Water supply capacity evaluation system and method for water supply network based on multi-source data analysis

Abstract

The application discloses a water supply capacity evaluation system and method for a water supply network based on multi-source data analysis, relates to the technical field of water supply network monitoring and fault diagnosis, and comprises the following steps: collecting historical and real-time time series data of each monitoring node; determining a historical dynamic normal fluctuation interval and calculating a real-time abnormal state; analyzing the spatial variation gradient of the abnormality between adjacent monitoring nodes and the transmission speed along the network topology; constructing a dynamic abnormality correlation network; and finally locating a fault core area according to the abnormal state, abnormal time sequence and correlation strength attenuation gradient of the node. The system comprises a historical data processing module, a real-time abnormality analysis module, a dynamic network construction module and a fault core positioning module, and is used for executing the above method. The application realizes dynamic correlation analysis of the abnormal state of the network and accurate positioning of the fault source.

Description

Technical Field

[0001] This invention relates to the field of water supply network monitoring and fault diagnosis technology, specifically a water supply network water supply capacity assessment system and method based on multi-source data analysis. Background Technology

[0002] Water supply networks are critical infrastructure for urban operations, and accurate assessment of their water supply capacity is essential for ensuring water supply security. Traditional assessment methods often rely on fixed thresholds for single indicators such as pressure or flow rate for alarms, which has significant limitations: static thresholds are difficult to adapt to dynamic load changes in the network, easily leading to false alarms and missed alarms; and they only focus on whether data at a single point exceeds limits, failing to analyze the propagation relationship of anomalies in the network space, making it difficult to effectively distinguish between the source of the anomaly and its secondary effects, resulting in insufficient fault location capabilities.

[0003] With the development of monitoring technology, multi-source time-series data such as pressure, flow, and vibration have enabled more refined condition assessments. However, existing methods still have shortcomings in multi-source information fusion and dynamic anomaly propagation analysis, failing to effectively construct real-time correlation models of abnormal states within the pipeline network topology. Therefore, a new method is needed that can dynamically model normal fluctuations, comprehensively quantify node anomalies, and accurately locate the core fault area based on the spatiotemporal propagation characteristics of anomalies, in order to achieve a more intelligent and accurate assessment of the pipeline network's water supply capacity. Summary of the Invention

[0004] The purpose of this invention is to provide a system and method for assessing the water supply capacity of a water supply network based on multi-source data analysis, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system and method for assessing the water supply capacity of a water supply network based on multi-source data analysis. The methods include: Step S1: Collect historical time-series data of each monitoring node in the water supply network under normal operation without faults. The historical time-series data includes historical pressure time-series data, historical flow time-series data and historical vibration time-series data. Based on the historical time-series data, obtain the historical dynamic normal fluctuation range of each monitoring node in each cycle through statistical analysis. Step S2: Collect real-time time-series data of the water supply network at each monitoring node. The real-time time-series data includes real-time pressure time-series data, real-time flow time-series data, and real-time vibration time-series data. Compare the real-time time-series data with the comprehensive historical normal fluctuation range of each monitoring node obtained in Step S1. Calculate the deviation of the real-time data of each monitoring node from its historical normal fluctuation range to generate real-time abnormal state values ​​for each monitoring node. Calculate the difference between the abnormal state values ​​of any two adjacent monitoring nodes as the spatial gradient. Analyze the chronological order of the abnormal state values ​​at different monitoring nodes and, combined with the network topology distance, calculate the transmission speed along the network topology. Step S3: Treat each monitoring node as a dynamic monitoring node, and the state of each dynamic monitoring node is represented by the real-time abnormal state value of each monitoring node generated in Step S2; calculate the correlation strength between the abnormal states of adjacent dynamic monitoring nodes based on the spatial change gradient obtained in Step S2; dynamically adjust the correlation strength based on the transmission speed obtained in Step S2 to obtain the adjusted correlation strength; construct a dynamic abnormal correlation network based on all dynamic monitoring nodes and their adjusted correlation strength, where nodes represent the real-time degree and location of the abnormality, and the edges between nodes represent the real-time strength and direction of the abnormality propagation. Step S4: In the dynamic anomaly association network constructed in step S3, select dynamic monitoring nodes whose real-time anomaly values ​​are higher than a set threshold; for each selected dynamic monitoring node, check whether the occurrence time of the anomaly value of the dynamic monitoring node is earlier than the occurrence time of the anomaly value of its adjacent downstream node, and whether the association strength of the edge connected to the dynamic monitoring node shows a gradient distribution that significantly decreases from the dynamic monitoring node outward; determine the dynamic monitoring node that meets the above conditions as the core fault area.

[0006] Furthermore, step S1 includes: Step S1-1: The water supply network contains k monitoring nodes, forming a monitoring node set S={S1,S2,…,S…} i ,...,S k}, where S1, S2, ..., S i ,...,S k Let S represent the 1st monitoring node, the 2nd monitoring node, ..., the i-th monitoring node, ..., the k-th monitoring node, respectively; collect historical time-series data of each monitoring node in the monitoring node set S under normal operating conditions without faults over a continuous M natural days; for the i-th monitoring node S in the monitoring node set S... i From the i-th monitoring node S i Extract the historical pressure time series data P of the i-th monitoring node from the historical time series data respectively. hist (i) Historical traffic time series data F of the i-th monitoring nodehist (i) and the historical vibration time series data V of the i-th monitoring node hist (i); Step S1-2: Calculate the S of the i-th monitoring node i Historical pressure time series data P hist (i) The average value μ of historical pressure time series data within the M natural day time period P (i) Standard deviation σ of historical pressure time series data P (i) and determine the normal fluctuation range of historical pressure dynamics based on the 3σ criterion: I P (i)=[μ P (i)-3σ P (i),μ P (i)+3σ P (i)]; Step S1-3: Calculate the S of the i-th monitoring node i Historical traffic time series data F hist (i) The average value μ of historical flow time series data within the M natural day time period F (i) Standard deviation σ of historical traffic time series data F (i) and determine the normal fluctuation range of historical flow based on the 3σ criterion: I F (i)=[μ F (i)-3σ F (i),μ F (i)+3σ F (i)]; Step S1-4: Calculate the S of the i-th monitoring node i Historical vibration time series data V hist (i) The average value μ of historical vibration time series data within the M natural day time period V (i) Standard deviation σ of historical vibration time series data V (i) and determine the normal fluctuation range of the vibration history dynamics based on the 3σ criterion: I V (i)=[μ V (i)-3σ V (i),μ V (i)+3σ V (i)].

