Municipal pipe network health condition analysis method and system based on computer analysis

Abstract

The invention relates to the technical field of pipe network management, and provides a municipal pipe network health condition analysis method and system based on computer analysis. The method comprises the following steps: determining a first acquisition time length and a second acquisition time length based on inspection data and sensor monitoring data of a municipal pipe network area, and respectively acquiring corresponding first road monitoring data and first network evaluation data based on the first acquisition time length and the second acquisition time length; a feature extraction network is used for extracting a first health condition feature from the inspection data and the sensor monitoring data, extracting a second health condition feature from the first road monitoring data and the first network evaluation data, and obtaining a third health condition feature through feature fusion; and analyzing and predicting the third health condition characteristics by using a health assessment model to obtain a first pipe network health condition level of the municipal pipe network area. The analysis accuracy of the health state of the municipal pipe network area is improved through the multi-source data.

Description

Technical Field

[0001] The present invention relates to the technical field of pipe network management. Specifically, it relates to a method and system for analyzing the health status of municipal pipe networks based on computer analysis. Background Art

[0002] With the rapid advancement of urbanization, the scale of the municipal pipe network system has been continuously expanding and its structure has become increasingly complex. The municipal pipe network covers various types of pipelines such as water supply, drainage, gas, and heat, and its health status is directly related to the normal operation of the city and the quality of life of residents. Traditional methods for detecting the health status of municipal pipe networks, such as manual inspections and simple physical detections, have problems such as low efficiency, limited detection scope, and difficulty in discovering potential hidden dangers. Moreover, for a large and complex pipe network system, relying solely on the method of manually analyzing data not only consumes a large amount of manpower and time but also easily leads to omissions and errors.

[0003] In recent years, with the rapid development of computer technology, new ways have been provided for analyzing the health status of municipal pipe networks. By utilizing the powerful data processing and analysis capabilities of computers and combining advanced algorithms and models, real-time monitoring and efficient analysis of pipe network operation data can be carried out, thereby more accurately evaluating the health status of the pipe network and promptly discovering potential faults and safety hazards. However, this method currently mainly monitors the pipe network operation data based on direct data such as manual inspection data and sensor monitoring data. Due to reasons such as the efficiency of manual inspections and the density of sensor installations, these direct data generally have problems of insufficient data, resulting in insufficient accuracy of the pipe network health status obtained through computer analysis. Summary of the Invention

[0004] In order to at least solve the technical problems existing in the above background art, the present invention provides a method, system, electronic device, storage medium, and computer program product for analyzing the health status of municipal pipe networks based on computer analysis.

[0005] The present invention provides a method for analyzing the health status of municipal pipe networks based on computer analysis, and the method includes the following steps:

[0006] Receiving the inspection data and sensor monitoring data of the municipal pipe network area, determining a first acquisition duration and a second acquisition duration based on the inspection data and the sensor monitoring data, and respectively acquiring first road monitoring data and first network evaluation data corresponding to the municipal pipe network area based on the first acquisition duration and the second acquisition duration;

[0007] Using a first feature extraction network to extract first health status features from the inspection data and the sensor monitoring data, and using a second feature extraction network to extract second health status features from the first road monitoring data and the first network evaluation data;

[0008] Fuse the first health status feature and the second health status feature to obtain a third health status feature, and use a health assessment model based on Transformer to analyze and predict the third health status feature to obtain the first pipe network health status level of the municipal pipe network area.

[0009] Optionally, determining the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data includes:

[0010] Determine the second pipe network health status level of the municipal pipe network area based on the inspection data, the sensor monitoring data, and a simple evaluation rule;

[0011] If the second pipe network health status level is higher than the first preset level or lower than the second preset level, set the first acquisition duration to a first value and the second acquisition duration to a second value; wherein, the first value is less than the second value.

