Bridge tower construction progress intelligent identification system and method based on adaptive weight

Abstract

The invention discloses a bridge tower construction progress intelligent identification method based on adaptive weight. The method comprises the following steps: S1, sensor data acquisition and preprocessing; s2, collecting and analyzing machine vision data; s3, establishing a confidence evaluation model; s4, carrying out adaptive weight fusion; s5, multi-modal data fusion decision making; and S6, performing anomaly detection and error correction. Through dual verification of sensor data and visual data, complementary advantages of different data sources are fully utilized, compared with a single data source method, the system accuracy can be improved, and meanwhile, through an anomaly detection and cross verification mechanism, the influence of a single-point fault on the whole system is effectively prevented.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent identification of bridge tower construction progress. More particularly, the present application relates to a system and method for intelligent identification of bridge tower construction progress based on adaptive weights. BACKGROUND

[0002] Steel-concrete composite bridge tower construction includes the sequential construction of multiple groups of bridge tower segments, each group of bridge tower segments including a tower column steel segment hoisting construction process and a tower column concrete pouring construction process. Among them, the tower column steel segment hoisting construction process includes a hoisting construction preparation sub-process, a steel segment hoisting and positioning sub-process, and a steel segment connection sub-process; the tower column concrete pouring construction process includes a pouring construction preparation sub-process, a concrete pouring sub-process, and a concrete curing sub-process.

[0003] Traditional construction progress monitoring methods rely on manual inspection and recording, which has the problems of low efficiency, strong human subjectivity, and difficulty in realizing real-time monitoring and automatic management. In recent years, many domestic and foreign scholars have studied intelligent identification technology for construction progress, but there are the following problems: Single data source problem: existing intelligent methods mostly rely on a single data source, such as making judgments based only on sensor data or only on visual data. Although sensor data has high precision, it lacks semantic understanding of complex construction site conditions; although machine vision data is rich in information, it is easily affected by environmental factors such as light and weather.

[0004] Lack of cross-validation mechanism: a single data source is prone to failure in the event of equipment failure or environmental anomalies, and existing methods lack effective cross-validation and anomaly detection mechanisms, which cannot guarantee the reliability of the discrimination results.

[0005] Poor environmental adaptability: the construction site environment is complex and variable, and traditional methods are difficult to adapt to different lighting conditions, weather conditions, and changes in construction processes, etc., and the recognition accuracy significantly decreases in complex environments.

[0006] Therefore, there is an urgent need to develop an intelligent construction progress discrimination method that integrates multi-source heterogeneous data, fully utilizes the complementary advantages of different data sources, and improves the accuracy and reliability of bridge construction progress monitoring. SUMMARY

[0007] In order to achieve these objects and other advantages of the present application, a preferred embodiment of the present application provides a method for intelligent identification of bridge tower construction progress based on adaptive weights, comprising the following steps: S1, sensor data acquisition and preprocessing Synchronize the acquisition of the running state of the climbing frame, the tower crane hoisting weight information, and the position trajectory of the key equipment, perform sliding average filtering processing on the climbing frame height data, and establish a hoisted object type identification model based on the hoisting weight information and the hoisted object image. S2, Machine Vision Data Acquisition and Analysis Multi-dimensional visual information of the construction site is acquired through video image acquisition equipment to identify the work types and locations of construction workers, and target detection and trajectory analysis are performed on key equipment. S3. Establish a confidence assessment model Establish confidence assessment models for sensor data and machine vision data respectively; S4, Adaptive Weight Fusion The basic weights are calculated based on the confidence scores of sensor data and machine vision data output by S3. The sensor weights are then dynamically adjusted by combining the environmental condition scores. Finally, the fusion weights are obtained by optimizing the historical accuracy feedback of sensor and vision data, thus achieving dynamic adaptation of data quality and environmental conditions. S5, Multimodal Data Fusion Decision Making A hierarchical decision-making model is established. The first layer is based on the preprocessed climbing formwork status and current construction height of S1 to initially determine the possible construction sub-process. The second layer uses the final fusion weights obtained from S4 to perform confidence-weighted fusion of the sensor data discrimination results of S1 and the visual data discrimination results of S2, and outputs the final construction progress discrimination result by combining the comprehensive confidence threshold. S6. Anomaly Detection and Error Correction Based on the sensor data discrimination results of S1 and the visual data discrimination results of S2, a dual-modal data consistency check is performed. When the consistency is lower than the threshold, anomaly detection is triggered. Combined with the confidence index of S3, the sensor anomaly, vision system anomaly, or environmental interference type is identified. Error correction is completed through data correction, weight reallocation, and historical data assistance strategies to ensure the reliability of the discrimination results.

