A pipeline direct drinking water remote monitoring method and system based on an internet of things

Abstract

The application provides a pipeline direct drinking water remote monitoring method and system based on Internet of Things, comprising: constructing a digital graph model of a pipeline network; fusing static topology and dynamic working condition data to quantitatively evaluate the risk of each node; using a multi-objective optimization algorithm to solve an optimal sensor deployment scheme considering risk coverage and cost, and determining the final scheme through iterative verification; and deploying sensors according to the scheme for real-time monitoring. The method also includes a closed-loop feedback mechanism, which automatically triggers re-optimization when the change amount of the pipeline topology or water use mode exceeds the preset threshold. In addition, the application also provides a system for executing the method. The application solves the problem of unscientific sensor distribution and fixed scheme in the prior art, realizes dynamic optimization of monitoring points and maximization of cost-effectiveness, and improves monitoring accuracy.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) application technology, and more specifically, to a remote monitoring technology for building internal water supply systems, particularly an IoT-based remote monitoring method and system for piped drinking water. Background Technology

[0002] With the development of society and the economy and the improvement of living standards, piped drinking water systems are becoming increasingly popular in modern residential buildings, office buildings, commercial complexes, and other buildings. To ensure the safety of users' drinking water, it is usually necessary to establish a remote monitoring system to monitor the water quality and operating status of the pipe network.

[0003] In existing technologies, a common approach to implementing remote monitoring systems involves installing online water quality monitors or sensors at key locations within the water supply network. These locations are typically chosen based on past experience in water management, such as branch points in the main pipeline, outlets of regional booster pump stations, outlets of secondary water supply facilities, or the main water meter at the entrance of large buildings. Each monitor uploads the collected water quality data to a central monitoring platform at preset time intervals via an embedded wireless communication module, such as GPRS or 4G. The monitoring platform is responsible for receiving, storing, and visualizing the data, and includes a threshold alarm mechanism that notifies management personnel when monitoring data exceeds normal limits. This approach, however, only installs sensors on main pipelines or at fixed locations, ignoring the topological characteristics of the pipeline network and the spatial differences within buildings. This static deployment method cannot adapt to varying monitoring needs arising from different floor heights, branch pipeline distributions, or changes in water load. For example, in high-rise buildings, differences in floor height can cause water pressure fluctuations, affecting water quality stability, while complex connections in branch pipelines increase the potential risk of leaks or contamination. In addition, existing methods often lack dynamic assessment of risk factors and are difficult to adjust monitoring strategies based on information such as pipeline material aging or user complaints, which may lead to high-risk areas being overlooked.

[0004] In the field of piped drinking water monitoring, the core technical challenge lies in how to achieve efficient and dynamic sensor deployment within complex pipe networks and building spaces. Determining sensor placement locations is crucial because pipe networks have complex topologies and numerous branch points, making it difficult to cover all critical nodes simply by relying on experience. For example, in a multi-branched pipe system, water quality anomalies at a critical branch point may go undetected due to improper sensor placement. Another related challenge is how to comprehensively consider various dynamic factors during sensor placement, such as the spatiotemporal variations in water load and the degree of pipe aging. Fluctuations in water load across different time periods and floors require sensors to adapt flexibly, while aging pipe materials increase the risk of water contamination. These two issues are interconnected: inappropriate sensor placement can lead to monitoring blind spots, while ignoring dynamic factors may prevent the timely detection of anomalies at high-risk locations.

[0005] Therefore, in real-world business scenarios, scientifically determining the deployment location and number of sensors in complex building piping networks has become a critical issue that urgently needs to be addressed. For example, in a high-rise mixed-use building, the piping system may span multiple floors with dozens of branch points, and the water demand in different areas varies significantly. If the sensor deployment cannot cover key branch points or high-risk areas, such as basements with dense aging pipes or peak water usage areas, water pollution or leakage problems may be missed, thereby affecting residents' health and the stability of the water supply system.

[0006] In summary, the focus is on how to optimize sensor placement in complex building spaces and pipe networks, taking into account topology, floor differences, and dynamic water demand. Simultaneously, it's crucial to address how to dynamically assess risk factors such as pipe aging and contamination probability to adjust monitoring strategies and ensure timely detection of anomalies in high-risk areas. These are all pressing technical challenges in the field of remote monitoring of piped drinking water. Summary of the Invention

[0007] (a) Technical problems to be solved

[0008] The main objective of this invention is to overcome the shortcomings of existing remote monitoring systems for piped drinking water in terms of the scientific nature, dynamic adaptability, cost-effectiveness, and risk prediction capabilities of sensor deployment strategies. This invention provides a remote monitoring method and system for piped drinking water based on the Internet of Things, enabling scientific planning of monitoring points, dynamic adaptation to changes in risks, and improved monitoring accuracy and response speed.

[0009] (II) Technical Solution

[0010] To address the aforementioned technical problems, this invention provides a remote monitoring method for piped drinking water based on the Internet of Things (IoT), characterized by comprising the following steps:

[0011] Obtain the pipeline network design data of the preset building and parse the data to extract the topological connection relationship, geometric attributes and physical attributes of the pipeline; based on the topological connection relationship, geometric attributes and physical attributes, construct a digital graph model of the pipeline network; wherein, the graph model is a weighted graph, the nodes in the graph represent the branch points or terminal points in the pipeline network, and the edges in the graph represent the pipeline segments connecting the nodes;

[0012] Based on the aforementioned digital graph model, static topology risk and dynamic operating condition risk are integrated, and a comprehensive risk score is quantitatively calculated for each node in the graph model.

[0013] Using the comprehensive risk score of the node as input, a multi-objective optimization function is established with the goal of maximizing the monitoring coverage of high-risk nodes and minimizing the total cost of sensor deployment; and a preset global optimization algorithm is used to solve the optimization function to generate a set of candidate monitoring point schemes that include the location and number of sensor deployments.

[0014] The candidate monitoring point schemes are iteratively revised and simulated to determine the final monitoring point deployment scheme.