[0012] Optionally, determining the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data further includes:

[0013] If the second pipe network health status level is lower than the first preset level and higher than the second preset level, obtain the second road monitoring data and the second network evaluation data that are in the same time period as the inspection data or the sensor monitoring data;

[0014] Use a semantic analysis model to extract first health semantic information from the second road monitoring data and second health semantic information from the second network evaluation data, and calculate the semantic similarity between the first health semantic information and the second health semantic information;

[0015] Determine that the first acquisition duration is a third value and the second acquisition duration is a fourth value according to the semantic similarity.

[0016] Optionally, fusing the first health status feature and the second health status feature to obtain a third health status feature includes:

[0017] Perform feature clustering on the first health status feature and the second health status feature respectively to obtain a refined third health status feature and a fourth health status feature respectively;

[0018] Concatenate and fuse the third health status feature and the fourth health status feature to form a new feature vector, which is the third health status feature.

[0019] Optionally, performing feature clustering on the first health status feature and the second health status feature respectively to obtain a refined third health status feature and a fourth health status feature respectively, including:

[0020] Retrieving a first clustering intensity corresponding to the first health status feature and a second clustering intensity corresponding to the second health status feature; wherein, the first clustering intensity is less than the second clustering intensity;

[0021] Clustering the first health status feature based on the first clustering intensity, and clustering the second health status feature based on the second clustering intensity.

[0022] The present invention also provides a computer - based analysis system for the health status of municipal pipe networks, characterized in that the system includes an acquisition unit, a feature extraction unit, and a prediction unit;

[0023] The acquisition unit is configured to receive inspection data and sensor monitoring data of a municipal pipe network area, determine a first acquisition duration and a second acquisition duration based on the inspection data and the sensor monitoring data, and respectively acquire first road monitoring data and first network evaluation data corresponding to the municipal pipe network area based on the first acquisition duration and the second acquisition duration;

[0024] The feature extraction unit is configured to extract a first health status feature from the inspection data and the sensor monitoring data using a first feature extraction network, and extract a second health status feature from the first road monitoring data and the first network evaluation data using a second feature extraction network;

[0025] The prediction unit is configured to perform feature fusion on the first health status feature and the second health status feature to obtain a third health status feature, and use a health assessment model based on Transformer to analyze and predict the third health status feature to obtain a first pipe network health status level of the municipal pipe network area.

[0026] Optionally, the acquisition unit is specifically configured to:

[0027] Determine a second pipe network health status level of the municipal pipe network area based on the inspection data, the sensor monitoring data, and a simple evaluation rule;

[0028] If the second pipe network health status level is higher than a first preset level or lower than a second preset level, set the first acquisition duration to a first value and the second acquisition duration to a second value; wherein, the first value is less than the second value.

[0029] Optionally, the acquisition unit is specifically configured to:

[0030] Determining the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data further includes:

[0031] If the health status level of the second pipe network is lower than the first preset level and higher than the second preset level, then obtain the second road monitoring data and the second network evaluation data that belong to the same time period as the inspection data or the sensor monitoring data;

[0032] Use a semantic analysis model to extract the first health semantic information from the second road monitoring data, and extract the second health semantic information from the second network evaluation data, and calculate the semantic similarity between the first health semantic information and the second health semantic information;

[0033] Determine that the first acquisition duration is a third value and the second acquisition duration is a fourth value according to the semantic similarity.

[0034] Optionally, the prediction unit is specifically configured to:

[0035] Perform feature clustering on the first health status feature and the second health status feature respectively to obtain a refined third health status feature and a fourth health status feature respectively;

[0036] Concatenate and fuse the third health status feature and the fourth health status feature to form a new feature vector, that is, the third health status feature.

[0037] Optionally, the prediction unit is specifically configured to:

[0038] Retrieve the first clustering intensity corresponding to the first health status feature and the second clustering intensity corresponding to the second health status feature; wherein, the first clustering intensity is less than the second clustering intensity;

[0039] Perform clustering on the first health status feature based on the first clustering intensity, and perform clustering on the second health status feature based on the second clustering intensity.

[0040] The present invention also provides an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method as described in any one of the preceding items.

