A salt drainage fault early warning system in saline soil areas based on multi-source data

Abstract

This invention relates to the field of fault early warning technology and discloses a fault early warning system for desalination in saline soil areas based on multi-source data. The system includes a preprocessing module, a prediction module, a working condition monitoring module, a display module, and an optional fault location module for extended functions. The system predicts crystallization risk by monitoring multi-source parameters such as temperature, flow rate, concentration, and pressure head within the desalination pipe, and issues anomaly warnings before a fault occurs. This overcomes the shortcomings of existing technologies that cannot effectively monitor crystallization risk within the desalination pipe, and is of great significance for ensuring the reliable operation of the desalination system and the long-term stability of saline soil subgrades.

Description

Technical Field

[0001] This invention relates to the field of fault early warning technology, specifically, to a salt drainage fault early warning system for saline soil areas based on multi-source data. Background Technology

[0002] The high salt content in the soil of saline-alkali areas can easily lead to diseases such as roadbed salt swelling and frost heave, seriously endangering the safety of road engineering.

[0003] To control soil salinity, underground drainage pipes are often laid to construct a drainage system, which removes saline water from the soil. Monitoring systems are also used to monitor the operational status of the drainage project and the risk of geological hazards. Existing technologies, such as Chinese invention patent CN103792340A, disclose an IoT-based saline soil monitoring and early warning system and method. Measurement terminals are deployed at various monitoring points in the area to be monitored to sample and detect the soil moisture content, temperature, and electrical conductivity at each point. The measurement data generated by the terminals is then transmitted via the IoT to a remote monitoring center, where the soil salinity distribution is determined and displayed visually on a screen.

[0004] Groundwater in saline soil areas typically contains Cl - SO4 2- Na + Ca 2+ Mg 2+ Plasma, total mineralization 5–35 g / L. High-concentration brine flows within the drainage pipe. When there are significant temperature fluctuations or substantial water evaporation within a short period, the solubility of substances such as NaCl, Na₂SO₄, CaSO₄, and CaCO₃ decreases, causing crystals to precipitate on the pipe wall. Over time, these crystals adhere to the pipe wall, gradually forming a hard scale layer, resulting in loss of water flow cross-section and even functional blockage. However, the initial stage of crystallization is highly concealed and difficult to detect in a timely manner. Existing technologies primarily focus on monitoring macroscopic indicators such as soil salinity or drainage volume, lacking real-time sensing and early warning methods for the crystallization state inside the drainage pipe, making it difficult to detect the aforementioned highly concealed crystallization risks. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies in effectively monitoring the risk of crystallization within salt drainage pipes by providing a salt drainage fault early warning system based on multi-source data in saline soil areas. By monitoring multi-source parameters such as temperature, flow rate, concentration, and pressure head within the salt drainage pipe, the system predicts the risk of crystallization and can issue anomaly warnings before a fault occurs. This is of great significance for ensuring the reliable operation of the salt drainage system and the long-term stability of saline soil subgrades.

[0006] This invention provides an early warning system for salt drainage failure in saline soil areas based on multi-source data; the system includes a preprocessing module, a prediction module, a working condition monitoring module, and a display module.

[0007] The preprocessing module is used to acquire measured data including real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head; acquire identification IDs composed of monitoring point numbers and sampling times; package measured data from the same monitoring point and the same sampling time with the corresponding identification IDs into a single measured record; and send it to the prediction module, operating condition monitoring module, and display module. It is also used to construct a topology map of the brine discharge system based on the mapping relationship between monitoring points and the spatial coordinate system.

[0008] The prediction module is used to predict the crystallization prediction value at each monitoring point of the salt discharge pipe under different operating conditions based on the measured data, saturation concentration lookup table and crystallization prediction analysis formula in the measured record, and to package the associated identification ID and crystallization prediction value into a crystallization prediction result and send it to the operating condition monitoring module.

[0009] The operating condition monitoring module is used to configure the brine concentration warning level based on the comparison between the real-time brine concentration and the brine concentration threshold, configure the crystallization warning level based on the comparison between the crystallization prediction value and the crystallization probability threshold, analyze and integrate the over-threshold results and the overall warning level, generate feedback information and send it to the display module.

[0010] The display module is used to visually output the running status based on feedback information.

[0011] To better realize the present invention, a salt discharge fault early warning system based on multi-source data in saline soil areas further includes a fault location module; when the fault location module is added to the system, the feedback information generated by the operating condition monitoring module is sent to the fault location module and the display module respectively; the fault location module is used to parse the received feedback information and filter out the abnormal operation information, determine the abnormal pipeline segment according to the identification ID in the abnormal operation information, and then generate fault location information by combining the confidence model and send it to the display module.

[0012] To better realize the present invention, the fault location module further includes a signal analysis unit, a fault point analysis unit, a fault section analysis unit, and a fault location output unit.

[0013] The signal analysis unit is used to receive feedback information and parse out the monitoring point number, sampling time, operating condition label, comprehensive early warning level, salt solution concentration early warning level, and crystallization probability early warning level corresponding to each identification ID. It filters out abnormal information with the value of "1" for the operating condition label, generates an abnormal point information table, and sends it to the fault point analysis unit. Each piece of information in the abnormal point information table is an abnormal information corresponding to an identification ID, and the monitoring point associated with the abnormal information is an abnormal point.

[0014] The fault point analysis unit is used to receive the abnormal point information table, obtain the coordinates of the abnormal point, the measured data of the N sampling times before the abnormal point, the feedback information of the N sampling times before the abnormal point, and the measured data of the N sampling times before the upstream and downstream adjacent monitoring points of the abnormal point from the preprocessing module, the prediction module, and the working condition monitoring module, and input all of them into the confidence model. The confidence model is used to calculate the abnormal confidence score of each abnormal point, and the abnormal point with an abnormal confidence score greater than the preset confidence score is marked as a fault point, and the coordinates of the abnormal point are marked as the fault point location. It is also used to store the abnormal confidence score, fault point location, and sampling time of the fault point into the fault point information table, and send it to the fault section analysis unit and the fault location output unit; N is a positive integer.

[0015] The fault section analysis unit is used to receive the fault point information table, extract the abnormal confidence score and fault point location of each fault point at the same sampling time, analyze the spatial distribution relationship of multiple fault points according to the topology map of the salt discharge system, and aggregate continuously adjacent fault points according to pipeline segments to obtain at least one fault section. It also extracts the coordinates of the fault point at the starting point and the coordinates of the fault point at the ending point of the fault section to obtain the fault section location. The unit is also used to store the fault section location in the fault section information table and send it to the fault location output unit.

[0016] The fault location output unit is used to receive the fault point information table and the fault section information table, and to package the abnormal confidence score of the fault point, the fault point location, and the fault section location at the same sampling time into a fault location information, store it in the fault information table, and send it to the display module.

[0017] To better implement the present invention, the confidence model is further configured to calculate an anomaly confidence score for each outlier according to a confidence scoring strategy. The confidence scoring strategy includes multiple confidence scoring rules, each confidence scoring rule is matched with a confidence weight, and the confidence weights corresponding to all confidence scoring rules are summed to obtain the anomaly confidence score.

[0018] To better realize the present invention, the display module is further configured to display a digital twin of the salt removal system.

[0019] Compared with the prior art, the present invention has the following advantages and beneficial effects.

[0020] (1) The present invention discloses a salt drainage fault early warning system based on multi-source data in saline soil areas, which fills the gap in the existing technology of no crystallization monitoring method in the salt drainage pipe. The system predicts crystallization risk by monitoring multi-source parameters in the salt drainage pipe, provides support for proactive maintenance decisions, and is of great significance for ensuring the reliable operation of the salt drainage system and the long-term stability of saline soil subgrade.

[0021] (2) The present invention discloses a salt discharge fault early warning system based on multi-source data in saline soil areas. The system collects and analyzes the temperature, flow rate, concentration and pressure head in the salt discharge pipe in real time through the preprocessing module and the prediction module. It also makes an early and quantitative prediction of the most critical fault, pipe blockage, by predicting the crystallization prediction value. This changes the maintenance mode from "post-event remediation" to "pre-event early warning", turning passive into active.

[0022] (3) The present invention discloses a salt drainage fault early warning system based on multi-source data in saline soil areas. By adding a fault location module, the fault area can be quickly located, which makes it easier for maintenance personnel to eliminate crystallization hazards as soon as possible, reduces the difficulty of maintenance personnel's work, and solves the problem of difficult fault location.

