional urban underground pipeline network safety monitoring network. In order to coordinate the resources of multiple pipeline networks and propose rapid collaborative processing of pipeline network risk hazards, a method for monitoring urban lifeline underground pipeline network data needs to be proposed. Summary of the Invention

[0003] The present invention integrates the safety operation data of the existing infrastructure (including important bridges) of gas, water supply, drainage, and municipal management offices, associates this data with mature processing solutions for pipeline network anomalies / failures, and constitutes an online operation monitoring system for urban lifelines, enabling timely response and rapid collaboration of each department when pipeline network anomalies / failures occur, and providing a method, system, electronic device, and computer-readable storage medium for monitoring urban lifeline underground pipeline network data.

[0004] According to a first aspect of the present invention, there is provided a method for monitoring urban lifeline underground pipeline network data. S1: Obtain historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data;

[0005] S2: Respectively perform data label calibration on the historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data. The data labels include data types and status types. The data types include water supply data type, drainage data type, gas data type, and bridge data type. The status types include normal, abnormal, and faulty;

[0006] S3. Construct a water supply anomaly data curve for the water supply network data with a data type of water supply data class and a status type of anomaly, a water supply fault data curve for the water supply network data with a data type of water supply data class and a status type of fault, a drainage anomaly data curve for the drainage network data with a data type of drainage data class and a status type of anomaly, a drainage fault data curve for the drainage network data with a data type of drainage data class and a status type of fault, a gas anomaly data curve for the gas network data with a data type of gas data class and a status type of anomaly, a gas fault data curve for the gas network data with a data type of gas data class and a status type of fault, a bridge anomaly data curve for the bridge network data with a data type of bridge data class and a status type of anomaly, and a bridge fault data curve for the bridge network data with a data type of bridge data class and a status type of fault;

[0007] S4. Associate the processing scheme for water supply network anomalies with the water supply anomaly data curve, the processing scheme for water supply network faults with the water supply fault data curve, the processing scheme for drainage network anomalies with the drainage anomaly data curve, the processing scheme for drainage network faults with the drainage fault data curve, the processing scheme for gas network anomalies with the gas anomaly data curve, the processing scheme for gas network faults with the gas fault data curve, the processing scheme for bridge network anomalies with the bridge anomaly data curve, and the processing scheme for bridge network faults with the bridge fault data curve;

[0008] S5. Input the historical water supply network data, drainage network data, gas network data, and bridge network data labeled with data tags into a deep learning model for training in a classification task;

[0009] S6. Obtain real-time water supply network data, drainage network data, gas network data, and bridge network data;

[0010] S7. Input the real-time water supply network data, drainage network data, gas network data, and bridge network data into a comprehensive deep learning model to obtain the data type and status type of these real-time data, construct the data curves of these real-time data, and based on the data type and status type of these real-time data, call the anomaly data curve or fault data curve stored in the database, calculate the curve distance between the real-time data curve and the anomaly data curve or fault data curve, and output an anomaly or fault warning message and push the processing scheme associated with the anomaly data curve or fault data curve when the curve distance meets the predicted value.

[0011] In some embodiments, the water supply network data includes water supply point pipeline data, water supply point pressure data, water supply point flow data, and water supply point flow velocity data. The sewer network data includes sewer point pipeline data, sewer point liquid level data, sewer point flow data, and sewer point water quality data. The gas pipeline network data includes gas pipeline data, gas pressure regulating box monitoring data, and gas well monitoring data. The bridge pipeline network data includes bridge point temperature data and bridge point strain data.

[0012] In some embodiments, the water supply network data further includes water supply point location data, the sewer network data further includes sewer point location data, the gas pipeline network data further includes gas point location data, and the bridge pipeline network data further includes bridge point location data. Wherein, if the absolute value of the difference between any two location data is less than a preset value, the abnormal data curves or fault data curves of the points corresponding to the two location data are associated.

[0013] In some embodiments, when a point corresponding to a point location data outputs an abnormal or fault warning message and a processing solution, the associated corresponding points also output an abnormal or fault warning message and a processing solution.

[0014] In some embodiments, the processing solutions associated with their respective abnormal data curves in multiple same time periods are associated, and the processing solutions associated with their respective fault data curves in multiple same time periods are associated.

[0015] In some embodiments, in step S3, abnormal data curves and fault data curves are constructed in time sequence. In step S7, real-time data curves are constructed in time sequence. The curve distance between the real-time data curve and the abnormal data curve is calculated based on the DTW algorithm, or the curve distance between the real-time data curve and the fault data curve is calculated based on the DTW algorithm.

[0016] In some embodiments, the curve change rate of the real-time data curve is calculated. When the absolute value of the difference between the curve change rate of the real-time data curve and the curve change rate of the abnormal data curve or the fault data curve is less than a preset value, an abnormal or fault warning message is output and the processing solution associated with the abnormal data curve or the fault data curve is pushed.

