Repair processing method, system, medium and program product based on intelligent customer service

Abstract

This application provides a repair processing method, system, medium, and program product based on intelligent customer service, relating to the field of electrical digital data processing technology. The method includes: obtaining user repair information and determining a set of related repair information, extracting geographic location information and fault phenomenon keywords of the related repair points, and generating a three-level pipe network association map; obtaining water consumption data by time period within the map in real time to determine areas with abnormal water consumption; obtaining hardware status data in the abnormal area and determining the fault type based on pressure fluctuations and sound wave spectra; and generating an intelligent work order based on the fault type, including the fault type, geographic coordinates, pipe network level, and a list of repair tools and accessories. Implementing this method enables intelligent and precise repair processing.

Description

Technical Field

[0001] The present application relates to the field of electronic digital data processing technology, and in particular to a repair processing method, system, medium and program product based on intelligent customer service. Background Art

[0002] In the water service repair reporting scenario, traditional customer service systems only generate work orders based on users' immediate descriptions and can only electronically record repair information. As a result, the work order content only remains on the surface, forcing maintenance personnel to check each work order on site one by one. The handling process is inefficient and lacks data support, making it difficult to meet the needs of smart water services for data-driven precise operation and maintenance.

[0003] To address these shortcomings, existing technologies have proposed a repair report classification method based on keyword matching. By pre-setting keywords such as "water leak," "low water pressure," and "abnormal water quality," user descriptions are automatically categorized into standardized work orders. This technology enables rapid classification of work orders through simple text processing, significantly improving dispatch efficiency.

[0004] However, the core technical problem with existing technologies is that they are essentially passive response models, relying entirely on users' subjective descriptions of fault symptoms and lacking the ability to deeply analyze big data. Upon arrival at the site, maintenance personnel must recollect data to identify the underlying cause, resulting in a poor match between work orders and the actual cause of the fault, and a waste of manpower and time. Especially in complex pipe networks, this dispatching model, without the support of data analysis, makes it difficult to quickly locate the root cause of a fault and accurately allocate maintenance resources. Summary of the Invention

[0005] This application provides a repair processing method, system, medium and program product based on intelligent customer service, which is used to realize intelligent management of the entire process of user repair reporting, improve the efficiency of pipeline fault diagnosis and the accuracy of maintenance scheduling.

[0006] In the first aspect, the present application provides a repair processing method based on intelligent customer service, which is applied to an intelligent customer service system. The method includes: obtaining user repair information and determining an associated repair information set; extracting the geographical location information and fault phenomenon keywords of the associated repair point in the associated repair information set through a natural language processing algorithm; combining the geographical location information, generating a three-level pipe network association map covering all the associated repair points and extending outward at a set distance, the three-level pipe network association map includes a primary associated pipeline, a secondary associated pipeline and a tertiary associated pipeline, the primary associated pipeline is a pipeline directly connected to the repair point, the secondary associated pipeline is a branch pipeline connected to the primary associated pipeline, and the tertiary associated pipeline is a branch pipeline connected to the primary associated pipeline. The secondary associated pipeline is connected to the main pipeline with a diameter greater than or equal to a preset diameter threshold; the time-division water consumption data of each pipeline area in the three-level pipeline network association map is obtained in real time through the intelligent metering equipment; the pipeline area with abnormal water consumption is determined based on the time-division water consumption data; the real-time hardware status data of the pipeline area with abnormal water consumption is obtained, and the hardware status data at least includes continuous pressure fluctuation data, valve torque, switching frequency, pressure curve and acoustic spectrum signal; the fault type is determined based on the continuous pressure fluctuation data and the acoustic spectrum signal; a corresponding intelligent work order is generated according to the fault type, and the work order at least includes the fault type, the geographical coordinates and pipeline network level of the pipeline area with abnormal water consumption, and a list of tools and accessories required for maintenance.

[0007] By employing the above technical solution, we first obtain user repair report information and identify related repair report information sets. Natural language processing algorithms are then used to extract key information from this information, representing a preliminary screening and mining of user repair report big data. A three-level pipeline network correlation map is generated, visually presenting the pipeline network structure. Smart metering devices are used to obtain time-based water consumption data. This data is combined with abnormal areas to identify abnormal areas. Hardware status data is then obtained to determine the fault type and generate a work order. This entire process conducts in-depth analysis of multi-source big data, progressing from repair report information to pipeline network data, water consumption data, and hardware status data. This overcomes the lack of in-depth big data analysis capabilities and enables precise fault location and resolution.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the steps of obtaining user repair information and determining an associated repair information set specifically include: extracting the time data and cell information in the user repair information; extracting the fault keywords in the user repair information; if the similarity of the fault keywords reaches a set threshold, and the cell information is the same or within a set distance range, and the time data is within a set time window, the corresponding user repair information is merged and determined as an associated repair information set.

[0009] By employing this technical solution, when obtaining user repair report information to determine the associated repair report information set, time data, cell information, and fault keywords are extracted. When the fault keyword similarity meets the requirements, the cell information is similar, and the time is within the set window, the related repair reports are merged. This consolidates scattered, potentially related repair report information, avoids duplicate processing, and reduces unnecessary workflows. Furthermore, the integrated information is more comprehensive, providing richer and more accurate data for subsequent analysis of geographic location and fault symptoms, improving the accuracy and efficiency of fault analysis and providing a better understanding of the overall fault situation.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of generating a three-level pipe network association map covering all the associated repair points and extending outward at a set distance in combination with the geographic location information specifically includes: based on the geographic location information, generating a minimum closed irregular space envelope through a minimum circumscribed polygon algorithm; using the minimum closed irregular space envelope as a reference, extending outward at a set distance to generate a standard irregular space envelope; obtaining all pipeline data in the standard irregular space envelope, the pipeline data including at least the pipeline ID, pipeline spatial coordinates and pipe diameter; and constructing a three-level pipe network association map based on the standard irregular space envelope and the pipeline data.

[0011] By adopting the above technical solution, the minimum circumscribed polygon algorithm can accurately frame the scope of associated repair points. The extended standard envelope can cover the surrounding pipelines that may be affected. By obtaining detailed pipeline data, it can accurately construct a map to clearly display the connection and distribution relationship of pipelines at all levels, providing an intuitive and accurate pipeline network model for subsequent analysis of pipeline water usage and fault location.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the step of determining the abnormal water usage pipeline area in combination with the time-divided water consumption data specifically includes: calculating the deviation rate in real time in combination with the time-divided water consumption data; if the deviation rate exceeds a preset deviation rate threshold, determining the pipeline area corresponding to the deviation rate as the abnormal water usage pipeline area.

[0013] By implementing this technical solution, time-of-day water consumption data can reflect actual pipeline usage at different times, and deviation rate calculation can quantify the difference between actual and normal usage. By setting appropriate thresholds, pipeline areas with abnormal water usage can be quickly and accurately identified, providing clear targets for subsequent acquisition of hardware status data and identification of fault types, thereby improving the targetedness and efficiency of troubleshooting.

[0014] In combination with some embodiments of the first aspect, in some embodiments, the step of determining the fault type by combining the continuous pressure fluctuation data and the acoustic wave spectrum signal specifically includes: if the deviation rate is greater than a first deviation rate threshold, and the pressure curve drops by more than the first pressure drop threshold within the set time, and the acoustic wave spectrum is greater than the first frequency within the set continuous time, then the fault type is determined to be a pipeline rupture; if the deviation rate is less than a second deviation rate threshold, and the valve torque exceeds the standard value by a magnitude greater than the first valve torque magnitude, and the switching frequency increases abnormally, then the fault type is determined to be a pipeline blockage; if the deviation rate is less than a third deviation rate threshold, and the pressure curve amplitude is greater than the first pressure curve amplitude, and the proportion of low-frequency components in the acoustic wave spectrum is greater than the proportion of the first acoustic wave spectrum, then the fault type is determined to be a valve failure.