[0015] According to the final monitoring point deployment plan, water quality sensors are installed at the corresponding physical locations in the pipeline network, and wireless communication links are established. Through the water quality sensors and wireless communication links, water quality parameters at each monitoring point are continuously collected and uploaded to the remote monitoring center to achieve real-time remote monitoring of the pipeline drinking water network.

[0016] The system continuously monitors the topology data and real-time water load data of the pipeline network. When the change in the topology data exceeds a first preset threshold, or the change characteristic value of the real-time water load data exceeds a second preset threshold within a statistical period, the system automatically triggers and re-executes the aforementioned construction, quantification calculation, solution, and determination steps.

[0017] On the other hand, the present invention also provides a remote monitoring system for piped drinking water based on the Internet of Things, characterized in that it includes:

[0018] The data acquisition module is used to acquire pipeline network design data of the pre-designed building, collect water load data and water quality parameters in real time from smart sensing devices deployed in the pipeline network, and acquire historical aging data of the pipeline.

[0019] A data processing server, communicatively connected to the data acquisition module, includes:

[0020] A digital modeling unit is configured to receive the pipeline network design data and construct a digital graphical model of the pipeline network.

[0021] The risk assessment unit is configured to perform a quantitative calculation of a comprehensive risk score for each node in the graph model based on the digital graph model, water load data, and historical aging data.

[0022] The deployment optimization unit is configured to establish and solve a multi-objective optimization function aimed at maximizing monitoring coverage and minimizing deployment cost, and to determine the final monitoring point deployment scheme through iterative correction and simulation verification.

[0023] The monitoring and operation unit is configured to generate deployment instructions based on the final monitoring point deployment plan, receive and process real-time water quality parameters, and execute a closed-loop feedback mechanism. The closed-loop feedback mechanism is configured to trigger the digital modeling unit, risk assessment unit, and deployment optimization unit to perform re-optimization when the amount of change in the topology data of the pipeline network exceeds a first preset threshold, or the change characteristic value of the real-time water load data within a statistical period exceeds a second preset threshold.

[0024] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0025] (III) Technical Effects

[0026] This invention moves the experience-based, back-end sensor deployment stage of traditional remote monitoring systems to the forefront and restructures it into an intelligent monitoring network planning process driven by data, algorithmic decision-making, and capable of closed-loop optimization. First, the complex physical network is precisely abstracted into a computable digital graph model. Then, by creatively integrating multi-dimensional heterogeneous information such as static topology information, dynamic real-time operating conditions, and historical data reflecting physical health status, a precise and quantitative profile of the risk of each potential node in the monitoring network is achieved. Based on this solid data foundation, global optimization algorithms and multiple simulation verification techniques are used to fundamentally transform the sensor deployment problem into a solvable optimization problem, thereby solving the technical problems of poor monitoring performance and lack of dynamic adjustment.

[0027] Compared to the prior art, this invention, by constructing a sophisticated digital graph model and combining it with multi-dimensional risk assessment, can identify key risk nodes that are easily overlooked by traditional empirical methods. This allows for more targeted sensor deployment, effectively reducing monitoring blind spots and improving the accuracy and comprehensiveness of the monitoring network. The closed-loop feedback mechanism included in this invention enables the monitoring network to respond to actual changes in building pipe network structure or water usage patterns, and to iteratively adjust and optimize, ensuring the long-term effectiveness of the monitoring strategy and overcoming the performance degradation problem of traditional solutions over time. Furthermore, by introducing a multi-objective optimization algorithm, this invention can maximize risk coverage within a given cost budget, or minimize costs while meeting preset safety levels, providing decision-makers with an optimal deployment scheme that balances safety and economy. This method also extends risk management from reactive post-event response to proactive pre-event prediction. By analyzing factors such as pipe aging and contamination probability, it achieves forward-looking risk identification and defense, enhancing the safety resilience of the entire water supply system. Finally, the high degree of automation in the entire planning process replaces tedious manual surveys and subjective decisions, providing a more efficient and scientific implementation path for the construction of direct drinking water monitoring systems in large and complex buildings. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall process of a remote monitoring method for piped drinking water based on the Internet of Things according to an embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of the overall architecture of a remote monitoring system for piped drinking water based on the Internet of Things, according to an embodiment of the present invention.

[0030] Figure 3 This is an example diagram used to illustrate the process of constructing a digital graphical model of a building pipeline network in an embodiment of the present invention.

[0031] Figure 4 This is a schematic diagram illustrating the principle of using a genetic algorithm to optimize monitoring points in embodiments of the present invention.

[0032] Figure 5 This is a block diagram illustrating the relationships between various functional subsystems in a monitoring system according to an embodiment of the present invention.

[0033] Figure 6 This is a schematic diagram of hardware interaction of a monitoring system according to an embodiment of the present invention.

[0034] Figure 7 This is a schematic diagram illustrating a specific application scenario according to an embodiment of the present invention. Detailed Implementation

[0035] The following will refer to the appendices in the embodiments of the present invention. Figures 1 to 7 The technical solutions in the embodiments of the present invention will be clearly and completely described herein. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] It should be noted that the terms "connection" and "linking" mentioned in the embodiments of the present invention can refer to a direct physical connection or a logical indirect connection. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.

[0037] Example 1

[0038] This embodiment details the specific implementation steps of a remote monitoring method for piped drinking water based on the Internet of Things (IoT). The overall process is as follows: Figure 1 As shown.

[0039] Step 1: Obtain the pipeline network design data of the preset building and parse the data to extract the topological connection relationship, geometric attributes and physical attributes of the pipeline; based on the topological connection relationship, geometric attributes and physical attributes, construct a digital graph model of the pipeline network; wherein, the graph model is a weighted graph, the nodes in the graph represent the branch points or terminal points in the pipeline network, and the edges in the graph represent the pipeline segments connecting the nodes.