[0017] In some embodiments, based on the historical water supply network data, sewer network data, gas pipeline network data, and bridge pipeline network data, the status type of the data 46 to 90 days before the fault of the points corresponding to the pipeline network data is marked as abnormal, and the status type of the data 1 to 45 days before the fault of the points corresponding to the pipeline network data is marked as faulty.

[0018] In some embodiments, the method further includes: Step S8, inputting real-time water supply network data, drainage network data, gas pipeline network data, and bridge pipeline network data into a time prediction model, generating prediction data through data prediction based on time series, constructing a prediction data curve based on the prediction data, where the prediction data curve is used to calculate the curve distance by comparing with the abnormal data curve or the fault data curve, and outputting a prediction warning message and pushing a processing solution associated with the abnormal data curve or the fault data curve when the curve distance meets the predicted value.

[0019] In some embodiments, the deep learning model is an LSTM model, and the LSTM model includes a data type module and a state type classification module.

[0020] A monitoring system based on urban lifeline underground pipeline network data according to the second aspect of the present invention includes: a water supply network data acquisition module for acquiring historical water supply network data and real-time water supply network data; a drainage network data acquisition module for acquiring historical drainage network data and real-time drainage network data; a gas pipeline network data acquisition module for acquiring historical gas pipeline network data and real-time gas pipeline network data; a bridge pipeline network data acquisition module for acquiring historical bridge pipeline network data and real-time bridge pipeline network data; a label calibration module for calibrating data labels for the historical water supply network data, historical drainage network data, historical gas pipeline network data, and historical bridge pipeline network data, where the data labels include a data type and a state type, the data type includes water supply data type, drainage data type, gas data type, and bridge data type, and the state type includes normal, abnormal, and fault; a deep learning module for training the classification tasks of the historical water supply network data, historical drainage network data, historical gas pipeline network data, and historical bridge pipeline network data with calibrated data labels, and classifying the real-time water supply network data, real-time drainage network data, real-time gas pipeline network data, and real-time bridge pipeline network data; a data curve generation and comparison module for constructing a data curve, and constructing and calling an abnormal data curve and a fault data curve stored in a database according to the data curve, data type, and state type of real-time data.

[0021] An electronic device according to the third aspect of the present invention, the electronic device includes a processor and a memory coupled to the processor; the memory is used to store a computer program, and the processor is used to execute the computer program to implement the monitoring method of urban lifeline underground pipeline network data described in any one of the above.

[0022] A computer-readable storage medium provided according to the fourth aspect of the present invention, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for monitoring urban lifeline underground pipeline network data described in any one of the above.

[0023] The beneficial effects of a method for monitoring urban lifeline underground pipeline network data provided by the present invention are as follows: Compared with the prior art, this method integrates the safety operation data of existing infrastructure (including important bridges) of gas, water supply, drainage and the municipal management department, classifies and deeply learns the data, enables the system to learn to distinguish data, and does not limit the format of subsequent data input. Since the data formats input by each pipeline network may be different, intelligent identification is required for data classification. Then, many data interfaces can be extended to connect to different data sources, and it has strong versatility. When solving the situation of multiple data inputs and identifications, this method also constructs a relationship table between the abnormal data curve - processing method and the fault data curve - processing method based on historical data, compares the data curve constructed based on real-time data with the abnormal / fault data curve in the database, so as to quickly identify which one of the historical abnormalities / faults it is, and then call the processing scheme for the abnormality / fault. Each of these processing schemes has reference significance. Whether it is the processing method or the assistance opinions of relevant units that need to be combined, it can quickly promote the solution of the hidden dangers of abnormalities / faults, realizing a systematic and dynamic underground pipeline network safety management and ensuring a long-term mechanism. This method realizes the concepts of unified efficiency, reliable command, and rapid coordination, and forms a new model for urban underground pipeline network safety management and active risk prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic diagram of a method for monitoring urban lifeline underground pipeline network data provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the invention herein are for the purpose of describing specific embodiments only and are not intended to limit the invention. The singular forms "a", "the", and "said" used in the present invention and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms "first", "second", "third", etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other.

[0028] The methods and devices adopted in the present invention, unless otherwise specified, are conventional reagents, methods, and devices in this technical field.

[0029] The present invention will be further described in detail below with reference to the accompanying drawings.

[0030] Reference Figure 1 , according to the first aspect of the present invention, a method for monitoring urban lifeline underground pipeline network data is provided:

[0031] S1. Obtain historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data;

[0032] S2. Respectively perform data label calibration on the historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data. The data labels include data types and status types. The data types include water supply data type, drainage data type, gas data type, and bridge data type. The status types include normal, abnormal, and faulty;

[0033] S3. Construct a water supply anomaly data curve for water supply network data with a data type of water supply data class and a status type of anomaly, a water supply fault data curve for water supply network data with a data type of water supply data class and a status type of fault, a drainage anomaly data curve for drainage network data with a data type of drainage data class and a status type of anomaly, a drainage fault data curve for drainage network data with a data type of drainage data class and a status type of fault, a gas anomaly data curve for gas network data with a data type of gas data class and a status type of anomaly, a gas fault data curve for gas network data with a data type of gas data class and a status type of fault, a bridge anomaly data curve for bridge network data with a data type of bridge data class and a status type of anomaly, and a bridge fault data curve for bridge network data with a data type of bridge data class and a status type of fault;

