Pressure detection system based on regional Internet of Things water meter data analysis and early warning method thereof

Abstract

The invention discloses a pressure detection system based on regional Internet of Things water meter data analysis and an early warning method thereof, and relates to the field of water meter pressure monitoring, the system comprises a data acquisition layer, a network layer, a data processing layer and an application layer; the data acquisition layer comprises a large number of Internet of Things water meters distributed at all nodes of a regional water supply network, sensors are arranged in the water meters, and pressure data of the positions where the water meters are located are collected in real time while the water consumption is accurately metered. According to the method, the regions can be accurately divided according to the water consumption and pressure characteristics of different regions, a highly personalized pressure model is established for each category, and through cooperative application of a Logistic regression model, K-Means clustering analysis and an association rule mining algorithm, a data relationship is deeply mined, so that the pressure of the regions can be accurately determined. Pressure models of different areas are accurately established, an early warning threshold value is dynamically adjusted, abnormity is pre-judged in advance, and the accuracy and timeliness of pressure detection and early warning are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water meter pressure monitoring, and in particular to a pressure detection system and an early warning method thereof based on regional Internet of Things water meter data analysis. Background Art

[0002] In urban water supply systems, the stability of water pressure is crucial to ensure the normal water use of residents and the safe operation of water supply facilities. With the acceleration of urbanization, urban population continues to grow, urban scale continues to expand, and various industrial and commercial activities are also increasingly prosperous. This has led to the continuous expansion of the coverage of urban water supply systems, and water demand has become diversified and dynamic. On the one hand, the amount of domestic water used by residents has not only continued to rise, but also the requirements for the stability of water quality and water pressure have become higher and higher; on the other hand, the water demand for industrial and commercial water fluctuates greatly due to factors such as production scale and production period. This complex and changeable water demand has brought great challenges to the stable control of water supply pressure.

[0003] At present, traditional water supply pressure detection often relies on a limited number of fixed monitoring points, which are sparsely distributed and have limited coverage. This makes it impossible to fully and real-time grasp the water supply pressure conditions in the entire area. When the pressure is abnormal, it is difficult to locate the problem area in a timely and accurate manner, and it is impossible to provide strong support for rapid response measures. In addition, it is impossible to conduct in-depth analysis of the IoT water meter data distributed in the area, and it is difficult to provide early warning of changes in water pressure in different areas. Summary of the invention

[0004] The purpose of the present invention is to provide a pressure detection system and an early warning method based on regional Internet of Things water meter data analysis, so as to solve the problem that the traditional water supply pressure detection proposed in the above background technology cannot comprehensively and real-time grasp the water supply pressure status in the entire area. When the pressure is abnormal, it is difficult to locate the problem area in time and accurately, and it is impossible to provide strong support for rapid response measures. In addition, it is impossible to conduct in-depth analysis of the Internet of Things water meter data distributed in the area, and it is difficult to provide early warning for changes in water pressure in different areas.

[0005] To achieve the above object, the present invention provides the following technical solutions: a pressure detection system based on regional Internet of Things water meter data analysis, including a data acquisition layer, a network layer, a data processing layer and an application layer;

[0006] The data collection layer includes a large number of IoT water meters distributed at various nodes of the regional water supply network. The water meters have built-in sensors, which can accurately measure water consumption and collect pressure data at the location of the water meters in real time. The collected data is uploaded to the network layer via wireless communication;

[0007] The network layer is used for data transmission and aggregation. It receives data from various IoT water meters and transmits the data to the cloud server via 4G / 5G or a dedicated network. During data transmission, encryption technology is adopted to ensure the security and integrity of the data, preventing data from being stolen or tampered with. At the same time, the network layer also has data caching and error correction functions to cope with possible network failures and ensure that data is not lost;

[0008] The data processing layer is used to, after the cloud server receives the data, first clean and preprocess the data, remove abnormal data, and perform interpolation or complementation processing on missing data. Then, a Logistic regression model is established for the preprocessed data to predict the probability of abnormal water meter pressure and detect abnormal water meter pressure data; perform K-Means clustering analysis on the water meter pressure data in different regions to dynamically adjust the warning threshold; at the same time, use the association rule mining algorithm to discover the potential relationships between different variables, add variables related to the mined association rules to train the Logistic regression model, and when relevant conditions are monitored, anticipate pressure anomalies in advance and issue warnings in a timely manner by combining the probability calculated by the model;

[0009] The application layer is used to provide a visual operation interface for the water supply management department, operation and maintenance personnel, etc. On this interface, users can view the pressure data, pressure change curves, and pressure anomaly alarm information of each monitoring point in the area in real time. At the same time, the system also provides data analysis reports to help users understand the long-term trend of water supply pressure, so as to formulate reasonable water supply scheduling strategies and pipe network maintenance plans. It also supports data interaction and integration with the GIS system to achieve more efficient water supply management.

