Method and system for measuring distance of hazardous sources at construction sites based on error correction model

Abstract

The present invention provides a method and system for measuring the distance of dangerous sources at construction sites based on an error correction model, including: retrieving training samples based on a database, screening the training samples, parsing the screened training samples, and extracting corresponding distance measurement results and error evaluations; training a machine learning model based on the distance measurement results and error evaluations, building an error correction model, and associating the error correction model with a distance measurement terminal for an execution process; controlling the distance measurement terminal to perform differential distance measurement on dangerous sources based on environmental characteristics of the construction site, and performing error correction on the differential distance measurement results based on the error correction model associated with the execution process. The accuracy and reliability of distance detection of dangerous sources are ensured, and it is also convenient to accurately and effectively understand the distribution of dangerous sources based on the distance measurement results, thereby ensuring the safe operation of the construction site.

Description

Technical Field

[0001] The present invention relates to the technical field of distance measurement, and in particular to a method and system for measuring distance of dangerous sources at a construction site based on an error correction model. Background Art

[0002] There are various danger sources at the construction site, such as falling from heights, being hit by objects, etc., which pose a threat to the safety of workers. Effective distance measurement of danger sources can improve the safety factor of the construction site and ensure the safety of workers.

[0003] However, the traditional distance measurement method mainly relies on the combination of machines and personnel to perform the distance measurement of hazardous sources, and is easily affected by external environmental factors, which leads to errors in the distance measurement of hazardous sources and affects the accurate assessment of hazardous sources.

[0004] Therefore, in order to overcome the above-mentioned defects, the present invention provides a method and system for measuring the distance between dangerous sources at a construction site based on an error correction model. Summary of the invention

[0005] The present invention provides a construction site hazard source distance measurement method and system based on an error correction model, which are used to retrieve corresponding training samples from a database, screen and analyze the training samples, and accurately and effectively determine the distance measurement results and error evaluations of the training samples, thereby providing data support for building an error correction model. Secondly, a machine learning model is trained through the obtained distance measurement results and error evaluations, thereby reliably building the error correction model, and providing a guarantee for error correction of distance measurement results. Finally, a distance measurement terminal is controlled to perform differential distance measurement on the hazard source according to environmental characteristics of the construction site, and the obtained differential distance measurement results are effectively corrected through the error correction model, thereby ensuring the accuracy and reliability of distance detection on the hazard source, facilitating accurate and effective understanding of the distribution of the hazard source according to the distance measurement results, and ensuring safe operation of the construction site.

[0006] The present invention provides a method for measuring the distance of a hazard source at a construction site based on an error correction model, comprising:

[0007] Step 1: Retrieve training samples based on the database, screen the training samples, parse the screened training samples, and extract the corresponding ranging results and error evaluation;

[0008] Step 2: Train the machine learning model based on the ranging results and error evaluation, build the error correction model, and associate the error correction model with the ranging terminal for execution process;

[0009] Step 3: Based on the environmental characteristics of the construction site, the distance measurement terminal is controlled to perform differential distance measurement on the hazardous source, and the error correction of the differential distance measurement result is performed based on the error correction model associated with the execution process.

[0010] Preferably, a method for measuring the distance of hazardous sources at a construction site based on an error correction model, in step 1, extracting training samples based on a database, includes:

[0011] Obtain sample retrieval tasks based on the management terminal, analyze the sample retrieval tasks, determine the retrieval dimensions of the training samples and the retrieval indicators under each retrieval dimension;

[0012] Determine the database corresponding to each dimension based on the retrieved dimension, and traverse the cached data in the corresponding database based on the retrieved index;

[0013] Based on the traversal result, the cache data to be retrieved in each database is locked, and the cache data to be retrieved is compressed, and dimension tags are generated based on the retrieved dimension to dimensionally mark the data compression results;

[0014] Based on the dimension labeling results, the cached data to be retrieved in each database is retrieved in a distributed manner to obtain training samples.