[0034] S4. Associate the handling solutions for water supply network anomalies with the water supply anomaly data curve, the handling solutions for water supply network faults with the water supply fault data curve, the handling solutions for drainage network anomalies with the drainage anomaly data curve, the handling solutions for drainage network faults with the drainage fault data curve, the handling solutions for gas network anomalies with the gas anomaly data curve, the handling solutions for gas network faults with the gas fault data curve, the handling solutions for bridge network anomalies with the bridge anomaly data curve, and the handling solutions for bridge network faults with the bridge fault data curve;

[0035] S5. Input the historical water supply network data, drainage network data, gas network data, and bridge network data labeled with data tags into a deep learning model for training in a classification task;

[0036] S6. Obtain real-time water supply network data, drainage network data, gas network data, and bridge network data;

[0037] S7. Input the real-time water supply network data, drainage network data, gas network data, and bridge network data into the comprehensive deep learning model to obtain the data types and status types of these real-time data, construct the data curves of these real-time data, based on the data types and status types of these real-time data, call the anomaly data curve or fault data curve stored in the database, calculate the curve distance between the real-time data curve and the anomaly data curve or fault data curve, and output an anomaly or fault warning message if the curve distance meets the predicted value and push the handling solution associated with the anomaly data curve or fault data curve.

[0038] In some embodiments, the water supply network data includes water supply point pipeline data, water supply point pressure data, water supply point flow rate data, and water supply point flow velocity data. The sewer network data includes sewer point pipeline data, sewer point liquid level data, sewer point flow rate data, and sewer point water quality data. The gas pipeline network data includes gas pipeline data, gas pressure regulating box monitoring data, and gas well monitoring data. The bridge pipeline network data includes bridge point temperature data and bridge point strain data. Here, it should be noted that abnormal or faulty conditions such as the aging and leakage of water supply and drainage pipelines can be judged by the flow rate and flow volume detected in the pipelines. What the present invention needs to do is to calibrate the normal, abnormal, and faulty conditions based on the pipeline data provided by the water supply company and the drainage company. Specific values are not elaborated here. Additionally, the same applies to gas data and bridge data. It is necessary to confirm the specific normal, abnormal, and faulty values with companies or personnel in the corresponding fields, and then calibrate the data based on the values. Therefore, it can be understood that what this method needs to focus on is to calibrate the data to form a training set based on professional data in the technical field, and the values corresponding to these states can be continuously adjusted according to the actual situation. Therefore, specific values are not elaborated here. It should also be noted that the water supply point pipeline data, water supply point pressure data, water supply point flow rate data, water supply point flow velocity data, sewer point pipeline data, sewer point liquid level data, sewer point flow rate data, sewer point water quality data, gas pipeline data, gas pressure regulating box monitoring data, gas well monitoring data, bridge pipeline network data including bridge point temperature data and bridge point strain data can all generate their respective corresponding data curves during abnormal or faulty conditions. Among them, pipeline data such as other text information like the year of use can be digitized first and then the data curve can be constructed. Among them, in step S7, judging abnormal and faulty conditions based on the comparison results of the data curves can be done by judging a single data curve or by combining multiple data curves. Among them, the real-time data can be collected by the following methods: Gas data is mainly collected through combustible gas intelligent monitors. According to the device data volume measurement, an NB-IOT flow card is used. Drainage data is collected through liquid level gauges, water level gauges, rain gauges, and flow meters. According to the device data volume measurement, the front-end data upload uses a 4G monthly flow card. Water supply data is mainly collected through flow meters, pressure gauges, and online leakage monitoring devices. According to the device data volume measurement, the front-end data upload uses a 4G monthly flow card. Bridge data is mainly collected through monitoring devices such as strain gauges, acceleration sensors, deflection gauges, displacement gauges, and thermometers. According to the data volume measurement, a 20M dedicated line is used to access the monitoring center computer room.

[0039] Among them, the data related to pipelines or pipelines can specifically include the basic information of the pipeline / pipe, the planar position and burial depth, the ownership unit, the pipeline material, the construction year, the pipe diameter, the service capacity of the pipeline network node, the service scope, the type of adjacent pipelines, and the spacing, which is convenient for tracing the pipeline / pipe.

[0040] It should be noted that the described processing solution is the processing method and details for each accident archived by each relevant management department. Based on this method, it can be obtained from the existing processing solutions what kind of processing method for what kind of accident will cause what kind of impact on people's livelihood. The accumulated experience can enable the management department to respond to accidents more efficiently. This method adopts the technical architecture of public safety Internet of Things for sensing, transmitting, knowing, and applying. Based on the risk assessment of urban underground pipe networks, major risks are monitored in real time to sense the changes in risks and give early warnings in a timely manner. At the same time, based on a large amount of monitoring data, a public safety technology model for urban underground pipe networks is used to analyze and evaluate the safe operation status of urban underground pipe networks, analyze the secondary and derivative relationships of emergencies, and accurately judge and locate the accident points.