[0010] Preferably, the data acquisition layer includes a base meter, a sensor module, a microprocessor, a storage module, and a wireless communication module;

[0011] The base meter is used to accurately measure the water flow rate through the water meter and is the basis for obtaining water consumption data;

[0012] The sensor module integrates a pressure sensor and is responsible for real-time sensing of the water supply pressure at the location of the water meter;

[0013] The microprocessor is used to preliminarily process the pressure signal collected by the sensor and the water consumption data of the base meter, and convert the analog signal into a digital signal for subsequent processing and storage;

[0014] The storage module is used to temporarily store the preliminarily processed data. A non-volatile memory is adopted to ensure that the data will not be lost in case of power failure of the water meter, etc., and store the data within a certain time interval;

[0015] The wireless communication module is responsible for uploading the data in the storage module to the network layer, and the NB-IoT protocol is selected to build the wireless communication of the regional water meter monitoring system.

[0016] Preferably, the network layer includes a communication base station, a core network module, and a data encryption and caching module;

[0017] The communication base station establishes a connection with the wireless communication module of the Internet of Things water meter, receives data from the water meter, and accesses it to the core network of the operator. For a private network, independent base station equipment may be deployed and reasonably arranged according to the regional scale and data transmission requirements to ensure that data can be transmitted efficiently and stably;

[0018] The core network module converges, routes, and forwards the data from each base station and transmits the data to the cloud server;

[0019] The data encryption and caching module is used to encrypt the transmitted data using an advanced encryption algorithm during the data transmission process to prevent the data from being stolen or tampered with during transmission. At the same time, distributed caching technology is adopted. When the network experiences a short interruption or congestion, the data can be temporarily stored in the cache and then transmitted after the network returns to normal to ensure that the data is not lost.

[0020] Preferably, the data processing layer includes a cloud server cluster module, a data preprocessing module, a data analysis module, and a database management module;

[0021] The cloud server cluster module consists of multiple servers and realizes parallel processing of large-scale data through a distributed computing framework;

[0022] The data preprocessing module is used to clean and preprocess the original data, identify and remove abnormal data caused by sensor failures, communication interference, etc. For missing data, linear interpolation based on time series is used for prediction and completion;

[0023] The data analysis module is used to perform in-depth clustering analysis on the preprocessed data, establish pressure models for commercial areas, residential areas, and industrial parks, analyze the pressure change rules in different regions and different time periods, accurately calculate the actual water supply pressure in each region and predict the pressure change trend, adjust the warning threshold and strategy according to the characteristics of different regions, anticipate pressure anomalies in advance in combination with the model calculation probability, and issue warnings in a timely manner;

[0024] The database management module is used to store and manage the processed and analyzed data, including historical pressure data, water consumption data, and model parameters.

[0025] Preferably, the application layer includes a visualization interface, a data analysis report generation module, and an integration interface module;

[0026] The visualization interface is constructed using Web technology or mobile application development technology, providing an intuitive and user-friendly operation interface for water supply management departments, operation and maintenance personnel, etc. Through graphical display, users can view the pressure data, pressure change curves, and pressure anomaly alarm information of each monitoring point in the area in real time;

[0027] The data analysis report generation module is used to extract relevant data from the database according to user needs and generate various data analysis reports. The report content includes pressure statistical analysis for different time periods, statistics of pressure anomaly events, and correlation analysis between water supply pressure and water consumption, providing data support for users to formulate reasonable water supply scheduling strategies and pipe network maintenance plans;

[0028] The integrated interface module is used to provide a standardized interface for data interaction and integration with the GIS system, and visually display the relationship between the pipe network layout and pressure distribution using the GIS map, providing more comprehensive information for pipe network planning and maintenance.

[0029] A pressure detection and warning method based on the analysis of regional Internet of Things water meter data includes the following steps:

[0030] S1. Collect the pressure data from the Internet of Things water meters in the area, covering the relevant data of water consumption, pressure change range, and geographical location, and clean and preprocess the collected data;

[0031] S2. Use the Logistic regression method to build a model to predict the probability P of water meter pressure anomaly, where P ∈ [0, 1]. The basic formula of the Logistic regression model is as follows:

[0032] Z = B 0 + B 1 X 1 + B 2 X 3 +…+ B n X n ;

[0033]

[0034] Among them, X 1 , X 2 ,…, X n are the selected independent variables, and B 1 , B 2 ,…, B n are the regression coefficients to be estimated.