[0008] Preferably, the operating status of the climbing scaffold includes the elevation of the top of the climbing scaffold. and running status ; The elevation of the top of the climbing scaffold is monitored in real time using a climbing scaffold data acquisition device. and the operating status of the climbing frame ; The following formula is used to apply a moving average filter to the top elevation of the climbing scaffold: , in For the size of the filter window, The height of the climbing frame after filtering; The following formula is used to determine the operating status of the climbing scaffold: , in For height variation threshold, Indicates the climbing status. Indicates a stopped state.

[0009] Preferably, in step S2, identifying the work type and location of the construction personnel specifically includes: S21. The following formula is used to detect the location of construction workers: , in Let be the coordinates of the i-th person. For confidence level The number of people detected; S22. Classification and identification of construction workers' work: Based on time-series feature analysis, personnel behavior is classified into welding operations, steel reinforcement construction, and concrete pouring operations.

[0010] Preferably, in step S2, target detection and trajectory analysis are performed on key equipment, including concrete mixer trucks, truck-mounted pumps, concrete pouring platforms, welding platforms, transport ships, and steel sections. S23. The following formula is used to distinguish the types of key equipment using target detection: , in Let i be the bounding box of the i-th critical device. For the type of key equipment, To test the confidence level; S24. Analyze the trajectory of key equipment using the following formula: , Then, based on the trajectory data, the motion status and operation mode of key equipment are determined.

[0011] Preferably, step S3, establishing a confidence assessment model, specifically includes: S31, Sensor Data Confidence Modeling The confidence level of the sensor data is calculated using the following formula; , in Cs The confidence level of the sensor data. For accuracy coefficients, The stability coefficient, This is the timeliness coefficient. These are the weighting coefficients; The accuracy coefficient is calculated using the following formula: , in Measure the standard deviation of the sensor; The stability coefficient is calculated using the following formula: , in The variance of the current sensor data. This represents the variance under normal operating conditions. The timeliness coefficient is calculated using the following formula: , in The time decay coefficient, This is the current timestamp. timestamps for sensor data ； S32. Calculate the confidence level of machine vision data using the following formula; , in The confidence level of machine vision data. Image quality coefficient, For the recognition accuracy coefficient, For environmental adaptability coefficient, These are the weighting coefficients; The image quality coefficient is calculated using the following formula: , Among them, brightness quality ,in Average brightness of the image; Contrast quality ,in The standard deviation of pixels; Clarity Quality ;in The variance of the image after processing with the Laplacian operator; The recognition accuracy coefficient is calculated using the following formula: , in Let be the confidence level of the i-th detection result. Overlap degree; The environmental adaptability coefficient is calculated using the following formula: , These are the weighting coefficients for sunlight and weather. The light intensity score ranges from [0,1] and is calculated as follows: On a sunny day with sufficient sunlight... On a cloudy day with normal sunlight At night or when there is insufficient light ; The weather score is set, with a value range of [0,1]. The calculation method is as follows: sunny day with no rain or snow. At times; during light rain or fog During heavy rain, dense fog, or rain and snow. .

[0012] Preferably, S4, multimodal data fusion decision-making, specifically includes the following steps: S41. Calculate the basic weights using the following formula: , in As the basic weight for sensor data, As the basic weight for machine vision data, The confidence level of the sensor data calculated for S31. Calculate the confidence level of the machine vision data for S32; S42. Calculate the sensor weights after environmental adjustment using the following formula. , in, For the sensor weights adjusted for the environment, Based on the sensor weights, Rate the environmental conditions; S43, Historical Performance Feedback Adjustment: , in For adjustment coefficients, These represent the historical accuracy rates of sensor and vision data, respectively.