[0015] By adopting the above technical solution, the fault judgment rules are constructed through multi-parameter combination conditions (deviation rate + pressure curve + acoustic spectrum), and the threshold model is trained based on historical data. The technical problems in the existing technology, such as low reliability of single indicator diagnosis, reliance on manual experience in fault type identification, and inability to accurately distinguish similar fault characteristics, are effectively solved. The system achieves the effect of establishing fault feature vectors through joint analysis of multi-dimensional data, automatically matching historical fault patterns, and accurately identifying specific types such as pipeline rupture, blockage, and valve failure, thereby improving the accuracy and intelligence level of fault diagnosis.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of generating a corresponding intelligent work order according to the fault type, it also includes: obtaining current time information; if the time information is within the set working hours, obtaining the maintenance personnel's capability information and positioning information; combining the capability information and positioning information, determining the maintenance personnel who is closest to the geographical coordinates of the abnormal water pipeline area and has matching capabilities according to the fault type.

[0017] By employing these technical solutions, location information prioritizes dispatching personnel closest to the fault point, while capability tags ensure that maintenance skills match the fault type, avoiding inefficiencies caused by remote dispatch or skill mismatches. This process deeply integrates personnel data (location, skills) with fault data, enabling dynamic matching of optimal resources with optimal tasks through real-time calculations, enhancing the real-time decision-making capabilities of big data in resource scheduling.

[0018] In combination with some embodiments of the first aspect, in some embodiments, it also includes: if the time information is outside the set working hours, obtaining the maintenance personnel's ability information and home address information; combining the ability information and home address information, determining the maintenance personnel who is closest to the geographical coordinates of the abnormal water pipeline area and has matching abilities according to the fault type.

[0019] By adopting the above technical solutions, this strategy solves the problem of missing real-time positioning data during non-working hours, uses historically stored address data to continue the principle of spatial priority, and at the same time maintains the professionalism of capability matching, ensuring that fault responses are not disconnected at night or on holidays, and improving the adaptability of big data analysis to complex time scenarios and its full-time coverage capabilities.

[0020] In a second aspect, the present application provides an intelligent customer service system, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the intelligent customer service system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, the present application provides a computer-readable storage medium comprising instructions, which, when executed on an intelligent customer service system, causes the intelligent customer service system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, the present application provides a computer program product, which, when running on an intelligent customer service system, enables the intelligent customer service system to execute the method described in the first aspect and any possible implementation of the first aspect.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0024] 1. By adopting the natural language processing algorithm to extract key data of repair information, construct a three-level pipe network association map to integrate space and pipe attributes, and conduct correlation analysis of multi-source dynamic data (water consumption data by time period, hardware status data), the system effectively solves the technical problems in existing technologies such as fragmentation of user repair data, insufficient visualization of pipe network structure, and reliance on manual experience for fault diagnosis, resulting in low analysis efficiency and poor positioning accuracy. It then achieves the technical effects of automatically mining fault correlation features from multi-dimensional data, constructing a precise pipe network topology model, and realizing intelligent judgment of fault types through data-driven algorithms, significantly improving the automation level of repair processing and fault handling efficiency.

[0025] 2. By adopting the minimum circumscribed polygon algorithm to generate the spatial envelope, setting a distance to expand the analysis range, and constructing a three-level pipeline network hierarchy based on pipeline ID and pipe diameter, the method effectively solves the technical problems of the existing technology, such as the vague pipeline network analysis scope, unclear pipeline association hierarchy, and the separation of spatial data and physical properties. It also achieves the technical effects of converting the geographical distribution of repair points into a computable pipeline network topology model, highlighting the key role of trunk pipelines through hierarchical division, and providing accurate spatial indexes for subsequent data monitoring, making it possible to visualize the pipeline network structure and quickly locate the fault area.

[0026] 3. By adopting the technical means of constructing fault judgment rules based on multi-parameter combination conditions (deviation rate + pressure curve + acoustic wave spectrum) and training threshold models based on historical data, the system effectively solves the technical problems in existing technologies such as low reliability of single-indicator diagnosis, reliance on manual experience in fault type identification, and inability to accurately distinguish similar fault characteristics. It then achieves the technical effects of establishing fault feature vectors through joint analysis of multi-dimensional data, automatically matching historical fault patterns, and accurately identifying specific types of faults such as pipeline rupture / blockage / valve failure, thereby improving the accuracy and intelligence level of fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flowchart of a repair processing method based on intelligent customer service in an embodiment of the present application;

[0028] Figure 2 This is another flowchart of the repair processing method based on intelligent customer service in an embodiment of the present application;

[0029] Figure 3 This is a schematic diagram of the physical device structure of the intelligent customer service system in an embodiment of the present application. DETAILED DESCRIPTION

[0030] The terms used in the following examples of the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in the specification and appended claims of the present application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to and encompasses any or all possible combinations of one or more of the listed items.

[0031] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0032] For ease of understanding, the following describes the process of the method provided by this implementation. Figure 1 , which is a flow chart of the repair processing method based on intelligent customer service in an embodiment of the present application.

[0033] S101, obtaining user repair information and determining a related repair information set;

[0034] User repair report information refers to fault reporting data submitted by users through the intelligent customer service system. It includes information in various forms, such as text descriptions, voice recordings, images, or videos, and is used to describe the pipe network problems encountered by users, such as "Low water pressure in Building 1 of a certain community" or "Water leak downstairs in Unit 3." A related repair report information set is a group of repair reports that share temporal and spatial correlations and similar fault characteristics. By analyzing correlations across dimensions such as time, location, and fault type, scattered individual repair report information is aggregated into a statistically significant set. For example, multiple reports of "water leaks" within a 24-hour period from the same community can be considered a related repair report information set.

[0035] This step occurs when a user initiates a repair request through the intelligent customer service system. The application scenario is the initial data collection and preprocessing phase for water pipe network faults. Specifically, the intelligent customer service system first collects user-submitted repair information in real time through multiple channels, including online chat, voice calls, and the app's repair portal. Each report is then cleaned to remove duplicate or malformed data. Next, the system enters the correlation analysis phase, constructing a three-dimensional screening model based on a sliding time window (e.g., the past hour, 12 hours), geofences (e.g., cell-based, 500-meter radius), and fault keywords (e.g., "leak," "blockage," and "abnormal water pressure"). In the temporal dimension, repair reports submitted within the same time period are included in the candidate set. In the geographic dimension, an address resolution algorithm converts the user's address into longitude and latitude coordinates, calculates spatial distance, and determines whether it falls within the predefined geographic range. In the fault dimension, a keyword matching algorithm is used to calculate the semantic similarity between the repair report content and the predefined fault type. When a repair report meets the requirements of being within a time window, overlapping in geographic scope, and having a fault keyword similarity exceeding a threshold (e.g., 80%), the system merges it with the existing associated set to form a complete set of associated repair report information. This process effectively filters out isolated and invalid data, focusing on high-frequency fault areas and providing a structured data set for subsequent pipe network analysis.