[0035] S3. Use K-Means clustering to perform clustering analysis on the pressure data, divide the detection area into commercial areas, residential areas, and industrial parks, and determine the pressure anomaly characteristics of the three different areas;

[0036] S4. Dynamically adjust the warning threshold of the Logistic regression model according to the pressure characteristics of different regions;

[0037] S5. Use the association rule mining algorithm to find the potential association relationships between different variables in the historical data related to water meters. In the variable setting of the Logistic regression model, add variables related to the mined association rules, retrain the Logistic regression model, and adjust the regression coefficients;

[0038] S6. In real-time monitoring, collect the pressure data of water meters in each region and the data of related influencing factors in real time, and input these data into the Logistic regression model whose threshold has been adjusted for different regions. The model calculates the probability P of pressure anomaly;

[0039] S7. When P exceeds the pre-set threshold, the system triggers the warning mechanism, or when it is monitored that the antecedent conditions in the association rules are met, immediately input the relevant data into the Logistic regression model incorporating the association rules to calculate the probability P of pressure anomaly. Even if P has not reached the conventional warning threshold, as long as it is close to the threshold, an early warning is issued, and relevant personnel are notified by means such as text messages, emails or pop-ups on the monitoring platform, informing the specific abnormal area, reasons and scope of influence.

[0040] Preferably, in step S2, the construction of the model using the Logistic regression method includes the following steps:

[0041] S21. Divide the processed water meter pressure data into a training set and a test set according to a ratio, with 70% of the data for training and 30% for testing;

[0042] S22. Use the training set data to solve the regression coefficients by the maximum likelihood estimation method to minimize the prediction error of the model on the training set;

[0043] S23. Use the test set data to evaluate the trained model;

[0044] S24. Optimize the model according to the evaluation results.

[0045] Preferably, in step S3, the clustering analysis of the pressure data using K-Means clustering includes the following steps:

[0046] S31. Set K = 3, randomly select 3 data points as the initial clustering centers, and calculate the Euclidean distances from the data points in each region to these 3 clustering centers. The Euclidean distance calculation formula is:

[0047]

[0048] Where: F and G are two n-dimensional data points, and the dimensions of the data points include water consumption, pressure change range, and geographical location-related parameters. For example, for the data points in two regions, the water consumption in one region is f 1 , the pressure change range is f 2 , and the distance from the water supply source is f 3 ; the corresponding data for the other region is g 1 , g 2 , g 3 The distance between them can be calculated through this formula to judge their similarity degree;

[0049] S32. Assign the data points to the category where the nearest cluster center is located, and then recalculate the cluster center of each category, that is, the mean value of each feature of all data points in this category;

[0050] In the K-Means clustering process, after each re-assignment of data points, recalculate the center of each cluster. Suppose there are m data points T j in the jth cluster C 1 , T 2 , T 3 , and each data point T i =(t i1 , t i2 ,, t in ), then the new center M j =(m j1 , m j2 ,…, m jn ) is calculated by the formula:

[0051]

[0052] where k = 1, 2, …, n, representing the dimension of the data point;

[0053] S33. Continuously repeat this process until the cluster center no longer changes or reaches the preset number of iterations;

[0054] S34. Determine the pressure anomaly characteristics in different regions:

[0055] Commercial area: Analyze the historical pressure data and related influencing factors in the commercial area to clarify its characteristic of large pressure fluctuations;

[0056] Residential area: Analyze the data in the residential area and the pressure change situation during the peak water consumption period;

[0057] Industrial park: Analyze the pressure change range under relatively stable water consumption in the industrial park and the pressure change situation during equipment failures or production plan adjustments.

[0058] Preferably, in step S5, the method of using the association rule mining algorithm to find potential association relationships between different variables in the historical data related to water meters includes the following steps:

[0059] S51. Collect historical data related to water meters covering water meter pressure, water consumption, time, surrounding environmental temperature, and pipeline service life, and perform cleaning and preprocessing on the data;

[0060] S52. Use the Apriori algorithm to set appropriate minimum support and minimum confidence. The minimum support indicates the minimum frequency at which the association rule appears in all data records, and the minimum confidence indicates the probability of the consequent occurring when the antecedent condition is met.

[0061] Preferably, in step S5, the variables added related to the mined association rules include associated water consumption variables, associated time variables, and associated environmental variables;

[0062] Associated water consumption variables:

[0063] Duration of high water consumption: This variable can be set when the association rule mining finds that the water consumption in a certain area continuously higher than the normal level for a certain period of time will affect the pressure;

[0064] Association between total water consumption in the area and pressure: If a rule is mined that when the total water consumption in a certain area reaches a certain threshold, it will have a significant impact on the overall pressure, it can be set as a variable;

[0065] Associated time variables:

[0066] Pressure fluctuation flag during peak water consumption periods: If the association rule indicates that the probability of abnormal pressure fluctuation during certain peak water consumption periods is relatively high, this variable can be set;

[0067] Associated environmental variables:

[0068] Temperature-pressure association variable: According to the mined association rule, when the environmental temperature is higher or lower than a specific value, it will affect the pressure.