[0040] In one alternative implementation, the goal of this step is to transform the physical world's network of pipes into a computer-processable digital graph model, such as... Figure 3 As shown. First, the Building Information Model (BIM) data of the preset building is obtained, specifically an IFC format file. Using libraries such as ifcopenshell, this file is automatically parsed, traversing entities related to pipes such as IfcPipeSegment, IfcFitting, and IfcFlowTerminal to extract their spatial coordinates, connection relationships, and physical attributes such as material, nominal diameter, and installation year. Then, a directed graph object is created using graph calculation libraries such as NetworkX. In this graph object, each pipe fitting (e.g., tee 303) and water terminal (e.g., terminals 304 and 305) is created as a node, storing its attributes; each pipe segment (e.g., pipe segment 306) is created as a weighted edge connecting two nodes, with the weight information including the pipe segment's length, material, diameter, and installation year.

[0041] Step 2: Based on the digital graph model, integrate static topology risk and dynamic operating condition risk, and perform a quantitative calculation of the comprehensive risk score for each node in the graph model.

[0042] This step calculates a comprehensive risk score for each node on the constructed graph model. This score is a weighted combination of two parts.

[0043] The first part is the static topology risk score R_static(v), which reflects the node's risk level. v The inherent importance of nodes in pipeline structures. This score can be obtained by calculating the degree centrality and betweenness centrality of nodes, followed by normalization and weighted summation. The calculation formula can be expressed as:

[0044] In this context, v represents a node in the graph model; R_static(v) is the static topological risk score of node v; C_D_norm(v) is the normalized degree centrality value of node v; C_B_norm(v) is the normalized betweenness centrality value of node v; α and β are preset, non-negative weight coefficients that sum to 1, used to adjust the importance of degree centrality and betweenness centrality in the score.

[0045] The second part is dynamic operating condition risk assessment. R dynamic ( v This score reflects the risk arising from node v's real-time operational status and physical health. This score can be composed of two factors:

[0046] 1) Water load fluctuation Vwater(v) is an indicator designed to quantify the stability of hydraulic conditions in a pipe segment near a node. Drastic and frequent water load fluctuations can lead to the shedding of deposits (such as rust and biofilm) from the pipe walls, resulting in instantaneous water quality deterioration; simultaneously, frequent water hammer effects increase the risk of fatigue leaks at pipe connections. The calculation of this indicator relies on instantaneous flow data continuously collected and uploaded by smart water meters 602 in the data acquisition module at a small time granularity (e.g., time interval Δt = 5 minutes). For each node v in the graphical model, the system first identifies its directly downstream region in the pipe network topology. Then, all flow measurements of this region over the past statistical period T (e.g., T = 7 days) are extracted to form a time series {q1,q2,...,qk}, where k = T / Δt.

[0047] To quantify the volatility of this sequence, this embodiment calculates its coefficient of variation (CV), a dimensionless value that allows comparison of volatility at different average flow levels. The calculation formula is as follows:

[0048] , where σ_q is the standard deviation of {q1,q2,...,qk}; μ_q is the arithmetic mean of the time series. The larger the value of Vwater(v), the stronger the fluctuation of water load at that node, and the higher the potential risk.

[0049] 2) Historical Aging Index (Aaging(v)): This index quantifies the inherent risks associated with pipe aging, including material corrosion, scaling, and joint aging, all significant sources of secondary water pollution. The calculation of this index integrates the service life of the pipe and its material characteristics. First, based on industry standards and expert experience, an aging coefficient γmaterial is preset for pipes of different materials. This coefficient is a value greater than 0, reflecting the aging resistance of different materials. For example, γmaterial for stainless steel can be set to 0.6, copper pipe to 0.7, PPR (random copolymer polypropylene) pipe to 1.0, while galvanized steel pipes that may exist in some early buildings can be set to 1.8. The higher the value, the easier it is to age. For a node v in the graph model, the system searches for all edges originating from that node (i.e., pipe segments flowing out of that node). For each edge ei, its aging degree can be calculated as follows:

[0050] Where Yearcurrent is the current year; Yearinstall(ei) is the installation year of pipe edge ei; and γmaterial(ei) is the aging coefficient corresponding to the material of that pipe edge. The aging risk of a node is usually determined by the most vulnerable pipe segment it is connected to. Therefore, the historical aging index Aaging(v) of node v is defined as the maximum aging degree among all outgoing edges connected to it: A_aging(v) = max{Aging(e_i)}. The larger the value of Aaging(v), the more serious the aging problem in the downstream pipe connected to the node, and the higher the risk.

[0051] After calculating Vwater(v) and Aaging(v) for all nodes, the values ​​are min-max normalized to obtain water load volatility and historical aging index. These two factors are then normalized and weighted to obtain the dynamic operating condition risk score. R dynamic ( v ).

[0052] Finally, the static and dynamic risk scores are weighted and combined again to obtain the final comprehensive risk score for each node. Rcomp ( v ).

[0053] Step 3: Using the comprehensive risk score of the node as input, establish a multi-objective optimization function with the goal of maximizing the monitoring coverage of high-risk nodes and minimizing the total cost of sensor deployment; and use a preset global optimization algorithm to solve the optimization function to generate a set of candidate monitoring point schemes that include the location and number of sensor deployments.

[0054] After quantifying the risk of all nodes, this step employs a Genetic Algorithm (GA) to solve for the optimal sensor deployment scheme. A Genetic Algorithm is a heuristic search algorithm that simulates the biological evolution process in nature. It is particularly suitable for solving complex, nonlinear, multi-objective combinatorial optimization problems, such as the sensor deployment problem faced here. The algorithm efficiently searches for near-optimal solutions in a vast space of possible solutions by simulating the natural selection mechanism of survival of the fittest.

[0055] The implementation of the genetic algorithm in this embodiment is as follows: Figure 4 As shown. First, a deployment scheme is encoded as a binary string chromosome 401, where a value of 1 indicates deployment and a value of 0 indicates no deployment. Then, an initial population 402 containing multiple chromosomes is randomly generated using a uniform random generation method to ensure that the "1"s and "0"s in the initial population are roughly balanced.