[0023] (4) The present invention discloses a salt discharge failure early warning system in a saline soil area based on multi-source data. Through the working condition monitoring module, it is configured with two early warnings: salt concentration and crystallization, which avoids misjudgment of single indicators and makes the early warning results more reliable.

[0024] (5) The salt drainage fault early warning system in the saline soil area disclosed in this invention, based on multi-source data, can provide solid data support for subsequent in-depth analysis of the long-term performance evolution of saline soil subgrade through the measured records saved by the preprocessing module. Attached Figure Description

[0025] Figure 1 This is a basic functional module diagram of the salt drainage fault early warning system in saline soil areas based on multi-source data, as described in this invention.

[0026] Figure 2 The functional module diagram of the salt drainage fault early warning system in the saline soil area based on multi-source data of the present invention after adding fault location function.

[0027] Figure 3 This is a schematic diagram of the preprocessing module.

[0028] Figure 4 This is a schematic diagram of the prediction module.

[0029] Figure 5 This is a schematic diagram of the prediction module's workflow.

[0030] Figure 6 This is a structural diagram of the operating condition monitoring module.

[0031] Figure 7 This is a structural diagram of the fault location module.

[0032] Figure 8 The diagram shows three typical layout methods for salt drainage pipelines; among which: Figure 8 (a) is shaped like an "E". Figure 8(b) is in the shape of "non", Figure 8 (c) is in the shape of "fishbone".

[0033] Figure 9 It is a partial schematic diagram of the topology of the salt drainage system.

[0034] Figure 10 It is a schematic diagram of the main working process of the salt drainage fault warning system in the saline soil area based on multi-source data described in the present invention.

[0035] Figure 11 It is an architecture diagram of the salt drainage system platform in Embodiment 8.

[0036] Figure 12 It is a schematic diagram of the hardware deployment of the salt drainage system platform in Embodiment 8.

[0037] Among them: 101, data acquisition instrument; 201, Internet of Things base station; 301, central server; 401, display terminal. Specific implementation manners

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0039] Embodiment 1:

[0040] This embodiment provides a salt drainage fault warning system in the saline soil area based on multi-source data. As Figure 1 shown, it includes: a preprocessing module, a prediction module, a working condition monitoring module, and a display module.

[0041] The preprocessing module is used to obtain the measured data including real-time temperature, real-time salt solution flow rate, real-time salt solution concentration, and real-time pressure head, obtain the identification ID composed of the number of the monitoring point and the sampling time, package the measured data and the corresponding identification ID at the same monitoring point and the same sampling time into a measured record, and send it to the prediction module, the working condition monitoring module, and the display module; it is used to construct the topology diagram of the salt drainage system according to the mapping relationship between the monitoring points and the spatial coordinate system.

[0042] The prediction module is used to predict the crystallization probability at each monitoring point of the salt drainage pipe under different working conditions according to the measured data, the saturation concentration lookup table, and the crystallization prediction analysis formula in the measured record, and package the associated identification ID and the crystallization prediction value into a crystallization prediction result and send it to the working condition monitoring module.

[0043] The operating condition monitoring module is used to configure the brine concentration warning level based on the comparison between the real-time brine concentration and the brine concentration threshold, configure the crystallization warning level based on the comparison between the crystallization prediction value and the crystallization probability threshold, analyze and integrate the over-threshold results and the overall warning level, generate feedback information and send it to the display module.

[0044] The display module is used to visually output the running status based on feedback information.

[0045] It should be noted that when the system provided in this embodiment issues a fault warning, it does not monitor for complete blockage of the brine drain pipe. Instead, it focuses on monitoring for "crystal formation on the inner wall of the brine drain pipe leading to a significant reduction in the water flow area." Unless otherwise specified, "functional blockage" in this embodiment refers to a reduction in the water flow area of ​​30%. When the system detects "functional blockage" and issues a severe warning, maintenance personnel need to intervene manually as soon as possible to remove or dissolve the crystals to maintain the normal operation of the brine drain system.

[0046] This embodiment provides a system that adopts a modular design approach. Each module implements preprocessing, prediction, operating condition monitoring, and display functions, which conforms to the principle of high cohesion and low coupling in software engineering, and facilitates system maintenance and functional expansion.

[0047] Example 2:

[0048] This embodiment further optimizes the scheme based on Embodiment 1. For example... Figure 3 As shown, the preprocessing module includes a measured database and a topology graph construction unit.

[0049] The measured database is used to store measured records;

[0050] The topology construction unit is used to retrieve the GIS pipeline network map of the salt drainage system. First, it establishes a "monitoring point - pipeline segment" mapping table based on the correlation between the coordinates of the monitoring points in the spatial coordinate system, the salt drainage pipe laying path, the monitoring point number, and the pipeline segment number. Then, it obtains the topology map of the salt drainage system by using the location of the monitoring point as the node and the salt drainage pipe laying path as the edge.

[0051] This embodiment provides a factual measurement database table (fact_measurement) to store factual measurement records; the table structure of this factual measurement database table is shown in Table 1:

[0052] Table 1. Table structure of the measured database tables

[0053]

[0054] The pipeline segment number can be obtained through the "monitoring point - pipeline segment" mapping table, or it can be stored directly.

[0055] Primary key (mp_id, sample_time): This is used as the unique identifier for each test record.

[0056] Indexes; a fast time-range scan used to create indexes on mp_id and sample_time respectively, to optimize the performance of queries by time and by monitoring point.

[0057] Although the primary key (mp_id, sample_time) is already an index, it sorts primarily by mp_id and then by sample_time. This is highly efficient for retrieving the latest data for a specific measurement point. However, for queries that don't specify mp_id and only filter by time range, such as retrieving all measurement point data for a specific day or retrieving the latest 100 records, the primary key index may not be optimal. In this case, the Indexes (ts_desc) created solely on the sample_time field can play a significant role, allowing the database to quickly locate data within a specific time range and greatly improving query performance. In other words, Indexes (ts_desc) is a separate, optimized structure created to improve the speed of querying and sorting by time fields.

[0058] The preprocessing module continuously receives real-time data packets from various monitoring points; after parsing, the system combines the temperature, flow rate, concentration, pressure head data of the same monitoring point number and the same sampling time into a single measured record, which is then structured and stored in the measured database.

[0059] It should be noted that after the preprocessing module parses the real-time data packets collected from each monitoring point, it also performs data cleaning such as outlier removal and sensor drift correction to avoid false alarms due to data quality issues. Data cleaning is a conventional technique in this field, and this embodiment does not involve any improvement in this aspect, so it will not be described in detail here.

[0060] In another specific embodiment, a method for constructing a topology map of a salt drainage system is provided. First, the design drawing file of the salt drainage pipeline network is imported. Then, in the geographic information system software, the pipeline elements in the drawing are converted into vector line data with spatial coordinates. Next, in the spatial database, based on the spatial relationship between the coordinates of monitoring points and the spatial positions of pipeline segments, a spatial association query is performed to establish a "monitoring point - pipeline segment" mapping table. Then, using a graph computing library, the node data composed of monitoring points and the edge data composed of pipeline segments are read to generate a directional topology map structure. In this topology map, the node attributes contain spatial coordinates, and the edge attributes contain pipeline physical parameters, which can be used by the subsequent fault location module.

[0061] To implement the above method for constructing the topology map of the salt drainage system, the technical path of "DWG→ArcGIS Pro→PostGIS→NetworkX" can be adopted. First, import the *.dwg pipeline map provided by the design unit, with the coordinate system CGCS2000 / 3° zone, and use the ArcGIS Pro "CAD to Geodatabase" tool to generate the feature class PipeCenterline; then, execute SQL in PostGIS to establish a "monitoring point - pipeline segment" mapping table; next, use NetworkX to read the node-edge CSV and generate a directed graph G, with node attributes including (x, y, z) and edge attributes including pipe length, pipe diameter, and material, which can be used by the subsequent fault location module.

[0062] It should be noted that the technical path of "DWG→ArcGIS Pro→PostGIS→NetworkX" is widely used in the fields of Geographic Information Systems (GIS) and network analysis. The improvement in this embodiment does not lie in this path, and therefore will not be elaborated upon. On the other hand, constructing a topology map of the salt drainage system does not necessarily require the use of commercial software / libraries such as "ArcGIS Pro," "PostGIS," and "NetworkX." Open-source alternatives such as QGIS, FME, Oracle, Spatial, and Neo4j can be used to reduce costs. Other self-developed tools can also be used, as long as they can achieve the same technical effect; therefore, they will not be elaborated upon further.

