Data fusion-based blow filling pipeline damage early warning method and system and readable medium

Abstract

The application discloses a data fusion-based warning method and system for damage of a blow-fill pipeline and a readable medium, and belongs to the technical field of data processing of the blow-fill pipeline. The method comprises the following steps: collecting damage data and influence factor data on the blow-fill pipeline and storing the data in association; constructing a damage prediction model according to a data association algorithm, analyzing pipeline damage distribution rules and influence factor diffusion characteristics, obtaining expected influence factor data along the extension direction of the blow-fill pipeline in combination with the influence factor diffusion characteristics, respectively generating two groups of expected damage data according to the damage distribution rules and the damage prediction model, and then obtaining theoretical damage data through weighted algorithm fusion, and outputting warning information based on the theoretical damage data. Through the above scheme, the damage conditions of each point of the pipeline can be quickly evaluated, a targeted inspection scheme can be provided, the inspection efficiency can be effectively improved, and the stable and efficient development of the blow-fill construction can be ensured.

Description

Technical Field

[0001] This application belongs to the field of data processing technology for dredged and filled pipelines, and relates to a method, system and readable medium for early warning of damage to dredged and filled pipelines based on data fusion. Background Technology

[0002] Numerous dredging and remediation projects are undertaken along the coast to improve water environments and maintain ecosystems, requiring the laying of a large number of reclamation pipelines. These pipelines are categorized into three types based on their deployment method: floating pipes, submerged pipes, and shore pipes. Due to external factors (such as land subsidence and water flow impact) or inherent pipeline parameters (such as connection methods and wall thickness), these three types of reclamation pipelines are frequently damaged during construction. For example, floating pipes, adrift on the water's surface, are highly susceptible to damage due to harsh marine conditions, periodic tidal erosion, and impacts from passing ships and large floating objects.

[0003] To ensure the integrity of pipelines and avoid impacting construction efficiency, it is typically necessary to assign dedicated personnel to regularly inspect the dredging and reclamation pipelines. Current manual inspection methods require personnel to be on-site for close observation, or to use binoculars for remote observation in areas difficult to reach. This results in blind spots and prevents a complete inspection of the pipeline's condition. Furthermore, the significant time lag between manual inspections hinders timely location and troubleshooting of anomalies.

[0004] Overcoming the limitations of the manual inspection mode and promptly detecting or predicting pipeline damage is key to improving construction efficiency and effectiveness. Summary of the Invention

[0005] To address the issue of ineffective damage monitoring and early warning for dredged reclamation pipelines using manual inspection methods, this application aims to provide a data fusion-based damage early warning method for dredged reclamation pipelines. This method selects key inspection nodes on the pipeline and acquires the associated influencing factor data and damage data. It then establishes a correlation model between each influencing factor and the damage type and probability of occurrence. Furthermore, by collecting and observing the distribution patterns of influencing factors or existing damage conditions, it infers the type and probability of damage occurring at various points on the dredged reclamation pipeline. This makes pipeline inspection more targeted, enables early warning of damage occurrence, reduces the probability of damage, and improves construction efficiency. To implement the above damage monitoring method, this application also aims to propose a data fusion-based damage early warning system for dredged reclamation pipelines. Finally, to facilitate the widespread use of this damage monitoring method, a computer-readable storage medium loaded with a computer program for implementing the above damage monitoring method is proposed. The specific scheme is as follows: A data fusion-based method for early warning of damage in dredging pipelines includes: Multiple sampling sites were selected on the dredging pipeline, and damage data and influencing factor data at each sampling site were obtained and stored in association with the dredging pipeline information and sampling site number. Based on damage data and impact factor data, the correlation between damage data and each impact factor data is obtained by analyzing the data association algorithm according to the set data, and a damage prediction model is generated. The distribution patterns of various types of damage in the evacuation and filling pipeline were analyzed and obtained. The diffusion characteristics of each influencing factor in the laying direction of the dredging pipeline were analyzed and obtained. Based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor, the expected impact factor data for each point on the dredging pipeline is generated. Based on the currently determined damage types and the distribution location patterns, generate the first expected damage data for each point on the purging pipeline; Based on the expected impact factor data and the damage prediction model, second expected damage data for each point on the purging pipeline is generated. The first expected damage data and the second expected damage data are fused together using a weighted algorithm to generate theoretical damage data at each point on the blowing pipeline. Based on the aforementioned theoretical damage data, early warning information is output; The damage data includes damage type and damage degree. The damage type includes pipe wall wear, pipe wall perforation, structural deformation, pipe fracture, pipe corrosion and pipe aging. The set data association algorithm includes multiple linear regression algorithm, logistic regression algorithm and deep learning algorithm; The theoretical damage data includes damage type, damage degree, and damage occurrence probability; The influencing factors include pipe shape, pipe material, pipe installation process, pipe pressure, characteristics of backfill material, settlement of pipe support surface, water flow impact, corrosion, collision with foreign objects, and other related damage.