[0007] Furthermore, step S2 includes: Step S2-1: Collect the data from the i-th monitoring node S in the set of monitoring nodes S. i The real-time time series data is obtained from the i-th monitoring node S. i Extract the i-th monitoring node S from the real-time time series data respectivelyi Real-time pressure time series data P real (i) The i-th monitoring node S i Real-time traffic time series data F real (i) and the i-th monitoring node S i Real-time vibration time series data V real (i); for the i-th monitoring node S i Calculate the degree of deviation of the real-time time series data from the historical dynamic normal fluctuation range obtained in step S1: real-time pressure deviation δ P (i): When P real (i)∈[μ P (i)-3σ P (i),μ P (i)+3σ P When (i)], δ P (i)=0; otherwise, δ P (i)=min{|P real (i)-(μ P (i)+3σ P (i))|,|P real (i)-(μ P (i)-3σ P (i))|} / (6σ P (i)); Real-time deviation of traffic δ F (i): When F real (i)∈[μ F (i)-3σ F (i),μ F (i)+3σ F When (i)], δ F (i)=0; otherwise, δ F (i)=min{|F real (i)-(μ F (i)+3σ F (i))|,|F real (i)-(μ F (i)-3σ F (i))|} / (6σ F (i)); Real-time deviation of vibration δ V (i): When V real (i)∈[μ V (i)-3σ V (i),μV (i)+3σ V When (i)], δ V (i)=0; otherwise, δ V (i)=min{|V real (i)-(μ V (i)+3σ V (i))|,|V real (i)-(μ V (i)-3σ V (i))|} / (6σ V (i)); Step S2-2: For the i-th monitoring node S i According to the real-time deviation δ of the pressure P (i) The real-time deviation of the traffic flow δ F (i) and the real-time deviation of the vibration δ V (i) The real-time abnormal state value A(i) = w is calculated. P ×δ P (i)+w F ×δ F (i)+w V ×δ V (i), where w P w F w V These are preset weighting coefficients for real-time pressure deviation, real-time flow deviation, and real-time vibration deviation; for any two adjacent monitoring nodes S i and S j Calculate the S of any two adjacent monitoring nodes. i and S j The spatial gradient G(i,j): G(i,j) = |A(i) - A(j)|; Step S2-3: For the i-th monitoring node S i Record the i-th monitoring node S i The abnormal state value A(i) exceeds the preset threshold A for the first time. th The data collection time is taken as the i-th monitoring node S. i The time T of the abnormal occurrence i For any two adjacent monitoring nodes S i and S j When any two adjacent monitoring nodes S i and S j Anomalies occurred in all cases, and the times of occurrence of the anomalies were T. i With T j And T i ≠T jAt that time, according to the pipeline topology from S i To S j The pipe length L(i,j) is given by the formula U(i,j)=L(i,j) / |T j -T i Calculate the propagation speed U(i,j) of the abnormal signal between two points.

[0008] Furthermore, step S3 includes: Step S3-1: Define each monitoring node in the monitoring node set S as a dynamic monitoring node, and the state value of each monitoring node is the real-time abnormal state value A(i) generated in step S2; for any two adjacent dynamic monitoring nodes S i and S j Based on the spatial change gradient G(i,j) obtained in step S2, calculate the dynamic monitoring nodes S for any two adjacent nodes. i and S j Initial correlation strength R o (i,j)=1 / (1+G(i,j)); Step S3-2: For any two adjacent dynamic monitoring nodes S i and S j Based on the abnormal signal propagation speed U(i,j) obtained in step S2, calculate the adjusted correlation strength R(i,j): if U(i,j) exists, then R(i,j) = R o (i,j)×(U(i,j) / U o If U(i,j) does not exist, then R(i,j) = R o (i,j); among them, U o The preset reference transmission speed; Step S3-3: Based on all dynamic monitoring nodes and the adjusted association strength R(i,j), construct a dynamic anomaly association network; in the dynamic anomaly association network, for any two adjacent dynamic monitoring nodes S i and S j If any two adjacent dynamic monitoring nodes S i and S j The time of occurrence of the anomaly satisfies T i <T j Then establish a path by S i Pointing to S j If a directed edge is formed, the value of R(i,j) is assigned to that edge as its weight; otherwise, a directed edge is formed by S. j Pointing to S i The directed edges are assigned the value of R(i,j) as the weight; the nodes represent the real-time degree and location of the anomaly, the weight of the edge represents the real-time intensity of the anomaly propagation, and the direction of the edge represents the temporal direction of the anomaly propagation.