[0013] Preferably, step S5 specifically includes the following steps: A hierarchical decision-making model was established for the specific construction procedures of steel-concrete composite bridge towers: S51, First-level decision: Based on the preprocessed climbing formwork status in S1 and the current construction height, perform foundation assessment to determine the current possible construction stage: ①if That is, the height of the climbing frame is located at the bottom of the steel segment, and the construction process may be the steel segment hoisting and positioning, steel segment connection and concrete pouring construction preparation. ②if This means the climbing scaffold is ascending, and the current construction process is likely preparation for concrete pouring. ③if That is, the height of the climbing frame is located at the top of the steel segment, and the construction process may be the hoisting and positioning of the steel segment, preparation for concrete pouring, concrete pouring and concrete curing; in This represents the current height of the steel structure at the top of the tower. This is the current elevation of the top of the climbing scaffold; This indicates that the climbing scaffold is in a stopped state. This indicates that the climbing scaffold is in the climbing phase; S52, Second-level decision-making: Accurate discrimination by fusing sensor and visual data, achieving multi-source data collaboration based on the final weights obtained in S4: , in, To integrate the discrimination results, The final sensor weights output by S4. For the final machine vision weights, The process discrimination result of sensor data S1 The process discrimination result for S2 visual data; S53, Overall Confidence Level: , in To assess the overall confidence level, The sensor confidence level of S31, The visual confidence level of S32; S54. Confidence-weighted fusion, based on the reliability of the comprehensive confidence judgment result: , in To assess the overall confidence level, The sensor confidence level of S31, The visual confidence level of S32.

[0014] Preferably, step S6 specifically includes the following steps: S61. Perform a consistency check between the sensor discrimination result based on S1 and the visual discrimination result based on S2 to determine whether the multi-source data are coordinated: when At that time, the procedures were completely identical; when At that time, adjacent processes may be in the process transition phase; when At that time, the process intervals exceeded 1, indicating a clear abnormality; Where Ps is the result of the sensor-based discrimination process, Pv is the result of the vision-based discrimination process, and the process number is a discrete value from 1 to 6; when When an anomaly is triggered; S62, Anomaly Type Identification: Accurately locate the source of anomalies by combining the confidence index of S3; Sensor malfunction: ;in The standard deviation of sensor measurements in S31, This is the standard deviation threshold for when the sensor is operating normally; Visual system abnormalities: Where Qv is the image quality coefficient in S32; Environmental interference: ; where Ev is the environmental adaptability coefficient in S32.

[0015] On the other hand, another preferred embodiment of the present invention provides an intelligent recognition system for bridge tower construction progress based on adaptive weights, comprising: Data acquisition module: used to acquire sensor data and machine vision data; Data preprocessing module: used to filter sensor data and perform personnel recognition, key equipment detection and trajectory analysis on machine vision data; Confidence assessment module: used to calculate the confidence level of sensor data and machine vision data; Adaptive weight fusion module: used to dynamically adjust the fusion weights; Fusion Decision Module: Used to establish a hierarchical decision model and output construction progress judgment results; Anomaly detection and correction module: used to identify anomaly types and execute intelligent error correction strategies.

[0016] This invention offers at least the following advantages: By employing dual verification with both sensor and visual data, it fully leverages the complementary advantages of different data sources, improving system accuracy compared to single-data-source methods. Furthermore, through anomaly detection and cross-validation mechanisms, it effectively prevents single-point failures from impacting the entire system. The system utilizes adaptive weight adjustment technology, automatically optimizing the discrimination strategy based on environmental conditions. It maintains high accuracy even under adverse weather and complex lighting conditions, significantly enhancing environmental adaptability. This method supports the establishment of digital progress archives during the construction of steel-concrete composite bridge towers, providing a scientific basis for project management and enabling intelligent progress tracking. Through highly automated processing, it significantly reduces the workload of manual monitoring, improves monitoring efficiency, and effectively lowers the cost of manual monitoring, providing crucial technical support for the intelligent management of modern bridge construction.