[0006] Through the above technical solution, inspection personnel can quickly assess the expected damage data at other locations on the dredging pipeline based on the damage data and influencing factor data obtained from the current sampling point. Targeted inspection actions can be carried out based on the expected damage data, which can greatly improve inspection efficiency, reduce the probability of damage, and maintain the efficient operation of the dredging construction.

[0007] Optionally, based on damage data and impact factor data, a damage prediction model is generated, including: Obtain damage data and influencing factor data at each sampling point of the current blowing pipeline, and determine whether the amount of damage data and influencing factor data has reached the set threshold. If the amount of data reaches a set threshold, the damage data and influencing factor data of the current buoyancy pipeline are used as source data. Multiple linear regression and logistic regression algorithms are used to analyze and obtain the correlation between each influencing factor data and the damage data, and the damage prediction model is generated. If the amount of data does not reach the set threshold, a set number of sample data will be selected from the damage data and impact factor data of the current purging pipeline and temporarily stored. Search the construction database for data on dredged and filled pipelines that are similar to the current dredged and filled pipeline, and store them together with the corresponding data of the current dredged and filled pipeline as a temporary reference database; Damage data and impact factor data from the temporary reference database are obtained as source data. Deep learning algorithms are used to analyze and obtain the correlation between each impact factor data and damage data, and a transition model is generated. Based on the sample data, the transition model is adjusted using a transfer learning algorithm to generate the damage prediction model; This includes searching the construction database for data on dredged filling pipelines similar to the current dredged filling pipeline, including: Obtain current damage data and influencing factor data for the dredging and filling pipeline; The correlation coefficients between each damage type and each influencing factor or their combination were obtained using a correlation analysis algorithm. Influence factors with correlation coefficients greater than a set value are selected as judgment reference factors, and judgment weights are assigned to each judgment reference factor according to the magnitude of the correlation coefficient. Based on the aforementioned influencing factors and their weighting data, similar dredging pipeline data were obtained using the cosine similarity algorithm.

[0008] The above technical solution enables the accurate generation of damage prediction models by leveraging data from similar dredging pipelines, thereby improving the accuracy of current dredging pipeline damage prediction.

[0009] Optionally, the distribution patterns of various types of damage on the evacuation and backfill pipeline can be analyzed, including: A one-dimensional reference coordinate axis is set along the length of the dredging pipeline with the mud pump as the starting coordinate. Based on the temporary reference database, the coordinate positions of various types of damage on the dredging pipeline are obtained and associated with the damage type for storage, forming a damage coordinate database. Based on the damage coordinate database, a segmented density statistical algorithm is used to analyze and obtain the distribution location patterns of various types of damage on the buoyancy pipeline.

[0010] The above technical solution can clearly show the possible distribution of various types of damage on the current purging pipeline. Based on the coordinates of the above distribution locations, the expected damage sites can be obtained, which can improve the efficiency of inspection.

[0011] Optionally, the purging pipeline damage early warning method further includes: Establish and store an inspection plan reference model to reflect the correspondence between various damage data and their corresponding inspection plans; Acquire and, based on the theoretical damage data of each point on the current purging pipeline, generate and output the inspection plan for each point according to the inspection plan reference model. The inspection plan includes one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, drone aerial inspection, video remote image recognition, and pipe wall ultrasonic testing.

[0012] Through the above technical solutions, inspection personnel can quickly obtain inspection plans suitable for various points on the dredging pipeline, enabling them to carry out inspection actions more effectively, improving inspection efficiency and results, reducing the probability of accidents or malfunctions, and improving the efficiency of dredging construction.

[0013] Optionally, the purging pipeline damage early warning method further includes an inspection plan alternative step: Based on the damage prediction model and the already determined damage types and location coordinates, the expected damage sites within a set range are generated. Based on the expected damage sites mentioned above, an alternative inspection plan for the expected damage sites is generated and output via the inspection plan reference model; Damage data for the expected damage sites are obtained according to the alternative inspection scheme; Based on the actual damage data of the expected damage sites, and according to the damage prediction model and the distribution pattern of various damages on the purging pipeline, the target inspection data is obtained.

[0014] With the above technical solution, when a certain point on the blowing pipeline is suspected of having damage but the generated inspection plan is difficult to implement, other types of damage can be identified through the generated alternative inspection plan, and then the inspection data can be indirectly calculated through the damage prediction model.