[0063] Furthermore, the data structure and process for constructing this topology graph are explained. First, a node table is designed to store information about monitoring points, including fields such as node_id (node ​​ID), x_coord (X-coordinate), y_coord (Y-coordinate), and z_coord (elevation). Second, an edge table is designed to store information about pipe segments, including fields such as edge_id (edge ​​ID), from_node (starting node), to_node (ending node), length (pipe length in meters), diameter (pipe diameter in millimeters), and material (material). Third, the node and edge tables are used as input to generate the topology graph G using a graph computation library. Node attributes include coordinate values; edge attributes include their physical parameters and direction, such as water flow from from_node to to_node. This topology graph structure can be directly called by the fault location module.

[0064] The preprocessing module provided in this embodiment collects multi-dimensional data such as temperature, flow rate, concentration, pressure head, etc. in real time, combines GIS to construct a topology map, adopts a "monitoring point-pipeline section" mapping table and spatial association technology, and integrates and stores actual measurement records to provide data support for subsequent prediction and operating condition monitoring.

[0065] The other parts of this embodiment are the same as those in Embodiment 1, and will not be described again.

[0066] Example 3:

[0067] This embodiment is a further optimized scheme based on Embodiment 1 or Embodiment 2. For example... Figure 4 As shown, the prediction module includes a historical database and a prediction model; its operation is as follows: Figure 5 As shown.

[0068] The historical database is used to store a saturation concentration lookup table. The saturation concentration lookup table contains several historical records, each of which consists of a set of associated temperature, brine flow rate, and brine saturation concentration. For example, multiple "temperature-flow rate-salt saturation concentration" records obtained from multiple laboratory simulation tests are stored as historical records in the SQLite table tbl_sat to obtain the saturation concentration lookup table.

[0069] The prediction model is used to analyze each measured record, extract the identification ID, real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head, obtain the reference brine saturation concentration according to the saturation concentration lookup table, calculate the saturation coefficient according to the ratio of the real-time brine concentration to the reference brine saturation concentration, calculate the crystallization prediction value according to the crystallization prediction analysis formula, and package the associated identification ID and crystallization prediction value and send them to the operating condition monitoring module.

[0070] First, it should be noted that the saturation concentration of a single salt solution in the laboratory is mainly affected by temperature. In actual engineering, the composition of salt solutions is complex and influenced by factors such as the slope of the drainage pipe, the smoothness of the pipe material, and the flow rate of the salt solution. Therefore, when crystallization occurs, the actual salt solution concentration differs significantly from the saturation concentration of a single salt solution in the laboratory. For example, the supersaturated concentration of pure NaCl aqueous solution in the laboratory is 315 g / L, but the brine produced during the salt drainage process has a complex composition, with Na... + Ca 2+ Mg 2+ K + CL - SO4 2- CO4 2- HCO3 -Plasma mainly exists in brine in the form of compounds such as NaCl, MgCl2, Na2SO4, MgSO4, CaSO4, and CaCO3, and precipitates out when supersaturated. The brine is the salt solution described in this embodiment. CaSO4 and CaCO3, with low solubility, precipitate from the salt solution and adhere to the pipe walls, forming scale. They can also form mixed scale with other salts, and over time, the deposition intensifies, leading to crystallization blockage in the salt drainage pipe. On-site measurements at a certain engineering project revealed that crystallization was observed at the underground pipe port of the saline soil drainage system with a concentration of 210-240 g / L; at 5℃, the common crystallization starting point for brine with high CaSO4 and CaCO3 content was as low as approximately 190 g / L. Furthermore, at pipe bends and small-diameter sections where flow velocity increases, CO2 escapes from the water, increasing the pH value and promoting the formation of CaCO3, which crystallizes out when supersaturated. CCTV monitoring of a certain test section revealed that the crystal layer thickness reached 2mm after 30 days. After one year, crystallization and siltation caused a loss of more than 50% of the water flow cross-section. At this point, the desalination system was determined to have a serious malfunction, and maintenance personnel needed to be notified as soon as possible for repair.

[0071] When building digital models for simulation or constructing test areas for experimental measurements, the slope and material of the drain pipe in the salt drainage system are usually designed according to common parameters. Therefore, in simulation or experimental measurements, the slope and material of the drain pipe are treated as fixed parameters, and the focus is on studying the salt saturation concentration corresponding to different combinations of temperature and flow rate. A set of correlated temperature, flow rate, and salt saturation values ​​corresponds to a historical record.

[0072] In another specific implementation, two prediction models are provided. Initially, the workflow designed for the prediction model involved indexing historical records with consistent temperature and flow rate from a historical database based on measured data. This allowed for the direct acquisition of the salt saturation concentration corresponding to that temperature and flow rate, which was then used as a reference salt saturation concentration for analyzing crystallization probability. When there were sufficient historical records with high coverage, the prediction model could iterate through the saturation concentration lookup table based on real-time temperature and flow rate from the measured records to find the directly corresponding salt saturation concentration. However, when there were few historical records with low coverage, situations arose where the salt saturation concentration could not be directly matched, resulting in system errors due to no index results. Pre-storing more historical records could reduce the number of errors due to no index results, but acquiring a massive amount of historical records was difficult and still couldn't absolutely cover all actual engineering environments. Therefore, two versions of the prediction model were designed sequentially.

[0073] In the first version of the prediction model, when executing the subroutine "obtain reference brine saturation concentration", it searches the saturation concentration lookup table for historical records that match the real-time temperature and real-time brine flow rate in the real-time measurement data, extracts the brine saturation concentration value from the historical record, and assigns it to the reference brine saturation concentration.

[0074] The second version of the prediction model, when executing the subroutine of "obtaining reference brine saturation concentration", uses a step-by-step matching method to match in the order of temperature and flow rate, and searches the saturation concentration lookup table for the reference brine saturation concentration that matches the real-time temperature and real-time brine flow rate in the measured records.

[0075] When calculating the reference brine saturation concentration C_sat_ref under the current conditions using the stepwise matching method, the priority is: temperature matching > flow rate matching.

[0076] The first step is temperature matching and filtering. The system searches the saturation concentration lookup table for the historical data set with the smallest absolute difference between the current and real-time temperatures.

[0077] The second step is flow rate matching and filtering. From the optimal results obtained by temperature matching and filtering, we find the set of historical records with the smallest absolute difference between the flow rate and the real-time flow rate.

[0078] The third step is to assign the value of the brine saturation concentration in the optimal result of flow rate matching screening to the reference brine saturation concentration if there is only one historical record in the optimal result of flow rate matching screening; if there are ≥2 historical records in the optimal result of flow rate matching screening, the average value of the brine saturation concentration in the optimal result is calculated and the average value of the saturation concentration is assigned to the reference brine saturation concentration.

[0079] Furthermore, the crystallization prediction analysis formula is as follows:

[0080] When S≤1, then Pr=0.

[0081] When S > 1, then ;

[0082] Where Pr is the predicted crystallization value, S is the saturation coefficient, and S max This represents the oversaturation limit.

[0083] The supersaturation limit S for different salt compositions max The values ​​are different, for example: S of CaSO4 at 5℃ max The value is around 1.05. The S of NaCl at 5℃ max The value is approximately 1.25; furthermore, considering the inhibitory effect of flow rate on crystallization rate, a correction factor also needs to be introduced, such as increasing it by 20% when the flow rate is greater than 0.3 m / s. The improvement in this embodiment does not lie in how to obtain S. maxInstead of using the value of , we can substitute it as a known parameter into the crystallization prediction analysis formula for calculation, so we will not go into details.

[0084] Taking the saline-alkali soil test area in Northwest China as an example, during the initial irrigation and drainage period, the real-time measurement data obtained at a certain sampling time at a certain monitoring point are as follows: real-time temperature 5.5℃, real-time salt solution flow rate 0.32m / s, real-time salt solution concentration 210.6g / L. The process of calculating the crystallization probability is as follows:

[0085] (1) Obtain the reference brine saturation concentration: Based on the real-time measurement data, index the historical record that best matches the temperature and flow rate in the real-time measurement data from the saturation concentration lookup table. The brine saturation concentration in this historical record is 195.6 g / L, that is, the reference brine saturation concentration C_sat_ref=195.6 g / L;

[0086] (2) Calculate the saturation coefficient S: S = 210.6 / 195.6 = 1.077, which meets the condition S > 1;

[0087] (3) Obtain the oversaturation limit S max :S max The value is 1.30;

[0088] (4) Calculate the crystallization probability: Set S=1.077 and S max Substituting 1.30 into the formula for calculating Pr when S > 1:

[0089] We get 0.257, which is 25.7%.