[0017] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the framework structure of the intelligent recognition system for bridge tower construction progress based on adaptive weights in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0020] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious modifications will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0021] It is understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0022] like Figure 1 As shown, a preferred embodiment of the present invention provides a method for intelligent identification of bridge tower construction progress based on adaptive weights, comprising the following steps: S1, Sensor Data Acquisition and Preprocessing The system synchronously collects the operating status of the climbing formwork, the lifting weight information of the tower crane, and the location trajectory of key equipment. It also performs moving average filtering on the climbing formwork height data to reduce errors caused by data fluctuations. Based on the lifting weight information and images of the lifted objects, a lifting object type recognition model is established to accurately distinguish the types of lifted objects. S2, Machine Vision Data Acquisition and Analysis By acquiring multi-dimensional visual information of the construction site through video image acquisition equipment, the operation type and location of construction personnel can be identified, the distribution and working status of construction personnel on the construction site can be clarified, and target detection and trajectory analysis can be performed on key equipment to understand the working status of key equipment. S3. Establish a confidence assessment model Confidence assessment models for sensor data and machine vision data were established separately. The models comprehensively considered multiple influencing factors such as data accuracy, stability, and timeliness. Through reasonable weight allocation, the confidence scores of the two types of data were calculated, providing a basis for subsequent weight fusion.

[0023] S4, Adaptive Weight Fusion The basic weights are calculated based on the confidence scores of sensor data and machine vision data output by S3. The sensor weights are then dynamically adjusted by combining the environmental condition scores. Finally, the fusion weights are obtained by optimizing the historical accuracy feedback of sensor and vision data, thus achieving dynamic adaptation of data quality and environmental conditions. S5, Multimodal Data Fusion Decision Making A hierarchical decision-making model is established. The first layer is based on the preprocessed climbing formwork status and current construction height of S1 to initially determine the possible construction sub-process. The second layer uses the final fusion weights obtained from S4 to perform confidence-weighted fusion of the sensor data discrimination results of S1 and the visual data discrimination results of S2, and outputs the final construction progress discrimination result by combining the comprehensive confidence threshold. S6. Anomaly Detection and Error Correction Based on the sensor data discrimination results of S1 and the visual data discrimination results of S2, a dual-modal data consistency check is performed. When the consistency is lower than the threshold, anomaly detection is triggered. Combined with the confidence index of S3, the sensor anomaly, vision system anomaly, or environmental interference type is identified. Error correction is completed through data correction, weight reallocation, and historical data assistance strategies to ensure the reliability of the discrimination results.

[0024] This method comprehensively collects multi-source data during bridge tower construction and performs effective processing and analysis. Utilizing a confidence assessment model and adaptive weight fusion technology, it achieves efficient integration of sensor and machine vision data, improving the rationality and accuracy of data utilization. The application of a hierarchical decision model makes construction progress assessment more logical and targeted, accurately distinguishing different construction stages. Anomaly detection and correction mechanisms further ensure the reliability of the assessment results, effectively avoiding misjudgments caused by data anomalies or environmental interference. This method achieves intelligent identification of bridge tower construction progress, providing accurate and reliable progress data support for construction management, helping to improve construction efficiency and ensure smooth construction progress.

[0025] In another technical solution, the operating status of the climbing scaffold includes the elevation of the top of the climbing scaffold. and running status ; The elevation of the top of the climbing scaffold is monitored in real time using a climbing scaffold data acquisition device. and the operating status of the climbing frame , The following formula is used to apply a moving average filter to the top elevation of the climbing scaffold: , in For the size of the filter window, The height of the climbing frame after filtering; The following formula is used to determine the operating status of the climbing scaffold: , in For height variation threshold, Indicates the climbing status. Indicates a stopped state.

[0026] Throughout the implementation process, the climbing scaffold data acquisition device continuously collects data, and the data processing terminal performs real-time filtering and operational status determination, feeding the results back to the construction management system in real time to ensure that managers can keep abreast of the working status of the climbing scaffold.

[0027] In another technical solution, step S2, identifying the work type and location of construction personnel, specifically includes: S21. The following formula is used to detect the location of construction workers: , in Let be the coordinates of the i-th person. For confidence level The number of people detected; S22. Classification and identification of construction workers' work: Based on time-series feature analysis, personnel behavior is classified into welding operations, steel reinforcement construction, and concrete pouring operations.