[0015] Preferably, in a method for measuring the distance between dangerous sources at a construction site based on an error correction model, in step 1, the training samples are screened, including:

[0016] Acquire the obtained training samples, and acquire data screening events for different types of training samples from the management terminal based on sample attributes of the training samples;

[0017] Determine event center features and screening indicators based on data screening events, and analyze corresponding types of training samples based on the screening indicators to obtain target distances between sample centers and event center features of different training samples;

[0018] The attribution degree of each training sample in different types of training samples relative to the event center feature is determined based on the target distance, and each training sample in different types of training samples is screened based on the relative size relationship between the attribution degree and a preset threshold.

[0019] Preferably, a method for measuring the distance of hazardous sources at a construction site based on an error correction model, in step 1, the screened training samples are parsed to extract the corresponding distance measurement results and error evaluation, including:

[0020] Obtain the screened training samples, and perform structural splitting on the training samples based on the text structure of the training samples;

[0021] Extract key data of each part based on the structural splitting results, and perform semantic analysis on the key data based on neural network to determine the summary information of the key data;

[0022] The corresponding ranging results and error evaluations in the training samples are determined based on the summary information, and the ranging results and error evaluations corresponding to each training sample are bound based on a single alignment principle.

[0023] Preferably, a method for measuring the distance of hazardous sources at a construction site based on an error correction model, in step 2, training a machine learning model based on the distance measurement results and error evaluation to construct an error correction model, including:

[0024] Based on the model building requirements, the machine learning model structure is retrieved from the model library, and the parameters of the machine learning model structure are initialized to obtain the target model structure;

[0025] Determine an error measurement indicator and a regularization parameter value for the error correction model based on the model building task, and generate an initial model loss function of the target model structure based on the error measurement indicator;

[0026] Combine the regularization parameter value with the initial model loss function to obtain the model loss function;

[0027] The distance measurement results and error evaluations of different training samples are analyzed respectively to obtain the distance representation values ​​and distance error values ​​corresponding to the distance measurement results and error evaluations, and the distance representation values ​​and distance error values ​​of different training samples of the same distance measurement type are displayed differently in the same coordinate system;

[0028] Obtain the distance measurement index value of each distance measurement type, and use the distance measurement index value as an associated influencing factor to map and display the same object in the difference display result;

[0029] Based on the same-object mapping display results, a parameter control group of different training samples of the same distance measurement type is obtained, and based on the parameter control group, the distance representation value, the distance error value and the interaction weights between the associated influencing factors in each training sample are determined;

[0030] Based on the parameter control group, the interaction weight of each training sample is weighted and comprehensively analyzed to obtain the distance error correction strategy for each distance measurement type;

[0031] Iteratively train the target model structure based on the distance error correction strategy of each distance measurement type, and obtain the value range distribution characteristics of the model loss function based on the iterative training process;

[0032] The minimum value of the model loss function is determined based on the value range distribution characteristics, and the iterative training result corresponding to the minimum value is used as the error correction level of the current ranging type. The error correction levels corresponding to different ranging types are encapsulated in the target model structure to obtain the error correction model.

[0033] Preferably, a method for measuring the distance of a hazardous source at a construction site based on an error correction model, in step 2, the error correction model is associated with an execution process of a distance measurement terminal, including:

[0034] Obtaining the obtained error correction model, and determining a deployment environment for the error correction model based on operation requirements;

[0035] Acquire environmental characteristics of the deployment environment, and determine environmental adaptation parameters for the error correction model based on the environmental characteristics;

[0036] Performing first parameter adaptation on the error correction model based on the environment adaptation parameter, and at the same time, obtaining device attributes of the ranging terminal, and performing second parameter adaptation on the model interface of the error correction model based on the device attributes;

[0037] The execution logic of the error correction model and the ranging terminal is determined based on the ranging execution logic, and the error correction model after the first parameter adaptation and the second parameter adaptation is associated with the ranging terminal through an execution process based on the execution logic.