[0090] The prediction module provided in this embodiment uses a crystallization prediction model based on the saturation coefficient S and the supersaturation limit S. max The prediction of crystallization probability conforms to the solubility equilibrium law, and the problem of sparsity of historical data is solved by hierarchical matching method, resulting in highly reliable prediction results.

[0091] The other parts of this embodiment are the same as those in Embodiment 1 or Embodiment 2, and will not be described again.

[0092] Example 4:

[0093] This embodiment is a further optimization based on any one of Embodiments 1-3. For example... Figure 6 As shown, the operating condition monitoring module includes a threshold storage unit, a salt solution concentration monitoring unit, a crystallization monitoring unit, an anomaly feedback unit, and an operating database.

[0094] The threshold storage unit is used to store the salt concentration threshold and the crystallization probability threshold.

[0095] The brine concentration monitoring unit is used to parse and identify the ID and real-time brine concentration from the measured records, call the brine concentration threshold, determine whether the real-time brine concentration exceeds the brine concentration threshold and output the judgment result A, and configure the brine concentration warning level according to the relationship between the real-time brine concentration and the brine concentration threshold.

[0096] The crystallization monitoring unit is used to parse and identify the ID and crystallization prediction value from the crystallization prediction result, call the crystallization probability threshold, determine whether the crystallization prediction value exceeds the crystallization probability threshold and output the judgment result B, and configure the crystallization probability warning level according to the relationship between the crystallization prediction value and the crystallization probability threshold.

[0097] Examples of salt solution concentration warning levels and crystallization probability warning levels are shown in Tables 2 and 3:

[0098] Table 2 Examples of Salt Solution Concentration Warning Levels

[0099]

[0100] Table 3 Examples of Crystallization Probability Warning Levels

[0101]

[0102] Where C_th represents the salt concentration threshold and P_th represents the crystallization probability threshold.

[0103] When the parameter values ​​for both the brine concentration warning level and the crystallization probability warning level are 0, it indicates that both are within the threshold and the system is operating normally. When the parameter values ​​for both the brine concentration warning level and the crystallization probability warning level are not 0, it indicates that there is a situation exceeding the threshold, and a comprehensive warning level information will be generated.

[0104] The anomaly feedback unit is used to analyze judgment results A and B corresponding to the same identification ID. When judgment result A or judgment result B exceeds the threshold, the operating condition label is assigned a value of "1", and when judgment result A and judgment result B are within the threshold range, the operating condition label is assigned a value of "0". It is also used to generate comprehensive warning level information based on the highest level among the salt solution concentration warning level and the crystallization probability warning level. Furthermore, it is used to generate feedback information based on the operating condition label, comprehensive warning level, salt solution concentration warning level, crystallization probability warning level, and associated identification ID, and send it to the display module. Feedback information with an operating condition label value of "1" indicates abnormal operation information, and the associated monitoring point is an abnormal point. Feedback information with an operating condition label value of "0" indicates normal operation information.

[0105] The following are some typical scenarios when obtaining comprehensive early warning level information:

[0106] If the salt concentration warning level is 0 and the crystallization probability warning level is 0, then the overall warning level is 0.

[0107] If the salt concentration warning level is 0 and the crystallization probability warning level is 1, then the overall warning level is 1.

[0108] If the salt concentration warning level is 1 and the crystallization probability warning level is 0, then the overall warning level is 1.

[0109] If the salt solution concentration warning level is 1 and the crystallization probability warning level is 3, then the overall warning level is 3.

[0110] Other cases follow the same principle as described above and will not be listed individually.

[0111] The operating database is used to store feedback information.

[0112] For salt concentration warnings, crystallization probability warnings, and comprehensive anomaly warnings, two indicator areas will be designed on the display interface. The first area directly reflects the comprehensive anomaly situation through operating status lights, while the second area displays information on monitoring points exceeding the threshold, the items exceeding the threshold, and the corresponding warning levels through an information table. Furthermore, based on actual needs, the second area can be set to a collapsible mode; clicking "Anomaly Details" expands the view, while clicking "Hide Details" collapses the window.

[0113] Table 4 shows examples of system warning methods corresponding to each comprehensive warning level:

[0114] Table 4 Examples of System Warning Methods Corresponding to Comprehensive Warning Level Information

[0115]

[0116] The colors "green, solid light", "yellow, flashing", "yellow, solid light", and "red, solid light" correspond to levels "0", "1", "2", and "3" respectively.

[0117] The operational condition monitoring module disclosed in this embodiment first performs early warnings for salt concentration and crystallization probability, and then summarizes them into a comprehensive early warning level according to the highest-level comprehensive rule, considering two key but different risk dimensions. Compared with the single-index early warning mechanism, the dual-threshold early warning mechanism can reduce false alarms and provide important reference for maintenance teams to select solutions. Furthermore, a hierarchical threshold analysis method is used to divide each early warning level into levels L0-L3, and matching early warning methods such as pop-ups, SMS, and buzzers are matched. Through the hierarchical response mechanism, both false alarm interference can be avoided and serious faults can be promptly and effectively communicated to maintenance personnel, which conforms to the design specifications of industrial monitoring systems.

[0118] In another specific embodiment, the salt concentration threshold stored in the threshold storage unit is updated periodically or irregularly using a dynamic update mechanism, and the crystallization probability threshold stored is a fixed preset.

[0119] In the formula for calculating the probability of crystallization, S>1 indicates that precipitation is thermodynamically possible. The crystallization probability threshold represents the physical critical point and is independent of season and region. Generally, the crystallization probability threshold is set to 0.15, at which point the corresponding saturation coefficient S is 1.045.

[0120] The salt concentration threshold needs to be dynamically adjusted according to environmental conditions. As the salt drainage project progresses, the salt concentration should generally show a downward trend; however, influenced by factors such as rainfall, irrigation, and evaporation, the salt concentration may fluctuate significantly within a short period. For example, real-time measurement data collected at a monitoring point in the experimental area showed that the real-time salt concentration dropped from 20.8 g / L to 16.6 g / L before and after heavy rainfall. After the rain, as evaporation continued, the real-time salt concentration collected at this monitoring point gradually increased. For instance, based on data released by the meteorological bureau, if the evaporation rate in the experimental area exceeded 10 mm / d for three consecutive days, the originally set salt concentration threshold was raised from 21.3 g / L to 23.8 g / L to avoid false alarms. It should be noted that the salt concentration threshold can be set based on experience or the results calculated by other systems. The improvement of this invention does not lie in how to determine the salt concentration threshold, but only in obtaining the salt concentration threshold and using it as the basis for early warning judgment; therefore, it will not be elaborated further.

[0121] Therefore, this embodiment uses a fixed crystallization probability threshold to remain sensitive to "initial crystallization" and a dynamic salt concentration threshold to remain robust to "environmental background". The combination of the two achieves the early warning effect of "reporting when required and not reporting randomly".

[0122] The other parts of this embodiment are the same as any one of Embodiments 1-3, and will not be described again.

[0123] Example 5:

[0124] Compared with any one of Embodiments 1-4, this embodiment expands the fault location module. By screening fault points through a confidence model and combining it with the topology map to aggregate segments, it achieves a technical leap from "what the problem is" to "where the problem is", which is a key technical point of this invention.

[0125] Specifically, such as Figure 2 As shown, this embodiment provides a salt drainage fault early warning system for saline soil areas based on multi-source data, including: a preprocessing module, a prediction module, an operating condition monitoring module, a fault location module, and a display module.

[0126] The preprocessing module is used to acquire measured data including real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head; acquire identification IDs composed of the monitoring point number and sampling time; package the measured data of the same monitoring point and the same sampling time with the corresponding identification ID into a single measured record; and send it to the prediction module, the operating condition monitoring module, and the display module. It is also used to construct a topology map of the brine discharge system based on the mapping relationship between the monitoring points and the spatial coordinate system.

[0127] The prediction module is used to predict the crystallization probability at each monitoring point of the salt discharge pipe under different operating conditions based on the measured data, saturation concentration lookup table and crystallization prediction analysis formula in the measured records, and to package the associated identification ID and crystallization prediction value into a crystallization prediction result and send it to the operating condition monitoring module.