[0056] Next, a fitness function is designed to evaluate the performance of each solution. This function aims to balance the two objectives of risk coverage (O_coverage) and deployment cost (O_cost), and can be expressed as:

[0057] In this context, W_cov and W_cost are the weighting coefficients of the two objectives, for example, W_cov = 0.7 and W_cost = 0.3; B_max is the maximum budget constraint used for cost normalization; P(O_cost) is a penalty function that takes a larger value when O_cost exceeds B_max to reduce the fitness of the corresponding solution. O_coverage is calculated as the ratio of the sum of the risk scores of the nodes covered by the scheme to the total risk score of the entire network. O_cost is the sum of the equipment and installation costs of all sensors in the scheme.

[0058] Where O_coverage is the risk coverage ratio, calculated using the following formula:

[0059] , where si is the value of the i-th position of the chromosome (0 or 1), Rcomp(i) is the comprehensive risk score of node i, and N is the total number of nodes.

[0060] O_cost represents the deployment cost metric, and its calculation formula is:

[0061] Where C_sensor is the unit price of the sensor, and Cinstall(i) is the installation cost of node i.

[0062] P(O_cost) is a penalty function used to handle infeasible solutions that exceed the budget. It is defined as: Where M is a very large positive number (penalty factor). This means that once the cost of a program exceeds the budget, its fitness will be subject to a huge negative penalty, making it almost impossible for it to be selected during the evolutionary process.

[0063] Finally, the population is iteratively evolved through a series of genetic operations such as selection, crossover, and mutation. In each generation, the scheme with high fitness has a greater probability of being retained and producing offspring.

[0064] Genetic Operation 403: Iterative evolution of the population, each generation including the following operations: Selection: This embodiment uses a tournament selection strategy. This strategy randomly selects k individuals (e.g., k=3) from the population each time, and then selects the individual with the highest fitness from these k individuals to enter the next generation. This process is repeated Psize times. Compared to roulette wheel selection, tournament selection better avoids super-individuals dominating the population prematurely, maintaining population diversity. Crossover: After the selection operation, the new population is paired with a certain crossover probability pc (e.g., pc=0.8). For each selected pair of parent chromosomes, uniform crossover is used. Specifically, a random binary template consisting of 0s and 1s, the same length as the chromosome, is generated. Then, based on the value of the corresponding bit in the template, it is determined which parent the offspring chromosome inherits the gene from at that position. For example, if the template bit is 0, offspring one inherits the gene from parent one, and offspring two inherits the gene from parent two; and vice versa. Uniform crossover allows for better combination of gene segments from the parents. Mutation: For all offspring individuals produced by crossover, each gene on their chromosome is traversed with a small mutation probability pm (e.g., pm=0.02). If a gene is selected for mutation, its value is flipped (0 becomes 1, 1 becomes 0). Mutation is crucial for the algorithm to explore new solution spaces and prevent premature convergence. Elitism: To prevent the loss of optimal solutions during evolution, before each generation begins, several individuals with the highest fitness (e.g., 1 to 2) from the previous generation are directly copied into the next generation.

[0065] Repeat the above genetic operations until the preset maximum number of generations is reached, such as 100 generations, or the increase in the optimal fitness value of the population is less than a very small threshold ϵ within 100 consecutive generations. When the algorithm terminates, the chromosome with the highest fitness in history is the final output set of one or more sets (if the fitness is the same) of candidate monitoring point schemes.

[0066] Step 4: Iteratively revise and simulate the proposed set of candidate monitoring point schemes to determine the final monitoring point deployment scheme.

[0067] This step aims to select and confirm the optimal solution from the candidate solutions. It begins with iterative refinement, using a Bayesian Network (BN) model. A Bayesian Network is a representation and inference tool based on a probabilistic graphical model. It represents a set of random variables and their conditional dependencies through a directed acyclic graph, excelling particularly at probabilistic inference under conditions of incomplete information. It is well-suited for incorporating multiple factors to assess uncertain events such as pipeline contamination.

[0068] In this embodiment, a Bayesian network is constructed, where nodes can include pipeline age, pipeline material, water pressure fluctuations, etc., as parent nodes, and "contamination probability" as child nodes. The dependencies between nodes are defined using a conditional probability table (CPT), where probabilities can be derived from domain expert knowledge or historical data statistics. By inputting specific attributes of each node (such as pipe age of 10 years, material PPR, etc.) as evidence into the network, the posterior probability of contamination at that node can be inferred. This probability is used to adjust the overall risk score of the nodes, thereby correcting the ranking of candidate solutions and giving higher monitoring priority to nodes with higher contamination probabilities.

[0069] Next, simulation verification is performed using the Monte Carlo method. This is a computational method that approximates the numerical solution of complex problems based on a large number of random samples. Thousands of simulations are conducted in a digital graph model, with each simulation randomly selecting a high-risk node as a virtual pollution source and simulating the diffusion of pollutants in the pipeline network. The average detection time and success rate of the modified scheme across all simulations are statistically evaluated. If the evaluation indicators do not reach the preset safety threshold, the system automatically identifies slow-responding "weak monitoring paths," increases the sensor deployment weights on the nodes along these paths, and then returns to the optimization solution from the previous step and the verification in this step, until all simulation indicators of the final scheme meet the standards.

[0070] Step 5: According to the final monitoring point deployment plan, install water quality sensors at the corresponding physical locations in the pipeline network and establish wireless communication links; continuously collect water quality parameters at each monitoring point through the water quality sensors and wireless communication links and upload them to the remote monitoring center to achieve real-time remote monitoring of the pipeline drinking water network.

[0071] Once the final plan is determined, the system generates a deployment list containing precise installation locations. Based on this list, the construction team installs integrated water quality sensors at the corresponding physical locations on the building. These sensors incorporate multiple parameters such as turbidity, residual chlorine, TDS, and pH levels. A wireless communication network is then built using low-power wide-area network technologies such as LoRaWAN. After the system is operational, all sensors will collect water quality data at a preset frequency, such as every 15 minutes, and upload it wirelessly to a remote monitoring center. The monitoring center then performs real-time analysis, display, and storage of the data.