[0038] Preferably, a method for measuring distance of hazardous sources at a construction site based on an error correction model, in step 3, controlling the distance measuring terminal to perform differential distance measurement on the hazardous sources based on the environmental characteristics of the construction site, comprises:

[0039] Obtaining environmental images of the construction site, analyzing the environmental images, and determining the construction operations and environmental structure of the construction site;

[0040] Based on the construction business and environmental structure, the environmental characteristics of the construction site are obtained. At the same time, based on the environmental image analysis results, the recorded objects in the environmental image are extracted, and the object attributes of the recorded objects are analyzed based on the business execution standards to obtain the hazard sources;

[0041] The corresponding distance measurement scheme is determined based on the environmental characteristics and the hazard source, and the distance measurement terminal is controlled to perform differential distance measurement on the corresponding hazard source according to the distance measurement scheme.

[0042] Preferably, a method for measuring the distance of a hazard source at a construction site based on an error correction model, in step 3, error correction is performed on the difference distance measurement result based on the error correction model associated with the execution process, including:

[0043] Obtain the obtained difference ranging result, and determine the ranging solution source corresponding to the difference ranging result;

[0044] The difference ranging result and the corresponding ranging scheme source are respectively input into the error correction model, and based on the input result, the corresponding error correction layer is called according to the ranging scheme source to analyze the difference ranging result, and the correction value of the difference ranging result is determined;

[0045] The difference distance measurement result is corrected based on the correction value, and the corrected difference distance measurement result is marked based on the identity information of the hazard source;

[0046] The corrected difference ranging result is fed back to the management terminal based on the attribution marking result.

[0047] Preferably, a method for measuring the distance of a hazard source at a construction site based on an error correction model feeds back the corrected difference distance measurement result to a management terminal based on the attribution marking result, including:

[0048] Extract multi-dimensional features of the construction site based on environmental features to obtain a three-dimensional structure of the construction site, and virtually reconstruct the construction site based on the three-dimensional structure to obtain a three-dimensional simulation model of the construction site;

[0049] Acquire the actual geographic location of the ranging terminal at the construction site, and make a first mark in the three-dimensional simulation model based on the actual geographic location;

[0050] Obtaining the corrected difference ranging result, and determining the relative position of the danger source and the ranging terminal based on the first marking result and the corrected difference ranging result;

[0051] Determining a target position of the hazard source based on the relative position, and performing a second marking in the three-dimensional simulation model based on the target position;

[0052] A three-dimensional hazard source visualization model is obtained based on the second marking result, and the three-dimensional hazard source visualization model is recorded and archived.

[0053] The present invention provides a construction site hazard source distance measurement system based on an error correction model, comprising:

[0054] The sample acquisition and processing module is used to retrieve training samples based on the database, screen the training samples, analyze the screened training samples, and extract the corresponding ranging results and error evaluation;

[0055] A model building module is used to train the machine learning model based on the ranging results and error evaluation, build an error correction model, and associate the error correction model with the ranging terminal for execution process;

[0056] The distance measurement and correction module is used to control the distance measurement terminal to perform differential distance measurement on the hazardous source based on the environmental characteristics of the construction work, and to perform error correction on the differential distance measurement results based on the error correction model associated with the execution process.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1. By retrieving the corresponding training samples from the database, screening and analyzing the training samples, the ranging results and error evaluations of the training samples can be accurately and effectively determined, providing data support for building an error correction model. Secondly, the machine learning model is trained through the obtained ranging results and error evaluations to achieve reliable construction of the error correction model, providing a guarantee for error correction of the ranging results. Finally, the ranging terminal is controlled to perform differential ranging of the hazardous source according to the environmental characteristics of the construction site, and the differential ranging results obtained are effectively corrected through the error correction model, ensuring the accuracy and reliability of the distance detection of the hazardous source, and facilitating the accurate and effective understanding of the distribution of the hazardous source based on the ranging results, ensuring the safe operation of the construction site.