[0041] For example, when a gas leak occurs at a certain network point based on a curve, trace back the gas pipeline section where the leak may occur. According to the results of the leak traceability analysis and the comparison with the processing solutions of similar curves, combined with the actual boundary conditions around the leak point, give the diffusion trend of the leaked gas. With the help of the system function, the range of the "pollution area" after the gas leak can be obtained, so as to provide a reference and a quick processing solution for the division of the control area in the accident emergency disposal.

[0042] In some embodiments, the water supply pipe network data further includes water supply network point location data, the drainage pipe network data further includes drainage network point location data, the gas pipe network data further includes gas network point location data, and the bridge pipe network data further includes bridge network point location data. Among them, if the absolute value of the difference between any two location data is less than a preset value, then the abnormal data curves or fault data curves of the network points corresponding to the two location data are associated. It should be noted that since the laying of water supply, drainage, gas, and bridge pipelines in the city is complex and intertwined, once an abnormality / fault occurs at the intersection of these pipelines, it is necessary to jointly handle the abnormality / fault with other relevant departments. And the point to be solved in this step is that when an abnormality / fault occurs at a certain network point, quickly notify the other network points associated with the abnormal / fault network point, and based on the associated processing solution, a collaborative processing solution for all departments can be quickly drawn up with reference to the historical processing solution, or even directly refer to the historical processing solution. The impact between the urban lifeline underground pipe networks can be large or small, so it is necessary to obtain an effective and quick solution. In this embodiment, the preset value can be the default 50-meter distance, or it can be adjusted according to actual operations, which will not be elaborated here.

[0043] In fact, the data involved cover a vast range. Therefore, based on the network point location data and in combination with the BIM / GIS geographic information system, an underground pipeline network map of urban lifelines can be constructed, associating the data, curves, and processing solutions of each network point. Then, the underground pipeline network data can be further monitored through the macroscopic map, and the situation of each network point can also be intuitively understood. Based on the BIM / GIS system, the present invention uses technical means such as permission management and data exchange to open up the pipeline data sharing channels among relevant departments. At the same time, a three-dimensional model data of the above-ground buildings is established, matching the above-ground urban buildings with the underground pipeline network data to form a comprehensive management mode of "one map" for the underground and above-ground pipeline networks, breaking the information silos, establishing a unified standard, unified storage, unified update, and unified sharing of pipeline big data for the whole city's pipeline network. Based on big data, deeply understand the safe operation rules of the urban underground pipeline network, timely discover various risk hazards, realize the transparency of urban underground pipeline network safety supervision and the automation of service management, escort the safe and healthy operation of the urban underground pipeline network, improve the urban disaster reduction and prevention capabilities, and enhance the public's life safety index.

[0044] Based on the geographic information system, the existing underground pipeline network data and two- and three-dimensional basic data can be sorted out and converted to establish an underground pipeline database, meeting the requirements of the urban underground pipeline network geographic information system and the safety operation monitoring system project for pipeline data. The purpose of building the database is to effectively organize all relevant data, establish a unified spatial index according to their geographical distribution, and then quickly dispatch the data within any range in the database, achieving seamless roaming and management of the entire terrain and distribution services. According to the size of the display range, different levels of data can be automatically and flexibly loaded, enabling a panoramic view as well as seeing the minute details of local areas. All data should be able to be dispatched and browsed under a unified interface, and various scales and types of data should be able to be nested or superimposed on each other to form an integrated pipeline database, quickly locating accident points and delineating the affected areas.

[0045] In some embodiments, when a network point corresponding to a certain network point location data outputs an abnormal or fault warning message and a processing solution, the associated corresponding network points also output an abnormal or fault warning message and a processing solution. To prevent a network point from causing abnormalities in other network points when not collaborating with other departments in the abnormal / fault processing solution, this step introduces a multi-party participation method, allowing the responsible parties of the network points associated with the abnormal / fault points to participate in the abnormal / fault processing, preventing problems such as unclear understanding, unexpected situations, and ineffective management.

[0046] In some embodiments, the processing solutions associated with respective abnormal data curves in multiple same time periods are associated, and the processing solutions associated with respective fault data curves in multiple same time periods are associated. It should be noted that the association of processing solutions in the same time period is to find the associated processing solutions more quickly. For example, in the overlapping section of a water supply pipeline and a gas pipeline, when maintenance or repair is required for either the water supply pipeline or the gas pipeline, it is necessary to shut down the other pipeline for cooperation. Therefore, in the same time period of this maintenance or repair, the data curves and processing solutions of these two pipelines are necessarily associated. Here, the association is made according to the existing processing solutions to make the subsequent push of processing solutions faster and more accurate.