[0072] Step 6. Further, this method also includes a closed-loop feedback step, continuously monitoring the topology data and real-time water load data of the pipeline network; when the change in the topology data exceeds the first preset threshold, or the change characteristic value of the real-time water load data exceeds the second preset threshold within a statistical period, the aforementioned construction, quantification calculation, solution and determination steps are automatically triggered and re-executed.

[0073] This step ensures the long-term adaptability of the monitoring system, which continuously monitors two key types of changes that may affect risk distribution. The first type is changes to the pipeline topology. For example, when a building's BIM model is updated, or maintenance records show pipeline modifications, the system calculates the difference between the old and new models—the ratio of the total number of added or removed nodes and edges to the total number of nodes and edges in the original model. If this ratio exceeds a preset threshold, such as 5%, re-optimization is triggered. The second type is changes in water load patterns. The system periodically, such as weekly, calculates statistical characteristics like the standard deviation of water load data for each area and compares it with the values ​​from the previous period. If the difference exceeds a preset threshold, re-optimization is also triggered. Once triggered, the entire process is re-executed with the latest data to generate a new deployment plan or adjustment recommendations that match the current situation.

[0074] It should be noted that the first and second preset thresholds can be adjusted according to the security level requirements and operation and maintenance costs of the actual application scenario. In a preferred, non-limiting embodiment, the first preset threshold can be set to 5%, meaning that when the total number of nodes and edges in the pipeline topology changes by more than 5%, a significant change requiring reassessment is considered to have occurred. The second preset threshold can be set to 15% of the standard deviation of water load data in the current statistical period compared to the standard deviation of the previous statistical period. That is, when the volatility of water usage patterns changes by more than 15%, it is considered that new risks may have arisen, requiring re-optimization. Those skilled in the art can set other reasonable values ​​according to specific circumstances.

[0075] Example 2

[0076] This embodiment provides a specific implementation scheme for a remote monitoring system for piped drinking water based on the Internet of Things (IoT). This system serves as the physical carrier and execution environment for the method described in Embodiment 1. (Refer to...) Figure 2 The system architecture shown is mainly composed of three core components, both physically and logically: a data acquisition module, a data processing server, and terminal devices.

[0077] I. Data Acquisition Module

[0078] The data acquisition module is the data source for the entire system. Its core functions are to acquire the pre-designed building's pipeline network design data, collect real-time water load data and water quality parameters from intelligent sensing devices deployed in the pipeline network, and acquire historical aging data of the pipelines. The intelligent sensing devices include at least smart water meters for collecting water load data and multi-parameter water quality sensors for collecting water quality parameters. In this embodiment, the module is a distributed collection containing multiple sub-units:

[0079] 1. Piping Network Design Data Acquisition Unit: This unit is functionally responsible for acquiring architectural design documents in formats such as BIM or CAD. In actual deployment, this can be a software interface that allows administrators to upload files in IFC or DWG formats; or it can be an automated script that periodically scans a specified project file server to obtain the latest version of the design data.

[0080] 2. Dynamic Operating Condition Data Acquisition Unit: This unit consists of a series of smart water meters 602 deployed in the pipeline network. These smart water meters, such as ultrasonic water meters, are responsible for measuring the instantaneous flow rate of the pipeline in real time to quantify the water load. Each water meter has an embedded wireless communication module, such as a LoRaWAN module, which reports the collected flow data at a preset frequency.

[0081] 3. Water quality parameter acquisition unit, which consists of a series of multi-parameter water quality sensors 601 deployed at the finally determined monitoring points. These sensors are responsible for real-time measurement of multiple water quality indicators such as turbidity, residual chlorine, TDS, and pH value. Similarly, each sensor also reports the collected data at a preset frequency through an embedded wireless communication module.

[0082] 4. Historical Aging Data Acquisition Unit: This unit is responsible for acquiring information on the physical health status of the pipelines. It can be an interface that connects to the Building Management System (BMS) database to automatically read information such as the installation year and maintenance records of the pipelines; or it can be a database module for administrators to manually enter and manage data.

[0083] Each of the above-mentioned acquisition units communicates with the data processing server through the network layer, such as the LoRaWAN network described in Embodiment 1, and continuously transmits the various raw data collected to the brain of the system.

[0084] II. Data Processing Server

[0085] The data processing server is the core of the entire system, communicating with the data acquisition module and responsible for all data processing, intelligent analysis, and decision generation. It is typically deployed in the cloud and consists of the following core functional units, whose internal relationships are as follows: Figure 5 As shown:

[0086] 1. Digital Modeling Unit 501, configured to receive the pipeline network design data and construct a digital graphical model of the pipeline network. When the data acquisition module obtains a new BIM or CAD file, the file is transmitted to this unit. The unit's internal parsing engine automatically extracts the topological, geometric, and physical properties of the pipelines and uses the graph construction engine to generate or update the digital graphical model, which is structured and stored in the graph database.

[0087] 2. Risk Assessment Unit 502 is configured to perform quantitative calculations of a comprehensive risk score for each node in the digital graph model based on the aforementioned digital graph model, water load data, and historical aging data. It obtains the latest graph model from the digital modeling unit and real-time water load data and historical aging data from the data acquisition module. The unit internally incorporates the risk assessment algorithm detailed in Embodiment 1, capable of calculating the static topological risk and dynamic operating condition risk of each node, and ultimately obtaining a weighted comprehensive risk score. The calculation results are appended to the corresponding nodes of the graph model as attributes, forming a dynamically updated risk heatmap.

[0088] 3. Deployment optimization unit 503 is configured to establish and solve a multi-objective optimization function aimed at maximizing monitoring coverage and minimizing deployment costs, and to determine the final monitoring point deployment scheme through iterative correction and simulation verification. This is the core of the system's intelligent decision-making. It receives a graphical model with risk scores generated by the risk assessment unit, as well as constraints such as budget input by the user. Its integrated genetic algorithm engine, Bayesian network inference engine, and Monte Carlo simulation engine work together to fully execute the multi-objective optimization, scheme correction, and simulation verification process described in Example 1, ultimately outputting an economical and reliable optimal sensor deployment scheme.