[0041] The present invention also provides a storage medium, on which a computer program is stored, and when the computer program is run by a processor, it executes the method as described in any one of the preceding items.

[0042] The present invention also provides a computer program product, which, when run by a processor, causes the processor to execute the method described in any of the previous items.

[0043] The beneficial effects of the present invention are as follows: Based on direct data such as inspection data and sensor monitoring data, the present invention also comprehensively analyzes the health status of the municipal pipe network area based on indirect data such as road monitoring data and network evaluation data. This not only solves the problem of insufficient data for computer analysis of the health status but also improves the accuracy of analyzing the health status of the municipal pipe network area through multi-source data. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0045] Figure 1 is a schematic flowchart of a method for analyzing the health status of a municipal pipe network based on computer analysis disclosed in an embodiment of the present invention;

[0046] Figure 2 is a schematic structural diagram of a second feature extraction network disclosed in an embodiment of the present invention;

[0047] Figure 3 is a schematic structural diagram of a system for analyzing the health status of a municipal pipe network based on computer analysis disclosed in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The following describes exemplary embodiments of the present disclosure, including various details of the embodiments of the present disclosure to assist understanding. It should be considered that they are merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, the description below omits the description of well-known functions and structures.

[0049] In the field of analyzing the health status of municipal pipe networks, current computer analysis mainly relies on direct data such as manual inspection data and sensor monitoring data to monitor the operation data of the pipe network. However, due to low manual inspection efficiency, insufficient sensor layout density, etc., data is insufficient, which in turn affects the accuracy of the pipe network health status obtained by computer analysis.

[0050] For the above technical problems, refer to Figure 1As shown in the figure, an embodiment of the present invention discloses a method for analyzing the health status of municipal pipe networks based on computer analysis. The method includes the following steps:

[0051] S10, Receive the inspection data and sensor monitoring data of the municipal pipe network area, determine the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data, and acquire the first road monitoring data and the first network evaluation data corresponding to the municipal pipe network area respectively based on the first acquisition duration and the second acquisition duration.

[0052] Among them: After receiving the traditional inspection data of the municipal pipe network area (the inspection results of the pipe network uploaded by inspectors, such as health ratings, abnormal conditions, etc.) and direct data such as sensor (pressure sensor, video sensor, flow sensor, etc.) monitoring data, indirect data that can be used to assist in analyzing the health status of the pipe network in the municipal pipe network area is then acquired. Specifically, based on the pipe network health-related status indicated by the inspection data and the sensor monitoring data, the first acquisition duration of the first road monitoring data and the second acquisition duration of the first network evaluation data are determined.

[0053] The road monitoring data mainly refers to the data detected by various monitoring cameras deployed on urban roads (including traffic monitoring cameras, public security monitoring cameras, etc.). By docking with the urban traffic management department and other departments, the above data can be acquired. Information related to the health status of the pipe network in the municipal pipe network area may be obtained from the road monitoring data. For example, abnormal conditions such as water accumulation, collapse, uplift, and abnormal deceleration of vehicle driving on the road surface may imply problems such as water leakage, rupture, or deformation of the underground pipe network. By analyzing the video data and using object detection algorithms to identify these abnormal features, it provides important clues for evaluating the health status of the pipe network.

[0054] The network evaluation data refers to the evaluation, complaint, and feedback information related to each municipal pipe network area published by citizens on various network platforms, such as social media, the citizen message board of the government official website, and specialized urban service APPs. Citizens are more likely to discover some pipe network problems in their daily lives, such as strange smells and unexplained noises. These information are reflected through the network evaluation data, which helps to discover potential pipe network failures.

[0055] In view of the fact that the first road monitoring data and the first network evaluation data may contain more information that is irrelevant or repetitive to the health status of the municipal pipe network, it is necessary to determine the appropriate acquisition duration for the above indirect data, which will be specifically explained in the following content.

[0056] S20, use the first feature extraction network to extract and obtain the first health status feature from the inspection data and the sensor monitoring data, and use the second feature extraction network to extract and obtain the second health status feature from the first road monitoring data and the first network evaluation data.