[0015] Optionally, the purging pipeline damage early warning method further includes: Obtain theoretical damage data for each point on the blowing pipeline, and extract damage type and damage occurrence probability data; If the probability of damage occurs exceeds a set value, the above sites will be marked as risk sites; Based on the damage type corresponding to each risk site, temporary sensing devices are installed at the aforementioned risk sites to acquire monitoring data at those risk sites.

[0016] The above technical solutions enable rapid and accurate detection based on the expected damage type, thereby improving the accuracy of pipeline damage detection.

[0017] Optionally, the purging pipeline damage early warning method further includes: Get real-time images of the current purging pipeline or its location; Based on AR technology, a virtual status image of the blowing pipeline is generated on the human-computer interaction interface and overlaid on the real-time image for display. The virtual state image includes the blowing pipeline body, sensor data of each point on the pipeline, theoretical damage data of each point on the pipeline, and corresponding inspection plans.

[0018] With the above technical solutions, inspection personnel can use AR equipment to quickly and accurately locate risk points on the purging pipeline, which helps to improve inspection efficiency and accuracy.

[0019] To implement the above-mentioned method for early warning of damage to dredged pipelines, this application also discloses a data fusion-based early warning system for early warning of damage to dredged pipelines, comprising: The data acquisition module is configured to collect damage data and influencing factor data at each sampling point on the purging pipeline, and to associate and store the damage data, influencing factor data, purging pipeline information and sampling point number. The model building module is configured to analyze and obtain the correlation between damage data and various influencing factor data based on a set data association algorithm, and generate a damage prediction model. The pattern analysis module is configured to analyze the distribution patterns of various types of damage in the dredging pipeline, as well as the diffusion characteristics of various influencing factors in the laying direction of the dredging pipeline. The data extrapolation module is configured to generate expected impact factor data for each point on the dredging pipeline based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor; generate first expected damage data for each point on the dredging pipeline based on the determined damage type and the distribution pattern; and generate second expected damage data for each point on the dredging pipeline based on the expected impact factor data and the damage prediction model. The data fusion module is configured to fuse the first expected damage data and the second expected damage data by setting a weighting algorithm to generate theoretical damage data at each point on the shovel-fill pipeline. The early warning output module is configured to output early warning information based on the theoretical damage data. The damage data includes damage type and damage degree. The damage type includes pipe wall wear, pipe wall perforation, structural deformation, pipe fracture, pipe corrosion and pipe aging. The influencing factors include pipe shape, pipe material, pipe installation process, pipe pressure, characteristics of dredged material, settlement of pipe support surface, water flow impact, corrosion, collision with foreign objects, and other related types of damage. The theoretical damage data includes damage type, damage degree, and damage occurrence probability; The set data association algorithms include multiple linear regression, logistic regression, and deep learning algorithms.

[0020] Optionally, the purging pipeline damage early warning system further includes: The inspection plan matching module has a built-in inspection plan reference model that reflects the correspondence between various damage data and the appropriate inspection plan. It acquires and generates corresponding inspection plans based on the theoretical damage data of each point in the blowing pipeline and outputs them. The inspection alternative module is configured to generate expected damage sites based on the damage prediction model and the already determined types of damage and their location coordinates, generate and output alternative inspection schemes through the inspection scheme reference model, and obtain damage data of the expected damage sites based on the alternative inspection schemes. The inspection plan includes one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, drone aerial inspection, video remote image recognition, and pipe wall ultrasonic testing. Obtaining damage data for expected damage sites according to the alternative inspection scheme includes: obtaining target inspection data based on the actual damage data of the obtained expected damage sites, the damage prediction model, and the distribution patterns of various types of damage on the purging pipeline.

[0021] A computer-readable storage medium having a computer program module loaded thereon, which, when executed by a processor, is used to implement the data fusion-based early warning method for purging pipeline damage as described above.

[0022] The above technical solutions will help promote the use of the aforementioned early warning method for damage to backfill pipelines.

[0023] This application includes at least one of the following beneficial effects: (1) The inspection personnel can quickly assess the expected damage data at other locations on the dredging pipeline based on the damage data and influencing factor data obtained from the current sampling point. Based on the expected damage data, targeted inspection actions can be carried out, which can greatly improve the inspection efficiency and reduce the probability of damage, thus maintaining the efficient operation of the dredging construction. (2) The inspection personnel can quickly obtain inspection plans that are suitable for each point on the dredging pipeline, and can carry out inspection actions in a more targeted manner, which will improve the inspection efficiency and the inspection effect, reduce the probability of accidents or failures, and ensure the efficiency of dredging construction. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the early warning method for dredging pipeline damage in this application; Figure 2 A schematic diagram illustrating the method for generating a damage prediction model for this application; Figure 3 This is a schematic diagram showing the connection of the functional modules of the dredging pipeline damage early warning system in this application.