[0028] In another technical solution, step S2 involves target detection and trajectory analysis of key equipment, including concrete mixer trucks, truck-mounted pumps, concrete pouring platforms, welding platforms, transport ships, and steel sections. S23. The following formula is used to distinguish the types of key equipment using target detection: , in Let i be the bounding box of the i-th critical device. For the type of key equipment, To test the confidence level; S24. Analyze the trajectory of key equipment using the following formula: , Then, based on the trajectory data, the motion status and operation mode of key equipment are determined.

[0029] Based on trajectory data sets, the positional change trends of key equipment are analyzed to determine its motion status, such as whether it is moving, stationary, or moving slowly. Combined with information such as the distribution of work areas and the arrangement of construction procedures at the construction site, the operational patterns of key equipment are further analyzed. For example, whether concrete mixer trucks stop to unload in the unloading area, and whether flatbed trucks shuttle between the material storage area and the work area. Through continuous tracking and analysis of the trajectories of key equipment, a comprehensive understanding of its operational dynamics is achieved.

[0030] In another technical solution, S3, establishing a confidence assessment model, specifically includes: S31, Sensor Data Confidence Modeling The confidence level of the sensor data is calculated using the following formula; , in Cs The confidence level of the sensor data. For accuracy coefficients, The stability coefficient, This is the timeliness coefficient. These are the weighting coefficients; The accuracy coefficient is calculated using the following formula: , in Measure the standard deviation of the sensor; The stability coefficient is calculated using the following formula: , in The variance of the current sensor data. This represents the variance under normal operating conditions. The timeliness coefficient is calculated using the following formula: , in The time decay coefficient, This is the current timestamp. timestamps for sensor data ； S32. Calculate the confidence level of machine vision data using the following formula; , in The confidence level of machine vision data. Image quality coefficient, For the recognition accuracy coefficient, For environmental adaptability coefficient, These are the weighting coefficients; The image quality coefficient is calculated using the following formula: , Among them, brightness quality ,in Average brightness of the image; Contrast quality ,in The standard deviation of pixels; Clarity Quality ;in The variance of the image after processing with the Laplacian operator; The recognition accuracy coefficient is calculated using the following formula: , in Let be the confidence level of the i-th detection result. Overlap degree; The environmental adaptability coefficient is calculated using the following formula: , in Rate the light intensity. Rate the weather; These are the weighting coefficients for sunlight and weather. The light intensity score ranges from [0,1] and is calculated as follows: On a sunny day with sufficient sunlight... On a cloudy day with normal sunlight At night or when there is insufficient light ; The weather score is set, with a value range of [0,1]. The calculation method is as follows: sunny day with no rain or snow. At times, during light rain or fog During heavy rain, dense fog, or rain and snow. .

[0031] In the above technical solution, by constructing a confidence assessment model, the reliability of sensor data and machine vision data can be comprehensively and objectively quantified. The sensor data confidence score Cs comprehensively considers the accuracy, stability, and timeliness of the data, accurately reflecting the quality level of the sensor-acquired data. The machine vision data confidence score Cv is evaluated from three dimensions: image quality, recognition accuracy, and environmental adaptability, comprehensively reflecting the reliability of the visual data. The accurate calculation of the confidence scores for both types of data provides a scientific basis for subsequent adaptive weight fusion, enabling dynamic adjustment of weight allocation based on data quality. This improves the rationality and accuracy of multi-source data fusion, providing a strong guarantee for the reliability of the final construction progress recognition results.

[0032] In another technical solution, S4, multimodal data fusion decision-making, specifically includes the following steps: S41. Calculate the basic weights using the following formula: , in, As the basic weight for sensor data, As the basic weight for machine vision data, The confidence level of the sensor data calculated for S31. Calculate the confidence level of the machine vision data for S32; S42. Calculate the sensor weights after environmental adjustment using the following formula. , in, For the sensor weights adjusted for the environment, Based on the sensor weights, Rate the environmental conditions; S43, Historical Performance Feedback Adjustment: , in For adjustment coefficients, These represent the historical accuracy rates of sensor and vision data, respectively.