[0069] Compared with the prior art, the beneficial effects of the present invention are:

[0070] 1. The present invention has high accuracy and reliability in predicting the probability of abnormal water meter pressure by using the Logistic regression model; through in-depth learning and analysis of a large amount of historical data and various influencing factors, the model can accurately capture the potential trends and characteristics of abnormal pressure, and predict possible abnormal pressure situations in advance, with a high early warning accuracy rate.

[0071] 2. The present invention can accurately divide regions according to the water usage and pressure characteristics of different regions and establish highly personalized pressure models for each category. By synergistically applying the Logistic regression model, K-Means clustering analysis, and association rule mining algorithms, it deeply explores data relationships, accurately establishes pressure models for different regions, dynamically adjusts warning thresholds, and anticipates anomalies in advance, greatly improving the accuracy and timeliness of pressure detection and warning.

[0072] 3. The present invention provides real-time and intuitive pressure information display for management personnel through the visualization interface at the application layer, facilitating a quick understanding of the regional pressure situation. The data analysis report generation module provides comprehensive and in-depth data analysis reports, helping to formulate scientific and reasonable water supply scheduling and pipe network maintenance strategies. Integration with the GIS system further enhances the intuitiveness of information display, provides strong support for pipe network planning and maintenance, and effectively improves the refinement level of water supply management and the scientific nature of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 is a system block diagram of the pressure detection system based on regional Internet of Things water meter data analysis according to the present invention;

[0074] Figure 2 is a flowchart of the pressure detection and warning method based on regional Internet of Things water meter data analysis according to the present invention.

[0075] In the figure: 1. Data acquisition layer; 2. Network layer; 3. Data processing layer; 4. Application layer. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0076] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0077] Refer to Figure 1 and Figure 2 As shown: The pressure detection system based on regional Internet of Things water meter data analysis includes a data acquisition layer 1, a network layer 2, a data processing layer 3, and an application layer 4;

[0078] The data acquisition layer 1 includes a large number of Internet of Things water meters distributed at various nodes of the regional water supply network. The water meters are equipped with built-in sensors, which can not only accurately measure the water consumption but also collect the pressure data at the location of the water meter in real time. The collection frequency is 5 seconds per time. The collected data is uploaded to the network layer through NB-IoT wireless communication. The large number of Internet of Things water meters distributed at various nodes of the regional water supply network ensure the comprehensiveness and representativeness of data collection, and can capture the subtle pressure changes at different locations. Specifically, it includes a base meter, a sensor module, a microprocessor, a storage module, and a wireless communication module;

[0079] The base meter is used to accurately measure the water flow passing through the water meter; the sensor module integrates a pressure sensor to sense the water supply pressure at the location of the water meter in real time; the microprocessor is used to preliminarily process the pressure signal collected by the sensor and the water consumption data of the base meter, and convert the analog signal into a digital signal for subsequent processing and storage; the storage module is used to temporarily store the preliminarily processed data. It uses a non-volatile memory to ensure that the data will not be lost in case of power failure of the water meter and stores the data within a certain time interval; the wireless communication module is responsible for uploading the data in the storage module to the network layer. The NB-IoT protocol is selected to build the wireless communication of the regional water meter monitoring system. The NB-IoT protocol shows strong adaptability in the regional water meter monitoring system. Its low-power consumption feature enables the water meter to operate stably for a long time;

[0080] The network layer 2 is used for data transmission and aggregation. It receives data from each Internet of Things water meter and transmits the data to the cloud server through 4G / 5G or a dedicated network. During the data transmission process, the data is encrypted, and at the same time, data caching and error correction functions are provided. Specifically, it includes a communication base station, a core network module, and a data encryption and caching module;

[0081] The communication base station establishes a connection with the wireless communication module, receives data from the water meter and accesses it to the operator's core network. The dedicated network may deploy independent base station equipment, which is reasonably arranged according to the regional scale and data transmission requirements to ensure that the data can be transmitted efficiently and stably;

[0082] The core network module is used to set up hubs in different regions, aggregate, route, and forward the data from each base station, and aggregate and transmit the data to the cloud server;

[0083] The data encryption and caching module is used to encrypt the transmitted data using advanced encryption algorithms during the data transmission process, and at the same time use distributed caching technology to provide temporary data caching. The distributed caching technology of the data encryption and caching module can dynamically adjust the caching strategy according to network load and data traffic. When the network has a short-term congestion or failure, the data can be quickly stored and retrieved in the cache, ensuring that the data will not be lost or delayed due to network problems. At the same time, the error correction function of this module is powerful, and it can automatically identify and correct a small amount of error data that appears during the transmission process, ensuring the quality of the data received by the cloud server and providing a reliable data source for subsequent data processing.