[0059] 2. By parsing the sample retrieval tasks obtained by the management terminal, the retrieval dimensions and the retrieval indicators under each retrieval dimension can be accurately and effectively determined. Finally, the cached data to be retrieved under each dimension can be effectively retrieved according to the retrieval dimensions and retrieval indicators, and the training samples can be accurately and effectively acquired, providing reliable data support for building the error correction model.

[0060] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0061] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0063] Figure 1 It is a flow chart of a method for measuring the distance of dangerous sources at a construction site based on an error correction model in an embodiment of the present invention;

[0064] Figure 2 This is a flowchart of step 1 in a method for measuring distance to hazardous sources at a construction site based on an error correction model in an embodiment of the present invention;

[0065] Figure 3 It is a structural diagram of a construction site hazard source distance measurement system based on an error correction model in an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0067] Embodiment 1:

[0068] This embodiment provides a method for measuring the distance of dangerous sources at a construction site based on an error correction model. Figure 1 As shown, including:

[0069] Step 1: Retrieve training samples based on the database, screen the training samples, parse the screened training samples, and extract the corresponding ranging results and error evaluation;

[0070] Step 2: Train the machine learning model based on the ranging results and error evaluation, build the error correction model, and associate the error correction model with the ranging terminal for execution process;

[0071] Step 3: Based on the environmental characteristics of the construction site, the distance measurement terminal is controlled to perform differential distance measurement on the hazardous source, and the error correction of the differential distance measurement result is performed based on the error correction model associated with the execution process.

[0072] In this embodiment, the distance measurement result refers to the specific distance measurement corresponding to each training sample, that is, the determined distance of the danger source relative to the distance measurement device.

[0073] In this embodiment, the error evaluation refers to the error between the actual distance determined based on the distance measurement result and the measured distance.

[0074] In this embodiment, the machine learning model is a pre-set model framework.

[0075] In this embodiment, the execution process association refers to connecting the error correction model with the ranging terminal, so as to facilitate directly inputting the ranging result obtained by the ranging terminal into the error correction model for result correction processing.

[0076] In this embodiment, the environmental characteristics refer to information such as the construction type of the construction site and the type of buildings included in the construction site.

[0077] In this embodiment, the hazard source refers to the hazard factors that exist at the construction site and affect normal construction, such as fire.

[0078] In this embodiment, the difference ranging refers to selecting a suitable ranging method to perform corresponding ranging according to the environmental characteristics and the type of the hazard source, including ultrasonic ranging, laser ranging, infrared ranging, and wireless signal ranging.

[0079] The working principle and beneficial effects of the above technical solution are: by retrieving corresponding training samples from the database, screening and parsing the training samples, the ranging results and error evaluations of the training samples can be accurately and effectively determined, providing data support for building an error correction model; secondly, the machine learning model is trained through the obtained ranging results and error evaluations to achieve reliable construction of the error correction model, providing a guarantee for error correction of the ranging results; finally, the ranging terminal is controlled to perform differential ranging of the hazardous source according to the environmental characteristics of the construction site, and the obtained differential ranging results are effectively corrected through the error correction model, ensuring the accuracy and reliability of the distance detection of the hazardous source, and also facilitating the accurate and effective understanding of the distribution of the hazardous source based on the ranging results, ensuring the safe operation of the construction site.

[0080] Embodiment 2:

[0081] Based on Example 1, this example provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. Figure 2 As shown, in step 1, training samples are retrieved based on the database, including:

[0082] Step 101: obtaining a sample retrieval task based on a management terminal, parsing the sample retrieval task, and determining a retrieval dimension for training samples and a retrieval index under each retrieval dimension;

[0083] Step 102: Determine the database corresponding to each dimension based on the retrieved dimension, and traverse the cached data in the corresponding database based on the retrieved index;

[0084] Step 103: based on the traversal result, the cache data to be retrieved in each database is locked, and data compression is performed on the cache data to be retrieved, and dimension tags are generated based on the retrieved dimension to dimensionally mark the data compression result;

[0085] Step 104: Based on the dimension labeling results, distributed retrieval is performed on the cached data to be retrieved in each database to obtain training samples.