[0025] Attached reference numerals: 1. Data acquisition module; 2. Model building module; 3. Pattern analysis module; 4. Data extrapolation module; 5. Data fusion module; 6. Early warning output module; 7. Construction big data database. Detailed Implementation

[0026] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0027] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0028] A data fusion-based method for early warning of damage in dredging pipelines mainly includes the following steps: S100: Select multiple sampling sites on the purging pipeline, obtain damage data and influencing factor data at each sampling site, and store them in association with the purging pipeline information and sampling site number. S200, based on damage data and influencing factor data, analyzes the correlation between damage data and each influencing factor data according to a set data association algorithm, and generates a damage prediction model; S300, analyze and obtain the distribution location pattern of various types of damage in the blowing pipeline; S400, analyze and obtain the diffusion characteristics of each influencing factor in the laying direction of the dredging pipeline; S500 generates expected impact factor data for each point on the shovel-fill pipeline based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor. S600, based on the currently determined damage type and the distribution location pattern, generate the first expected damage data for each point on the shovel-fill pipeline; S700, Based on the expected impact factor data and the damage prediction model, generate the second expected damage data for each point on the shovel-fill pipeline; S800, the first expected damage data and the second expected damage data are fused together using a set weighting algorithm to generate theoretical damage data at each point on the shovel-fill pipeline; S900 outputs early warning information based on the aforementioned theoretical damage data.

[0029] In step S100 above, the damage data includes the damage type and the degree of damage. The damage type specifically includes pipe wall wear, pipe wall perforation, structural deformation, pipe fracture, pipe corrosion, and pipe aging. The degree of damage is determined by evaluation criteria based on pipe parameters. For example, if the pipe wall thickness wears down from an initial 15mm to 8mm, the degree of damage can be defined as moderate damage, or it can be directly expressed as the proportion of the worn portion, i.e., the degree of damage is marked as 47%.

[0030] Influencing factors include pipe shape, pipe material, pipe installation process, pipe pressure, characteristics of backfill material, settlement of pipe support surface, water flow impact, corrosion, collision with foreign objects, and other related damage. The aforementioned other damage refers to damage associated with a specific damage type. For example, factors affecting pipe corrosion, besides water salinity, also include the pipe's own material, installation process, or cracking condition.

[0031] In step S200, a damage prediction model is generated based on the damage data and influencing factor data, such as... Figure 2 As shown, it specifically includes: S210, acquire the damage data and influencing factor data corresponding to each sampling point in the current purging pipeline, and determine whether the amount of damage data and influencing factor data has reached the set threshold: S211, if the amount of data reaches a set threshold, then: S2110, using the current damage data and influencing factor data of the buoyancy pipeline as source data, the correlation between each influencing factor data and the damage data is obtained by using multiple linear regression algorithm and logistic regression algorithm to generate the damage prediction model.

[0032] S212, if the amount of data has not reached the set threshold, then: S2120: Select a set number of sample data from the current damage data and impact factor data of the purging pipeline and store them temporarily. S2121: Search for data on dredged and filled pipelines similar to the current dredged and filled pipeline in the construction database, and store them together with the corresponding data of the current dredged and filled pipeline as a temporary reference database; S2122, obtain damage data and impact factor data from the temporary reference database as source data, use deep learning algorithm to analyze and obtain the correlation between each impact factor data and damage data, and generate a transition model; S2123, Based on the sample data, the transition model is adjusted using a transfer learning algorithm to generate the damage prediction model.

[0033] In step S210, the damage data and influencing factor data of the dredged pipeline are stored in the construction database. The damage data and influencing factor data are configured to be collected and stored periodically. For dredged pipelines with more recent installation time, the amount of data is small, and the correlation between various data cannot be obtained from it.

[0034] The output data of the damage prediction model obtained in step S2110, namely the second expected damage data, includes damage type, damage degree and damage occurrence probability.

[0035] The damage prediction model includes two sets of calculation formulas: The formula for the multiple linear regression algorithm is as follows:

[0036] in, This represents the quantification value of the damage severity for the m-th type of damage; For all impact factor data x1, x2...x K When all values ​​are 0, this represents the degree of natural foundation damage inherent in the dredging pipeline. Let be the regression coefficient of the i-th influencing factor; Let be the regression coefficient of the interaction term of the i-th and j-th influence factor combination; This represents the random error of the model.