[0009] Furthermore, step S4 includes: Step S4-1: From all dynamic monitoring nodes in the dynamic anomaly association network constructed in step S3, select those nodes whose real-time anomaly state value A(i) is greater than a preset threshold A. th The dynamic monitoring nodes constitute the candidate monitoring node set C={C1,C2,…,C…} m ,...,C n}, where C1, C2, ..., C m ,...,C n Let represent the 1st candidate monitoring node, the 2nd candidate monitoring node, ..., the mth candidate monitoring node, ..., the nth candidate monitoring node, respectively; Step S4-2: For the m-th candidate monitoring node C in the candidate monitoring node set C m Perform timing condition checks: Obtain the m-th candidate monitoring node C m All downstream adjacent monitoring nodes in the dynamic anomaly association network constitute the downstream adjacent monitoring node set D. m ={D m_1 D m_2 ,…,D m_q}, where D m_1 D m_2 ,…,D m_p, …,D m_q Let C represent the first downstream neighboring monitoring node of the m-th candidate monitoring node, the second downstream neighboring monitoring node of the m-th candidate monitoring node, ..., the q-th downstream neighboring monitoring node of the m-th candidate monitoring node; check the m-th candidate monitoring node C. m The time of occurrence of the anomaly T(C) m Is it earlier than its downstream adjacent monitoring node set D? m The time of an anomaly occurrence for each node is recorded; if the condition is met, the m-th candidate monitoring node C is determined. m Check by timing conditions; Step S4-3: For the m-th candidate monitoring node C that passed the timing condition check in step S4-2... m Perform correlation strength gradient condition check: obtain the m-th candidate monitoring node C m The set of monitoring nodes adjacent to its downstream side D m The weights of the edges between adjacent downstream monitoring nodes are the adjusted association strengths R(i,j) from step S3-2, where i is the m-th candidate monitoring node C. m Let j be the index of the downstream adjacent monitoring node in the monitoring node set S; calculate the standard deviation σ of the adjusted correlation strength R(i,j). R (C mIf σ R (C m The gradient decay threshold σ is less than the preset threshold value. th Then the m-th candidate monitoring node C is determined. m The correlation strength between the m-th candidate monitoring node C and its downstream adjacent monitoring nodes is presented by the correlation strength between the m-th candidate monitoring node C and its downstream adjacent monitoring nodes. m The gradient distribution that significantly attenuates towards adjacent downstream monitoring nodes is examined using the correlation intensity gradient condition. Step S4-4: Candidate monitoring nodes that pass both the timing condition check in step S4-2 and the correlation intensity gradient condition check in step S4-3 are identified as fault core region nodes; all nodes identified as fault core region nodes constitute the fault core region F.

[0010] The system includes: a historical data processing module, a real-time anomaly analysis module, a dynamic network construction module, and a fault core location module; The historical data processing module is used to determine the historical dynamic normal fluctuation range based on the historical time-series data of the monitoring nodes; The real-time anomaly analysis module is used to calculate the real-time anomaly state based on real-time time series data and the historical dynamic normal fluctuation range, and to analyze the spatial change gradient and transmission speed of the anomaly. The dynamic network construction module is used to construct a dynamic anomaly association network based on the real-time anomaly state, spatial change gradient, and transmission speed. The fault core location module is used to locate the fault core area in the dynamic anomaly association network.

[0011] The historical data processing module includes: The historical data acquisition unit is used to collect historical pressure time-series data, historical flow time-series data, and historical vibration time-series data. The historical interval calculation unit is used to determine the historical dynamic normal fluctuation range of pressure, the historical dynamic normal fluctuation range of flow, and the historical dynamic normal fluctuation range of vibration.

[0012] The real-time anomaly analysis module includes: The real-time deviation calculation unit is used to calculate the degree of deviation of real-time data from the historical dynamic normal fluctuation range. An abnormal state value calculation unit is used to calculate a real-time abnormal state value based on the degree of deviation. The gradient and velocity calculation unit is used to calculate the spatial gradient and anomaly propagation velocity.

[0013] The dynamic network building blocks include: The initial association strength calculation unit is used to calculate the initial association strength between adjacent monitoring nodes; A correlation strength adjustment unit is used to adjust the initial correlation strength according to the transmission speed; The network construction unit is used to construct the dynamic anomaly association network based on the adjusted association strength and the anomaly occurrence sequence.

[0014] The core fault location module includes: The candidate monitoring node filtering unit is used to filter nodes whose abnormal state values ​​are higher than the threshold. The timing condition check unit is used to check the timing sequence of anomalies. The gradient condition check unit is used to check the gradient decay distribution of the correlation strength. The fault determination unit is used to determine the nodes in the core fault area.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention replaces the traditional fixed threshold by establishing a dynamic normal fluctuation range based on historical time-series data, enabling the judgment criteria to adapt to changes in pipeline load and reducing false alarms. Simultaneously, the method integrates the deviation levels of multi-source real-time data such as pressure, flow, and vibration to generate a comprehensive real-time abnormal state value, achieving refined and comprehensive quantification of node anomalies and avoiding reliance on misjudgments based on a single indicator.

[0016] 2. This invention quantifies the intensity and speed of anomaly propagation in a pipeline network by calculating the spatial variation gradient between adjacent monitoring nodes and the anomaly propagation speed based on the time difference of occurrence. The dynamic anomaly association network constructed in this way uses nodes and directed edges to characterize the intensity, location, propagation direction, and association strength of anomalies in real time, thus clearly presenting the spatiotemporal propagation pattern of anomalies.