[0047] In some embodiments, in step S3, an abnormal data curve and a fault data curve are constructed in chronological order, and in step S7, a real-time data curve is constructed in chronological order. The curve distance between the real-time data curve and the abnormal data curve is calculated based on the DTW algorithm, or the curve distance between the real-time data curve and the fault data curve is calculated based on the DTW algorithm. It should be noted that since the data used in this method is all related to time series, and time series data may be affected by various factors, resulting in different stretching or deformation of the sequence on the time axis. For example, the same event may occur at different time points, but the overall pattern remains unchanged. However, DTW can find the optimal alignment between two time series by allowing non-linear deformation on the time axis, even if they have different paces in time. Therefore, based on DTW, this method has strong robustness to noise and local time deviation. It can be understood that under different data types and conditions, it can be between 0 and 10, and the specific distance value judgment is defined according to the actual situation, which will not be elaborated here.

[0048] In some embodiments, the curve change rate of the real-time data curve is calculated. When the absolute value of the difference between the curve change rate of the real-time data curve and the curve change rate of the abnormal data curve or the fault data curve is less than a preset value, an abnormal or fault warning message is output and the processing solution associated with the abnormal data curve or the fault data curve is pushed. It should be noted that in addition to calculating the curve distance based on DTW, the normal, abnormal, and fault changes of the data curve can also be reflected by calculating the curve change rate.

[0049] In some embodiments, based on the historical water supply network data, drainage network data, gas pipeline network data, and bridge network data, the status type of the data 46 to 90 days before the failure of the network points corresponding to the network data is marked as abnormal, and the status type of the data 1 to 45 days before the failure of the network points corresponding to the network data is marked as a fault. The most common anomalies in pipelines and bridges are data changes caused by material aging. Therefore, it is necessary to define the number of days of data before the fault to infer whether they have entered the aging period. When anomalies occur, personnel need to perform inspections and maintenance to prevent risks and eliminate potential hazards. It should be noted that, according to the actual situation and experience, the actual number of days can be adjusted.

[0050] In some embodiments, the method further includes: Step S8, inputting the real-time water supply network data, drainage network data, gas pipeline network data, and bridge network data into a time prediction model, generating prediction data based on time series data prediction, constructing a prediction data curve based on the prediction data, and using this prediction data curve to calculate the curve distance by comparing with the abnormal data curve or the fault data curve. When the curve distance meets the predicted value, a prediction warning message is output, and a processing plan associated with the abnormal data curve or the fault data curve is pushed. It should be noted that data prediction and the construction of prediction curves can make prevention and control more forward-looking.

[0051] In some embodiments, the deep learning model is an LSTM model, and this LSTM model includes a data type module and a status type classification module. The following is a fragment of code for constructing this LSTM model using Python as the processing language:

[0052] # Each file corresponds to a set of data

[0053] num_time_steps = 50

[0054] num_features_per_step = 10 # Adjust according to actual data

[0055] def load_and_preprocess_data(file_path, category, status_label_col):

[0056] data = pd.read_csv(file_path)

[0057] X = data.drop(status_label_col, axis = 1).values

[0058] y_category = np.full((X.shape[0],), category)

[0059] y_status = data[status_label_col].values

[0060] return X, y_category, y_status

[0061] # Load all data

[0062] X_water, y_water_category, y_water_status = load_and_preprocess_data('water_supply_data.csv', 'water','status')

[0063] X_sewage, y_sewage_category, y_sewage_status = load_and_preprocess_data('sewage_system_data.csv','sewage','status')

[0064] X_gas, y_gas_category, y_gas_status = load_and_preprocess_data('gas_network_data.csv', 'gas','status')

[0065] X_bridge, y_bridge_category, y_bridge_status = load_and_preprocess_data('bridge_network_data.csv', 'bridge','status')

[0066] # Merge all data

[0067] X = np.concatenate([X_water, X_sewage, X_gas, X_bridge], axis = 0)

[0068] y_category = np.concatenate([y_water_category, y_sewage_category, y_gas_category, y_bridge_category], axis = 0)

[0069] y_status = np.concatenate([y_water_status, y_sewage_status, y_gas_status, y_bridge_status], axis = 0)

[0070] # Label Encoding

[0071] category_encoder = LabelEncoder()

[0072] y_category_encoded = category_encoder.fit_transform(y_category)

[0073] y_category_categorical = to_categorical(y_category_encoded)

[0074] status_encoder = LabelEncoder()

[0075] y_status_encoded = status_encoder.fit_transform(y_status)

[0076] y_status_categorical = to_categorical(y_status_encoded)

[0077] # Shape adjustment

[0078] X = X.reshape((X.shape[0], num_time_steps, num_features_per_step))

[0079] # Split data into training set and test set

[0080] X_train, X_test, y_category_train, y_category_test, y_status_train, y_status_test = train_test_split(X, y_category_categorical, y_status_categorical, test_size = 0.2, random_state = 42)

[0081] # Build LSTM model

[0082] input_shape=(num_time_steps, num_features_per_step)

[0083] inputs = Input(shape = input_shape)