[0036] In some embodiments, determining sets of associated repair report information can be achieved in a variety of ways: Optionally, the system first extracts the timestamp, cell name or address coordinates, and fault keywords (e.g., by matching regular expressions with radicals such as "leak," "block," and "pressure") from each repair report. It then creates a three-dimensional index table based on time, geography, and keywords, hashes all repair reports into buckets, and automatically labels information within the same bucket as a potential associated set. Finally, the DBSCAN density clustering algorithm is used to calculate the density of the data within the buckets, clustering samples with a reachable density into a single associated set. For example, if a cell contains five repair reports containing the keyword "leakage" within three hours, and the address coordinates are less than 200 meters apart, the system will cluster them into a single leak associated set. Optionally, a rule-based association determination method can be used, pre-setting multiple determination rules. For example, "same cell + same fault type + ≥3 repair reports within 24 hours" automatically triggers the creation of an associated set, or "adjacent cells + similar fault keywords + time interval ≤1 hour" is considered a cross-regional associated set. For example, between 9:00 AM and 10:00 AM, adjacent communities A and B reported three reports of "low water pressure" and two reports of "low water flow." Through semantic analysis, the system determined that "low water pressure" and "low water flow" belong to the same fault category, with close temporal and spatial proximity, and merged them into a single water pressure anomaly association set. It's understandable that machine learning-based association prediction models could also be employed, training classifiers (such as random forests and neural networks) using historical association set data. This input takes in the temporal, geographic, and textual features of a new repair report and outputs a probability value that it belongs to an existing association set. These reports are automatically merged when the probability exceeds a threshold, though this is not a limitation here.

[0037] S102, extracting geographical location information and fault phenomenon keywords of the associated repair points in the associated repair information set through a natural language processing algorithm;

[0038] Natural language processing algorithms represent a collection of technologies used to process, understand, and generate human language. They refer to methods that analyze the syntax, semantics, and pragmatics of natural language through computer programs, such as word segmentation, named entity recognition, and keyword extraction. Associated repair points represent the physical location corresponding to each specific repair event in the associated repair information set. These geographic coordinates (such as longitude and latitude) are obtained through address resolution or user location, and are used to identify the specific location of the fault, such as the coordinates of the green belt east of Building 1 in a residential complex. Geographic location information refers to the spatial location data associated with the associated repair point, including the detailed address, administrative division, and geographic coordinates. These data represent the spatial scope of the fault, such as "Building 1, XX Community, XX Street, XX District, XX City" and its corresponding longitude and latitude coordinates. Fault phenomenon keywords are words or phrases that characterize the fault type and characteristics. These keywords are extracted from repair report content through text analysis, such as "water leak," "rapid water flow," and "abnormal pipe noise." These keywords are used for subsequent fault classification and pipe network analysis.

[0039] This step is performed after the associated repair report information set is constructed. It is used in the preprocessing phase to extract key feature data from unstructured text data. Specifically, the intelligent customer service system first performs sentence and word segmentation on each repair report in the associated repair report information set, using word segmentation tools to break down the natural language text into its smallest semantic units (e.g., words). It then proceeds to named entity recognition (NER), using a pretrained model to identify geographic location entities (e.g., neighborhood names, road names, house numbers, etc.) and fault phenomenon entities (e.g., "leakage," "blockage," "water pressure," etc.) in the text. To extract geographic location information, the system first identifies address keywords, such as "XX neighborhood" and "XX Road, No. XX." It then converts the text address into precise latitude and longitude coordinates using an address encoding interface (e.g., the Baidu Maps API) and records the address confidence level (e.g., whether the matching accuracy is at the house number, building, or neighborhood level). To extract fault symptom keywords, a preset algorithm calculates word importance scores, selecting words with scores above a threshold as keywords. These keywords are then normalized using a synonym dictionary. For example, "water pipe leaks" and "pipeline water seepage" are uniformly mapped to the keyword "leakage." Furthermore, the system identifies modifiers in the text (such as "serious," "continuous," and "sudden") for subsequent fault severity assessment. This entire process automatically converts user natural language descriptions into structured geographic coordinates and fault labels, providing fundamental data support for generating pipeline network association maps and fault diagnosis.

[0040] In some embodiments, the extraction of geographic location information and fault phenomenon keywords can be achieved in a variety of ways: optionally, a rule-based template extraction method is adopted, address templates (such as "[administrative district][street][community][building number][unit number]") and fault templates (such as "[fault type]+[degree word]+[location]") are pre-defined, and regular expressions are used to match strings in the text that meet the template pattern.

[0041] S103. Based on the geographic location information, generate a three-level pipe network association map that covers all associated repair points and extends outward at a set distance. The three-level pipe network association map includes primary associated pipes, secondary associated pipes, and tertiary associated pipes. The primary associated pipes are pipes directly connected to the repair points. The secondary associated pipes are branch pipes connected to the primary associated pipes. The tertiary associated pipes are trunk pipes connected to the secondary associated pipes and have a diameter greater than or equal to a preset pipe diameter threshold.

[0042] The geographic location information represents the spatial data, such as the geographic coordinates and detailed address, of the associated repair point extracted in step S102, and is used to locate the specific location of the fault. The three-level pipeline network association map is a multi-layered pipeline network topology model constructed based on geographic information and the pipeline network structure. It graphically displays the spatial associations between the repair point and each level of pipelines, including the hierarchical structure and connection relationships of primary, secondary, and tertiary pipelines. Primary associated pipelines refer to pipelines directly physically connected to the repair point. These are typically branch pipes within a user's home or building, with a smaller diameter (e.g., DN20-DN50), such as those connecting to faucets in a home. Secondary associated pipelines refer to branch pipes connected to primary pipelines. These are typically branch pipes between buildings or within a residential complex, with a medium diameter (e.g., DN80-DN150), such as the water supply branch pipes connecting buildings within a residential complex. Tertiary associated pipelines refer to trunk pipelines connected to secondary pipelines. These are typically main water supply pipes under municipal roads, with a larger diameter (≥ a preset threshold, such as DN200) and primarily responsible for water transport, such as DN300 water supply trunk pipes under urban arterials. The set distance represents the spatial range extending outward from the associated repair point, which is used to determine the analysis range of the map. For example, if the distance is set to 200 meters, it means that the map covers all pipelines within a radius of 200 meters with the repair point as the center.

[0043] This step is performed after obtaining geographic location information and fault phenomenon keywords. It is applied during the construction of a pipe network topology model to support subsequent water usage data monitoring and fault location. Specifically, the intelligent customer service system first uses the minimum circumscribed polygon algorithm based on the geographic coordinates of the associated repair point to generate a minimum closed irregular spatial envelope encompassing all repair points. This envelope tightly encloses all repair points, forming the initial analysis scope. Using this minimum envelope as a baseline, the system then evenly expands outward at set distances (e.g., 100 meters, 200 meters, etc.) to generate a standard irregular spatial envelope, ensuring that it covers the potentially affected pipeline areas surrounding the repair point. Next, the system retrieves data for all pipelines within the standard envelope from the pipe network's Geographic Information System (GIS), including attributes such as pipeline ID, spatial coordinates, diameter, material, and burial depth. Based on the pipeline hierarchy, pipelines directly connected to the repair point are marked as primary associated pipelines, branches connected to the primary pipeline are marked as secondary associated pipelines, and pipelines connected to the secondary pipeline with a diameter ≥ a preset threshold are marked as tertiary associated pipelines. Finally, using a graph database or GIS visualization tool, the pipelines at all levels and their connections are displayed in a map. Nodes represent pipeline nodes (such as valves and water meter wells), and edges represent pipeline connections. Pipelines at different levels are distinguished by different colors or line types (e.g., primary pipelines are blue dashed lines, secondary pipelines are green solid lines, and tertiary pipelines are red thick lines). The map is annotated with the locations of repair points and fault keywords, forming an intuitive three-level pipeline network association map. This map not only shows the spatial distribution of repair points but also reveals their physical connections with pipelines at all levels, providing a spatial index for subsequent time-based water consumption data monitoring and fault diagnosis.