[0057] Among them: Different feature extraction networks are adopted in the present invention to extract features that can be used to characterize the health status of the municipal pipe network area from direct data (inspection data and sensor monitoring data) and indirect data (road monitoring data and network evaluation data) respectively.

[0058] Specifically, the first feature extraction network can be a convolutional neural network (CNN), and an object detection algorithm is used to identify and extract features related to abnormal conditions such as pipe rupture and valve leakage from the direct data. As Figure 2 shown, the second feature extraction network can be a composite convolutional neural network embedded with a word vector model. The composite convolutional neural network itself can be used to extract features related to abnormal conditions such as road surface collapse and waterlogging from the road monitoring data. At the same time, the composite convolutional neural network can combine the word vector model to extract relevant features from the network evaluation data, that is, the word vector model uses natural language processing technology to extract keyword features in the feedback from citizens, such as "leakage", "strange smell", etc.

[0059] S30, fuse the first health status feature and the second health status feature to obtain a third health status feature, and use a health assessment model based on Transformer to analyze and predict the third health status feature to obtain the first pipe network health status level of the municipal pipe network area.

[0060] Among them: The first health status feature and the second health status feature extracted above are fused to obtain a more comprehensive third health status feature, and then it is analyzed and predicted by a health assessment model based on Transformer, so as to obtain the first pipe network health status level. The first pipe network health status level is, for example, excellent, good, normal, general, poor, etc.

[0061] The health assessment model based on Transformer is mainly composed of parts such as multi-head attention mechanism (Multi-HeadAttention), feed-forward neural network (Feed-Forward Neural Network), layer normalization (LayerNormalization) and residual connection (Residual Connection). Specifically, it contains multiple parallel attention heads, and each head independently calculates the attention distribution of the input third health status feature, that is, generates three vectors of query, key, and value according to the input.

[0062] The feedforward neural network consists of fully connected layers, including an input layer, a hidden layer, and an output layer. After receiving the above vectors output by the multi-head attention mechanism, the feedforward neural network will perform further feature transformation and mapping on them. The feature transformation is a non-linear transformation, which can enhance the expressive ability of the model through non-linear transformation, so as to obtain a more discriminative feature representation.

[0063] The layer normalization operation involves each layer of the health assessment model, calculates the mean and variance of each sample in each dimension, and then normalizes the data to unify the features into a similar distribution range, avoiding difficulties in model training caused by excessive differences in data distribution, and improving the training efficiency and accuracy of the model.

[0064] The residual connection is to directly add the input of a certain layer of the model to the output of this layer to form a new output. In the health assessment model, the residual connection runs through between various modules. Through the residual connection, the model can better retain and transmit the original feature information, avoid losing important information during the multi-layer feature extraction and transformation process, and ensure the accurate assessment of the health status of the pipe network by the model.

[0065] Optionally, determining the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data includes:

[0066] Determine the second pipe network health status level of this municipal pipe network area based on the inspection data, the sensor monitoring data, and the simple evaluation rule;

[0067] If the second pipe network health status level is higher than the first preset level or lower than the second preset level, set the first acquisition duration to a first value and the second acquisition duration to a second value; where the first value is less than the second value.

[0068] In this embodiment, in order to reduce the interference of irrelevant data on the analysis of the health status of the pipe network, it is necessary to pre-determine the size of the acquisition time span of the first road monitoring data and the first network evaluation data. Specifically:

[0069] First, analyze and process the collected direct data, namely inspection data and sensor monitoring data, and obtain the current second pipe network health status level of this municipal pipe network area through an evaluation method or model based on a simple evaluation rule. The simple evaluation rule here is, for example, a quantitative analysis of the intensity and quantity of positive and negative evaluation information in the inspection data, as well as a quantitative analysis of whether there are abnormal data beyond the normal range, their quantity, and the duration in the sensor monitoring data. Based on the quantitative analysis results, the second pipe network health status level is obtained. Among them, the quantitative analysis is a part of the simple evaluation rule, and the simple evaluation rule should at least also include the pipe network health status levels corresponding to different quantitative analysis results.