[0084] The data processing layer 3 is used to clean and preprocess the data after the cloud server receives the data, remove abnormal data and perform interpolation or complementation processing on missing data. Use the preprocessed data to establish a Logistic regression model to predict the probability of abnormal water meter pressure and perform abnormal detection on the water meter pressure data. Perform K-Means clustering analysis on the water meter pressure data in different regions to dynamically adjust the warning threshold. At the same time, use the association rule mining algorithm to discover the potential relationships between different variables, add variables related to the mined association rules to train the Logistic regression model, and when relevant conditions are monitored, anticipate pressure abnormalities in advance and issue warnings in a timely manner based on the probability calculated by the model. Specifically, it includes a cloud server cluster module, a data preprocessing module, a data analysis module, and a database management module.

[0085] The cloud server cluster module consists of multiple servers and realizes parallel processing of large-scale data through a distributed computing framework.

[0086] The data preprocessing module is used to clean and preprocess the original data, identify and remove abnormal data, and use linear interpolation based on time series to predict and complement missing data.

[0087] The data analysis module is used to perform in-depth clustering analysis on the preprocessed data, establish pressure models for commercial areas, residential areas, and industrial parks, analyze the pressure change laws in different regions and different time periods, accurately calculate the actual water supply pressure in each region and predict the pressure change trend, adjust the warning threshold and strategy according to the characteristics of different regions, anticipate pressure abnormalities in advance, and issue warnings in a timely manner based on the probability calculated by the model.

[0088] The database management module is used to store and manage the processed and analyzed data, including historical pressure data, water consumption data, and model parameters.

[0089] The application layer 4 is used to provide a visual operation interface for the water supply management department, operation and maintenance personnel, etc., to view the pressure data, pressure change curves, and pressure anomaly alarm information of each monitoring point in the area in real time, and at the same time support data interaction and integration with the GIS system; specifically, it includes a visual interface, a data analysis report generation module, and an integration interface module;

[0090] The visual interface is used to provide an intuitive and easy-to-use operation interface. Through graphical display, users can view the pressure data, pressure change curves, and pressure anomaly alarm information of each monitoring point in the area in real time;

[0091] The data analysis report generation module is used to extract relevant data from the database according to user needs and generate various data analysis reports. The report content includes pressure statistical analysis for different time periods, pressure anomaly event statistics, and correlation analysis between water supply pressure and water consumption, providing data support for users to formulate reasonable water supply scheduling strategies and pipe network maintenance plans;

[0092] The integration interface module is used to provide a standardized interface for data interaction and integration with the GIS system, and use the GIS map to intuitively display the relationship between the pipe network layout and pressure distribution, providing comprehensive information for pipe network planning and maintenance.

[0093] The pressure detection and early warning method based on the analysis of regional Internet of Things water meter data includes the following steps:

[0094] Step 1: Collect the pressure data from the Internet of Things water meters in the area, covering relevant data such as water consumption, pressure change range, and geographical location, and clean and preprocess the collected data;

[0095] Step 2: Use the Logistic regression method to build a model to predict the probability P of water meter pressure anomaly, P ∈ [0, 1]. The basic formula of the Logistic regression model is as follows:

[0096] Z = B 0 + B 1 X 1 + B 2 X 3 +…+ B n X n ;

[0097]

[0098] where X 1 , X 2 ,…, X n are the selected independent variables, and B 1 , B 2 ,…, B n are the regression coefficients to be estimated;

[0099] Among them, the process of constructing the model using the Logistic regression method is as follows: First, the processed water meter pressure data is divided into a training set and a test set according to a certain proportion, with 70% of the data used for training and 30% for testing; then, using the training set data, the regression coefficients are solved through the maximum likelihood estimation method to minimize the prediction error of the model on the training set; the trained model is evaluated using the test set data; and the model is optimized according to the evaluation results.

[0100] Step 3: Use K-Means clustering to perform clustering analysis on the pressure data, divide the detection area into commercial areas, residential areas, and industrial parks, and determine the pressure anomaly characteristics of the three different areas.

[0101] Among them, using K-Means clustering to perform clustering analysis on the pressure data includes the following steps:

[0102] 31) Set K = 3, randomly select 3 data points as the initial cluster centers, and calculate the Euclidean distance from each data point in the area to these 3 cluster centers. The Euclidean distance calculation formula is:

[0103]

[0104] In the formula: F and G are two n-dimensional data points, and the dimensions of the data points include water consumption, pressure change range, and geographical location-related parameters. For example, for the data points in two areas, the water consumption in one area is f 1 , the pressure change range is f 2 , and the distance from the water supply source is f 3 ; the corresponding data for the other area is g 1 , g 2 , g 3 The distance between them can be calculated through this formula to judge their similarity degree.