[0044] In some embodiments, generating a three-level pipe network association map can be achieved in a variety of ways: optionally, a rule-based hierarchical division method is adopted, firstly, the connection relationship between the repair point and the pipeline is determined through spatial topology analysis (such as determining whether the repair point is on a certain pipeline through point-line spatial intersection analysis), and the directly connected pipeline is set as the primary; then, the end node of the primary pipeline is used as the starting point, and the downstream pipeline connected to it is searched, and the pipes with a diameter less than a preset threshold are set as secondary, and those with a diameter greater than or equal to the threshold are set as tertiary; repeat this process until all pipelines are traversed.

[0045] S104, using smart metering equipment to obtain real-time time-divided water consumption data for each pipeline area in the three-level pipe network association map;

[0046] Smart metering devices, such as ultrasonic flowmeters, electromagnetic flowmeters, and smart water meters, are IoT devices installed in the pipe network that have data collection, transmission, and processing capabilities. These devices monitor water flow in the pipes in real time. For example, an electromagnetic flowmeter installed on a DN200 main pipe section can collect real-time flow data. Time-of-day water consumption data refers to water consumption data for each pipe area, collected at preset time intervals (such as 15 minutes or 1 hour). This data reflects water consumption patterns and abnormal fluctuations during different time periods. For example, the time-of-day water consumption for a residential branch pipe is 50 cubic meters per hour between 7:00 a.m. and 9:00 a.m. (peak water consumption), and 5 cubic meters per hour at night. Pipeline areas refer to the spatial extents of pipelines at different levels, as delineated according to the three-level pipe network association map. For example, primary-level associated pipes correspond to building branch pipe areas, secondary-level associated pipes correspond to community branch pipe areas, and tertiary-level associated pipes correspond to municipal trunk pipe areas.

[0047] This step is performed after the three-level pipe network association map is generated. Its application scenario is the data collection phase for real-time monitoring of water usage in the pipe network. Specifically, the intelligent customer service system first locates smart metering devices (such as flow meters and water meters) installed on each pipe based on the pipe ID and spatial coordinates in the three-level pipe network association map. A real-time data connection is established via IoT communication protocols such as LoRa and NB-IoT. Flow data from each device is then periodically collected at a preset time granularity (e.g., every 15 minutes), and water consumption for the corresponding pipe area is calculated based on pipe diameter and length. For pipe areas without direct metering devices (such as secondary branches), the system indirectly calculates water consumption by taking the difference between the upstream main pipe flow data and the downstream branch flow data. For example, if the main pipe flow is 100 cubic meters per hour and the sum of the known downstream branch flow rates is 80 cubic meters per hour, the water consumption for the intermediate branch area is 20 cubic meters per hour. After denoising (e.g., removing out-of-range data and correcting for sensor errors), the collected data is stored in a time series database by pipe area and timestamp, forming a time-based water consumption dataset. This dataset is not only used to monitor current water usage in real time, but can also be used as historical data to train water usage pattern models and provide a benchmark for anomaly detection.

[0048] S105. Determine the pipeline area with abnormal water consumption based on the water consumption data of each time period;

[0049] The time-based water consumption data represents the statistical data on water consumption for each pipeline area during different time periods, obtained in step S104. It is used to reflect normal water consumption patterns and abnormal fluctuations. An abnormal water consumption pipeline area refers to a pipeline area where water consumption deviates from the normal fluctuation range, including situations where water consumption suddenly increases or decreases, or is continuously above or below a threshold. For example, if water consumption in a trunk pipeline area during off-peak hours increases by 30% compared to the same period historically, this may indicate a pipeline leak or illegal water use.

[0050] This step is performed after the collection of time-based water consumption data is complete, and its application scenario is the pipeline network anomaly detection stage based on historical and real-time data. Specifically, the intelligent customer service system first establishes a water consumption baseline model for each pipeline area. By analyzing historical time-based data (such as the past 30 days), it calculates the average water consumption, standard deviation, and fluctuation range for each period, and determines the threshold range for normal water consumption (such as the average value ± 2 times the standard deviation). For the real-time time-based data, the system calculates the deviation rate of current water consumption from the baseline model for each period, using the formula: Deviation rate = (real-time water consumption - average water consumption) / average water consumption × 100%. When the deviation rate exceeds a preset threshold (such as ±15%), an anomaly warning is triggered, and the corresponding pipeline area is marked as a potential anomaly area. To avoid misjudgments, the system also performs multi-dimensional verification: ① Time dimension verification, checking whether anomalies occur for multiple consecutive periods (e.g., the deviation rate exceeds the threshold for two consecutive periods); ② Spatial dimension verification, analyzing whether water usage data in adjacent pipeline areas is synchronously abnormal (e.g., when a trunk pipeline anomaly occurs, checking whether flow anomalies in downstream branches occur simultaneously); and ③ Business logic verification, adjusting thresholds based on external factors such as holidays and weather (e.g., increasing the peak water usage threshold by 10% during high summer temperatures). After these multiple layers of verification, the pipeline area with abnormal water usage is ultimately identified and highlighted in red on the three-level pipeline network association map, providing a clear target for subsequent fault diagnosis.

[0051] In some embodiments, identifying pipeline areas with abnormal water usage can be achieved in a variety of ways: Optionally, a statistically based threshold detection method can be used. First, historical time-based data for each pipeline area is tested for normality. If it conforms to a normal distribution, the anomaly threshold is set using the 3σ principle (mean ± 3 standard deviations). If it does not conform to a normal distribution, a quantile method (such as the 95th percentile) is used to set the upper threshold and the 5th percentile is used to set the lower threshold. For example, if the historical hourly water consumption data for a branch pipe in a residential area shows a skewed distribution, with a 95th percentile of 60 cubic meters / hour and a 5th percentile of 10 cubic meters / hour, periods where real-time water consumption exceeds 60 cubic meters or falls below 10 cubic meters are considered anomalies. Optionally, an LSTM anomaly detection model from machine learning can be used. By training the model to learn the characteristic representations of normal water usage patterns, an anomaly score is calculated for the real-time data, and an anomaly is determined when the score exceeds a threshold. For example, if an LSTM model is trained on a time-divided water consumption sequence for a trunk pipeline, and the model predicts that the next moment's water consumption will be 80 cubic meters per hour, while the actual collected value is 120 cubic meters per hour, the anomaly score exceeds the threshold, indicating that the water consumption in that area is abnormal. It is understood that dynamic thresholds can also be set based on expert experience, such as temporarily raising the water consumption threshold for a specific area during a fire drill to avoid false alarms. This is not limited here.