[0084] x = LSTM(50)(inputs)

[0085] x = Dropout(0.5)(x)

[0086] # Classification part

[0087] category_output = Dense(y_category_categorical.shape[1], activation='softmax', name='category_output')(x)

[0088] # Determine the status based on the category output

[0089] status_input = Concatenate()([x, category_output])

[0090] status_output = Dense(y_status_categorical.shape[1], activation='softmax', name='status_output')(status_input)

[0091] model = Model(inputs = inputs, outputs = [category_output, status_output])

[0092] model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])

[0093] # View the model architecture

[0094] model.summary()

[0095] # Train the model

[0096] history = model.fit(X_train, [y_category_train, y_status_train], epochs = 20, batch_size = 32, validation_split = 0.2)

[0097] # Model evaluation

[0098] loss, category_loss, status_loss, category_accuracy, status_accuracy = model.evaluate(X_test, [y_category_test, y_status_test])

[0099] print(f'Test Loss: {loss}')

[0100] print(f'Category Accuracy: {category_accuracy}')

[0101] print(f'Status Accuracy: {status_accuracy}')

[0102] # Data prediction

[0103] new_data = np.array([...]).reshape((1, num_time_steps, num_features_per_step))

[0104] category_pred, status_pred = model.predict(new_data)

[0105] predicted_category = category_encoder.inverse_transform([np.argmax(category_pred)])

[0106] predicted_status = status_encoder.inverse_transform([np.argmax(status_pred)])

[0107] It is understandable that the above code is only part of the reference content for building the LSTM model of this method, and it is not all the code involved in the method for monitoring urban lifeline underground pipeline network data provided by the present invention. In actual design, it is necessary to adjust the code. Those skilled in the art should understand that it will not be elaborated here.

[0108] According to a second aspect of the present invention, a monitoring system for urban lifeline underground pipeline network data is provided, including: a water supply pipeline network data acquisition module for acquiring historical and real-time water supply pipeline network data; a drainage pipeline network data acquisition module for acquiring historical and real-time drainage pipeline network data; a gas pipeline network data acquisition module for acquiring historical and real-time gas pipeline network data; a bridge pipeline network data acquisition module for acquiring historical and real-time bridge pipeline network data; a label calibration module for calibrating data labels for the historical water supply pipeline network data, historical drainage pipeline network data, historical gas pipeline network data, and historical bridge pipeline network data, where the data labels include data types and status types, the data types include water supply data type, drainage data type, gas data type, and bridge data type, and the status types include normal, abnormal, and faulty; a deep learning module for training the classification tasks of the historical water supply pipeline network data, historical drainage pipeline network data, historical gas pipeline network data, and historical bridge pipeline network data with calibrated data labels, and classifying the real-time water supply pipeline network data, real-time drainage pipeline network data, real-time gas pipeline network data, and real-time bridge pipeline network data; a data curve generation and comparison module for constructing data curves, and constructing and calling abnormal data curves and faulty data curves stored in the database according to the data curves, data types, and status types of real-time data.

[0109] According to a third aspect of the present invention, an electronic device is provided. The electronic device includes a processor and a memory coupled to the processor; the memory is used to store a computer program, and the processor is used to execute the computer program to implement the following method: S1. Acquire historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data;

[0110] S2. Respectively calibrate data labels for the historical water supply pipeline network data, drainage pipeline network data, gas pipeline network data, and bridge pipeline network data, where the data labels include data types and status types, the data types include water supply data type, drainage data type, gas data type, and bridge data type, and the status types include normal, abnormal, and faulty;

[0111] S3. Construct a water supply anomaly data curve for water supply network data with a data type of water supply data class and a status type of anomaly, a water supply fault data curve for water supply network data with a data type of water supply data class and a status type of fault, a drainage anomaly data curve for drainage network data with a data type of drainage data class and a status type of anomaly, a drainage fault data curve for drainage network data with a data type of drainage data class and a status type of fault, a gas anomaly data curve for gas network data with a data type of gas data class and a status type of anomaly, a gas fault data curve for gas network data with a data type of gas data class and a status type of fault, a bridge anomaly data curve for bridge network data with a data type of bridge data class and a status type of anomaly, and a bridge fault data curve for bridge network data with a data type of bridge data class and a status type of fault;

[0112] S4. Associate the processing scheme for water supply network anomalies with the water supply anomaly data curve, the processing scheme for water supply network faults with the water supply fault data curve, the processing scheme for drainage network anomalies with the drainage anomaly data curve, the processing scheme for drainage network faults with the drainage fault data curve, the processing scheme for gas network anomalies with the gas anomaly data curve, the processing scheme for gas network faults with the gas fault data curve, the processing scheme for bridge network anomalies with the bridge anomaly data curve, and the processing scheme for bridge network faults with the bridge fault data curve;

[0113] S5. Input the historical water supply network data, drainage network data, gas network data, and bridge network data labeled with data tags into a deep learning model for training in a classification task;

[0114] S6. Obtain real-time water supply network data, drainage network data, gas network data, and bridge network data;

[0115] S7. Input the real-time water supply network data, drainage network data, gas network data, and bridge network data into a comprehensive deep learning model to obtain the data type and status type of these real-time data, construct the data curves of these real-time data, and based on the data type and status type of these real-time data, call the anomaly data curve or fault data curve stored in the database, calculate the curve distance between the real-time data curve and the anomaly data curve or fault data curve, and output an anomaly or fault warning message and push the processing scheme associated with the anomaly data curve or fault data curve when the curve distance meets the predicted value.