[0089] 4. Monitoring and Operation Unit 504, configured to generate deployment instructions based on the final monitoring point deployment plan, receive and process real-time water quality parameters, and execute a closed-loop feedback mechanism. Its functions include:

[0090] The instruction generation transforms the deployment optimization unit's output scheme (node ​​ID, coordinates) into a construction deployment checklist that is readable and executable by construction personnel.

[0091] Real-time monitoring: It receives real-time data uploaded by the water quality parameter acquisition unit 24 / 7, stores and analyzes the data, compares it with preset thresholds, and immediately triggers an alarm if any abnormality is detected.

[0092] Closed-loop feedback execution: One of the core responsibilities of this unit is to execute a closed-loop feedback mechanism. When the amount of change in the topology data of the pipeline network exceeds a first preset threshold, or when the change characteristic value of the real-time water load data exceeds a second preset threshold within a statistical period, the digital modeling unit, risk assessment unit, and deployment optimization unit are triggered to perform further optimization.

[0093] In addition, the system in this embodiment may also include terminal devices, which are the medium for the system to interact with the physical world and users, mainly including two types:

[0094] 1. On-site execution terminals, including a multi-parameter water quality sensor 601 and a smart water meter 602 for collecting dynamic operating conditions, which are ultimately physically installed on the pipeline network according to the scheme determined by the deployment optimization unit.

[0095] 2. User Terminal 605: This is the human-computer interaction interface provided to end users (such as property managers or construction workers). In this embodiment, it is a web-based application platform. Through this terminal, users can intuitively view risk heat maps, review and confirm recommended deployment plans, monitor water quality in real time, and manage alarm information. All data seen and operations performed on the user terminal are achieved through real-time API interaction with various functional units on the data processing server.

[0096] Reference Figure 6 User terminal 605 interacts with data processing server 604 via the Internet. This terminal provides, but is not limited to, the following functions:

[0097] Situational Awareness and Visualization: The entire building's piping network is overlaid on an interactive 2D or 3D electronic sand table. Users can intuitively view the risk heat map calculated by the risk assessment unit 502 and rendered in different colors, gaining a clear understanding of the risk distribution of the entire building.

[0098] Intelligent planning and decision support: Displays recommended deployment plans generated by the deployment optimization unit 503 to authorized administrators. For example... Figure 7 As shown, the building visualization model 701 clearly marks the optimal locations for sensor installation, such as 702 and 703, along with key performance indicators such as the expected risk coverage and total cost of the solution, providing strong data support for managers' decision-making.

[0099] Real-time monitoring and historical data tracking: Water quality parameters from all deployed sensors are displayed in real-time using dashboards, graphs, and other formats. Users can view historical data for any time period to perform trend analysis and track abnormal events.

[0100] Alarm Management and Work Order Dispatch: Centrally displays all historical and real-time alarm information, and provides processing functions such as confirmation, silencing, and escalation. The terminal can also interface with the maintenance work order system to automatically create and dispatch repair or verification work orders when an alarm occurs.

[0101] Example 3

[0102] This embodiment aims to provide a more detailed and vivid explanation of the actual application process, effects, and advantages of the methods and systems disclosed in this invention through a specific and representative complex application scenario.

[0103] The application scenario in this embodiment is a newly built super high-rise urban complex with 50 floors above ground and 4 floors underground, and a total construction area of ​​approximately 250,000 square meters. The functional layout of this complex is extremely complex, as detailed below:

[0104] The basement levels B1 to B4 primarily house parking, equipment rooms (including the main booster pump room for the direct drinking water system and purification equipment rooms), and support areas. Levels 1 to 6 comprise a high-end shopping mall with numerous restaurants, boutique retail stores, and a fresh food supermarket. Levels 8 to 50 house Grade A office space and a luxury hotel, including guest rooms, an executive lounge, a fitness center, a spa, and a rooftop restaurant. Refuge and equipment levels are located on levels 7 and 26.

[0105] The complex's piped drinking water system has a complex network structure with over 5,000 nodes and a total pipe length exceeding 30 kilometers. Significant differences exist in water demand, peak water usage periods, pipe materials (e.g., copper pipes may be used in hotel guest rooms, while PPR pipes are used in office areas), and potential water quality risks across different functional areas. This presents a major challenge to traditional sensor deployment methods.

[0106] The application flow of the monitoring method and system of the present invention in this scenario is as follows:

[0107] Phase 1: Initial Deployment Planning

[0108] 1. Digital Modeling: After the completion and acceptance of the complex, the property management company imports the final version of the Building Information Model (BIM) data (IFC format) into the pipeline network design data acquisition unit of the system of this invention, and parses out all pipeline, fitting and terminal information, successfully constructing a digital graph model containing 5280 nodes and 5890 edges, and storing the spatial coordinates and floors of all nodes, as well as the length, material and installation year of all edges into the database.

[0109] 2. Initial Risk Assessment: As this is a newly constructed building, there is no historical water load data available. At this stage, the risk assessment unit primarily relies on static topological risk and preliminary dynamic risk based on design data for evaluation.

[0110] Static topological risk: Calculate the degree centrality and betweenness centrality of all 5280 nodes. Analysis revealed that several large branch nodes connecting the main trunk line of the shopping mall's food court to the branch trunk lines of various shops, as well as branch nodes on the vertical trunk lines connecting the main floors of the hotel, had significantly higher betweenness centrality scores than other areas.

[0111] Preliminary dynamic risk assessment: Based on the water consumption quotas for each functional area during the design phase, the water load volatility of different areas is estimated and assigned values ​​(e.g., the catering area is assigned a high volatility weight, and the office area is assigned a medium volatility weight). Simultaneously, a preliminary aging index is calculated based on the pipe material information in the BIM model (e.g., a small number of galvanized pipes remaining in the old renovation area of ​​the basement).