[0052] S106: Acquire real-time hardware status data of the abnormal water use pipeline area, where the hardware status data includes at least continuous pressure fluctuation data, valve torque, switching frequency, pressure curve, and acoustic spectrum signal;

[0053] Real-time hardware status data represents equipment operating parameters collected in real time by sensors installed in pipelines with abnormal water use. It reflects the physical status of the pipeline and its associated equipment (such as valves and pumps). Continuous pressure fluctuation data refers to the water pressure variation data within the pipeline continuously collected by pressure sensors. It is recorded in a time series format (e.g., one pressure value per second) and is used to detect abnormal pressure fluctuations (e.g., a sudden drop in pressure may indicate a pipeline rupture). Valve torque, measured by a torque sensor, is the torque required to open and close the valve and is used to determine whether the valve is stuck or damaged (e.g., a sudden increase in torque may indicate a foreign object blocking the valve body). Opening and closing frequency, recorded by a counter, is the number of times a valve opens and closes per unit time. It is used to monitor abnormal valve operating frequency (e.g., frequent opening and closing may cause valve disc wear). A pressure curve is a continuous curve plotted with time on the horizontal axis and pressure on the vertical axis. It is used to visually display pressure trends (e.g., a continuous drop in pressure may indicate a leak). Acoustic spectrum signals are acoustic signals generated by fluid or mechanical vibration within the pipeline, collected by acoustic sensors. The frequency distribution data, obtained through Fourier transform, is used to analyze vibration characteristics (e.g., high-frequency sound waves may indicate a pipeline leak, while low-frequency sound waves may indicate a valve malfunction).

[0054] S107, determining the fault type by combining the continuous pressure fluctuation data and the acoustic wave spectrum signal;

[0055] Among them, continuous pressure fluctuation data represents the continuous sequence data of the water pressure in the pipeline changing with time, collected in real time by the pressure sensor, which is used to reflect the dynamic change characteristics of the pressure. The acoustic spectrum signal refers to the frequency distribution data of the vibration signal in the pipeline collected by the acoustic sensor and processed by Fourier transform. It is used to characterize the fluid state in the pipeline or the operating characteristics of the mechanical components. Different fault types will correspond to specific frequency components (for example, pipeline rupture is often accompanied by high-frequency sound waves, and valve failures are mostly manifested as low-frequency vibrations). The fault type refers to the specific pipeline network fault category determined based on the characteristics of pressure and acoustic wave data. This application mainly includes three types: pipeline rupture, pipeline network blockage, and valve failure. Each type corresponds to a unique multi-parameter combination feature.

[0056] This step is executed after obtaining the real-time hardware status data of the abnormal water-using pipeline area, and the application scenario is the intelligent identification stage of the fault type based on multi-parameter data analysis. Specifically, the intelligent customer service system first pre-processes the continuous pressure fluctuation data, removes high-frequency noise through sliding average filtering, and calculates characteristic parameters such as the slope of the pressure curve (i.e., the pressure drop rate), amplitude, and duration. At the same time, the frequency domain analysis of the acoustic spectrum signal is performed to extract the amplitude and energy proportion of the main frequency and harmonic components, and identify the characteristic frequency range (such as the high frequency band 2000-4000Hz and the low frequency band 50-200Hz). Then, according to the preset fault judgment rule library, the pressure and acoustic wave features are jointly matched:

[0057] Pipeline rupture: A pipeline rupture is detected when the pressure curve drops by more than a first pressure drop threshold (e.g., 0.1 MPa) within a set timeframe (e.g., 5 minutes), the deviation rate of continuous pressure fluctuation data exceeds a first deviation rate threshold (e.g., 20%), and the high-frequency component (>1500 Hz) of the acoustic spectrum exceeds a first frequency threshold (e.g., 50 dB) for a continuous period (e.g., 3 minutes). High-frequency acoustic waves are generated by water jets striking the pipe wall, and the sudden pressure drop reflects pressure loss caused by fluid leakage.

[0058] Pipeline blockage: When the pressure curve amplitude is small but consistently low, the deviation rate is less than the second deviation rate threshold (e.g., -10%), the valve torque exceeds the standard value by a greater margin than the first valve torque amplitude (e.g., 30%), and the opening and closing frequency increases abnormally (e.g., twice the daily average), combined with a low high-frequency component and a slight increase in low-frequency noise in the acoustic spectrum, a pipe blockage is identified. Blockage increases flow resistance, requiring greater torque to open the valve. The increased opening and closing frequency may be due to frequent unclogging attempts.

[0059] Valve failure: A valve failure is identified when the pressure curve exhibits periodic fluctuations (such as pressure oscillations caused by pump startup and shutdown), the amplitude is greater than the amplitude of the first pressure curve (e.g., 0.05 MPa), the low-frequency component (<200 Hz) in the acoustic spectrum accounts for a greater proportion than the first acoustic spectrum (e.g., 40%), and the deviation rate is less than a third deviation rate threshold (e.g., ±5%). Low-frequency vibration is often caused by valve component wear, sticking, or poor sealing. Pressure fluctuations are related to the valve's opening and closing cycle.

[0060] The system uses fuzzy logic algorithms to comprehensively assess multiple parameters, allowing for a certain degree of parameter tolerance (e.g., ±5% for a pressure drop threshold) to avoid misjudgments caused by fluctuations in a single parameter. For example, if the pressure in a pipeline area drops by 0.09 MPa (close to the 0.1 MPa threshold), but the high-frequency energy of the acoustic wave continuously exceeds 55 dB and the deviation rate reaches 22%, it is still determined to be a pipeline rupture.

[0061] S108. Generate a corresponding intelligent work order according to the fault type. The work order at least includes the fault type, the geographical coordinates and pipe network level of the abnormal water-using pipeline area, and a list of tools and accessories required for maintenance.

[0062] Among them, the fault type indicates the specific fault category determined by step S107 (such as pipe rupture, pipe network blockage, valve failure), which is used to guide maintenance personnel to select targeted treatment plans. Geographic coordinates refer to the precise location of the abnormal water use pipeline area on the map (such as longitude and latitude), which is obtained through address resolution or GPS positioning to ensure that maintenance personnel can arrive at the scene quickly. The pipeline network level refers to the level (primary, secondary, tertiary) of the faulty pipeline in the three-level pipeline network association map, reflecting the importance of the pipeline and the priority of maintenance (such as the failure of the third-level trunk pipeline needs to be handled first). The list of tools and accessories required for maintenance refers to a list of tools and parts pre-configured according to the fault type. For example, if the pipeline ruptures, welding equipment, leak repair fixtures and pipes of the corresponding diameter are required, and if the valve fails, wrenches, sealing rings and spare valves are required.

[0063] This step is executed after the fault type is determined, and is applicable during the maintenance task generation and resource scheduling phase. Specifically, the intelligent customer service system first calls the corresponding work order template based on the fault type and automatically fills in the following information:

[0064] Fault type: directly reference the judgment result of S107, such as "pipeline rupture".

[0065] Geographic coordinates and pipe network hierarchy: Extract the center coordinates (e.g., 116.4810°E, 39.9219°N) and hierarchy labels (e.g., "third-level trunk pipe") of the water use anomaly area from the three-level pipe network association map, and mark the fault point on the map.

[0066] Repair tools and accessories list: Associate fault types with tool accessories through a rule mapping table, for example:

[0067] Tools corresponding to pipe rupture: pipe cutting machine, electric welder, pressure pump; accessories: DN200 steel pipe, leak-proof glue, sealing ring.

[0068] Tools corresponding to pipe blockage: pipe dredge, endoscope; accessories: dredge spring, filter.

[0069] Tools corresponding to valve failure: socket wrench, torque wrench; accessories: DN150 valve, sealing packing.

[0070] Furthermore, the work order automatically generates a link to a 3D pipe network model of the fault site, historical maintenance records (if available), and safety instructions for maintenance personnel to reference. Once a work order is generated, the system automatically assigns it to the nearest person with matching skills based on the current time (working hours / off-hours) and the maintenance personnel's status (on duty / on vacation). The system also notifies the person of the order via SMS or app push notifications.