[0070] The above-mentioned second pipeline network health status level is only an indicator representing the health status of the pipeline network quickly obtained from the perspective of direct data. However, the pipeline network health evaluation rules are usually fuzzy and not specific, resulting in a large deviation between the evaluation of the pipeline network health status in the inspection data uploaded by the inspection personnel and the actual situation; in addition, the normal range is mostly determined based on empirical values, and there are situations where it is too strict or too loose. The above-mentioned various factors may cause a deviation between the second pipeline network health status level obtained by the simple evaluation rule and the actual health status of the pipeline network.

[0071] In view of the above technical problems, the present invention sets a first preset level (such as better) representing the upper threshold of relatively good pipeline network health status and a second preset level (such as average) representing the lower threshold of relatively poor pipeline network health status. When the second pipeline network health status level is higher than the first preset level or lower than the second preset level, it indicates that the health status of the municipal pipeline network area is very clear, that is, the confidence level of the above-mentioned obtained second pipeline network health status level is high enough. At this time, the first acquisition duration and the second acquisition duration are directly set to preset fixed values. Among them, since when there are health problems in the pipeline network, it is easier and earlier to detect from the road monitoring video, while the feedback from the public on the pipeline network health problems is usually later, generally when the pipeline network health problems have occurred for a period of time and have not been reasonably maintained and solved, relevant problems will be reported on the Internet. Therefore, the present invention sets the above-mentioned first value (such as 1 day) to be less than the second value (such as 3 days), which is beneficial to fully obtain the road monitoring data and network evaluation data that can be used to analyze the pipeline network health status, and at the same time can effectively reduce the number of irrelevant data to improve the analysis efficiency.

[0072] Optionally, determining the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data further includes:

[0073] If the second pipeline network health status level is lower than the first preset level and higher than the second preset level, then obtain the second road monitoring data and the second network evaluation data that belong to the same time period as the inspection data or the sensor monitoring data;

[0074] Use a semantic analysis model to extract the first health semantic information from the second road monitoring data, and extract the second health semantic information from the second network evaluation data, and calculate the semantic similarity between the first health semantic information and the second health semantic information;

[0075] Determine that the first acquisition duration is a third value and the second acquisition duration is a fourth value according to the semantic similarity.

[0076] In this embodiment, in contrast to the foregoing situation, when the health status level of the second pipe network is between the first preset level and the second preset level, that is, in the intermediate range, it indicates that the health status of the pipe network is neither relatively good nor extremely worrying. There are certain potential risks, but the situation is not clear yet. Moreover, the probability that the obtained health status level of the second pipe network is affected by the wrong evaluation of the inspection personnel (due to the vague and non-specific pipe network health evaluation rules, which lead to cognitive errors of the inspection personnel) is greater, resulting in its confidence level not being high enough. In this case, it is not appropriate to use the above-mentioned preset and fixed first value and second value.

[0077] To address the above problems, the present invention is configured to obtain second road monitoring data and second network evaluation data that belong to the same time period (such as one day or half a day) as the inspection data or sensor monitoring data. The picture descriptions in the road monitoring data, the feedback statements of citizens in the network evaluation data, etc. all contain semantic information about the health of the pipe network. The present invention uses a semantic analysis model to extract first health semantic information and second health semantic information from the second road monitoring data and the second network evaluation data respectively, and both are semantic information that can be quantitatively analyzed. Then, the semantic similarity between the two is calculated. The semantic similarity can reflect the degree of consistency of the descriptions of the pipe network health status by the above two data sources. If the similarity is high, it means that the descriptions of the pipe network health status by the above two data sources are relatively consistent, and the data credibility is high; on the contrary, a low similarity indicates that there are differences between the above two data sources, and further analysis and judgment are required.