[0017] 3. This invention is based on a dynamic anomaly correlation network and uses two conditions for determination: "the anomaly occurs earlier than the downstream node" and "the correlation strength significantly decreases outwards." This method can effectively distinguish between the anomaly source and the area affected by propagation, thereby automatically and accurately identifying the core fault area from the network, providing a clear target for targeted maintenance. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the method for assessing the water supply capacity of a water supply network based on multi-source data analysis, as described in this invention. Figure 2 This is a schematic diagram of the water supply capacity assessment system for water supply networks based on multi-source data analysis, as described in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example: Figures 1-2 As shown, this invention provides a technical solution: a system and method for assessing the water supply capacity of a water supply network based on multi-source data analysis. The methods include: Step S1: Collect historical time-series data of each monitoring node in the water supply network under normal operation without faults. The historical multi-source time-series data includes historical pressure time-series data, historical flow time-series data and historical vibration time-series data. Based on the historical multi-source time-series data, obtain the historical dynamic normal fluctuation range of each monitoring node in each cycle through statistical analysis. Step S1-1: The water supply network contains 3 monitoring nodes, forming a monitoring node set S={S1,S2,S3}, where S1, S2, and S3 represent the 1st, 2nd, and 3rd monitoring nodes, respectively. Along the water flow direction, the topological relationship is S1 to S2 to S3. Historical time-series data of each monitoring node in the monitoring node set S is collected over a continuous period of M=30 natural days, confirming that it is in a fault-free and normal operating state. The data collection frequency is once per hour, with a total of 720 data points for each node and each indicator. For the first monitoring node S1 in the set of monitoring nodes S, extract the following from its historical time series data: Historical stress time series data, denoted as P hist (1) = {0.301, 0.299, 0.305, ..., 0.302} (unit: MPa), a total of 720 pressure data points; Historical traffic time-series data, denoted as F hist (1) = {101.2, 99.8, 102.1, ..., 100.5} (Unit: m) 3 / h), a total of 720 traffic data points; Historical vibration time series data, denoted as V hist (1) = {0.0102, 0.0098, 0.0105, ..., 0.0099} (unit: g), a total of 720 vibration data points; For the second monitoring node S2 in the monitoring node set S, extract the following from its historical time series data: Historical stress time series data, denoted as P hist(2) = {0.281, 0.279, 0.285, ..., 0.28} (unit: MPa), a total of 720 pressure data points; Historical traffic time-series data, denoted as F hist (2) = {99.0, 97.5, 100.2, ..., 98.3} (Unit: m) 3 / h), a total of 720 traffic data points; Historical vibration time series data, denoted as V hist (2) = {0.0092, 0.0088, 0.0095, ..., 0.0089} (unit: g), a total of 720 vibration data points; For the third monitoring node S3 in the monitoring node set S, extract the following from its historical time series data: Historical stress time series data, denoted as P hist (3) = {0.261, 0.259, 0.265, ..., 0.26} (unit: MPa), a total of 720 pressure data points; Historical traffic time-series data, denoted as F hist (3) = {97.1, 95.6, 98.3, ..., 96.4} (Unit: m) 3 / h), a total of 720 traffic data points; Historical vibration time series data, denoted as V hist (3) = {0.0082, 0.0078, 0.0085, ..., 0.0079} (unit: g), a total of 720 vibration data points; Step S1-2: Process the historical pressure time-series data P of the first monitoring node S1. hist (1) Calculate the average value μ of its historical pressure time series data over a 30-day period. P (1) Standard deviation σ of historical pressure time series data P (1); By calculating 720 pressure data points, we obtained: μ P (1) = 0.3 MPa; σ P (1) = 0.01 MPa; Based on the 3σ criterion, its historical dynamic normal fluctuation range of pressure is determined: I P (1)=[μ P (1)-3σ P (1),μ P (1)+3σ P (1)]=[0.3-3×0.01,0.3+3×0.01]=[0.27,0.33]MPa; Step S1-3: Analyze the historical traffic time-series data F of the first monitoring node S1. hist(1) Calculate the average value μ of its historical flow time series data over a 30-day period. F (1) Compared with historical traffic time series data standard σ F (1); By calculating 720 flow data points, we obtain: μ F (1) = 100m 3 / h;σ F (1) = 5m 3 / h; Based on the 3σ criterion, its historical dynamic normal fluctuation range of flow rate is determined: I F (1)=[μ F (1)-3σ F (1),μ F (1)+3σ F (1)]=[100-3×5,100+3×5]=[85,115]m 3 / h; Steps S1-4: Process the historical vibration time series data V of the first monitoring node S1. hist (1) Calculate the average value μ of its historical vibration time series data over a 30-day period. V (1) Standard deviation σ of historical vibration time series data V (1); By calculating 720 vibration data points, we obtained: μ V (1) = 0.010 g; σ V (1) = 0.001g; Based on the 3σ criterion, its vibration history dynamic normal fluctuation range is determined: I V (1)=[μ V (1)-3σ V (1),μ V (1)+3σ V (1)]=[0.010-3×0.001,0.010+3×0.001]=[0.007,0.013]g; Similarly, calculate the intervals of each indicator for monitoring nodes S2 and S3: Calculations are performed for the second monitoring node S2: Historical pressure time series data P of the second monitoring node S2 hist (2): By calculating P hist (2) 720 points, to obtain μ P (2) = 0.28 MPa, σ P (2) = 0.01 MPa, thus obtaining I P (2) = [0.25, 0.31] MPa; Historical traffic time series data F of the second monitoring node S2 hist (2): By calculating Fhist (2) 720 points, to obtain μ F (2) = 98m 3 / h, σ F (2) = 4m 3 / h, get I F (2) = [86, 110]m 3 / h; Historical vibration time series data V of the second monitoring node S2 hist (2): By calculating V hist (2) 720 points, to obtain μ V (2) = 0.009g, σ V (2) = 0.001g, yielding I V (2) = [0.006, 0.012] g; Calculations are performed for the third monitoring node S3: Historical pressure time series data P from the third monitoring node S3 hist (3): By calculating P hist (3) 730 points, to obtain μ P (3) = 0.26 MPa, σ P (3) = 0.01 MPa, thus obtaining I P (3) = [0.23, 0.29] MPa; Historical traffic time series data F of the third monitoring node S3 hist (3): By calculating F hist (3) 730 points, to obtain μ F (3) = 96m 3 / h, σ F (3) = 4m 3 / h, get I F (3) = [84, 108]m 3 / h; Historical vibration time series data V of the third monitoring node S3 hist (3): By calculating V hist (3) 730 points, to obtain μ V (3) = 0.008g, σ V (3) = 0.001g, yielding I V (3) = [0.005, 0.011]g.