[0033] In the above technical solution, the first-level decision-making, based on the climbing scaffold status and construction height, quickly achieves preliminary differentiation of construction stages, laying the foundation for subsequent accurate judgment. The second-level decision-making, through weighted fusion of sensor and visual data, fully utilizes the advantages of both types of data, improving the accuracy of the judgment results. The introduction of a comprehensive confidence level enables effective evaluation of the reliability of the judgment results. High-confidence results are output directly, medium-confidence results undergo manual review, and low-confidence results trigger anomaly handling, ensuring the reliability and accuracy of the final construction progress judgment results. This hierarchical decision-making model aligns with the construction process characteristics of steel-concrete composite bridge towers, accurately identifying construction progress and providing a scientific and reliable decision-making basis for construction management. It helps optimize construction scheduling and ensures that construction progress proceeds as planned.

[0034] In another technical solution, S5 specifically includes the following steps: A hierarchical decision-making model was established for the specific construction procedures of steel-concrete composite bridge towers: S51, First-level decision: Based on the preprocessed climbing formwork status in S1 and the current construction height, perform foundation assessment to determine the current possible stage: ①if That is, the height of the climbing frame is located at the bottom of the steel segment, and the construction process may be the steel segment hoisting and positioning, steel segment connection and concrete pouring construction preparation. ②if This means the climbing scaffold is ascending, and the current construction process is likely preparation for concrete pouring. ③if That is, the height of the climbing frame is located at the top of the steel segment, and the construction process may be the hoisting and positioning of the steel segment, preparation for concrete pouring, concrete pouring and concrete curing; in This represents the current height of the steel structure at the top of the tower. This is the current elevation of the top of the climbing scaffold; This indicates that the climbing scaffold is in a stopped state. This indicates that the climbing scaffold is in the climbing phase; S52, Second-level decision-making: Accurate discrimination by fusing sensor and visual data, achieving multi-source data collaboration based on the final weights obtained in S4: , in, To integrate the discrimination results, The final sensor weights output by S4. For the final machine vision weights, The process discrimination result of sensor data S1 The process discrimination result for S2 visual data; S53, Overall Confidence Level: , in To assess the overall confidence level, The sensor confidence level of S31, The visual confidence level of S32.

[0035] S54. Confidence-weighted fusion, based on the reliability of the comprehensive confidence judgment result: , in To assess the overall confidence level, The sensor confidence level of S31, The visual confidence level of S32.

[0036] In another technical solution, step S6 specifically includes the following steps: S61. Perform a consistency check between the sensor discrimination result based on S1 and the visual discrimination result based on S2 to determine whether the multi-source data are coordinated: when At that time, the procedures were completely identical; when At that time, adjacent processes may be in the process transition phase; when At that time, the process intervals exceeded 1, indicating a clear abnormality; Where Ps is the result of the sensor-based discrimination process, Pv is the result of the vision-based discrimination process, and the process number is a discrete value from 1 to 6; when When an anomaly is triggered; S62, Anomaly Type Identification: Accurately locate the source of anomalies by combining the confidence index of S3; Sensor malfunction: ;in The standard deviation of sensor measurements in S31, This is the standard deviation threshold for when the sensor is operating normally; Visual system abnormalities: Where Qv is the image quality coefficient in S32; Environmental interference: ; where Ev is the environmental adaptability coefficient in S32.

[0037] On the other hand, the present invention provides an intelligent recognition system for bridge tower construction progress based on adaptive weights, comprising: Data acquisition module: used to acquire sensor data and machine vision data; Data preprocessing module: used to filter sensor data and perform personnel recognition, key equipment detection and trajectory analysis on machine vision data; Confidence assessment module: used to calculate the confidence level of sensor data and machine vision data; Adaptive weight fusion module: used to dynamically adjust the fusion weights; Fusion Decision Module: Used to establish a hierarchical decision model and output construction progress judgment results; Anomaly detection and correction module: used to identify anomaly types and execute intelligent error correction strategies.