[0078] Specifically, if the semantic similarity is high, it means that the multi-source data can be mutually verified, the health status of the pipe network is relatively clear, the acquisition duration can be appropriately shortened, or the second road monitoring data and the second network evaluation data can be directly used, thereby improving the analysis efficiency. If the semantic similarity is low, it indicates that there are contradictions or inconsistencies in the data, and more comprehensive data needs to be collected for analysis. Therefore, the acquisition duration is appropriately extended to obtain more information to accurately evaluate the health status of the pipe network and ensure that potential problems are not missed. Similarly to the foregoing, the third value of the first acquisition duration (such as 1 day before and after the direct data indicates an abnormality, a total of 3 days) should also be less than the fourth value of the second acquisition duration (such as 2 days before and after the direct data indicates an abnormality, a total of 5 days).

[0079] Among them, the semantic similarity can be calculated based on the method of word vectors. For example, using Word2Vec to convert the first health semantic information and the second health semantic information into vectors respectively, and then calculating the similarity between the two vectors through the cosine similarity formula. Of course, other methods can also be used, such as the method based on deep learning, which will not be elaborated here.

[0080] Optionally, the feature fusion of the first health status feature and the second health status feature to obtain a third health status feature includes:

[0081] Perform feature clustering on the first health status feature and the second health status feature respectively to obtain a refined third health status feature and a fourth health status feature respectively;

[0082] Concatenate and fuse the third health status feature and the fourth health status feature to form a new feature vector, which is the third health status feature.

[0083] In this embodiment, due to the significant characteristics of complexity and variability in both road traffic conditions and network evaluation content, there will be a lot of duplicate content and pseudo-correlated content (i.e., content unrelated to the health status of the pipe network) in the first health status feature and the second health status feature obtained by the aforementioned extraction. It is necessary to remove these contents as much as possible to improve the confidence level of the first pipe network health status level predicted by the health assessment model.

[0084] Specifically, perform feature clustering on the first health status feature and the second health status feature respectively. For example, for the first health status feature, use a clustering algorithm (such as the K-Means algorithm) to divide it into different clusters according to the similarity between features. For example, for features such as abnormal pressure fluctuation features and pipe crack position features, features with similar pressure fluctuation amplitudes and crack severities are grouped into the same cluster. The same operation is also performed on the second health status feature. For example, features such as road surface water accumulation area features and concentrated areas of odor complaints are clustered according to dimensions such as water accumulation area size and odor complaint frequency. Through clustering, a large number of complex features can be refined and classified, highlighting key features and making the data more representative.

[0085] Concatenate and fuse the refined third health status feature (from the clustering result of the first health status feature) and the fourth health status feature (from the clustering result of the second health status feature) obtained after clustering, that is, connect two feature sets representing the key features of different data sources in a certain order to form a new feature vector, which is the final third health status feature.

[0086] The features after separate clustering in the present invention can better highlight the core information of their respective data sources. After concatenation and fusion, they can comprehensively reflect the health status of the municipal pipe network presented from different perspectives, provide higher-quality and more targeted inputs for the subsequent health assessment model, and improve the accuracy and reliability of the assessment of the pipe network health status.

[0087] Optionally, the performing feature clustering on the first health status feature and the second health status feature respectively to obtain a refined third health status feature and a fourth health status feature respectively includes:

[0088] Retrieve the first clustering intensity corresponding to the first health status feature and the second clustering intensity corresponding to the second health status feature; wherein, the first clustering intensity is less than the second clustering intensity;

[0089] Cluster the first health status feature based on the first clustering intensity and cluster the second health status feature based on the second clustering intensity.

[0090] In this embodiment, compared with indirect data, direct data contains less redundant data and irrelevant data, and the structural characteristics of direct data are more obvious. Therefore, its effective data is usually more than indirect data. Therefore, the present invention sets different clustering intensities when clustering the first health status feature and the second health status feature. Specifically, the first clustering intensity is set to be less than the second clustering intensity.

[0091] For example, taking the K-Means algorithm as an example, the clustering intensity is determined by adjusting the number of clustering clusters K and / or the distance metric.