[0021] Step S2: Collect real-time time-series data of the water supply network at each monitoring node. The real-time time-series data includes real-time pressure time-series data, real-time flow time-series data, and real-time vibration time-series data. Compare the real-time multi-source time-series data with the comprehensive historical normal fluctuation range of each monitoring node obtained in Step S1. Calculate the deviation of the real-time data of each monitoring node from its historical normal fluctuation range to generate real-time abnormal state values ​​for each monitoring node. Calculate the difference between the abnormal state values ​​of any two adjacent monitoring nodes as the spatial gradient. Analyze the chronological order of the abnormal state values ​​at different monitoring nodes and, combined with the network topology distance, calculate the transmission speed along the network topology. Step S2-1: Collect real-time time-series data of each node in the monitoring node set S; For the first monitoring node S1, extract the following from its real-time time series data: Real-time pressure time series data of the first monitoring node S1: P real (1) = 0.2 MPa; Real-time traffic time-series data of the first monitoring node S1: F real (1) = 130m 3 / h; Real-time vibration time series data of the first monitoring node S1: V real (1) = 0.02g; For the second monitoring node S2, extract the following from its real-time time series data: Real-time pressure time series data of the second monitoring node S2: P real (2) = 0.15 MPa; Real-time traffic time-series data of the second monitoring node S2: F real (2) = 140m 3 / h; Real-time vibration time-series data of the second monitoring node S2: V real (2) = 0.025g; For the third monitoring node S3, extract the following from its real-time time series data: Real-time pressure time series data of the third monitoring node S3: P real (3) = 0.18 MPa; Real-time traffic time-series data of the third monitoring node S3: F real (3) = 125m 3 / h; Real-time vibration time series data of the third monitoring node S3: V real (3) = 0.018g; Next, for each node, the deviation of its real-time data from the historical dynamic normal fluctuation range obtained in step S1 is calculated. Taking the first monitoring node S1, the calculation process is as follows: Calculate the real-time pressure deviation δ of the first monitoring node S1. P (1): Judgment: P real (1) = 0.2 MPa; μ P (1)-3σ P (1) = 0.27 MPa, μ P (1)+3σ P (1) = 0.33 MPa; Because 0.2 [0.27, 0.33], therefore P real (1) Not within the normal fluctuation range of historical pressure I P (1) Inside; Calculate the degree of deviation: Calculate the distance to the interval boundary: |P real (1)-(μ P (1)+3σ P (1))|=|0.2-0.33|=0.13; |P real (1)-(μ P (1)-3σ P (1))|=|0.2-0.27|=0.07; According to the formula: δ P (1) = min{|P real (1)-(μ P (1)+3σ P (1))|,|P real (1)-(μ P (1)-3σ P (1))|} / (6σ P (1)) =0.07 / (6σ P (1))=0.07 / (6×0.01)≈1.1667; Calculate the real-time deviation δ of the flow rate at the first monitoring node S1. F (1): Judgment: F real (1) = 130m 3 / h;μ F (1)-3σ F (1) = 85m 3 / h, μ F (1)+3σ F (1) = 115m 3 / h; Because 130∉[85,115], therefore F real (1) Not within the normal fluctuation range of historical traffic flow I F (1) Inside; Calculate the degree of deviation: Calculate the distance to the interval boundary: |F real (1)-(μ F (1)+3σ F (1))|=|130-115|=15; |F real (1)-(μ F (1)-3σ F (1))|=|130-85|=45; According to the formula: δ F (1)=min{|F real (1)-(μ F (1)+3σ F (1))|,|F real (1)-(μ F (1)-3σ F (1))|} / (6σ F (1)) =15 / (6σ F (1))=15 / (6×5)=0.5; Calculate the real-time vibration deviation δ of the first monitoring node S1. V (1): Judgment: V real (1) = 0.02 g; μ V (1)-3σ V (1) = 0.007 g, μ V (1)+3σ V (1) = 0.013g; Because 0.02 [0.007, 0.013], therefore V real (1) Not within the normal fluctuation range of the historical vibration I V (1) Inside; Calculate the degree of deviation: Calculate the distance to the interval boundary: |V real (1)-(μ V (1)+3σ V (1))|=|0.02-0.013|=0.007; |V real (1)-(μ V (1)-3σ V (1))|=|0.02-0.007|=0.013; According to the formula: δ V (1)=min{|V real (1)-(μ V (1)+3σ V (1))|,|V real (1)-(μ V (1)-3σ V (1))|} / (6σ V (1)) =0.007 / (6σ V (1))=0.007 / (6×0.001)≈1.1667; Similarly, calculate the degree of deviation between monitoring nodes S2 and S3: For the second monitoring node S2: P real (2) = 0.15 MPa I P (2), F real (2) = 140m 3 / h I F (2), V real (2) = 0.025g I V (2); Therefore, we can calculate: δ P (2)≈1.6667, δ F (2)≈1.25, δ V (2)≈2.1667; For the third monitoring node S3: P real (3) = 0.18 MPa I P (3), F_real(3) = 125m 3 / h I F (3), V real (3) = 0.018g I V (3); Therefore, we can calculate: δ P (3)≈0.8333, δ F (3)≈0.7083, δ V (3)≈1.1667; Step S2-2: For the i-th monitoring node S i Based on the calculated pressure, flow rate, and real-time deviation of vibration, the real-time abnormal state value A(i) is calculated; the pressure weighting coefficient w P =0.5, traffic weighting coefficient w F =0.3, vibration weighting coefficient w V =0.2; Calculated as follows: A(1)=w P ×δ P (1)+w F ×δ F (1)+w V ×δ V (1) = 0.5 × 1.1667 + 0.3 × 0.5 + 0.2 × 1.1667 = 0.9666; A(2)=w P ×δ P (2)+w F ×δ F (2)+w V ×δ V (2) = 0.5 × 1.6667 + 0.3 × 1.25 + 0.2 × 2.1667 = 1.6416; A(3)=w P ×δ P (3)+w F ×δ F (3)+w V ×δ V (3) = 0.5 × 0.8333 + 0.3 × 0.7083 + 0.2 × 1.1667 = 0.8625; Next, the difference between the abnormal state values ​​between any two adjacent monitoring nodes is calculated as the spatial change gradient G(i,j). The adjacent monitoring node pairs are (S1,S2) and (S2,S3); G(1,2)=|A(1)-A(2)|=|0.9666-1.6416|=0.675; G(2,3)=|A(2)-A(3)|=|1.6416-0.8625|=0.7791; Step S2-3: Continuously monitor the A(i) value of each monitoring node; anomaly detection threshold A th =0.5; Record the first time that the value of A(i) of each node exceeds A. th The time of occurrence of the anomaly is taken as T(i); based on historical monitoring records: The time of the anomaly occurrence of monitoring node S1 is T(1) = 10:00:00; The time of the anomaly occurrence at monitoring node S2 is T(2) = 10:00:05; The time of the anomaly occurrence at monitoring node S3 is T(3) = 10:00:15; For any two adjacent monitoring nodes S i and S j When both show abnormalities (i.e., T) i With T j (Both exist) and T i ≠Tj At that time, according to the pipeline topology from S i To S j Given the pipe length L(i,j), calculate the transmission speed U(i,j) of the abnormal signal between the two points; the pipe length is known to be L(1,2)=L(2,3)=1000 meters. U(1,2)=L(1,2) / |T(2)-T(1)|=1000 m / 5 sec=200 m / sec; U(2,3)=L(2,3) / |T(3)-T(2)|=1000m / 10s=100m / s.