[0037] The formula for the logistic regression algorithm is as follows:

[0038] Among them, P m Let m be the probability of the occurrence of the m-th type of injury; It represents the basic probability offset of the pipeline causing this type of damage due to its own inherent factors (such as material, manufacturing defects, natural aging, and base corrosion) under ideal conditions where all external influencing factors (such as water flow impact, collision, and settlement) are all 0. x i xj Let represent the data for the i-th and j-th impact factors, and K be the total number of impact factors; This represents the single-factor coefficient corresponding to the i-th impact factor data. This represents the interaction coefficient of the factor combination corresponding to the combination of influencing factors.

[0039] The damage prediction model described above can be used to predict the degree of damage and the probability of occurrence for each type of damage.

[0040] In step S2120, sample data refers to a data set that can accurately reflect the characteristics of the current buoyancy pipeline data, such as damage data and its influencing factor data of sampling sites with severe damage.

[0041] In step S2121, data on dredging pipelines similar to the current dredging pipeline are searched from the construction database, including: S21211, obtain current damage data and influencing factor data of the dredging and filling pipeline; S21212, the correlation coefficients between each damage type and each influencing factor or their combination are obtained through a correlation analysis algorithm; S21213, Select the influence factors with a correlation coefficient greater than the set value as the judgment reference factors, and configure the judgment weight for each judgment reference factor according to the correlation coefficient; S21214. Based on the above-mentioned influencing factors and their judgment weight data, similar dredging pipeline data are obtained through the cosine similarity algorithm.

[0042] The above technical solution first obtains the main influencing factors corresponding to various types of damage based on correlation analysis algorithms, such as the Pearson correlation coefficient algorithm. Then, using the above influencing factors as a reference, it can accurately generate a damage prediction model with the help of data from similar dredging pipelines, thereby improving the accuracy of damage prediction for the current dredging pipeline. Thus, even if the current dredging pipeline is a newly constructed pipeline, the model can be built with the help of historical inspection data from similar pipelines.

[0043] In step S300, the distribution patterns of various types of damage on the evacuation and filling pipeline are analyzed and obtained, including: S310, a one-dimensional reference coordinate axis is set along the length of the dredging pipeline with the mud pump as the starting coordinate. Based on the temporary reference database, the coordinate positions of various types of damage on the dredging pipeline are obtained and stored in association with the damage type to form a damage coordinate database. S320, Based on the damage coordinate database, a segmented density statistical algorithm is used to analyze and obtain the distribution location patterns of various types of damage on the buoyancy pipeline.

[0044] As can be seen from the distribution location pattern, the first expected damage data obtained based on the above distribution location pattern includes the damage type and the probability of damage occurrence.

[0045] In practical applications, thermal maps of various types of damage can be generated based on the above distribution patterns, making it easier for inspection personnel to intuitively understand the damage status of each section of the purging pipeline. Thus, the expected damage locations can be obtained based on the above distribution coordinates, which helps to improve the efficiency of inspections.

[0046] In actual construction, factors that prevent pipeline inspections from achieving the expected results also include inadequate inspection plans, such as failing to use ultrasonic detectors in timely detection of areas where the pipe wall has thinned due to corrosion. Therefore, in this application's embodiment, the dredging pipeline damage early warning method further includes: A100, establishes and stores an inspection plan reference model to reflect the correspondence between various damage data and their corresponding inspection plans; A200: Acquire and, based on the theoretical damage data of each point on the current purging pipeline, generate and output the inspection plan for each point according to the inspection plan reference model.

[0047] The above inspection plan includes one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, drone aerial inspection, video remote image recognition, and pipe wall ultrasonic testing.

[0048] Since theoretical damage data includes damage type, damage probability, and damage severity, inspection personnel can quickly obtain inspection plans suitable for each point on the dredging pipeline. This allows for more targeted inspection actions, improving inspection efficiency and effectiveness, reducing the probability of accidents or malfunctions, and increasing the efficiency of dredging construction.

[0049] In practice, inspection plans generated based on theoretical damage data are not always applicable to actual field conditions. For example, when obstructions prevent inspectors from approaching the dredging pipeline, the inspection plan, including tactile assistance, will be difficult to implement. Therefore, the dredging pipeline damage early warning method described in this application also includes an inspection plan replacement step, specifically including: A300, based on the damage prediction model and combined with the already determined types of damage and location coordinates, generates expected damage sites within a set range. A400, based on the expected damage sites mentioned above, generates and outputs alternative inspection plans for the expected damage sites via the inspection plan reference model; A500, obtains damage data of the expected damage sites according to the alternative inspection scheme; A600, based on the actual damage data of the expected damage sites, and according to the damage prediction model and the distribution pattern of various damages on the purging pipeline, obtains the target inspection data.