[0092] Adjustment of the number of clustering clusters K: The value of K directly determines the granularity of clustering, that is, the clustering intensity. In the new feature fusion scheme, the setting of the value of K determines how many categories the features are divided into. When the value of K is small, the clustering intensity is low, the number of clusters obtained is small, and the feature differences within each cluster are large, which means that more features with different degrees will be classified into one category. When clustering the first health status feature, if K is set to 2, the features of abnormal pressure fluctuations and pipeline crack positions can be simply divided into two categories: "relatively normal" and "abnormal". When the value of K increases, the clustering intensity increases, the number of clusters increases, and the features within each cluster are more similar, and the classification is more detailed. For example, when K is set to 5, according to factors such as the different degrees of abnormal pressure, the different sizes and positions of cracks, etc., the features are divided into 5 more specific categories, which helps to analyze the features more accurately.

[0093] Selection of distance metric: The K-Means algorithm determines the distance between feature points and the clustering center based on the distance metric, and then determines the cluster to which the feature points belong. Common distance metrics include Euclidean distance, Manhattan distance, etc.

[0094] The Euclidean distance calculates the straight-line distance between two points in space and can better reflect the similarity between features when the feature distribution is relatively uniform; the Manhattan distance calculates the sum of the absolute axis distances of two points on the standard coordinate system, and for some features with specific directions or distribution rules, it may better reflect their similarity. Selecting different distance metrics will result in different assignments of feature points during the clustering process, thereby affecting the clustering intensity. Specifically, when clustering the second health status features, if the features of the road surface water accumulation area and the concentrated area of odor complaints are relatively dispersed in spatial distribution, using the Euclidean distance for clustering may make the clustering result pay more attention to the overall spatial relationship, while using the Manhattan distance may focus more on the differences in certain specific directions of the features, resulting in different intensities of clustering results. The clustering intensity corresponding to different distance metrics can be determined in advance through measured data, so as to select the appropriate distance metric according to the required clustering intensity.

[0095] As Figure 3 shown, an embodiment of the present invention also discloses a computer analysis-based municipal pipe network health status analysis system, which is characterized in that the system includes an acquisition unit, a feature extraction unit, and a prediction unit;

[0096] The acquisition unit is configured to receive the inspection data and sensor monitoring data of the municipal pipe network area, determine a first acquisition duration and a second acquisition duration based on the inspection data and the sensor monitoring data, and respectively acquire first road monitoring data and first network evaluation data corresponding to the municipal pipe network area based on the first acquisition duration and the second acquisition duration;

[0097] The feature extraction unit is configured to extract first health status features from the inspection data and the sensor monitoring data using a first feature extraction network, and extract second health status features from the first road monitoring data and the first network evaluation data using a second feature extraction network;

[0098] The prediction unit is configured to perform feature fusion on the first health status features and the second health status features to obtain third health status features, and use a health assessment model based on Transformer to analyze and predict the third health status features to obtain the first pipe network health status level of the municipal pipe network area.

[0099] An embodiment of the present invention also discloses an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method as described in the foregoing embodiment.

[0100] An embodiment of the present invention also discloses a storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the method described in the foregoing embodiment.

[0101] An embodiment of the present invention also discloses a computer program product. When the computer program product is run by a processor, it causes the processor to execute the method described in the foregoing embodiment.

[0102] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0103] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0104] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0106] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the scope of the claims. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these changes and modifications therein.

Claims

1. A method for analyzing the health status of municipal pipe networks based on computer analysis, characterized in that, The method includes the following steps: Receiving inspection data and sensor monitoring data of a municipal pipe network area, determining a first acquisition duration and a second acquisition duration based on the inspection data and the sensor monitoring data, and acquiring first road monitoring data and first network evaluation data corresponding to the municipal pipe network area based on the first acquisition duration and the second acquisition duration respectively; Extracting first health status features from the inspection data and the sensor monitoring data using a first feature extraction network, and extracting second health status features from the first road monitoring data and the first network evaluation data using a second feature extraction network; Performing feature fusion on the first health status features and the second health status features to obtain third health status features, and analyzing and predicting the third health status features using a health assessment model based on Transformer to obtain a first pipe network health status level of the municipal pipe network area.