[0022] Step S3: Treat each monitoring node as a dynamic monitoring node, and the state of each dynamic monitoring node is represented by the real-time abnormal state value of each monitoring node generated in Step S2; calculate the correlation strength between the abnormal states of adjacent dynamic monitoring nodes based on the spatial change gradient obtained in Step S2; dynamically adjust the correlation strength based on the transmission speed obtained in Step S2 to obtain the adjusted correlation strength; construct a dynamic abnormal correlation network based on all dynamic monitoring nodes and their adjusted correlation strength, where nodes represent the real-time degree and location of the abnormality, and the edges between nodes represent the real-time strength and direction of the abnormality propagation. Step S3-1: Define each monitoring node in the monitoring node set S as a dynamic monitoring node, and the state value of each dynamic monitoring node is the real-time abnormal state value A(i) generated in step S2-2; for any two adjacent dynamic monitoring nodes S i and S j Based on the spatial variation gradient G(i,j) obtained in step S2-2, the initial correlation strength R is calculated. o (i,j)=1 / (1+G(i,j)); R o (1,2)=1 / (1+G(1,2))=1 / (1+0.675)≈0.597; R o (2,3)=1 / (1+G(2,3))=1 / (1+0.7791)≈0.5621; Step S3-2: For any two adjacent dynamic monitoring nodes S i and S j Based on the abnormal signal propagation speed U(i,j) obtained in step S2-3, calculate the adjusted correlation strength R(i,j); reference propagation speed U o =100 m / s; The adjustment rule is: if U(i,j) exists, then R(i,j) = R o (i,j)×(U(i,j) / U o); if U(i,j) does not exist, then R(i,j)=R o (i,j); For (S1,S2): U(1,2)=200 m / s exists, therefore R(1,2)=R o (1,2)×(U(1,2) / U o )=0.597×(200 / 100)=1.194; For (S2,S3): U(2,3)=100 m / s exists, therefore R(2,3)=R o (2,3)×(U(2,3) / U o )=0.5621×(100 / 100)=0.5621; Step S3-3: based on all dynamic monitoring nodes (S1,S2,S3) and their adjusted association strength R(i,j), construct a dynamic anomaly association network; When constructing the network, the direction of an edge is determined by the chronological order of anomaly occurrence: For dynamic monitoring nodes S1 and S2: since T1=10:00:00<T2=10:00:05, a directed edge from S1 to S2 is established, and the value R(1,2)=1.194 is assigned to this edge as its weight; For dynamic monitoring nodes S2 and S3: since T2=10:00:05<T3=10:00:15, a directed edge from S2 to S3 is established, and the value R(2,3)=0.5621 is assigned to this edge as its weight; In the dynamic anomaly association network, nodes represent the real-time degree and position of an anomaly, and directed edges between nodes represent the real-time intensity and direction of anomaly propagation.

[0023] Step S4: in the dynamic anomaly association network constructed in step S3, screen out dynamic monitoring nodes whose real-time anomaly state values are higher than a set threshold; for each screened dynamic monitoring node, check whether the occurrence time of the anomaly state value of the dynamic monitoring node is earlier than the occurrence time of the anomaly state value of its adjacent downstream nodes, and whether the association strength of edges connected to the dynamic monitoring node presents a gradient distribution that significantly attenuates outward from the dynamic monitoring node; a dynamic monitoring node that satisfies all the above conditions is determined as a core fault area; Step S4-1: among all dynamic monitoring nodes in the dynamic anomaly association network constructed in step S3, according to the A(i) values calculated in step S2-2: A(1)=0.9666, A(2)=1.6416, A(3)=0.8625, all are greater than the threshold A th=0.5; Therefore, the candidate monitoring node set C={C1,C2,C3}, where C1,C2,C3 represent the 1st, 2nd, and 3rd candidate monitoring nodes, respectively. These candidate monitoring nodes correspond to the nodes in the original monitoring node set S: candidate monitoring node C1 corresponds to monitoring node S1, candidate monitoring node C2 corresponds to monitoring node S2, and candidate monitoring node C3 corresponds to monitoring node S3. Step S4-2: For each candidate monitoring node in the candidate monitoring node set C, perform a timing condition check; obtain all downstream adjacent monitoring nodes of the candidate monitoring node in the dynamic anomaly association network, form a downstream adjacent monitoring node set, and check whether the anomaly occurrence time of the candidate monitoring node is earlier than the anomaly occurrence time of each node in its downstream adjacent monitoring node set.