[0071] In the embodiment of the present application, due to the technical means of using a natural language processing algorithm to extract key data of repair information, construct a three-level pipe network association map to integrate space and pipe attributes, and conduct correlation analysis of multi-source dynamic data (water consumption data by time period, hardware status data), the technical problems of fragmentation of user repair data, insufficient visualization of pipe network structure, and reliance on manual experience in fault diagnosis resulting in low analysis efficiency and poor positioning accuracy in the existing technology are effectively solved. The technical effects of automatically mining fault correlation features from multi-dimensional data, constructing an accurate pipe network topology model, and realizing intelligent judgment of fault types through data-driven algorithms are achieved, which significantly improves the automation level of repair processing and fault handling efficiency.

[0072] In some embodiments, after intelligent work order processing is completed, a series of quality control operations are performed to comprehensively and objectively measure the performance of the intelligent customer service system and take appropriate measures based on the evaluation results. First, the intelligent customer service system obtains user voice information. After the user completes the repair request and the work order is processed, the intelligent customer service system automatically collects voice recordings of the user's interaction with the intelligent customer service system. These voice messages contain direct user feedback on the repair request processing, which may include evaluations of the problem resolution and feelings about the customer service communication style. Next, the system uses automatic speech recognition technology to convert the voice information into text. The core principle of this technology is to analyze and process voice signals using acoustic models, language models, and other methods to convert sound signals into text that computers can understand and process. For example, if a user says, "Your repair speed is too slow. I've been waiting for a long time and no one has come to handle it," the speech recognition technology will accurately convert this into corresponding text, providing a basis for subsequent analysis.

[0073] The converted text information is then input into a large customer service quality inspection model. This model is constructed through deep learning, and multiple sensitive word sets and sentiment word sets with quality inspection scores were used during pre-training. The sensitive word set includes words that may indicate problems with the service, such as "complaint" and "too bad", while the sentiment word set covers various words that express emotions, such as "happy" and "angry". By learning from a large amount of such labeled data, the model can analyze the input text information, judge the performance of the intelligent customer service in handling repair reports from multiple dimensions such as vocabulary usage and sentence structure, and then give a preliminary quality inspection score. For example, if negative sensitive words appear multiple times in the text message, the model will give a lower preliminary quality inspection score accordingly.

[0074] At the same time, the system also uses emotion recognition models, combined with voice information, to determine the user's emotional state regarding the intelligent customer service handling of the repair request. This emotion recognition model is also based on deep learning, trained on multiple speech data sets annotated with emotion information. It can identify emotional features in speech, such as intonation, speaking speed, and volume changes, to determine the user's emotional state. It can identify at least three emotions: anger, satisfaction, and neutrality. If the user speaks in an excited tone, quickly, and loudly, the emotion recognition model is likely to identify anger. If the user speaks in a calm, relaxed tone, the model is likely to identify satisfaction or neutrality.

[0075] Next, the preliminary quality inspection score is adjusted based on the results of the emotion recognition model. If the emotion information indicates anger, it indicates that the user is extremely dissatisfied with the intelligent customer service. In this case, the preliminary quality inspection score will be deducted according to the preset deduction value to determine the final quality inspection score. For example, if the preliminary quality inspection score is 80 points and the preset deduction value is 20 points, the final quality inspection score will be 60 points. If the emotion information indicates satisfaction, it indicates that the user approves of the service. The preliminary quality inspection score will be added according to the preset bonus value to determine the final quality inspection score. For example, if the preliminary score is 80 points and the preset bonus value is 10 points, the final score will be 90 points. If the emotion information is neutral, it means that the user has no obvious preference or dislike for the service, and the preliminary quality inspection score will be directly determined as the final quality inspection score.

[0076] Finally, the final quality inspection score is compared with the set passing score threshold. If the final quality inspection score falls below the passing score, it indicates that the service quality of the intelligent customer service does not meet the standard, and the system triggers human customer service intervention. With their professional knowledge and rich experience, human customer service can communicate with users in a more in-depth and personalized manner, further understand their needs and dissatisfactions, and promptly resolve their issues, thus compensating for the shortcomings of the intelligent customer service, improving user satisfaction, and ensuring the high-quality completion of repair service.

[0077] After combining the above content, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , which is another flow chart of the repair processing method based on intelligent customer service in an embodiment of the present application.

[0078] S201, based on the geographic location information, generating a minimum closed irregular space envelope by a minimum circumscribed polygon algorithm;

[0079] The geographic location information represents the latitude and longitude coordinates or detailed address data of the associated repair points extracted through step S102, and is used to locate the set of spatial locations where the fault occurred, such as the coordinate point set {(x1, y1), (x2, y2), ..., (xn, yn)} corresponding to multiple repair points. The minimum circumscribed polygon algorithm refers to an algorithm that generates a minimum closed polygon containing all points by calculating the convex hull or concave hull of a plane point set. The minimum closed irregular spatial envelope refers to an irregular polygonal area generated by the algorithm that tightly encloses all associated repair points. The number of sides and shape of the area are determined by the distribution of the repair points. For example, when the repair points are linearly distributed, a narrow polygon is generated, and when they are dispersed, an irregular convex polygon is generated.

[0080] This step is executed after obtaining the geographic location information of the associated repair points, and the application scenario is the preliminary definition stage of the pipeline network analysis scope. Specifically, the intelligent customer service system first converts each repair point in the associated repair information set into geographic coordinates to form a coordinate point set. Then, the minimum circumscribed polygon algorithm (such as the convex hull algorithm based on the Qhull library) is called to process the coordinate point set: the algorithm traverses all points, finds the leftmost point as the starting point, and selects edge points in a counterclockwise direction to ensure that all repair points are located inside or on the edge of the polygon. Finally, the minimum closed polygon composed of a vertex sequence is generated. The envelope line is close to the edge of the repair point distribution and redundant space is eliminated. For example, if five repair points in a certain community are distributed in an L shape, the convex hull algorithm will generate a pentagon containing these five points, accurately reflecting the spatial boundary of the fault concentration area. The generated polygon is stored in GeoJSON format, containing attributes such as vertex coordinates, area, and perimeter, providing a benchmark for subsequent expansion of the analysis scope.

[0081] S202, using the minimum closed irregular space envelope as a reference, extending outward by a set distance to generate a standard irregular space envelope;

[0082] The set distance represents a preset spatial extension range based on the requirements of the pipeline network analysis. It is used to include potentially affected pipeline areas around the repair point in the analysis. The standard irregular spatial envelope is a polygonal area formed by uniformly expanding the set distance from the minimum envelope. Its shape is similar to the minimum envelope but its range is larger. For example, the original minimum envelope is a pentagon, which is expanded to a pentagon with expanded sides to ensure coverage of the pipeline network around the repair point.

[0083] This step is executed after the minimum closed irregular spatial envelope is generated, and its application scenario is the stage of expanding the pipeline network analysis scope to cover potential fault-related areas. Specifically, the intelligent customer service system first performs parallel extrapolation on each side of the minimum envelope, with a set distance, and rounds the vertices of the polygon (to avoid calculation errors caused by sharp angles) to generate an expanded polygon. For example, if the original minimum envelope is a square with a side length of 100 meters and the set distance is 50 meters, it will be expanded to a square with a side length of 200 meters, and the area will be expanded to 4 times the original. The expanded envelope is implemented through spatial topological operations (such as buffer analysis) to ensure that all spaces within the set distance from the repair point are covered. The generated standard envelope is used for subsequent pipeline data retrieval, such as querying all pipelines within the envelope from the GIS system, to ensure that the analysis scope neither misses related pipelines nor is over-expanded to cause data redundancy.