[0128] The operating condition monitoring module is used to configure the brine concentration warning level based on the comparison between the real-time brine concentration and the brine concentration threshold, and to configure the crystallization warning level based on the comparison between the crystallization prediction value and the crystallization probability threshold. It analyzes and integrates the over-threshold results and the overall warning level, and sends feedback information to the fault location module and the display module.

[0129] The fault location module is used to parse the received feedback information and filter out the abnormal operation information. Based on the identification ID in the abnormal operation information, it determines the abnormal pipeline segment, and then generates fault location information by combining the confidence model and sends it to the display module.

[0130] The display module is used to visualize the operating status based on feedback information. Based on the added fault location module, the system's overview interface has been enhanced and expanded in content compared to Example 1 to achieve precise visualization of fault locations. Key changes include: integrating fault location information into the global situation map and adding a fault location information panel.

[0131] Therefore, the overview interface in this embodiment is no longer a simple status monitor, but has evolved into an intelligent command center integrating "status monitoring, precise fault location, and operation and maintenance decision support". By transforming intangible data into tangible fault points and fault sections on a map, this invention greatly reduces the difficulty and time cost for operation and maintenance personnel in troubleshooting, and achieves a technological leap from "early warning" to "location".

[0132] In another specific implementation, such as Figure 7 As shown, the fault location module includes a signal analysis unit, a fault point analysis unit, a fault section analysis unit, and a fault location output unit.

[0133] The signal analysis unit is used to receive feedback information and parse out the monitoring point number, sampling time, operating condition label, comprehensive early warning level, salt solution concentration early warning level, and crystallization probability early warning level corresponding to each identification ID. It filters out abnormal information with an operating condition label value of "1", generates an abnormal point information table, and sends it to the fault point analysis unit. Each piece of information in the abnormal point information table is an abnormal information corresponding to an identification ID, and the monitoring point associated with the abnormal information is an abnormal point.

[0134] The fault point analysis unit is used to receive the abnormal point information table, obtain the coordinates of the abnormal point, the measured data of the N sampling times before the abnormal point, the feedback information of the N sampling times before the abnormal point, and the measured data of the N sampling times before the upstream and downstream adjacent monitoring points of the abnormal point from the preprocessing module, the prediction module, and the working condition monitoring module, and input all of them into the confidence model. The confidence model is used to calculate the abnormal confidence score of each abnormal point, and the abnormal point with an abnormal confidence score greater than the preset confidence score is marked as a fault point, and the coordinates of the abnormal point are marked as the fault point location. It is also used to store the abnormal confidence score, fault point location, and sampling time of the fault point into the fault point information table and send it to the fault section analysis unit and the fault location output unit; N is a positive integer.

[0135] The fault section analysis unit is used to receive the fault point information table, extract the abnormal confidence score and fault point location of each fault point at the same sampling time, analyze the spatial distribution relationship of multiple fault points according to the topology map of the salt discharge system, and aggregate continuously adjacent fault points according to pipeline segments to obtain at least one fault section. It also extracts the coordinates of the fault point at the starting point and the coordinates of the fault point at the ending point of the fault section to obtain the fault section location. The unit is also used to store the fault section location in the fault section information table and send it to the fault location output unit.

[0136] The fault location output unit is used to receive the fault point information table and the fault section information table, and to package the abnormal confidence score of the fault point, the fault point location, and the fault section location at the same sampling time into a fault location information, store it in the fault information table, and send it to the display module.

[0137] The main technical approach of the fault location module is as follows: parse the data sent by the operating condition monitoring module, execute the fault location subroutine, and output feedback information. When executing the fault location subroutine, the monitoring point number or identification ID is parsed from the abnormal operation information to obtain the coordinate values ​​of the abnormal monitoring point. Using the "abnormal monitoring point" as a seed monitoring point, and combining the comprehensive early warning level, topological connectivity, and historical abnormal frequency, an abnormal section confidence map is formed. Specifically, based on the salt discharge pipeline topology map and the identification ID of the abnormal monitoring point, the monitoring point numbers corresponding to the upstream and downstream neighbors of the abnormal monitoring point are obtained. The real-time brine flow rate and real-time pressure head of the upstream and downstream neighbors of the seed monitoring point at the same sampling time are retrieved from the preprocessing module. Simultaneously, the recent abnormal situations of the abnormal monitoring point are obtained, involving at least N measured records prior to the current sampling time. Then, an abnormal confidence score (0-1) is calculated for each abnormal monitoring point according to the abnormal confidence matching strategy. Monitoring points with high abnormal confidence scores are designated as abnormal points, and these points are aggregated according to the pipeline to obtain abnormal sections.

[0138] Furthermore, the anomaly confidence matching strategy includes flow velocity change rules, pressure head change rules, comprehensive early warning level rules, early warning item rules, and historical anomaly frequency rules.

[0139] Rule R1, Flow Velocity Variation Rule: At the same sampling time, if the real-time flow velocity at the seed monitoring point is significantly increased relative to the real-time flow velocity at the upstream monitoring point or the downstream monitoring point, a confidence weight of 0.3 is assigned. Generally, an increment > 30% is considered a significant increase; a change of no more than 5% is considered that the two are basically the same. Typical Case 1: At the same sampling time, comparing the real-time flow velocities Vup, Vseed, and Vdn at the upstream and downstream monitoring points, if the following conditions are met: the real-time flow velocities Vup and Vdn at the upstream and downstream monitoring points are basically the same, and the increase in the real-time flow velocity Vseed at the seed monitoring point is > 30% compared to the real-time flow velocities Vup or Vdn at the upstream or downstream monitoring point, then normal fluctuations in the water flow in the pipe can be ruled out, and the confidence level of functional blockage at the seed monitoring point is considered high. Typical Scenario 2: At the same sampling time, when comparing the real-time flow velocity Vup of the upstream monitoring point, the real-time flow velocity Vseed of the seed monitoring point, and the real-time flow velocity Vdn of the downstream monitoring point, if the following conditions are met: the real-time flow velocity Vseed of the seed monitoring point and the real-time flow velocity Vdn of the downstream monitoring point are basically the same and significantly higher than the real-time flow velocity Vup of the upstream monitoring point, the confidence level of functional blockage at the seed monitoring point is high, and functional blockage is also highly likely at the downstream monitoring point.

[0140] Rule R2, Pressure Head Variation Rule: At the same sampling time, if the pressure head at the seed monitoring point is significantly lower than the real-time pressure head at the upstream monitoring point or the downstream monitoring point, a confidence weight of 0.3 is assigned. Generally, a reduction > 30% is considered a significant decrease; a change of no more than 5% is considered that the two are basically the same. Typical Case 1: At the same sampling time, comparing the real-time pressure head PHup at the upstream monitoring point, the real-time pressure head PHseed at the seed monitoring point, and the real-time pressure head PHdn at the downstream monitoring point, if the following conditions are met: the real-time pressure head PHup at the upstream monitoring point and the real-time pressure head PHdn at the downstream monitoring point are basically the same, and the real-time pressure head PHseed at the seed monitoring point is reduced by > 30% compared to the real-time pressure head PHup at the upstream monitoring point or the real-time pressure head PHdn at the downstream monitoring point, then normal fluctuations in the water flow in the pipe can be ruled out, and the confidence level of functional blockage at the seed monitoring point is considered high. Typical Scenario 2: At the same sampling time, when comparing the real-time pressure head PHup of the upstream monitoring point, the real-time pressure head PHseed of the seed monitoring point, and the real-time pressure head PHdn of the downstream monitoring point, if the following conditions are met: the real-time pressure head PHseed of the seed monitoring point and the real-time pressure head PHdn of the downstream monitoring point are basically the same and significantly lower than the real-time pressure head PHup of the upstream monitoring point, the confidence level of functional blockage at the seed monitoring point is high, and functional blockage is also highly likely at the downstream monitoring point.

[0141] Rule R3, Comprehensive Early Warning Level Rule; Based on the comprehensive early warning level "1", "2" and "3" corresponding to the seed monitoring point, the confidence weights are assigned as 0.05, 0.1 and 0.15 respectively; The higher the comprehensive early warning level, the higher the anomaly confidence weight of the seed monitoring point.

[0142] Rule R4, Warning Item Rule: Based on whether the warning item of the seed monitoring point is salt concentration exceeding the threshold, crystallization probability exceeding the threshold, or both salt concentration and crystallization probability exceeding the threshold, assign confidence weights of 0.05, 0.05, and 0.1 respectively.