[0105] 32) Assign the data points to the category where the nearest cluster center is located, and then recalculate the cluster center of each category, that is, the mean value of each feature of all data points in this category; during the K-Means clustering process, after each re-assignment of data points, recalculate the center of each cluster. Suppose there are m data points T j in the jth cluster C 1 , T 2 , T 3 , and each data point T i = (t i1 , t i2 ,, t in ), then the new center M j of this cluster = (m j1 , m j2 , …, m jn) The calculation formula is as follows:

[0106]

[0107] where k = 1, 2, …, n, representing the dimension of data points;

[0108] 33) Continuously repeat this process until the cluster centers no longer change or the preset number of iterations is reached;

[0109] 34) Determine the pressure anomaly characteristics of different regions:

[0110] Commercial area: Analyze the historical pressure data and related influencing factors in the commercial area to clarify its characteristic of large pressure fluctuations;

[0111] Residential area: Analyze the data in the residential area and the pressure change situation during the peak water consumption period;

[0112] Industrial park: Analyze the pressure change range under relatively stable water consumption in the industrial park and the pressure change situation during equipment failures or production plan adjustments;

[0113] Step Four: Dynamically adjust the warning threshold of the Logistic regression model according to the pressure characteristics of different regions;

[0114] Step Five: Use the association rule mining algorithm to find the potential association relationships between different variables in the historical data related to water meters. In the variable setting of the Logistic regression model, add variables related to the mined association rules, retrain the Logistic regression model, and adjust the regression coefficients;

[0115] Among them, using the association rule mining algorithm to find the potential association relationships between different variables in the historical data related to water meters includes the following steps:

[0116] 51) Collect the historical data related to water meters covering water meter pressure, water consumption, time, ambient temperature of the surrounding environment, and service life of pipelines, and clean and preprocess the data;

[0117] 52) Use the Apriori algorithm to set appropriate minimum support and minimum confidence. The minimum support represents the minimum frequency at which this association rule appears in all data records, and the minimum confidence represents the probability of the consequent occurring when the antecedent condition is satisfied;

[0118] Adding variables related to the mined association rules includes associated water consumption variables, associated time variables, and associated environment variables;

[0119] Associated water variables: High water consumption duration: This variable can be set when association rule mining finds that the water consumption in a certain area continuously exceeds the normal level for a certain period of time and affects the pressure; Association between total water consumption in the area and pressure: If a rule is mined that when the total water consumption in a certain area reaches a certain threshold, it will have a significant impact on the overall pressure, it can be set as a variable;

[0120] Associated time variables: Pressure fluctuation flag during peak water consumption periods: If the association rule indicates that the probability of abnormal pressure fluctuation during some peak water consumption periods is relatively high, this variable can be set;

[0121] Associated environmental variables: Temperature-pressure association variable: According to the mined association rule, when the environmental temperature is higher or lower than a specific value, it will affect the pressure.

[0122] Step 6. In real-time monitoring, collect the pressure data of water meters in each area and the data of related influencing factors in real time, and input these data into the Logistic regression model with adjusted thresholds for different areas. The model calculates the probability of pressure abnormality P;

[0123] Step 7. When P exceeds the pre-set threshold, the system triggers the early warning mechanism, or when it is monitored that the antecedent conditions in the association rule are met, immediately input the relevant data into the Logistic regression model incorporating the association rule to calculate the probability of pressure abnormality P. Even if P has not reached the conventional early warning threshold, as long as it is close to the threshold, an early warning is issued in advance, and relevant personnel are notified by means such as text messages, emails, or pop-ups on the monitoring platform, informing the specific abnormal area, reasons, and scope of influence.

[0124] In the present invention, the probability prediction of abnormal water meter pressure by using the Logistic regression model has high accuracy and reliability; through in-depth learning and analysis of a large amount of historical data and various influencing factors, this model can accurately capture the potential trends and characteristics of abnormal pressure, predict in advance the possible abnormal pressure situations, and the early warning accuracy rate can reach more than 90%; in terms of K-Means clustering analysis, it can accurately divide regions into categories such as commercial areas, residential areas, and industrial parks according to the water use and pressure characteristics of different regions, and establish highly personalized pressure models for each category; these models can deeply analyze the pressure change rules in different regions and different time periods, not only can accurately calculate the actual water supply pressure in each region, but also can accurately predict the pressure change trend in the next few hours or even days, providing sufficient time for the water supply management department to formulate response strategies; at the same time, the potential relationships discovered by the association rule mining algorithm further enhance the prediction ability of the model; for example, when a strong association between a sharp increase in water consumption and a subsequent pressure drop in a specific time period in a certain region is discovered, the model can anticipate abnormal pressure in advance when monitoring the change in water consumption, and combine with the probability calculated by the Logistic regression model to issue an early warning in a timely manner, effectively reducing the risk of failures in the water supply system and ensuring the stability and safety of water supply.