[0038] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. An intelligent identification method for bridge tower construction progress based on adaptive weight, characterized in that, Comprise the following steps: S1, sensor data acquisition and pretreatment Synchronize the collection of climbing frame running state, tower crane load information, key equipment vehicle position trajectory, and perform sliding average filtering processing on the climbing frame height data, and establish a load type identification model based on the load information and the load image; S2, machine vision data acquisition and analysis Obtain multi-dimensional visual information of the construction site through video image acquisition equipment, identify the work type and position of the construction personnel, and perform target detection and trajectory analysis on the key equipment; S3, establish a confidence evaluation model Respectively establish a confidence evaluation model for sensor data and machine vision data; S4, adaptive weight fusion Calculate the basic weight based on the sensor data confidence and the machine vision data confidence output by S3, dynamically adjust the basic weight combined with the environmental condition score to calculate the sensor weight, and then optimize the final fusion weight through the historical accuracy feedback of the sensor and the vision data, so as to realize the dynamic adaptation of data quality and environmental conditions; S5, multi-modal data fusion decision Establish a hierarchical decision model, the first layer is based on the climbing frame state and the current construction height after S1 pretreatment, and preliminarily judges the possible construction sub-process currently; The second layer uses the final fusion weight obtained by S4 to perform confidence weighted fusion on the sensor data discrimination result of S1 and the vision data discrimination result of S2, and outputs the final construction progress discrimination result combined with the comprehensive confidence threshold; S6, anomaly detection and error correction Based on the sensor data discrimination result of S1 and the vision data discrimination result of S2, perform consistency test on the double modal data, when the consistency is lower than the threshold, trigger anomaly detection, identify the sensor anomaly, vision system anomaly or environmental interference type combined with the confidence index of S3, complete error correction through data correction, weight redistribution and historical data auxiliary strategy, and guarantee the reliability of the discrimination result.

2. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 1, characterized in that, Climbing frame operating state comprises climbing frame top level and operating state ; Real-time monitoring of the elevation of the top of the climbing frame using a climbing frame data collector and the operating state of the climbing frame , The following formula is used for sliding average filtering of the climbing frame top elevation: , wherein is a filter window size, is a filtered crawl height; The following formula is used for climbing frame running state determination: , wherein is a height change threshold value, represents a climbing state, represents a stop state.

3. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 1, characterized in that, In S2, the work type and position of the construction personnel are identified, specifically including: S21, the following formula is used to detect the position of the construction personnel: , wherein is the position coordinate of the i-th person, is the confidence, is the number of detected persons; S22, construction personnel work classification identification: based on time sequence feature analysis, the personnel behavior is classified into welding work, reinforcement construction, concrete pouring work.

4. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 3, characterized in that, In S2, the target detection and trajectory analysis of the key equipment are performed, and the key equipment includes concrete tank truck, automobile pump, concrete pouring platform, welding platform, transport ship and steel segment, S23, the following formula is used to distinguish the type of key equipment in the target detection area: , wherein is a bounding box of the i-th key device, is a type of the key device, is a detection confidence; S24, the following formula is used to analyze the trajectory of the key equipment: , And then judge the motion state and work mode of the key equipment based on the trajectory data.

5. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 1, characterized in that, In S3, the confidence evaluation model is established, specifically including: S31, sensor data confidence modeling The following formula is used to calculate the confidence of the sensor data; , wherein Cs is a confidence level of the sensor data, is a precision coefficient, is a stability coefficient, is a timeliness coefficient, is a weight coefficient; Wherein, the precision coefficient is calculated by the following formula: , wherein is the standard deviation of the sensor measurements; The stability coefficient is calculated by the following formula: , wherein is the variance of the current sensor data, is the variance in the normal operating state; The timeliness coefficient is calculated by the following formula: , wherein is a time decay coefficient, is a current timestamp, is a sensor data timestamp ； S32, the following formula is used to calculate the confidence of the machine vision data; , wherein a confidence of the machine vision data, is an image quality coefficient, is an identification accuracy coefficient, is an environmental adaptation coefficient, is a weight coefficient; The image quality coefficient is calculated by the following formula: , wherein the luminance quality wherein is the image average luminance; Contrast quality wherein is the pixel standard deviation; sharpness quality ; wherein is the variance value of the image after Laplacian operator processing; The recognition accuracy coefficient is calculated by the following formula: , wherein is the confidence of the ith detection result, is the overlap degree; The environmental adaptation coefficient is calculated by the following formula: , wherein is a light score, is a weather score; is the weight coefficient for light and weather; For the light score, the value range is [0, 1], and the calculation method is: when the weather is sunny and the light is sufficient ; when the weather is overcast and the light is general , at night or when the light is insufficient ; Weather score, range [0, 1], calculated as: sunny and no rain or snow when; light rain or mist when; heavy rain, heavy fog or rain and snow .

6. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 1, characterized in that, S4, multi-modal data fusion decision, specifically comprising the following steps: S41, calculate the basic weight using the following formula: , wherein, is a sensor data base weight, is a machine vision data base weight, is a sensor data confidence computed for S31, is a machine vision data confidence computed for S32; S42, calculate the sensor weight after environmental adjustment using the following formula, , wherein, is the sensor weight adjusted for the environment, is the base sensor weight, is the environmental condition score; S43, historical performance feedback adjustment: , wherein is a tuning coefficient, are the historical accuracy rates of the sensor and vision data, respectively.

7. The method for intelligent identification of construction progress of a tower based on adaptive weights according to claim 1, characterized in that, S5, specifically comprising the following steps: For the specific construction process of steel-concrete composite bridge tower, a hierarchical decision model is established: S51, first layer decision: based on the pre-processed climbing frame state and the current construction height, the basic discrimination is determined to determine the current possible construction stage: ①if If the height of the climbing frame is located at the bottom of the steel segment, the construction process may be steel segment hoisting and positioning, steel segment connection, and concrete pouring construction preparation. ii. if That is, the climbing frame is climbing, and the construction process in which it is located can be concrete pouring construction preparation. ③if That is, the climbing frame height is located at the top of the steel segment, and the construction process in which the climbing frame height can be located can be steel segment hoisting and positioning, concrete pouring construction preparation, concrete pouring, and concrete curing. wherein is the current tower top steel height; is the current climbing frame top elevation; denotes that the climbing frame is in a stop state, denotes that the climbing frame is in a climbing state; S52, second layer decision: precise discrimination of fusion sensor and visual data, multi-source data collaboration based on the final weight obtained by S4: , wherein, is the fusion discrimination result, is the final sensor weight output by S4, is the final machine vision weight, is the process discrimination result for S1 sensor data, is the process discrimination result for S2 vision data; S53, comprehensive confidence: , wherein is a comprehensive confidence, is a sensor confidence of S31, is a vision confidence of S32; S54, confidence weighted fusion, based on the reliability of the comprehensive confidence decision result: , wherein is a comprehensive confidence, is a sensor confidence of S31, is a vision confidence of S32. 8.The method of claim 1, wherein, S6, specifically comprising the following steps: S61, consistency check based on the sensor discrimination result of S1 and the visual discrimination result of S2, to determine whether the multi-source data is collaborative: When the procedure is exactly the same; When the adjacent process, can be in the process transition phase; When there is a clear anomaly when the process is separated by more than 1, Where Ps is the process result of sensor discrimination, Pv is the process result of visual discrimination, and the discrete value of process number is 1-6; When an exception is triggered; S62, abnormal type identification: accurately locate the abnormal source combined with the confidence index of S3; Sensor anomaly: ; wherein is a standard deviation of sensor measurements in S31, is a standard deviation threshold for when the sensor is working properly; Visual system anomaly: ; wherein Qv is the image quality factor in S32. Environmental interference: ; where Ev is the environmental adaptation coefficient in S32.

9. An intelligent identification system for construction progress of a bridge tower based on adaptive weights, characterized in that, Including: Data acquisition module: used for collecting sensor data and machine vision data; Data preprocessing module: used for filtering sensor data and identifying personnel, detecting key equipment and trajectory analysis of machine vision data; Confidence evaluation module: used for calculating the confidence of sensor data and machine vision data; Adaptive weight fusion module: used for dynamically adjusting the fusion weight; Fusion decision module: used for establishing a hierarchical decision model to output the construction progress discrimination result; Abnormality detection and error correction module: used for identifying abnormal types and executing intelligent error correction strategies.

Images