[0143] Rule R5 is a historical anomaly frequency rule. Based on the number of anomaly warnings appearing 1, 2, or 3 or more times in the last 5 sampling results of the seed monitoring point, confidence weights are assigned as 0.05, 0.1, and 0.15 respectively. The more anomalies in the past 3 samplings, the more likely a fault is, and the higher the confidence weight. Trend anomalies are captured based on time-series data from the previous N (N=5) sampling times, reducing interference from instantaneous fluctuations.

[0144] The sum of the confidence scores is used as the anomaly confidence score. An anomaly confidence score ≥ 0.6 is considered a high anomaly confidence score, and the corresponding monitoring point is the fault point.

[0145] In most cases, fault segments covering multiple anomaly monitoring points have their starting and ending monitoring points satisfying rules R1 and R2, so the anomaly confidence scores are usually relatively high.

[0146] The confidence model employs an anomaly confidence matching strategy to identify anomalies. It also assesses the credibility of anomalies by analyzing data from seed monitoring points and upstream / downstream monitoring points, effectively filtering out transient interference signals and improving the accuracy of fault location. In actual engineering, when salt drainage pipes experience functional blockage due to crystallization, it typically manifests as segmental crystallization blockage. Therefore, after locating the fault point, the faulty segment is aggregated based on the spatial relationships in the topology map, conforming to manual troubleshooting practices and effectively narrowing the scope and reducing the time cost of manual troubleshooting. Furthermore, combining this with a visual interface further enhances the system's practical value.

[0147] Example 6:

[0148] This embodiment is a further optimized version of any one of Embodiments 1-5. The display module is also used to display a digital twin generated based on the topology diagram of the salt drainage system, which facilitates a more intuitive presentation of the location of abnormal monitoring points and fault areas.

[0149] The system overview interface described in this embodiment typically displays a global situation map, a key indicator dashboard, a measured record display box, a fault alarm display box, a system data search box, and an export button. The key indicator dashboard centrally displays the comprehensive warning level, system operation status indicator lights, number and percentage of anomalies at the latest sampling time. It uses large numbers or level identifiers, such as L0-L3, to display the comprehensive warning level. It uses green / red colors combined with the text "Normal Operation" and "Crystallization Warning" for concise and eye-catching prompts. When the system detects normal operation, the system operation status indicator light is green; when the system detects an anomaly, the system operation status indicator light is red. The overview interface can be customized according to actual needs; this embodiment's improvement does not lie in this aspect and will not be elaborated upon further.

[0150] The other parts of this embodiment are the same as any one of Embodiments 1-5, and will not be described again.

[0151] Example 7:

[0152] This embodiment, based on the multi-source data-based early warning system for desalination faults in saline soil areas disclosed in Embodiment 6, illustrates a typical working mode of the system, and its main workflow is as follows: Figure 10 As shown.

[0153] Step S1: Obtain the measured data.

[0154] The system periodically acquires measured data from each monitoring point, including real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head. The acquired measured data is cached locally on the RTU for 7 days and can be resumed after network interruption.

[0155] Step S2: Perform preprocessing of the measured data using the preprocessing module.

[0156] First, the data packets are parsed through the edge gateway, then outlier removal is performed using the 3σ method, and after drift correction, the actual test record is generated.

[0157] Step S3: Calculate the crystallization probability using the prediction module.

[0158] First, the reference brine saturation concentration C_sat_ref is obtained by looking up the corresponding local SQLite table based on the real-time temperature and real-time brine flow rate. Then, the ratio of the real-time brine concentration to the reference brine saturation concentration is assigned to the saturation coefficient S. Finally, the supersaturation limit S is retrieved. max The value of mp_id is used to calculate the crystallization probability Pr, and finally the data such as mp_id, sample_time, and Pr are packaged and sent to the operating condition monitoring module.

[0159] Step S4: Calculate the crystallization probability using the operating condition monitoring module.

[0160] The brine concentration monitoring unit marks the brine concentration warning level with different levels from L0 to L3 based on the relationship between the real-time brine concentration and the brine concentration threshold.

[0161] The crystallization early warning monitoring unit marks the crystallization probability early warning level with different levels from L0 to L3 based on the relationship between the crystallization probability output by the prediction module and the crystallization probability threshold.

[0162] The abnormal feedback unit takes the highest level from the salt concentration warning level and the crystallization probability warning level, marks the comprehensive warning level with different levels from L0 to L3, and assigns a working condition label.

[0163] Step S5: Locate the fault point and fault area using the fault location module.

[0164] The signal analysis unit analyzes and filters out abnormal points with the condition label 1;

[0165] The fault point analysis unit calls the confidence model and calculates the confidence score according to the anomaly confidence matching strategy. Anomalies with a score ≥0.6 are identified as fault points.

[0166] The fault segment analysis unit aggregates consecutive fault points on the topology map and outputs the start-end coordinates to determine the start and end positions of the fault segment.

[0167] The fault location output unit packs the data including the fault point, fault section, and confidence level into JSON and pushes it to the display module.

[0168] Step S6: Visual display and warning push are performed through the display module.

[0169] If a digital twin of the desalination system is constructed, the fault point and fault section can be highlighted and flashed on the digital twin. When the alarm condition is triggered, local alarms can be made through pop-up windows, highlighting, etc.; enterprise WeChat messages can also be sent to inform the identification ID, comprehensive warning level, and disposal suggestions, such as flushing, backpressure, disassembly and replacement, etc. The function of the system can also be extended to enable maintenance personnel to click on the navigation on their mobile phones to directly reach the fault point or fault section.

[0170] Other parts of this embodiment are the same as those of the above embodiment, so they will not be described again.

[0171] Embodiment 8:

[0172] The salt-affected soil area desalination fault warning system based on multi-source data described in this embodiment is a product that combines software and hardware. Unless otherwise specified, the "system" in this text refers to the salt-affected soil area desalination fault warning system based on multi-source data. The software part of the system runs on the central server 301, and each module works together, relying on the sensor network deployed at the key nodes of the desalination pipe network in the salt-affected soil area for data collection. The hardware deployment of the system involves sensor types, monitoring point layout, data transmission methods, etc. The numerical values, materials, and equipment models involved in the embodiments are only for illustration, not for limitation.

[0173] A certain test section: K1235 + 400~K1236 + 200, with a total length of 800m; the groundwater salinity is 16.4g / L, the conductivity is 6 - 10mS / cm, and the main salt ions are SO4 2- 、Cl-, with medium-strong salt-affected soil characteristics. This test section can adopt the typical layout method of the desalination pipe network, such as Figure 8 shown, Figure 8 (a) is the "E" shape, Figure 8 (b) is the "non" shape, Figure 8(c) A "fishbone" shaped drainage network can also be used; other layout methods are also possible. The overall burial depth of the drainage network is 1.5m, with a slope of 2‰ to ensure water flow self-drainage; several transverse suction pipes are spaced outwards along the longitudinal main collection pipe to form an efficient collection system. The longitudinal main collection pipe and transverse suction pipes generally use high-strength, corrosion-resistant HDPE or PVC perforated corrugated pipes to ensure sufficient water flow capacity and structural strength. Stainless steel or concrete inspection wells are installed at the ends, bends, and at regular intervals of the longitudinal main collection pipe to facilitate future internal inspections using pipeline robots (CCTV) and high-pressure water jet flushing and maintenance. For example, with transverse suction pipes spaced 50m apart, monitoring points are set near the trisection points of each transverse suction pipe and at the connection points between the transverse suction pipes and the longitudinal main collection pipe, etc., at conventional locations. At the same time, monitoring points are densely deployed at key locations such as bends, confluence points, and low-lying areas of the pipeline to achieve focused monitoring of these high-risk areas. The salt drainage pipeline structure listed in this embodiment is for the purpose of illustrating the layout of the system data acquisition terminal, and does not limit the system described in Embodiments 1-7 to only be used in the salt drainage pipeline with the specific structure described above.

[0174] When applied to highway subgrade construction, the longitudinal water collection mains of the salt drainage system are typically located on both sides of the road shoulder or directly beneath the guardrail. Construction requirements for the salt drainage system: a construction period of 6 months; the system must be constructed and put into trial operation simultaneously with the main project; 12 months after construction, the soil salt content within 2m below the road surface should be ≤0.3%, i.e., electrical conductivity ≤2mS / cm; initial salt drainage is achieved through centralized flushing. A key function of the early warning system is to monitor the risk of "functional blockage" in the salt drainage system pipes due to crystallization.