[0024] For the first candidate monitoring node C1: Candidate monitoring node C1 corresponds to the original monitoring node S1; obtain all downstream adjacent monitoring nodes of candidate monitoring node C1 in the dynamic anomaly association network, forming the downstream adjacent monitoring node set D1={D 1_1}, where D 1_1 This represents the first downstream neighboring monitoring node of candidate monitoring node C1, corresponding to the original monitoring node S2; check whether the anomaly occurrence time T(C1)=T1=10:00:00 of candidate monitoring node C1 is earlier than that of its downstream neighboring monitoring node D. 1_1 The anomaly occurrence time T(S2) = 10:00:05; the condition 10:00:00 < 10:00:05 is met; therefore, candidate monitoring node C1 passes the timing condition check. For the second candidate monitoring node C2: Candidate monitoring node C2 corresponds to the original monitoring node S2; obtain all downstream adjacent monitoring nodes of candidate monitoring node C2 in the dynamic anomaly association network, forming the downstream adjacent monitoring node set D2={D 2_1}, where D 2_1 This represents the first downstream neighboring monitoring node of candidate monitoring node C2, corresponding to the original monitoring node S3; check whether the anomaly occurrence time T(C2)=T2=10:00:05 of candidate monitoring node C2 is earlier than that of its downstream neighboring monitoring node D. 2_1 The anomaly occurrence time T(S3) = 10:00:15; the condition 10:00:05 < 10:00:15 is met; therefore, candidate monitoring node C2 passes the timing condition check. For the third candidate monitoring node C3: Candidate monitoring node C3 corresponds to the original monitoring node S3; obtain all downstream adjacent monitoring nodes of candidate monitoring node C3 in the dynamic anomaly association network; S3 has no downstream adjacent monitoring nodes in the dynamic anomaly association network (no edge is drawn from S3); according to the method logic, candidate monitoring node C3 fails the timing condition check; The candidate monitoring nodes that pass the time-series condition check are C1 and C2, which correspond to the original monitoring nodes S1 and S2, respectively. Step S4-3: Perform an association intensity gradient condition check on the candidate monitoring nodes that have passed the time series check; For the candidate monitoring nodes (i.e., C1 and C2) that passed the time series condition check in step S4-2, perform an association intensity gradient condition check; gradient decay threshold σ th =0.05; For candidate monitoring node C1: Candidate monitoring node C1 corresponds to the original monitoring node S1; obtain the weights of the edges between candidate monitoring node C1 and each downstream adjacent monitoring node in the downstream adjacent monitoring node set D1, i.e., the adjusted association strength R(i,j) in step S3-2; where i is the index of candidate monitoring node C1 in the monitoring node set S (i.e., i=1), and j is the index of its downstream adjacent monitoring node in the monitoring node set S (for D... 1_1 (j=2); therefore, the weights to be obtained are R(1,2)=1.194; calculate the standard deviation σ of these weights. R (C1); Since there is only one node in D1, the weight set is {1.194}, and its standard deviation σ R (C1)=0; Determine: σ R (C1)=0<σ th (0.05), therefore, it is determined that the correlation strength between candidate monitoring node C1 (corresponding to S1) and its downstream adjacent monitoring node presents a gradient distribution that significantly decreases outward from candidate monitoring node C1 (S1); candidate monitoring node C1 passes the correlation strength gradient condition check. For candidate monitoring node C2: Candidate monitoring node C2 corresponds to the original monitoring node S2; obtain the weights of the edges between candidate monitoring node C2 and each downstream adjacent monitoring node in the downstream adjacent monitoring node set D2, i.e., the adjusted association strength R(i,j) in step S3-2; where i is the index of candidate monitoring node C2 in the monitoring node set S (i.e., i=2), and j is the index of its downstream adjacent monitoring node in the monitoring node set S (for D... 2_1 (j=3); therefore, the weights to be obtained are R(2,3)=0.5621; calculate the standard deviation σ of these weights. R (C2); Since there is only one node in D2, the weight set is {0.5621}, and its standard deviation σR (C2)=0; Determine: σ R (C2)=0<σ th (0.05), therefore, it is determined that the correlation strength between candidate monitoring node C2 (corresponding to S2) and its downstream adjacent monitoring node presents a gradient distribution that significantly decreases outward from candidate monitoring node C2 (S2); candidate monitoring node C2 passes the correlation strength gradient condition check; Step S4-4: The candidate monitoring nodes that pass both the timing condition check in step S4-2 and the correlation intensity gradient condition check in step S4-3 are identified as fault core area nodes. Based on the above examination results: Candidate monitoring node C1: checked by time series conditions and by correlation intensity gradient conditions; Candidate monitoring node C2: checked by time series conditions and by correlation intensity gradient conditions; Candidate monitoring node C3: Failed the timing condition check; Therefore, candidate monitoring nodes C1 and C2 are determined to be fault core area nodes; all nodes determined to be fault core areas constitute the fault core area F={C1,C2}, which corresponds to the original monitoring nodes S1 and S2.

[0025] The system includes: a historical data processing module, a real-time anomaly analysis module, a dynamic network construction module, and a fault core location module; The historical data processing module is used to determine the historical dynamic normal fluctuation range based on the historical time-series data of the monitoring nodes; The real-time anomaly analysis module is used to calculate the real-time anomaly state based on real-time time series data and the historical dynamic normal fluctuation range, and to analyze the spatial change gradient and transmission speed of the anomaly. The dynamic network construction module is used to construct a dynamic anomaly association network based on the real-time anomaly state, spatial change gradient, and transmission speed. The fault core location module is used to locate the fault core area in the dynamic anomaly association network.

[0026] The historical data processing module includes: The historical data acquisition unit is used to collect historical pressure time-series data, historical flow time-series data, and historical vibration time-series data. The historical interval calculation unit is used to determine the historical dynamic normal fluctuation range of pressure, the historical dynamic normal fluctuation range of flow, and the historical dynamic normal fluctuation range of vibration.

[0027] The real-time anomaly analysis module includes: The real-time deviation calculation unit is used to calculate the degree of deviation of real-time data from the historical dynamic normal fluctuation range. An abnormal state value calculation unit is used to calculate a real-time abnormal state value based on the degree of deviation. The gradient and velocity calculation unit is used to calculate the spatial gradient and anomaly propagation velocity.

[0028] The dynamic network building blocks include: The initial association strength calculation unit is used to calculate the initial association strength between adjacent monitoring nodes; A correlation strength adjustment unit is used to adjust the initial correlation strength according to the transmission speed; The network construction unit is used to construct the dynamic anomaly association network based on the adjusted association strength and the anomaly occurrence sequence.

[0029] The core fault location module includes: The candidate monitoring node filtering unit is used to filter nodes whose abnormal state values ​​are higher than the threshold. The timing condition check unit is used to check the timing sequence of anomalies. The gradient condition check unit is used to check the gradient decay distribution of the correlation strength. The fault determination unit is used to determine the nodes in the core fault area.