[0125] Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A pressure detection system based on regional Internet of Things water meter data analysis, characterized by: It includes data collection layer (1), network layer (2), data processing layer (3) and application layer (4); The data collection layer (1) includes a large number of IoT water meters distributed at various nodes of the regional water supply network. The water meters have built-in sensors that can accurately measure water consumption and collect pressure data at the location of the water meters in real time. The collected data is uploaded to the network layer via NB-IoT wireless communication; The network layer (2) is used for data transmission and aggregation, receives data from various IoT water meters, and transmits the data to a cloud server via 4G / 5G or a dedicated network, encrypts the data during the data transmission process, and provides data caching and error correction functions; The data processing layer (3) is used to clean and preprocess the data after the cloud server receives the data, remove abnormal data and interpolate or complete the missing data, use the preprocessed data to establish a Logistic regression model to predict the probability of abnormal water meter pressure, and perform abnormality detection on the water meter pressure data; Perform K-Means cluster analysis on water meter pressure data in different areas to dynamically adjust the warning threshold; At the same time, the association rule mining algorithm is used to discover the potential relationship between different variables, and the variables related to the mined association rules are added to train the Logistic regression model. When relevant conditions are monitored, pressure abnormalities are predicted in advance, and early warnings are issued in time based on the probability calculated by the model; The application layer (4) is used to provide a visual operation interface for water supply management departments and operation and maintenance personnel to view the pressure data, pressure change curves and abnormal pressure alarm information of each monitoring point in the area in real time, and supports data interaction and integration with the GIS system.

2. The pressure detection system based on regional Internet of Things water meter data analysis according to claim 1 is characterized in that: The data acquisition layer (1) comprises a base meter, a sensor module, a microprocessor, a storage module and a wireless communication module; The base meter is used to accurately measure the water flow through the water meter; The sensor module integrates a pressure sensor to sense the water supply pressure at the location of the water meter in real time; The microprocessor is used to perform preliminary processing on the pressure signal collected by the sensor and the water consumption data of the base meter, and convert the analog signal into a digital signal for subsequent processing and storage; The storage module is used to temporarily store the data that has been preliminarily processed, uses a non-volatile memory to ensure that the data will not be lost in the event of a power outage of the water meter, and stores data within a certain time interval; The wireless communication module is responsible for uploading the data in the storage module to the network layer, and the NB-IoT protocol is used to establish wireless communication of the regional water meter monitoring system.

3. The pressure detection system based on regional Internet of Things water meter data analysis according to claim 1 is characterized in that: The network layer (2) includes a communication base station, a core network module and a data encryption cache module; The communication base station receives data from the water meter and connects it to the operator's core network by establishing a connection with the wireless communication module. The dedicated network may deploy independent base station equipment and make a reasonable layout according to the regional scale and data transmission requirements to ensure that data can be transmitted efficiently and stably; The core network module is used to set up hubs in different areas to aggregate, route and forward data from various base stations, and aggregate and transmit the data to the cloud server; The data encryption and cache module is used to encrypt the transmitted data using advanced encryption algorithms during the data transmission process, and at the same time use distributed cache technology to provide temporary data cache.

4. The pressure detection system based on regional Internet of Things water meter data analysis according to claim 1 is characterized in that: The data processing layer (3) includes a cloud server cluster module, a data preprocessing module, a data analysis module and a database management module; The cloud server cluster module is composed of multiple servers, and realizes parallel processing of large-scale data through a distributed computing framework; The data preprocessing module is used to clean and preprocess the original data, identify and remove abnormal data, and use linear interpolation based on time series to predict and complete missing data; The data analysis module is used to perform in-depth cluster analysis on the pre-processed data, establish pressure models for commercial areas, residential areas and industrial parks, analyze the pressure change patterns in different areas and at different times, accurately calculate the actual water supply pressure in each area and predict the pressure change trend, adjust the warning threshold and strategy according to the characteristics of different areas, predict pressure anomalies in advance and calculate the probability based on the model, and issue warnings in time; The database management module is used to store and manage processed and analyzed data, including historical pressure data, water usage data and model parameters.