[0175] like Figure 9 As shown in the schematic diagram of a partial area of ​​the salt drainage pipeline network, the "△" symbol represents a monitoring point. This topology deployment constructs a sensing neural network covering the entire salt drainage pipeline network, ensuring that the risk of blockage due to crystallization at different locations within the salt drainage pipeline caused by changes in temperature, flow rate, and concentration can be captured in real time, providing a data foundation for systemic fault early warning.

[0176] Each monitoring point is equipped with a multi-functional data acquisition instrument 101, which integrates a temperature sensor, a flow meter, a conductivity sensor, and a pressure head meter.

[0177] The temperature sensor is used to collect real-time temperature data, such as a PT100 platinum resistance temperature sensor with a measurement range of -20℃ to 60℃ and an accuracy of ±0.1℃.

[0178] The flow meter is used to collect real-time brine flow rate, such as an electromagnetic flow meter with a measurement range of 0.5-5 m / s and an accuracy of 0.5 grade.

[0179] The conductivity sensor is equipped with a temperature compensation function, used to collect the conductivity value of the fluid, and calculate the real-time salt concentration using a pre-calibrated conductivity-concentration conversion formula.

[0180] The pressure head gauge is used to collect real-time pressure head, such as an NB-IoT wireless pressure head gauge with a range of 0–0.5 mH2O and an accuracy of ±0.25 %FS.

[0181] Each probe of the data acquisition unit 101 is inserted into the salt drainage pipe for periodic sampling. It is IP68 protected and transmits the collected multi-source data to the remote central server 301 via wireless communication modules such as NB-IoT and LoRa. Specific parameters for each acquisition device can be customized as needed. After purchase, the wireless communication module, battery pack, and other modules are connected, and a casing is added to integrate them into a single unit; therefore, further details are omitted.

[0182] Raw data such as pipeline GIS data, CGCS2000 national geodetic coordinate system data, pipeline design drawings, monitoring point coordinates, and pipeline segment attributes are stored in a resource database. After data cleaning, coordinate unification, and standardization, the GIS pipeline map is obtained. During GIS data cleaning and coordinate unification, the coordinates are transformed in one step using the GDAL / OGR command.

[0183] ogr2ogr -s_srs EPSG:4547 -t_srs EPSG:4479 -zscale 1 gis_route.shpraw_route.dxf

[0184] 4547 is the local urban construction coordinate, and 4479 is the CGCS2000 / 3° zone, ensuring consistency with the national benchmark.

[0185] Based on the above deployment, this embodiment provides a salt discharge monitoring platform to provide complete technical support for data acquisition, transmission, processing, and display. The salt discharge fault early warning system for saline soil areas based on multi-source data is the core business application system running on the salt discharge monitoring platform, relying on the data stream and basic services provided by the platform to realize its early warning function.

[0186] The salt discharge monitoring platform adopts a typical layered architecture of an Internet of Things (IoT) system. The IoT deployment architecture of the system is as follows: Figure 11 As shown, the system mainly includes: the perception layer, the transmission layer, the platform layer, and the presentation layer, involving the main data links from data acquisition to terminal display. The physical deployment topology of the system is as follows: Figure 12 The diagram shows the hardware layout of the data acquisition device 101 in the perception layer, the IoT base station 201 in the transmission layer, the central server 301 in the platform layer, and the display terminal 401 in the display layer.

[0187] The sensing layer, used for data acquisition, includes several data acquisition units 101 installed at various monitoring points. These monitoring points are deployed along the salt drainage pipeline network in the saline soil area. Each monitoring point integrates acquisition devices for temperature, flow rate, conductivity, and pressure head, used to collect multi-source data from the drainage pipeline in real time. The collected multi-source data can be initially aggregated and processed by a microprocessor inside the data acquisition unit 101 before being transmitted via the communication module, or it can be directly transmitted via the communication module of the data acquisition unit 101 for initial aggregation and processing by subsequent modules.

[0188] The transport layer, used for data transmission, includes various communication devices. The data acquisition device 101 wirelessly transmits encrypted monitoring data to the IoT base station 201 via low-power wide-area IoT communication technologies such as NB-IoT and LoRa, and then transmits it back to the remote central server 301 via the Internet or a dedicated network.

[0189] The platform layer, running on central server 301, provides core service functions through the early warning system. Received data is first parsed, correlated, and stored by the preprocessing module, which also manages the topology of the salt drainage system. The prediction module calculates the crystallization probability under current operating conditions based on a model fitted from historical data and real-time data. The operating condition monitoring module compares the salt concentration and crystallization probability with thresholds, providing feedback on normal system operation or system anomaly warning signals. The fault location module, after triggering an early warning event, locates the abnormal point / abnormal section by combining confidence scores.

[0190] The topology map construction unit in the preprocessing module can call data from the resource database; based on the actual GPS coordinates of the monitoring points, pipeline design drawings in "*.dwg" format, and the CGCS2000 / 3° coordinate system, it uses the ArcGIS Pro "CAD to Geodatabase" tool to generate the feature class PipeCenterline and establish a "monitoring point - pipeline segment" mapping table. Figure 9 For example: the pipe segments corresponding to monitoring points MP001L0001 and MP001L0002 are labeled PS001L01; the pipe segments corresponding to monitoring points MP002L0002 and MP002L0003 are labeled PS002L01; and the pipe segments corresponding to monitoring points MP001L0001 and MP002L0002 are labeled PS101. Ultimately, a topology map of the salt drainage system is generated, with monitoring points as nodes and pipe segments as edges, providing a spatial basis for visualization and fault location. The technical approach to constructing this topology map has been applied to pipeline network scenarios in water supply, gas supply, drainage, and petrochemical industries, and will not be elaborated further.

[0191] The prediction module's historical database stores a large number of {temperature, flow rate, brine saturation concentration} data sets acquired from laboratories and / or historical operations. A saturation concentration lookup table is generated through multivariate nonlinear regression fitting. This lookup table is essentially a multidimensional lookup table used to characterize the concentration value at which the brine reaches saturation under different temperature and flow rate conditions. Each {temperature, flow rate, brine saturation concentration} data set represents a historical record and is stored in the SQLite table tbl_sat.

[0192] The main technical approach of the operating condition monitoring module is as follows: It parses the data sent by the preprocessing module, executes a dual threshold comparison subroutine, compares the real-time brine concentration C_real with the brine concentration threshold C_th, and compares the crystallization prediction probability P_real with the crystallization probability threshold P_th. Based on the comparison results, it generates a brine concentration warning level lvl_C and a crystallization probability warning level lvl_P. Then, based on the warning level, it feeds back information indicating normal system operation or a system anomaly warning signal. The brine concentration threshold C_th changes dynamically; the latest value can be retrieved daily at 02:00 by the MySQL event scheduler, or it can be set manually. The crystallization probability threshold P_th has an initial value of 0.15, which can be changed via OTA but remains unchanged by default.

[0193] The comprehensive warning level can be determined using the following code:

[0194] def calc_level(C_real, P_real, C_th, P_th):

[0195] lvl_C = 0 if C_real<= C_th else \ (1 if C_real<= 1.2*C_th else

[0196] (2 if C_real<= 1.5*C_th else 3))

[0197] lvl_P = 0 if P_real<= P_th else \ (1 if P_real<= 1.2*P_th else

[0198] (2 if P_real<= 1.5*P_th else 3))

[0199] return max(lvl_C, lvl_P) #Comprehensive early warning level

[0200] If the overall warning level is "0", it indicates that the system is operating normally, and the normal operation information is fed back to the display module, with the operating condition label value being "0". If the overall warning level is ≥1, the abnormal operation information is fed back to the display module, with the operating condition label value being "1". The time taken for a single calculation is less than 0.2 ms.

[0201] The feedback information includes operating condition tags and monitoring point information. The operating condition tag indicates the type of feedback information: a value of "0" indicates normal operation, while a value of "1" indicates abnormal operation. The associated monitoring point number, sampling time, real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head are packaged together and sent to the display module along with the operating condition tag. The display module then provides a visual representation.

[0202] The presentation layer provides various user interfaces and interaction methods. The presentation terminal 401 can be selected as a large-screen terminal, web terminal, mobile terminal, etc., to display monitoring results in different ways as needed.

[0203] This early warning platform can provide maintenance personnel with comprehensive information from macro system status to micro fault points, supporting them in making efficient maintenance decisions, thus forming a closed loop.