2. The method for analyzing the health status of municipal pipe networks based on computer analysis according to claim 1, wherein: The determining of the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data includes: Determining a second pipe network health status level of the municipal pipe network area based on the inspection data, the sensor monitoring data, and a simple evaluation rule; If the second pipe network health status level is higher than a first preset level or lower than a second preset level, setting the first acquisition duration to a first value and the second acquisition duration to a second value; wherein, the first value is less than the second value.

3. The method for analyzing the health status of municipal pipe networks based on computer analysis according to claim 2, wherein: The determining of the first acquisition duration and the second acquisition duration based on the inspection data and the sensor monitoring data further includes: If the second pipe network health status level is lower than the first preset level and higher than the second preset level, acquiring second road monitoring data and second network evaluation data that are in the same time period as the inspection data or the sensor monitoring data; Extracting first health semantic information from the second road monitoring data and second health semantic information from the second network evaluation data using a semantic analysis model, and calculating the semantic similarity between the first health semantic information and the second health semantic information; Determining the first acquisition duration as a third value and the second acquisition duration as a fourth value according to the semantic similarity.

4. A method for analyzing the health status of municipal pipe networks based on computer analysis according to claim 1, characterized in that: The performing of feature fusion on the first health status features and the second health status features to obtain third health status features includes: Performing feature clustering on the first health status features and the second health status features respectively to obtain a refined third health status feature and a refined fourth health status feature respectively; Concatenating and fusing the third health status feature and the fourth health status feature to form a new feature vector, i.e., the third health status feature.

5. The method for analyzing the health status of municipal pipe networks based on computer analysis according to claim 4, wherein: The performing of feature clustering on the first health status features and the second health status features respectively to obtain a refined third health status feature and a refined fourth health status feature respectively includes: Retrieve a first clustering intensity corresponding to the first health status feature and a second clustering intensity corresponding to the second health status feature; wherein, the first clustering intensity is less than the second clustering intensity; Cluster the first health status feature based on the first clustering intensity and cluster the second health status feature based on the second clustering intensity.

6. A computer - based analysis system for the health status analysis of municipal pipe networks, characterized in that, The system includes an acquisition unit, a feature extraction unit, and a prediction unit; The acquisition unit is configured to receive inspection data and sensor monitoring data of a municipal pipe network area, determine a first acquisition duration and a second acquisition duration based on the inspection data and the sensor monitoring data, and acquire first road monitoring data and first network evaluation data corresponding to the municipal pipe network area based on the first acquisition duration and the second acquisition duration respectively; The feature extraction unit is configured to extract a first health status feature from the inspection data and the sensor monitoring data using a first feature extraction network, and extract a second health status feature from the first road monitoring data and the first network evaluation data using a second feature extraction network; The prediction unit is configured to perform feature fusion on the first health status feature and the second health status feature to obtain a third health status feature, and analyze and predict the third health status feature using a health assessment model based on Transformer to obtain a first pipe network health status level of the municipal pipe network area.

7. A computer - based analysis system for the health status analysis of municipal pipe networks according to claim 1, characterized in that: The acquisition unit is specifically configured to: Determine a second pipe network health status level of the municipal pipe network area based on the inspection data, the sensor monitoring data, and a simple evaluation rule; If the second pipe network health status level is higher than a first preset level or lower than a second preset level, set the first acquisition duration to a first value and the second acquisition duration to a second value; wherein, the first value is less than the second value.

8. An electronic device, comprising: A memory storing executable program code; A processor coupled to the memory; characterized in that: the processor calls the executable program code stored in the memory and executes the method according to any one of claims 1-5.

9. A storage medium, on which a computer program is stored, characterized in that: The computer program, when run by the processor, executes the method according to any one of claims 1-5.

10. A computer program product, when the computer program product is run by the processor, causes the processor to execute the method according to any one of claims 1-5.

Images