5. The pressure detection system based on regional Internet of Things water meter data analysis according to claim 1 is characterized in that: The application layer (4) includes a visualization interface, a data analysis report generation module and an integrated interface module; The visualization interface is used to provide an intuitive and easy-to-use operation interface. Through graphical display, users can view the pressure data, pressure change curves and pressure abnormality alarm information of each monitoring point in the area in real time; The data analysis report generation module is used to extract relevant data from the database according to user needs and generate various data analysis reports, the report content includes pressure statistics analysis in different time periods, pressure abnormality event statistics, and correlation analysis between water supply pressure and water consumption; The integrated interface module is used to provide a standardized interface for data interaction and integration with the GIS system, and use GIS maps to intuitively display the relationship between pipe network layout and pressure distribution, providing comprehensive information for pipe network planning and maintenance.

6. A pressure detection and early warning method based on regional Internet of Things water meter data analysis, characterized in that: The pressure detection system based on regional Internet of Things water meter data analysis according to any one of claims 1 to 5 comprises the following steps: S1. Collect pressure data from IoT water meters in the region, including water consumption, pressure variation and geographic location data, and clean and pre-process the collected data; S2. Use the Logistic regression method to build a model to predict the probability P of abnormal water meter pressure, P∈[0,1]. The basic formula based on the Logistic regression model is as follows: Z=B0+B1X1+B2X3+…+B n X n ; Among them, X1, X2, …, X n are the selected independent variables, B1, B2, …, B n is the regression coefficient to be estimated; S3. Use K-Means clustering to perform cluster analysis on the pressure data, divide the detection area into commercial area, residential area and industrial park, and determine the pressure anomaly characteristics of the three different areas; S4. Dynamically adjust the warning threshold of the Logistic regression model according to the pressure characteristics of different regions; S5. Use association rule mining algorithm to find potential associations between different variables in the historical data related to water meters. In the variable setting of the Logistic regression model, add variables related to the mined association rules, retrain the Logistic regression model, and adjust the regression coefficient. S6. In real-time monitoring, the pressure data of water meters in each area and the data of related influencing factors are collected in real time, and these data are input into the Logistic regression model with thresholds adjusted for different areas. The model calculates the pressure abnormality probability P; S7. When P exceeds the preset threshold, the system triggers the early warning mechanism, or immediately inputs the relevant data into the Logistic regression model integrated with the association rule to calculate the pressure abnormality probability P when the antecedent condition in the association rule is met. Even if P has not reached the conventional early warning threshold, as long as it approaches the threshold, an early warning will be issued, and relevant personnel will be notified by SMS, email or pop-up window of the monitoring platform to inform the specific abnormal area, cause and scope of impact.

7. The pressure detection and early warning method based on regional Internet of Things water meter data analysis according to claim 6 is characterized by: In step S2, the model construction using the Logistic regression method includes the following steps: S21, dividing the processed water meter pressure data into a training set and a test set according to the ratio, 70% of the data is used for training, and 30% of the data is used for testing; S22. Using the training set data, solve the regression coefficient by the maximum likelihood estimation method to minimize the prediction error of the model on the training set; S23. Use the test set data to evaluate the trained model; S24. Optimize the model according to the evaluation results.

8. The pressure detection and early warning method based on regional Internet of Things water meter data analysis according to claim 6 is characterized by: In step S3, the cluster analysis of the pressure data using K-Means clustering includes the following steps: S31, set K=3, randomly select 3 data points as initial cluster centers, and calculate the Euclidean distance from each regional data point to the 3 cluster centers; S32, assigning the data point to the category where the nearest cluster center is located, and then recalculating the cluster center of each category, that is, the mean of each feature of all data points in the category; S33, repeating this process continuously until the cluster center no longer changes or the preset number of iterations is reached; S34. Determine the abnormal pressure characteristics in different areas: Commercial area: Analyze the historical pressure data and related influencing factors of the commercial area to clarify the characteristics of large pressure fluctuations; Residential areas: Analyze residential area data and the pressure changes during peak water consumption periods; Industrial Park: Analyze the pressure change range when water consumption is relatively stable in industrial parks, as well as the pressure change when equipment fails or production plans are adjusted.

9. The pressure detection and early warning method based on regional Internet of Things water meter data analysis according to claim 6 is characterized by: In step S5, the method of using an association rule mining algorithm to find potential associations between different variables in water meter related historical data includes the following steps: S51, collecting water meter related historical data including water meter pressure, water consumption, time, ambient temperature, and pipe service life, and cleaning and preprocessing the data; S52. Use the Apriori algorithm to set appropriate minimum support and minimum confidence. The minimum support indicates the frequency of the association rule appearing at least once in all data records, and the minimum confidence indicates the probability of the consequent occurring when the antecedent condition is met.

10. The pressure detection and early warning method based on regional Internet of Things water meter data analysis according to claim 6 is characterized by: In step S5, the variables added that are related to the mined association rules include associated water use variables, associated time variables, and associated environment variables.

Images