[0050] The aforementioned alternative inspection scheme does not directly collect data at the target site. Instead, it collects data at sites associated with the target site and infers the target site data through the correlation between different types of damage in the damage prediction model.

[0051] In this embodiment of the application, the method for early warning of damage to the dredging pipeline further includes: B100, obtain theoretical damage data for each point on the blowing and filling pipeline, and extract damage type and damage occurrence probability data; B200, if the probability of damage occurs exceeds a set value, the above sites will be marked as risk sites; B300, based on the damage type corresponding to each risk site, temporarily sets up sensing devices at the aforementioned risk sites to acquire monitoring data at those risk sites.

[0052] To improve inspection efficiency and accuracy, the damage early warning method for blown pipelines described in this application also includes: C100 acquires real-time images of the current purging pipeline or its location. These real-time images are acquired by action cameras worn by inspection personnel or drone cameras.

[0053] The C200, based on AR technology, generates a virtual status image of the purging pipeline on the human-computer interaction interface and displays it overlaid with the real-time image.

[0054] C300, the virtual state image includes the blowing pipeline body, sensor data of each point on the pipeline, theoretical damage data of each point on the pipeline and its corresponding inspection plan.

[0055] In this embodiment of the application, the human-computer interaction interface includes AR glasses, which allow inspection personnel to quickly and accurately locate risk points on the blowing pipeline with the help of AR devices.

[0056] In step S900, the output warning information includes damage data and corresponding location coordinate data. This warning information is sent to the monitoring center and simultaneously displayed on the AR device worn by the inspection personnel.

[0057] To implement the aforementioned data fusion-based early warning method for dredged and filled pipeline damage, this application also discloses a data fusion-based early warning system for dredged and filled pipeline damage, such as... Figure 3 As shown, it includes a data acquisition module 1, a model building module 2, a pattern analysis module 3, a data extrapolation module 4, a data fusion module 5, and an early warning output module 6.

[0058] Data acquisition module 1 is configured to collect damage data and influencing factor data at various sampling points along the reclamation pipeline, and to associate and store the damage data and influencing factor data with the reclamation pipeline information and sampling point numbers. The aforementioned data acquisition module 1 includes various types of sensing devices, such as water flow velocity sensors, temperature sensors, and pressure sensors.

[0059] Model building module 2 is configured to analyze and obtain the correlation between damage data and various influencing factor data according to a set data association algorithm, and generate a damage prediction model. In this embodiment, the above-mentioned model building module 2 is configured in a specific server, such as the server of a data provider, and has a built-in set data association algorithm, such as multiple linear regression algorithm, logistic regression algorithm, and deep learning algorithm, for generating the damage prediction model.

[0060] The pattern analysis module 3 is configured to analyze the distribution pattern of various types of damage in the dredging pipeline, as well as the diffusion characteristics of various influencing factors in the laying direction of the dredging pipeline.

[0061] Data extrapolation module 4 is configured to generate expected impact factor data for each point on the dredging pipeline based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor. Based on the determined damage type and distribution pattern, it generates first expected damage data for each point on the dredging pipeline. Based on the expected impact factor data and the damage prediction model, it generates second expected damage data for each point on the dredging pipeline. Data fusion module 5 is configured to fuse the first and second expected damage data using a weighted algorithm to generate theoretical damage data for each point on the dredging pipeline. In practical applications, the weighting coefficients of the above weighting algorithm can be determined by the staff.

[0062] The early warning output module 6 is configured to output early warning information based on theoretical damage data. The early warning output module 6 has a built-in communication unit for sending theoretical damage data to the monitoring center and / or the smart terminal worn by the inspection personnel, such as a mobile phone or AR device.

[0063] To improve inspection efficiency, the damage early warning system for dredged pipelines in this application also includes an inspection scheme matching module and an inspection substitution module.

[0064] The inspection plan matching module has a built-in inspection plan reference model that reflects the correspondence between various damage data and suitable inspection plans. It acquires and generates corresponding inspection plans based on theoretical damage data for each point in the purging pipeline, and outputs them. The inspection substitution module is configured to generate expected damage sites based on the damage prediction model and the already determined types of damage and their location coordinates. It then generates and outputs alternative inspection plans using the inspection plan reference model, and obtains the damage data for the expected damage sites based on the alternative plans. The aforementioned inspection plans include one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, UAV aerial inspection, remote video image recognition, and ultrasonic testing of the pipe wall.

[0065] The damage data obtained from the alternative inspection plan for the expected damage sites includes: based on the actual damage data of the expected damage sites, the target inspection data is obtained according to the damage prediction model and the distribution pattern of various damages on the purging pipeline.