[0116] A computer-readable storage medium provided according to the fourth aspect of the present invention, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following method is implemented: S1. Obtain historical water supply network data, drainage network data, gas pipeline network data, and bridge network data;

[0117] S2. Respectively perform data label calibration on the historical water supply network data, drainage network data, gas pipeline network data, and bridge network data. The data labels include data types and status types. The data types include water supply data type, drainage data type, gas data type, and bridge data type. The status types include normal, abnormal, and faulty;

[0118] S3. Construct a water supply abnormal data curve for the water supply network data with a data type of water supply data type and a status type of abnormal, construct a water supply fault data curve for the water supply network data with a data type of water supply data type and a status type of faulty, construct a drainage abnormal data curve for the drainage network data with a data type of drainage data type and a status type of abnormal, construct a drainage fault data curve for the drainage network data with a data type of drainage data type and a status type of faulty, construct a gas abnormal data curve for the gas pipeline network data with a data type of gas data type and a status type of abnormal, construct a gas fault data curve for the gas pipeline network data with a data type of gas data type and a status type of faulty, construct a bridge abnormal data curve for the bridge network data with a data type of bridge data type and a status type of abnormal, construct a bridge fault data curve for the bridge network data with a data type of bridge data type and a status type of faulty;

[0119] S4. Associate the processing scheme when the water supply network is abnormal with the water supply abnormal data curve, associate the processing scheme when the water supply network fails with the water supply fault data curve, associate the processing scheme when the drainage network is abnormal with the drainage abnormal data curve, associate the processing scheme when the drainage network fails with the drainage fault data curve, associate the processing scheme when the gas pipeline network is abnormal with the gas abnormal data curve, associate the processing scheme when the gas pipeline network fails with the gas fault data curve, associate the processing scheme when the bridge network is abnormal with the bridge abnormal data curve, and associate the processing scheme when the bridge network fails with the bridge fault data curve;

[0120] S5. Input the historical water supply network data, drainage network data, gas pipeline network data, and bridge network data with calibrated data labels into a deep learning model for training of classification tasks;

[0121] S6. Obtain real-time water supply network data, drainage network data, gas pipeline network data, and bridge network data;

[0122] S7. Input the real-time water supply network data, drainage network data, gas pipeline network data, and bridge pipeline network data into the comprehensive deep learning model to obtain the data types and status types of these real-time data, construct the data curves of these real-time data, and based on the data types and status types of these real-time data, call the abnormal data curves or fault data curves stored in the database, calculate the curve distance between the real-time data curve and the abnormal data curve or fault data curve, and output abnormal or fault warning information and push the processing solutions associated with the abnormal data curve or fault data curve when the curve distance meets the predicted value.

[0123] The beneficial effects of a method for monitoring urban lifeline underground pipeline network data provided by the present invention are as follows: Compared with the prior art, this method integrates the safe operation data of existing infrastructure (including important bridges) of gas, water supply, drainage, and the municipal management department, classifies and deeply learns the data, enables the system to learn to distinguish data, and does not limit the format of subsequent data input. Since the data formats input by each pipeline network may be different, intelligent recognition is required for data classification, so many data interfaces can be extended to connect to different data sources, having strong versatility. When solving the problem of multiple data inputs and recognition, this method also constructs a relationship table of abnormal data curve - processing method and fault data curve - processing method based on historical data, compares the data curves constructed based on real-time data with the abnormal / fault data curves in the database, so as to quickly identify which type of historical abnormality / fault it is, and then call the processing solutions for the abnormality / fault. Each of these processing solutions is of reference significance. Whether it is the processing method or the assistance opinions of relevant units that need to be jointly involved, it can quickly promote the solution of abnormality / fault hidden dangers, realizing systematic and dynamic underground pipeline network safety management and ensuring a long-term mechanism. This method realizes the concepts of unified efficiency, reliable command, and rapid coordination, forming a new model for urban underground pipeline network safety management and active risk prevention and control.

[0124] In the embodiments of the present invention, unless otherwise specifically stated for the models of each device, the models of other devices are not limited, as long as the devices can perform the above functions.