[0112] Ultimately, the risk assessment unit generated the first comprehensive risk score for the entire building, which was then visualized on the 3D model by the application layer in the form of a risk heatmap. Figure 7As shown in the figure, 701 represents a partial 3D visualization model of the super high-rise complex building, whose interior can be represented by different colors or shaded areas to represent different levels of risk heat maps calculated by the risk assessment unit. 702 represents a sensor location determined and deployed on the vertical supply main road in the hotel room area according to the method of the present invention. 703 represents a sensor location determined and deployed on a key branch node in the high-risk catering service area according to the method of the present invention. 704 to 731, etc. (not all marked in the figure, only for illustration) represent other sensor deployment locations determined by the method of the present invention in other different functional areas within the building (such as equipment rooms, administrative offices, hotel rooms, etc.). The dashed circles in the figure schematically represent the theoretical monitoring coverage of each sensor, and the dotted lines schematically represent the wireless communication network connection between the sensor and the LoRaWAN gateway. Through this figure, it can be intuitively understood that the sensor deployment scheme generated by the present invention has obvious risk-oriented characteristics, that is, the deployment density is higher in high-risk areas (such as catering areas and high-rise areas), while the deployment density is lower in low-risk areas (such as office areas and underground equipment rooms), thus achieving optimal resource allocation.

[0113] 3. Multi-objective optimization and solution generation: The property management company set a total budget of 300,000 RMB for sensor deployment. The deployment optimization unit was activated, using this budget as the cost constraint B_max and the comprehensive risk score as input, and executed a genetic algorithm optimization for 1000 generations. After approximately 30 minutes of computation, the algorithm converged, generating three candidate deployment solutions with the highest fitness.

[0114] 4. Revision and Validation: The deployment optimization unit then revises and validates these three candidate solutions.

[0115] Bayesian Network Correction: The system's built-in Bayesian network model, based on information such as the use of non-PPR pipes in some areas (e.g., the basement) and the higher predicted water pressure fluctuations in the dining area, increases the prior contamination probability of nodes in these areas, thereby fine-tuning the sensor deployment weights for these areas in the candidate solutions.

[0116] Monte Carlo simulation verification: The system conducted 20,000 simulations on the revised optimal solution. Simulation results show that the average fault detection time of this solution is 25 minutes, and the detection success rate within 120 minutes is 99.8%, both of which meet the safety thresholds set by the property management.

[0117] Ultimately, the system determined an optimal deployment scheme consisting of 68 sensors. For example... Figure 7As shown, the locations of these sensors (such as 702 and 703) are precisely marked on the 3D model. It can be seen that their distribution is clearly concentrated in the high-risk catering area and the main vertical supply route of hotel rooms, while also covering key branch points in the office area. The distribution is well-organized and fully complies with the risk-oriented principle.

[0118] 5. Physical Deployment: Based on the deployment list generated by the system, which included precise location codes and installation requirements, the property's engineering team completed the installation and network commissioning of all 68 multi-parameter water quality sensors within two weeks.

[0119] Phase Two: Dynamic Monitoring and Adaptive Optimization During Operation. After the system is put into operation, it enters a phase of continuous dynamic monitoring and adaptive optimization.

[0120] 1. Daily Monitoring and Alarms: The monitoring unit began receiving water quality data from all 68 sensors 24 / 7. In the third month of operation, sensor 703, located in the 5th-floor dining area, suddenly reported an instantaneous turbidity value exceeding the preset threshold at 2:30 PM (a low water usage period). The system immediately sent an alarm to the property control room via the application layer and automatically dispatched a work order to the maintenance personnel's mobile app. The maintenance personnel quickly arrived at the scene and discovered that a restaurant upstream of the sensor had illegally modified its pipes during renovations, causing contaminants to enter the pipe network. Due to the timely alarm, the impact was contained to a very small area, preventing a potentially large-scale water pollution incident.

[0121] 2. Closed-loop feedback and dynamic re-optimization:

[0122] Triggered by changes in water usage patterns: After six months of system operation, the monitoring unit, during periodic data analysis, discovered that due to the move of a large financial company into the office area, the standard deviation of water load during the afternoon tea period (3:00 PM - 4:00 PM) exceeded the system's preset second threshold compared to the average of the previous statistical period (the previous three months). This indicated a significant change in the area's water usage pattern. The system automatically triggered a re-optimization process. The risk assessment unit recalculated the dynamic operating risk of all nodes using real water usage data from the past six months, finding a significant increase in the comprehensive risk score of several branch nodes on the floor where the financial company was located. The deployment optimization unit re-runs the process based on the new risk data, ultimately recommending the addition of two sensors to that floor to adapt to the new risk distribution.

[0123] Triggered by Topology Change: In the second year of system operation, some floors of the hotel underwent renovation, involving changes to the piping in some guest rooms. After the project was completed, the BIM model was updated. Upon detecting the BIM model version change, the monitoring unit automatically triggered re-optimization. The digital modeling unit obtained the new BIM file and calculated the change in topology data (the ratio of the total number of added and removed nodes and edges to the original graph model) to be 6%, exceeding the preset first threshold of 5%. The system then completely re-executed the entire process from modeling to verification. Based on the new pipeline structure, the deployment plan of the original 68 sensors was adjusted, recommending the relocation of 3 sensors with changed locations and the addition of 1 new sensor to ensure that the monitoring network was fully compatible with the renovated pipeline structure.

[0124] Through the complete demonstration of this embodiment, it can be clearly seen that the technical solution provided by the present invention can not only provide a scientific, economical and efficient sensor deployment plan for complex buildings in the initial construction stage, but also adjust dynamically with the entire life cycle of the building, thereby fundamentally improving the long-term operational safety of the piped drinking water system.

[0125] Although this application discloses the preferred embodiment as described above, it is not intended to limit this application. Any person skilled in the art can make many possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application shall be determined by the appended claims.

[0126] Although this application discloses the preferred embodiment as described above, it is not intended to limit this application. Any person skilled in the art can make many possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application shall be determined by the appended claims.

[0127] The foregoing has provided a detailed description of one embodiment of this application, but the description is merely a preferred embodiment and should not be construed as limiting the scope of this application. All equivalent variations and modifications made within the scope of this application should still fall within the patent coverage of this application.