[0030] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for assessing the water supply capacity of a water supply network based on multi-source data analysis, characterized in that: Includes the following steps: Collect historical time-series data of each monitoring node in the water supply network under normal operating conditions, and determine the historical dynamic normal fluctuation range of each monitoring node; Real-time time-series data of each monitoring node in the water supply network are collected and compared with the historical dynamic normal fluctuation range to calculate the real-time abnormal state of each monitoring node; based on the differences in the real-time abnormal states of adjacent monitoring nodes and the temporal relationship of the occurrence of abnormalities, the spatial change gradient of the abnormalities and the transmission speed along the network topology are analyzed. A dynamic anomaly association network is constructed based on the real-time abnormal status of each monitoring node, the spatial change gradient, and the transmission speed. Based on the dynamic anomaly association network, the core fault region is located according to the abnormal state of the nodes, the order of anomaly occurrence, and the attenuation gradient of the association strength.

2. The method for assessing the water supply capacity of a water supply network based on multi-source data analysis according to claim 1, characterized in that: The historical time-series data includes historical pressure time-series data, historical flow time-series data, and historical vibration time-series data; the real-time time-series data includes real-time pressure time-series data, real-time flow time-series data, and real-time vibration time-series data.

3. The method for assessing the water supply capacity of a water supply network based on multi-source data analysis according to claim 1, characterized in that: The determination of the historical dynamic normal fluctuation range for each monitoring node includes: For each monitoring node, the historical pressure time series data, historical flow time series data, and historical vibration time series data are used to calculate the average value and standard deviation within the time period. Based on the average value and standard deviation, the historical dynamic normal fluctuation range of pressure, the historical dynamic normal fluctuation range of flow, and the historical dynamic normal fluctuation range of vibration are determined.

4. The method for assessing the water supply capacity of a water supply network based on multi-source data analysis according to claim 1, characterized in that: The calculation of the real-time abnormal status of each monitoring node includes: The deviation of real-time pressure time-series data, real-time flow time-series data, and real-time vibration time-series data of each monitoring node from the historical dynamic normal fluctuation range is calculated, and the deviation is fused to obtain the real-time abnormal state.

5. The method for assessing the water supply capacity of a water supply network based on multi-source data analysis according to claim 1, characterized in that: The construction of the dynamic anomaly association network includes: The initial association strength between adjacent monitoring nodes is calculated based on the spatial change gradient, and the initial association strength is corrected using the transmission speed to obtain the adjusted association strength. Based on the temporal relationship of the occurrence of abnormal states, a directed connection between nodes is established using the adjusted association strength as the weight.

6. The method for assessing the water supply capacity of a water supply network based on multi-source data analysis according to claim 1, characterized in that: The core area for locating the fault includes: From the dynamic anomaly association network, monitoring nodes whose real-time anomaly status values ​​exceed a preset threshold are selected as candidate monitoring nodes. For each candidate monitoring node, it is determined whether it meets the timing condition and gradient distribution condition. The timing condition refers to the anomaly occurrence time of the candidate monitoring node being earlier than the anomaly occurrence time of all downstream adjacent monitoring nodes. The gradient distribution condition refers to the distribution of the association strength between the candidate monitoring node and each downstream adjacent monitoring node, which satisfies a gradient distribution that significantly decreases from the candidate monitoring node to the downstream adjacent monitoring nodes. Candidate monitoring nodes that simultaneously meet the timing condition and the gradient distribution condition are identified as the core fault region.

7. A water supply network capacity assessment system based on multi-source data analysis, used to execute the water supply network capacity assessment method based on multi-source data analysis as described in any one of claims 1-6, characterized in that: The system includes: Historical data processing module, real-time anomaly analysis module, dynamic network construction module, and fault core location module; The historical data processing module is used to determine the historical dynamic normal fluctuation range based on the historical time-series data of the monitoring nodes; The real-time anomaly analysis module is used to calculate the real-time anomaly state based on real-time time series data and the historical dynamic normal fluctuation range, and to analyze the spatial change gradient and transmission speed of the anomaly. The dynamic network construction module is used to construct a dynamic anomaly association network based on the real-time anomaly state, spatial change gradient, and transmission speed. The fault core location module is used to locate the fault core area in the dynamic anomaly association network.

8. The water supply capacity assessment system for water supply networks based on multi-source data analysis according to claim 7, characterized in that: The historical data processing module includes: The historical data acquisition unit is used to collect historical pressure time-series data, historical flow time-series data, and historical vibration time-series data. The historical interval calculation unit is used to determine the historical dynamic normal fluctuation range of pressure, the historical dynamic normal fluctuation range of flow, and the historical dynamic normal fluctuation range of vibration. The real-time anomaly analysis module includes: The real-time deviation calculation unit is used to calculate the degree of deviation of real-time data from the historical dynamic normal fluctuation range. An abnormal state value calculation unit is used to calculate a real-time abnormal state value based on the degree of deviation. The gradient and velocity calculation unit is used to calculate the spatial gradient and anomaly propagation velocity.

9. The water supply capacity assessment system for water supply networks based on multi-source data analysis according to claim 7, characterized in that: The dynamic network construction module includes: The initial association strength calculation unit is used to calculate the initial association strength between adjacent monitoring nodes; A correlation strength adjustment unit is used to adjust the initial correlation strength according to the transmission speed; The network construction unit is used to construct the dynamic anomaly association network based on the adjusted association strength and the anomaly occurrence sequence.

10. The water supply capacity assessment system for water supply networks based on multi-source data analysis according to claim 7, characterized in that: The fault location module includes: The candidate monitoring node filtering unit is used to filter nodes whose abnormal state values ​​are higher than the threshold. The timing condition check unit is used to check the timing sequence of anomalies. The gradient condition check unit is used to check the gradient decay distribution of the correlation strength. The fault determination unit is used to determine the nodes in the core fault area.

Images