[0084] S203, obtaining all pipeline data in the standard irregular space envelope, where the pipeline data at least includes pipeline ID, pipeline spatial coordinates, and pipe diameter;

[0085] Among them, pipeline data refers to pipeline attribute data stored in the pipeline network geographic information system (GIS), which is used to describe the physical characteristics and spatial location of the pipeline. Pipeline ID refers to the code that uniquely identifies each pipeline, such as "DN200-01-001", which is used for data association and tracking. Pipeline spatial coordinates refer to the latitude and longitude coordinates of the starting and ending points of the pipeline, such as the starting point (116.4810°E, 39.9219°N) and the end point (116.4820°E, 39.9225°N), which are used to draw the direction of the pipeline on the map. Pipe diameter refers to the nominal diameter of the pipeline (such as DN50, DN200), which is used to distinguish pipeline levels and water transmission capacity. Generally, the larger the pipe diameter, the higher the pipeline level (such as the diameter of the tertiary trunk pipe ≥ DN200).

[0086] This step is performed after the standard irregular spatial envelope is generated, and its application scenario is the stage of retrieving pipeline data for the target area from the GIS system. Specifically, the intelligent customer service system sends a request to the GIS database through the spatial query interface to retrieve all pipeline records that are completely or partially within the standard envelope. The query conditions include: the pipeline geometry intersects with the envelope polygon, and attributes such as pipeline ID, starting point coordinates, end point coordinates, pipe diameter, material, and burial depth are returned. For example, if a standard envelope covers an area of 1 square kilometer, the query results may include 100 primary pipelines (DN20-DN50), 50 secondary pipelines (DN80-DN150), and 20 tertiary pipelines (DN200-DN400). After deduplication and format conversion, the acquired data is stored as a structured dataset, such as a CSV file or JSON array, for subsequent map construction.

[0087] S204. Construct a three-level pipe network association map based on the standard irregular spatial envelope and pipe data.

[0088] The three-level pipe network association map is a pipe network topology model with a standard envelope as its spatial scope and a pipeline hierarchy as its structure. It is used to visualize the association between repair points and pipelines at all levels. Primary associated pipelines refer to branch pipes directly connected to the repair point, secondary associated pipelines refer to branch pipes connected to primary pipelines, and tertiary associated pipelines refer to main pipelines connected to secondary pipelines (pipe diameter ≥ a preset threshold). The map represents pipeline connection relationships through nodes and edges. For example, nodes represent valves and water meter wells, and edges represent pipelines. Different levels of pipelines are distinguished by different colors or line types (e.g., primary pipelines are blue dashed lines, tertiary pipelines are red solid lines).

[0089] This step is performed after obtaining the standard envelope and pipeline data, and is applied during the visualization modeling phase of the pipeline network topology. Specifically, the intelligent customer service system first divides the pipeline hierarchy based on pipe diameter and connection relationships. The pipeline closest to the repair point's coordinates is designated as the primary associated pipeline (e.g., the branch pipe at the user's home). The pipeline connection table (which stores upstream and downstream pipeline nodes) is then used to recursively search for secondary pipelines (upstream branches of the primary pipeline) and tertiary pipelines (upstream main pipelines of the secondary pipeline). A graph database or GIS visualization tool is then used to construct a map: each pipeline is represented as an edge, with nodes as pipeline endpoints, and attributes such as pipeline ID, diameter, and hierarchy are annotated. The standard envelope polygon and repair point coordinates are overlaid on the map, and the fault location is marked with an icon (e.g., a water drop icon represents a leak). For example, if a repair point is located on a DN50 branch pipe (primary), which connects to a DN100 branch pipe (secondary), which in turn connects to a DN300 main pipe (tertiary), the map displays the three-level pipeline hierarchy and annotates the branch pipe location with a repair icon.

[0090] In the embodiment of the present application, since the minimum circumscribed polygon algorithm is used to generate the minimum closed irregular space envelope, and the standard irregular space envelope is extended outward at a set distance, the problem of vague pipeline network analysis range and inability to accurately define the fault-related area in the existing technology is effectively solved, thereby achieving the technical effect of accurately framing the distribution range of the repair point through the geometric algorithm and covering the potentially affected pipelines through external expansion, providing a clear spatial benchmark for subsequent pipeline data retrieval and hierarchical division.

[0091] In some embodiments, after generating a corresponding smart work order based on the fault type, the process further includes: obtaining current time information; if the time information is within the set working hours, obtaining maintenance personnel's capability information and location information; and combining the capability information and location information to determine, based on the fault type, the maintenance personnel with the closest geographic coordinates to the area of the abnormal water pipe and matching their capabilities. The current time information represents the system time obtained in real time by the intelligent customer service system and is used to determine whether it is within the set working hours (e.g., Monday to Friday, 8:00 AM - 6:00 PM). Maintenance personnel capability information refers to a collection of maintenance personnel's skill tags, generated from historical maintenance records, training certifications, and other data, representing the maintenance personnel's professional skill range, such as "pipeline welding," "valve maintenance," and "GIS operation." Location information refers to the maintenance personnel's real-time geographic coordinates obtained via GPS, Beidou, or IoT base stations, and is used to calculate the spatial distance from the fault point. For example, a maintenance personnel's current coordinates are (116.4815°E, 39.9220°N). Geographic coordinates refer to the central longitude and latitude of the area where the water pipeline is experiencing abnormal water use. These coordinates are obtained from a three-level pipe network map. For example, the coordinates of the fault area are (116.4800°E, 39.9200°E). This step is performed after a smart work order is generated, and its application scenario is the real-time scheduling of maintenance resources. Specifically, the intelligent customer service system first obtains the current system time (e.g., 10:30 AM, April 26, 2025) and determines whether it falls within the set working hours (e.g., 8:00 AM to 6:00 PM). If it is within working hours, the system obtains the real-time location information of all on-duty maintenance personnel through a mobile app or vehicle-mounted positioning device and retrieves their capability tags from the personnel management database (e.g., Maintenance Personnel A's capability tags are "Pipeline Rupture Repair" and "Third-Level Pipeline Network Maintenance"). The system then selects a subset of maintenance personnel with corresponding capability tags based on the fault type (e.g., "Pipeline Rupture"), for example, filtering out personnel with capability tags including "Pipeline Welding" and "Leakage Repair." Next, a spatial distance calculation algorithm is used to calculate the straight-line distance between each qualified maintenance personnel and the fault point, prioritizing the personnel closest to the fault point. If there are multiple personnel at the same distance, the system further compares their current workload (such as the number of outstanding work orders) and selects the one with the lowest workload. For example, if maintenance worker B is 1.2 kilometers from the fault, has the "Pipeline Rupture Repair" tag, and currently has no pending work orders, the system will identify them as the best candidate and push a task notification through the work order system.

[0092] In some embodiments, if the time information falls outside of set working hours, the maintenance personnel's capability and home address information are obtained. Based on this capability and home address information, the maintenance personnel closest to the geographic coordinates of the water pipeline area with the abnormality and matching their capabilities are identified based on the fault type. "Outside of set working hours" refers to times outside the preset working hours, such as nighttime (6:00 PM - 8:00 AM the following day), weekends, or holidays. Home address information refers to the maintenance personnel's permanent address coordinates registered in the system and is used in place of real-time location data. Geographic coordinates are also the center coordinates of the fault area, maintaining spatial reference consistency. This step is performed after a smart work order is generated and the system determines that the current time falls outside of working hours (e.g., 10:00 PM on April 26, 2025). Specifically, the intelligent customer service system first confirms that the current time is outside of working hours and then retrieves the maintenance personnel's home address information from the personnel database (which must have been collected and encrypted in advance). Then, based on the fault type, a subset of personnel with corresponding capability tags is screened (e.g., "valve failure" requires screening those with the "valve maintenance" and "seal replacement" tag holders), and then the distance between each person's home address and the fault point is calculated using a preset formula, and the person closest is selected. For example, maintenance personnel D's home address is 3.5 kilometers away from the fault point, his capability tag includes "valve torque detection," and he is on standby during non-working hours. The system will identify him as a priority for scheduling. In addition, the system must also consider the standby status of maintenance personnel (such as "responsive" and "unresponsive"), and confirm whether he can accept orders through double verification via SMS or phone calls to avoid task delays due to poor communication.