Claims

1. A method for remote monitoring of piped drinking water based on the Internet of Things, characterized in that, Includes the following steps: Obtain the pipeline network design data of the preset building, and parse the data to extract the topological connection relationship, geometric attributes and physical attributes of the pipeline; Based on the aforementioned topological connections, geometric attributes, and physical attributes, a digital graph model of the pipeline network is constructed; wherein, the graph model is a weighted graph, where nodes represent branch points or terminal points in the pipeline network, and edges represent pipeline segments connecting nodes. Based on the aforementioned digital graph model, static topology risk and dynamic operating condition risk are integrated, and a comprehensive risk score is quantitatively calculated for each node in the graph model. Using the comprehensive risk score of the node as input, a multi-objective optimization function is established with the goal of maximizing the monitoring coverage of high-risk nodes and minimizing the total cost of sensor deployment; and a preset global optimization algorithm is used to solve the optimization function to generate a set of candidate monitoring point schemes that include the location and number of sensor deployments. The candidate monitoring point schemes are iteratively revised and simulated to determine the final monitoring point deployment scheme. According to the final monitoring point deployment plan, water quality sensors are installed at the corresponding physical locations in the pipeline network, and wireless communication links are established; through the water quality sensors and wireless communication links, water quality parameters at each monitoring point are continuously collected and uploaded to the remote monitoring center. The topology data and real-time water load data of the pipeline network are continuously monitored. When the change in the topology data exceeds the first preset threshold, or the change characteristic value of the real-time water load data exceeds the second preset threshold within a statistical period, the above-mentioned construction, quantification calculation, solution and determination steps are automatically triggered and re-executed. The steps of iteratively correcting and simulating the proposed set of candidate monitoring point schemes include: The iterative correction includes: constructing a Bayesian network model, inferring the prior contamination probability of a node based on its pipeline attributes, and adjusting the density of monitoring points in the candidate monitoring point scheme according to the prior contamination probability, so as to prioritize the coverage of nodes with high contamination probability. The simulation verification includes: using the Monte Carlo method to simulate various random pollution events in the digital graph model, evaluating the average detection time and detection success rate of the modified monitoring point scheme in the pollution events; if the average detection time or detection success rate does not reach the preset threshold, identifying the weak response path, increasing the sensor deployment weight on the path, returning and repeating the steps of solving the multi-objective optimization function and the iterative correction and simulation verification until the verification result meets the standard.

2. The method according to claim 1, characterized in that, The step of quantitatively calculating the comprehensive risk score for each node in the graph model includes: The static topological risk score of each node is calculated based on the graph theory centrality algorithm, wherein the centrality algorithm includes at least the degree centrality algorithm and the betweenness centrality algorithm. The dynamic operating condition risk score of each node is calculated based on the water load data collected in real time from the pipeline network and the recorded historical aging data. The static topology risk score and the dynamic operating condition risk score are weighted and combined to obtain the comprehensive risk score.

3. The method according to claim 1, characterized in that, The global optimization algorithm is a genetic algorithm; the multi-objective optimization function encodes the candidate monitoring point schemes as chromosomes and calculates their fitness based on the monitoring coverage and total deployment cost of the schemes.

4. The method according to claim 1, characterized in that, The change amount of the topology data refers to the ratio of the total number of newly added or removed nodes and edges to the total number of nodes and edges in the original graph model; the change characteristic value of the real-time water load data refers to the difference between the standard deviation of the water load data in the current statistical period and the standard deviation of the water load data in the previous statistical period.

5. The method according to claim 1, characterized in that, The pipeline network design data is either Building Information Modeling (BIM) data or Computer-Aided Design (CAD) data.

6. A remote monitoring system for piped drinking water based on the Internet of Things, characterized in that, include: The data acquisition module is used to acquire pipeline network design data of the pre-designed building, collect water load data and water quality parameters in real time from smart sensing devices deployed in the pipeline network, and acquire historical aging data of the pipeline. A data processing server, communicatively connected to the data acquisition module, includes: A digital modeling unit is configured to receive the pipeline network design data and construct a digital graphical model of the pipeline network. The risk assessment unit is configured to perform a quantitative calculation of a comprehensive risk score for each node in the graph model based on the digital graph model, water load data, and historical aging data. The deployment optimization unit is configured to establish and solve a multi-objective optimization function aimed at maximizing monitoring coverage and minimizing deployment cost, and to determine the final monitoring point deployment scheme through iterative correction and simulation verification. The monitoring and operation unit is configured to generate deployment instructions based on the final monitoring point deployment plan, receive and process real-time water quality parameters, and execute a closed-loop feedback mechanism. The closed-loop feedback mechanism is configured to trigger the digital modeling unit, risk assessment unit, and deployment optimization unit to perform further optimization when the amount of change in the topology data of the pipeline network exceeds a first preset threshold, or the change characteristic value of real-time water load data within a statistical period exceeds a second preset threshold. The deployment optimization unit is further configured to: A Bayesian network model is constructed, and the prior contamination probability of each node is inferred based on its pipeline attributes. The density of monitoring points in the candidate monitoring point scheme is then adjusted according to this prior contamination probability to prioritize coverage of nodes with high contamination probabilities. The Monte Carlo method is used to simulate various random pollution events in the digital graph model to evaluate the average detection time and detection success rate of the modified monitoring point scheme in the pollution events. If the average detection time or detection success rate does not reach the preset threshold, the weak response path is identified, and the sensor deployment weight is increased on the path. The steps of solving the multi-objective optimization function and the iterative correction and simulation verification are returned and repeated until the verification result meets the standard.

7. The system according to claim 6, characterized in that, The risk assessment unit is further configured as follows: The static topological risk score of each node is calculated using a graph-theoretic centrality algorithm. The dynamic operating condition risk score of each node is calculated based on the water load data and historical aging data. The static topology risk score and the dynamic operating condition risk score are weighted and combined to generate the comprehensive risk score.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

Images