[0204] The other parts of this embodiment are the same as those in the above embodiments, and will not be described again.

[0205] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A salt drainage fault early warning system for saline soil areas based on multi-source data, characterized in that: The system includes a preprocessing module, a prediction module, a working condition monitoring module, and a display module. The preprocessing module is used to acquire measured data including real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head; acquire identification ID composed of the monitoring point number and sampling time; package the measured data of the same monitoring point and the same sampling time with the corresponding identification ID into a measured record; and send it to the prediction module, the operating condition monitoring module, and the display module. Used to construct a topology map of the salt drainage system based on the mapping relationship between monitoring points and spatial coordinate system; The prediction module is used to predict the crystallization probability at each monitoring point of the salt discharge pipe under different operating conditions based on the measured data, saturation concentration lookup table and crystallization prediction analysis formula in the measured record, and to package the associated identification ID and crystallization prediction value into a crystallization prediction result and send it to the operating condition monitoring module. The operating condition monitoring module is used to configure the brine concentration warning level based on the comparison between the real-time brine concentration and the brine concentration threshold, configure the crystallization warning level based on the comparison between the crystallization prediction value and the crystallization probability threshold, analyze and integrate the over-threshold results and the overall warning level, generate feedback information and send it to the display module. The display module is used to visually output the running status based on feedback information; The prediction module includes a historical database and a prediction model; The historical database is used to store saturation concentration lookup tables; the saturation concentration lookup tables contain several historical records, each historical record consisting of a set of associated temperature, brine flow rate, and brine saturation concentration; The prediction model is used to analyze the measured records one by one, extract the identification ID, real-time temperature, real-time brine flow rate, real-time brine concentration, and real-time pressure head, obtain the reference brine saturation concentration according to the saturation concentration lookup table, calculate the saturation coefficient according to the ratio of the real-time brine concentration to the reference brine saturation concentration, calculate the crystallization prediction value according to the crystallization prediction analysis formula, and package the associated identification ID and crystallization prediction value and send them to the operating condition monitoring module. The crystallization prediction and analysis formula is as follows: When S≤1, then Pr=0. When S > 1, then ; Where Pr is the predicted crystallization value, S is the saturation coefficient, and S max This represents the oversaturation limit.

2. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 1, characterized in that: The preprocessing module includes a measured database and a topology graph construction unit; The measured database is used to store measured records; The topology map construction unit is used to retrieve the GIS pipeline map of the salt drainage system. First, it establishes a "monitoring point - pipeline segment" mapping table based on the correlation between the coordinates of the monitoring points in the spatial coordinate system, the salt drainage pipe laying path, the monitoring point number, and the pipeline segment number. Then, it obtains the topology map of the salt drainage system by using the location of the monitoring point as a node and the salt drainage pipe laying path as an edge.

3. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 1, characterized in that: When executing the subroutine of "obtaining reference brine saturation concentration", the prediction model uses a step-by-step matching method or a direct indexing method to obtain a reference historical record that matches the real-time temperature and real-time brine flow rate in the real-time measurement data from the saturation concentration lookup table, and obtains the reference brine saturation concentration based on the reference record. When using the step-by-step matching method, the matching is performed sequentially according to temperature and flow rate. The historical record with the highest matching degree between the saturated concentration and the measured real-time temperature and real-time brine flow rate is found in the saturated concentration lookup table and used as the reference historical record. The average value of the brine saturated concentration in the reference historical record or the interpolation value of the brine saturated concentration in multiple reference historical records is assigned to the reference brine saturated concentration. When using the direct indexing method, the historical records that match the real-time temperature and real-time brine flow rate in the measured records are indexed from the saturation concentration lookup table as reference historical records, and the brine saturation concentration values ​​in the reference historical records are assigned to the reference brine saturation concentration.

4. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 1, characterized in that: The operating condition monitoring module includes a threshold storage unit, a salt solution concentration monitoring unit, a crystallization monitoring unit, an anomaly feedback unit, and an operating database; The threshold storage unit is used to store the salt concentration threshold and the crystallization probability threshold; The brine concentration monitoring unit is used to parse and identify the ID and real-time brine concentration from the measured records, call the brine concentration threshold, determine whether the real-time brine concentration exceeds the brine concentration threshold and output the judgment result A, and also configure the brine concentration warning level according to the relationship between the real-time brine concentration and the brine concentration threshold. The crystallization monitoring unit is used to parse and identify the ID and crystallization prediction value from the crystallization prediction result, call the crystallization probability threshold, determine whether the crystallization prediction value exceeds the crystallization probability threshold and output the judgment result B, and also configure the crystallization probability warning level according to the relationship between the crystallization prediction value and the crystallization probability threshold. The anomaly feedback unit is used to analyze the judgment result A and judgment result B corresponding to the same identification ID. When judgment result A or judgment result B exceeds the threshold, the working condition label is assigned the value "1", and when judgment result A and judgment result B are within the threshold range, the working condition label is assigned the value "0". It is used to generate comprehensive early warning level information based on the highest level among the brine concentration early warning level and the crystallization probability early warning level; it is also used to generate feedback information based on the operating condition label, comprehensive early warning level, brine concentration early warning level, crystallization probability early warning level and associated identification ID, and send it to the display module; feedback information with an operating condition label value of "1" indicates abnormal operation information and associated monitoring point indicates abnormal point, while feedback information with an operating condition label value of "0" indicates normal operation information. The operating database is used to store feedback information.

5. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 4, characterized in that: The salt concentration threshold stored in the threshold storage unit is updated periodically or irregularly using a dynamic update mechanism, while the crystallization probability threshold is a fixed preset value.

6. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 1, characterized in that: It also includes a fault location module; When a fault location module is added to the system, the feedback information generated by the operating condition monitoring module is sent to both the fault location module and the display module. The fault location module is used to parse the received feedback information and filter out the abnormal operation information. Based on the identification ID in the abnormal operation information, it determines the abnormal pipeline segment, and then generates fault location information by combining the confidence model and sends it to the display module.

7. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 6, characterized in that: The fault location module includes a signal analysis unit, a fault point analysis unit, a fault section analysis unit, and a fault location output unit. The signal analysis unit is used to receive feedback information and parse out the monitoring point number, sampling time, operating condition label, comprehensive early warning level, salt solution concentration early warning level, and crystallization probability early warning level corresponding to each identification ID. It filters out abnormal information with the value of "1" for the operating condition label, generates an abnormal point information table, and sends it to the fault point analysis unit. Each piece of information in the abnormal point information table is an abnormal information corresponding to an identification ID, and the monitoring point associated with the abnormal information is an abnormal point. The fault point analysis unit is used to receive the abnormal point information table, obtain the coordinates of the abnormal point, the measured data of the N sampling times before the abnormal point, the feedback information of the N sampling times before the abnormal point, and the measured data of the N sampling times before the upstream and downstream adjacent monitoring points of the abnormal point from the preprocessing module, the prediction module, and the working condition monitoring module, and input all of them into the confidence model. The confidence model is used to calculate the abnormal confidence score of each abnormal point, and the abnormal point with an abnormal confidence score greater than the preset confidence score is marked as a fault point, and the coordinates of the abnormal point are marked as the fault point location. It is also used to store the abnormal confidence score, fault point location, and sampling time of the fault point into the fault point information table, and send it to the fault section analysis unit and the fault location output unit; N is a positive integer. The fault section analysis unit is used to receive the fault point information table, extract the abnormal confidence score and fault point location of each fault point at the same sampling time, analyze the spatial distribution relationship of multiple fault points according to the topology map of the salt discharge system, and aggregate continuously adjacent fault points according to pipeline segments to obtain at least one fault section. It also extracts the coordinates of the fault point at the starting point and the coordinates of the fault point at the ending point of the fault section to obtain the fault section location. The unit is also used to store the fault section location in the fault section information table and send it to the fault location output unit. The fault location output unit is used to receive the fault point information table and the fault section information table, and to package the abnormal confidence score of the fault point, the fault point location, and the fault section location at the same sampling time into a fault location information, store it in the fault information table, and send it to the display module.

8. The early warning system for salt drainage failure in saline soil areas based on multi-source data according to claim 7, characterized in that: The confidence model is used to calculate an anomaly confidence score for each outlier according to the confidence scoring strategy. The confidence scoring strategy includes multiple confidence scoring rules, each confidence scoring rule is matched with a confidence weight, and the anomaly confidence score is obtained by summing the confidence weights corresponding to all confidence scoring rules.

Images