[0093] The following describes the intelligent customer service system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of the intelligent customer service system in an embodiment of the present application.

[0094] It should be noted that Figure 3 The structure of the intelligent customer service system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0095] like Figure 3As shown, the intelligent customer service system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes, such as the methods described in the above embodiments, based on programs stored in a read-only memory (ROM) 302 or programs loaded from a storage unit 308 into a random access memory (RAM) 303. RAM 303 also stores various programs and data required for system operation. CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0096] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, push button switches, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, indicator lights, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the removable media can be installed in the storage section 308 as needed.

[0097] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from removable media 311. When executed by the central processing unit (CPU) 301, the computer program performs the various functions defined in the present invention.

[0098] It should be noted that specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0099] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0100] Specifically, the intelligent customer service system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the repair processing method based on intelligent customer service provided in the above embodiment is implemented.

[0101] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the intelligent customer service system described in the above embodiments, or may exist independently and not be incorporated into the intelligent customer service system. The storage medium carries one or more computer programs, which, when executed by a processor of the intelligent customer service system, enable the intelligent customer service system to implement the intelligent customer service-based repair processing method provided in the above embodiments.

[0102] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0103] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0104] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A repair processing method based on intelligent customer service, applied to an intelligent customer service system, characterized in that: The method comprises: obtaining user repair information and determining a related repair information set; Extracting geographical location information and fault phenomenon keywords of the associated repair points in the associated repair information set through a natural language processing algorithm; In combination with the geographic location information, a three-level pipe network association map is generated, covering all the associated repair points and extending outward at a set distance. The three-level pipe network association map includes primary associated pipes, secondary associated pipes, and tertiary associated pipes. The primary associated pipes are pipes directly connected to the repair points, the secondary associated pipes are branch pipes connected to the primary associated pipes, and the tertiary associated pipes are main pipes connected to the secondary associated pipes and having a pipe diameter greater than or equal to a preset pipe diameter threshold. Obtaining time-divided water consumption data of each pipeline area in the three-level pipe network association map in real time through smart metering equipment; Determine the pipeline area with abnormal water consumption based on the water consumption data of each time period; Acquire real-time hardware status data of the abnormal water use pipeline area, wherein the real-time hardware status data at least includes continuous pressure fluctuation data, valve torque, switching frequency, pressure curve and acoustic spectrum signal; Determining the fault type based on the real-time hardware status data; Generate a corresponding intelligent work order according to the fault type, the work order at least including the fault type, the geographical coordinates and pipe network level of the abnormal water pipe area, and a list of tools and accessories required for repair; The step of determining the fault type in combination with the real-time hardware status data specifically includes: If the deviation rate is greater than a first deviation rate threshold, and the pressure curve decreases by more than a first pressure drop threshold within a set time, and the acoustic wave spectrum is greater than a first frequency within a set continuous time, then the fault type is determined to be a pipeline rupture; If the deviation rate is less than the second deviation rate threshold, and the magnitude by which the valve torque exceeds the standard value is greater than the first valve torque magnitude, and the switching frequency increases abnormally, then the fault type is determined to be pipe network blockage; If the deviation rate is less than a third deviation rate threshold, the pressure curve amplitude is greater than the first pressure curve amplitude, and the proportion of low-frequency components in the sound wave spectrum is greater than the proportion of the first sound wave spectrum, then the fault type is determined to be a valve fault.

2. The method according to claim 1, characterized in that The steps of obtaining user repair information and determining the associated repair information set specifically include: Extracting time data and cell information from the user repair report information; Extracting fault keywords from the user repair report information; If the similarity of the fault keywords reaches a set similarity threshold, and the cell information is the same or within a set distance range, and the time data is within a set time range, the corresponding user repair information will be merged and determined as an associated repair information set.

3. The method according to claim 1, characterized in that The step of generating a three-level pipe network association map covering all the associated repair points and extending outward at a set distance in combination with the geographic location information specifically includes: Based on the geographic location information, a minimum closed irregular space envelope is generated by a minimum circumscribed polygon algorithm; Taking the minimum closed irregular space envelope as a reference, extending outward by a set extension distance to generate a standard irregular space envelope; Acquire all pipeline data in the standard irregular space envelope, wherein the pipeline data at least includes pipeline ID, pipeline spatial coordinates, and pipe diameter; A three-level pipe network association map is constructed according to the standard irregular spatial envelope and the pipe data.

4. The method according to claim 1, wherein The step of determining the pipeline area with abnormal water consumption based on the time-division water consumption data specifically includes: Calculate the deviation rate in real time based on the water consumption data of the time periods; If the deviation rate exceeds a preset deviation rate threshold, the pipeline area corresponding to the deviation rate is determined to be an abnormal water use pipeline area.

5. The method according to claim 1, wherein After the step of generating a corresponding intelligent work order according to the fault type, the method further includes: Get current time information; If the time information is within the set working hours, then obtain the maintenance personnel's capability information and location information; Combining the capability information and the location information, determining, based on the fault type, a maintenance personnel whose geographic coordinates are closest to the abnormal water pipe area and whose capability matches the area; If the time information is outside the set working hours, then obtaining the maintenance personnel's capability information and home address information; In combination with the capability information and the home address information, a maintenance personnel whose geographical distance to the abnormal water pipe area and whose capability matches the area is determined according to the fault type.

6. The method according to claim 1, characterized in that After the step of generating a corresponding intelligent work order according to the fault type, the method further includes: Acquire user's voice information and convert the voice information into text information through automatic speech recognition technology; Input the text information into a customer service quality inspection model to determine a preliminary quality inspection score. The customer service quality inspection model is constructed in advance through deep learning based on multiple sensitive word sets and emotional word sets annotated with quality inspection scores; Determining the user's emotional response to the intelligent customer service handling of the repair request based on the voice information using an emotion recognition model, wherein the emotion recognition model is constructed through deep learning using multiple voice information sets annotated with emotion information, wherein the emotion information includes at least anger, satisfaction, and neutrality; If the emotional information is anger, then the initial quality inspection score will be deducted according to the preset deduction value to determine the final quality inspection score; If the emotional information is satisfactory, then based on the preliminary quality inspection score, additional points are added according to the preset additional points value to determine the final quality inspection score; If the emotional information is neutral, the preliminary quality inspection score is determined as the final quality inspection score; If the final quality inspection score is lower than the set passing score threshold, manual customer service intervention will be triggered.

7. An intelligent customer service system, characterized in that: The intelligent customer service system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the intelligent customer service system to execute the method described in any one of claims 1-6.

8. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on the intelligent customer service system, the intelligent customer service system executes the method according to any one of claims 1 to 6.

9. A computer program product, characterized in that When the computer program product is run on an intelligent customer service system, the intelligent customer service system is enabled to execute the method according to any one of claims 1 to 6.

Images