[0125] The above are only some embodiments of the present invention. For those of ordinary skill in the art, without departing from the inventive concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A method for monitoring underground pipe network data based on urban lifeline, characterized in that: S1. Obtain historical water supply network data, drainage network data, gas network data and bridge network data; S2. Data label calibration is performed on the historical water supply network data, drainage network data, gas network data and bridge network data respectively, wherein the data label includes a data type and a status type, wherein the data type includes a water supply data class, a drainage data class, a gas data class and a bridge data class, and the status type includes normal, abnormal and fault; S3, constructing a water supply abnormal data curve for water supply network data whose data type is water supply data class and whose status type is abnormal, constructing a water supply fault data curve for water supply network data whose data type is water supply data class and whose status type is fault, constructing a drainage abnormal data curve for drainage network data whose data type is drainage data class and whose status type is abnormal, constructing a drainage fault data curve for drainage network data whose data type is drainage data class and whose status type is fault, constructing a gas abnormal data curve for gas network data whose data type is gas data class and whose status type is abnormal, constructing a gas fault data curve for gas network data whose data type is gas data class and whose status type is fault, constructing a bridge abnormal data curve for bridge network data whose data type is bridge data class and whose status type is abnormal, and constructing a bridge fault data curve for bridge network data whose data type is bridge data class and whose status type is fault; S4, associating the processing scheme when the water supply network is abnormal with the water supply abnormal data curve, associating the processing scheme when the water supply network fails with the water supply failure data curve, associating the processing scheme when the drainage network is abnormal with the drainage abnormal data curve, associating the processing scheme when the drainage network fails with the drainage failure data curve, associating the processing scheme when the gas network is abnormal with the gas abnormal data curve, associating the processing scheme when the gas network fails with the gas failure data curve, associating the processing scheme when the bridge network is abnormal with the bridge abnormal data curve, associating the processing scheme when the bridge network fails with the bridge failure data curve; S5. Inputting the historical water supply network data, drainage network data, gas network data, and bridge network data with calibrated data labels into a deep learning model for training of classification tasks; S6. Obtain real-time water supply network data, drainage network data, gas network data and bridge network data; S7. Input the real-time water supply network data, drainage network data, gas network data and bridge network data into the deep learning model to obtain the data type and status type of these real-time data, construct the data curves of these real-time data, and based on the data type and status type of these real-time data, call the abnormal data curve or fault data curve stored in the database for comparison, calculate the curve distance between the real-time data curve and the abnormal data curve or the fault data curve, output the abnormal or fault warning information according to the curve distance prediction value, and push the processing plan associated with the abnormal data curve or the fault data curve.

2. A method for monitoring urban lifeline underground pipe network data according to claim 1, characterized in that: The water supply network data includes water supply point pipeline data, water supply point pressure data, water supply point flow data and water supply point flow velocity data; the drainage network data includes drainage point pipeline data, drainage point liquid level data, drainage point flow data and drainage point water quality data; the gas network data includes gas pipeline data, gas pressure regulating box monitoring data and gas well monitoring data; the bridge network data includes bridge point temperature data and bridge point strain data.

3. A method for monitoring urban lifeline underground pipe network data according to claim 2, characterized in that: The water supply network data also includes water supply point location data, the drainage network data also includes drainage point location data, the gas network data also includes gas point location data, and the bridge network data also includes bridge point location data. If the absolute value of the difference between any two position data is less than a preset value, the abnormal data curves of the points corresponding to the two position data are associated, or the fault data curves are associated.

4. A method for monitoring urban lifeline underground pipe network data according to claim 3, characterized in that: When a network point corresponding to a certain network point location data outputs abnormality or fault warning information and a processing solution, the associated corresponding network points also output abnormality or fault warning information and a processing solution.

5. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: A plurality of processing schemes associated with respective abnormal data curves in the same time period are associated, and a plurality of processing schemes associated with respective fault data curves in the same time period are associated.

6. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: In step S3, an abnormal data curve and a fault data curve are constructed in time series. In step S7, a real-time data curve is constructed in time series, and a curve distance between the real-time data curve and the abnormal data curve is calculated based on the DTW algorithm, or a curve distance between the real-time data curve and the fault data curve is calculated based on the DTW algorithm.

7. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: Calculate the curve change rate of the real-time data curve. When the absolute value of the difference between the curve change rate of the real-time data curve and the curve change rate of the abnormal data curve or the fault data curve is less than a preset value, output abnormal or fault warning information and push a processing solution associated with the abnormal data curve or the fault data curve.

8. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: Based on historical water supply network data, drainage network data, gas network data and bridge network data, the status type of the data corresponding to the network data 46 to 90 days before the network point failure is marked as abnormal, and the status type of the data corresponding to the network data 1 to 45 days before the network point failure is marked as failure.

9. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: The method also includes: step S8, inputting real-time water supply network data, drainage network data, gas network data and bridge network data into a time prediction model, performing data prediction based on time series to generate prediction data, and constructing a prediction data curve based on the prediction data. The prediction data curve is used to compare with the abnormal data curve or the fault data curve to calculate the curve distance, and outputting prediction warning information and pushing a processing solution associated with the abnormal data curve or the fault data curve in accordance with the curve distance prediction value.

10. The method for monitoring urban lifeline underground pipe network data according to claim 1 is characterized in that: The deep learning model is an LSTM model, which includes a data type module and a state type classification module.

Images