[0066] Finally, to facilitate the widespread use of the aforementioned method for early warning of damage to reclamation pipelines, this application also discloses a computer-readable medium on which a computer program module is loaded. When the computer program module is executed by a processor, it is used to implement the data fusion-based early warning method for early warning of damage to reclamation pipelines as described above. The computer-readable storage medium includes, but is not limited to, disk storage, CD-ROM, optical storage, etc.

[0067] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for early warning of damage in dredging pipelines based on data fusion, characterized in that, include: Multiple sampling sites were selected on the dredging pipeline, and damage data and influencing factor data at each sampling site were obtained and stored in association with the dredging pipeline information and sampling site number. Based on damage data and impact factor data, the correlation between damage data and each impact factor data is obtained by analyzing the data association algorithm according to the set data, and a damage prediction model is generated. The distribution patterns of various types of damage in the evacuation and filling pipeline were analyzed and obtained. The diffusion characteristics of each influencing factor in the laying direction of the dredging pipeline were analyzed and obtained. Based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor, the expected impact factor data for each point on the dredging pipeline is generated. Based on the currently determined damage types and the distribution location patterns, generate the first expected damage data for each point on the purging pipeline; Based on the expected impact factor data and the damage prediction model, second expected damage data for each point on the shovel-fill pipeline is generated. The first expected damage data and the second expected damage data are fused together using a weighted algorithm to generate theoretical damage data at each point on the blowing pipeline. Based on the aforementioned theoretical damage data, early warning information is output; The damage data includes damage type and damage degree. The damage type includes pipe wall wear, pipe wall perforation, structural deformation, pipe fracture, pipe corrosion and pipe aging. The set data association algorithm includes multiple linear regression algorithm, logistic regression algorithm and deep learning algorithm; The theoretical damage data includes damage type, damage degree, and damage occurrence probability; The influencing factors include pipe shape, pipe material, pipe installation process, pipe pressure, characteristics of backfill material, settlement of pipe support surface, water flow impact, corrosion, collision with foreign objects, and other related damage.

2. The data fusion-based early warning method for damage in reclamation pipelines according to claim 1, characterized in that, Based on damage data and impact factor data, a damage prediction model is generated, including: Obtain damage data and influencing factor data at each sampling point of the current blowing pipeline, and determine whether the amount of damage data and influencing factor data has reached the set threshold. If the amount of data reaches a set threshold, the damage data and influencing factor data of the current buoyancy pipeline are used as source data. Multiple linear regression and logistic regression algorithms are used to analyze and obtain the correlation between each influencing factor data and the damage data, and the damage prediction model is generated. If the amount of data does not reach the set threshold, a set number of sample data will be selected from the damage data and impact factor data of the current purging pipeline and temporarily stored. Search the construction database for data on dredged and filled pipelines that are similar to the current dredged and filled pipeline, and store them together with the corresponding data of the current dredged and filled pipeline as a temporary reference database; Damage data and impact factor data from the temporary reference database are obtained as source data. Deep learning algorithms are used to analyze and obtain the correlation between each impact factor data and damage data, and a transition model is generated. Based on the sample data, the transition model is adjusted using a transfer learning algorithm to generate the damage prediction model; This includes searching the construction database for data on dredged filling pipelines similar to the current dredged filling pipeline, including: Obtain current damage data and influencing factor data for the dredging and filling pipeline; The correlation coefficients between each damage type and each influencing factor or their combination were obtained using a correlation analysis algorithm. Influence factors with correlation coefficients greater than a set value are selected as judgment reference factors, and judgment weights are assigned to each judgment reference factor according to the magnitude of the correlation coefficient. Based on the aforementioned influencing factors and their weighting data, similar dredging pipeline data were obtained using the cosine similarity algorithm.

3. The data fusion-based early warning method for damage in reclamation pipelines according to claim 2, characterized in that, The distribution patterns of various types of damage in the evacuation and backfill pipeline were analyzed and obtained, including: A one-dimensional reference coordinate axis is set along the length of the dredging pipeline with the mud pump as the starting coordinate. Based on the temporary reference database, the coordinate positions of various types of damage on the dredging pipeline are obtained and stored in association with the damage type to form a damage coordinate database. Based on the damage coordinate database, a segmented density statistical algorithm is used to analyze and obtain the distribution location patterns of various types of damage on the buoyancy pipeline.

4. The data fusion-based early warning method for damage in dredging pipelines according to claim 3, characterized in that, The method for early warning of damage to the purging pipeline also includes: Establish and store an inspection plan reference model to reflect the correspondence between various damage data and their corresponding inspection plans; Acquire and, based on the theoretical damage data of each point on the current purging pipeline, generate and output the inspection plan for each point according to the inspection plan reference model. The inspection plan includes one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, drone aerial inspection, video remote image recognition, and pipe wall ultrasonic testing.

5. The data fusion-based early warning method for damage in dredging pipelines according to claim 4, characterized in that, The method for early warning of damage to purging pipelines also includes alternative steps for inspection plans: Based on the damage prediction model and the already determined types of damage and their location coordinates, expected damage sites within a set range are generated. Based on the expected damage sites mentioned above, an alternative inspection plan for the expected damage sites is generated and output via the inspection plan reference model; Damage data for the expected damage sites are obtained according to the alternative inspection scheme; Based on the actual damage data of the expected damage sites, and according to the damage prediction model and the distribution pattern of various damages on the purging pipeline, the target inspection data is obtained.

6. The data fusion-based early warning method for damage in reclamation pipelines according to claim 1, characterized in that, The method for early warning of damage to the purging pipeline also includes: Obtain theoretical damage data for each point on the blowing pipeline, and extract damage type and damage occurrence probability data; If the probability of damage occurs exceeds a set value, the above sites will be marked as risk sites; Based on the damage type corresponding to each risk site, temporary sensing devices are installed at the aforementioned risk sites to acquire monitoring data at those risk sites.

7. The data fusion-based early warning method for damage in dredging pipelines according to claim 6, characterized in that, The method for early warning of damage to the purging pipeline also includes: Get real-time images of the current purging pipeline or its location; Based on AR technology, a virtual status image of the blowing pipeline is generated on the human-computer interaction interface and overlaid on the real-time image for display. The virtual state image includes the blowing pipeline body, sensor data of each point on the pipeline, theoretical damage data of each point on the pipeline, and corresponding inspection plans.

8. A data fusion-based early warning system for damage to dredged and filled pipelines, characterized in that, include: The data acquisition module (1) is configured to collect damage data and influence factor data at each sampling point on the blowing pipeline, and to associate and store the damage data, influence factor data, blowing pipeline information and sampling point number. The model building module (2) is configured to analyze and obtain the correlation between damage data and the data of each influencing factor according to the set data association algorithm, and generate a damage prediction model; The regularity analysis module (3) is configured to analyze the distribution pattern of various types of damage on the dredging pipeline and the diffusion characteristics of various influencing factors in the laying direction of the dredging pipeline. The data extrapolation module (4) is configured to generate expected impact factor data for each point on the dredging pipeline based on the impact factor data at each sampling point and the diffusion characteristics of each impact factor; generate first expected damage data for each point on the dredging pipeline based on the determined damage type and the distribution pattern; and generate second expected damage data for each point on the dredging pipeline based on the expected impact factor data and the damage prediction model. The data fusion module (5) is configured to fuse the first expected damage data and the second expected damage data by setting a weighting algorithm to generate theoretical damage data at each point on the shovel-fill pipeline. The early warning output module (6) is configured to output early warning information based on the theoretical damage data; The damage data includes damage type and damage degree. The damage type includes pipe wall wear, pipe wall perforation, structural deformation, pipe fracture, pipe corrosion and pipe aging. The influencing factors include pipe shape, pipe material, pipe installation process, pipe pressure, characteristics of dredged material, settlement of pipe support surface, water flow impact, corrosion, collision with foreign objects, and other related types of damage. The theoretical damage data includes damage type, damage degree, and damage occurrence probability; The set data association algorithms include multiple linear regression, logistic regression, and deep learning algorithms.

9. The data fusion-based early warning system for evacuation pipeline damage according to claim 8, characterized in that, The purging pipeline damage early warning system also includes: The inspection plan matching module has a built-in inspection plan reference model that reflects the correspondence between various damage data and the appropriate inspection plan. It acquires and generates corresponding inspection plans based on the theoretical damage data of each point in the blowing pipeline and outputs them. The inspection alternative module is configured to generate expected damage sites based on the damage prediction model and the already determined types of damage and their location coordinates, generate and output alternative inspection schemes through the inspection scheme reference model, and obtain damage data of the expected damage sites based on the alternative inspection schemes. The inspection plan includes one or more combinations of ground facility inspection, pipeline appearance inspection, surrounding environment investigation, auditory and tactile assisted inspection, drone aerial inspection, video remote image recognition, and pipe wall ultrasonic testing. Obtaining damage data for expected damage sites according to the alternative inspection scheme includes: obtaining target inspection data based on the actual damage data of the obtained expected damage sites, the damage prediction model, and the distribution patterns of various types of damage on the purging pipeline.

10. A computer-readable medium, characterized in that, It is loaded with a computer program module, which, when executed by a processor, is used to implement the data fusion-based early warning method for buoyancy pipeline damage as described in any one of claims 1-7.

Images