[0086] In this embodiment, the sample retrieval task is set in advance by the management terminal and is used to characterize the type of data to be retrieved and the corresponding retrieval amount.

[0087] In this embodiment, the retrieval dimension refers to the type of data that needs to be retrieved, which may specifically be data corresponding to ultrasonic ranging, data corresponding to infrared ranging, data corresponding to laser ranging, and data corresponding to wireless signal ranging.

[0088] In this embodiment, the retrieval index refers to the time information that needs to be retrieved under each retrieval dimension, etc., which is used as a reference for locking the data that needs to be retrieved.

[0089] In this embodiment, dimension marking refers to marking the data compression result with a corresponding type, so as to facilitate the distinction between different types of cache data to be retrieved.

[0090] The working principle and beneficial effects of the above technical solution are: by parsing the sample retrieval tasks obtained by the management terminal, the retrieval dimensions and the retrieval indicators under each retrieval dimension can be accurately and effectively determined, and finally the cached data to be retrieved under each dimension can be effectively retrieved according to the retrieval dimensions and retrieval indicators, so as to achieve accurate and effective acquisition of training samples, and provide reliable data support for building an error correction model.

[0091] Embodiment 3:

[0092] On the basis of Example 1, this embodiment provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. In step 1, the training samples are screened, including:

[0093] Acquire the obtained training samples, and acquire data screening events for different types of training samples from the management terminal based on sample attributes of the training samples;

[0094] Determine event center features and screening indicators based on data screening events, and analyze corresponding types of training samples based on the screening indicators to obtain target distances between sample centers and event center features of different training samples;

[0095] The attribution degree of each training sample in different types of training samples relative to the event center feature is determined based on the target distance, and each training sample in different types of training samples is screened based on the relative size relationship between the attribution degree and a preset threshold.

[0096] In this embodiment, the sample attribute refers to the sample type of the training sample.

[0097] In this embodiment, the data screening event refers to the requirement for screening different types of training samples, that is, the purpose of screening the data, for example, the requirement for data integrity, retained data parameters, etc.

[0098] In this embodiment, the event center feature refers to a feature that can characterize the core theme of the data screening event, that is, a parameter that can characterize the purpose of data screening.

[0099] In this embodiment, the screening index refers to information that needs to be referenced when screening training samples.

[0100] In this embodiment, the sample center refers to the core data feature of the training sample, so that it is convenient to determine whether the training sample can be retained through the sample center.

[0101] In this embodiment, the target distance is used to characterize the probability that the sample center belongs to the event center feature. The smaller the distance is, the more likely the corresponding training sample belongs to the data screening event.

[0102] In this embodiment, the attribution degree is used to characterize the probability of consistency between event center features of different training samples and data screening events, wherein the preset threshold is set in advance.

[0103] The working principle and beneficial effects of the above technical solution are: by determining the data screening events for training samples and determining the event center features corresponding to the data screening events, the target distance between the event center features and the sample centers of different training samples can be accurately and effectively determined. Finally, the degree of attribution of the training samples relative to the data screening events can be locked according to the target distance, thereby achieving accurate and effective screening of the training samples and ensuring the accuracy and reliability of the training samples.

[0104] Embodiment 4:

[0105] Based on Example 1, this example provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. In step 1, the screened training samples are parsed to extract the corresponding distance measurement results and error evaluation, including:

[0106] Obtain the screened training samples, and perform structural splitting on the training samples based on the text structure of the training samples;

[0107] Extract key data of each part based on the structural splitting results, and perform semantic analysis on the key data based on neural network to determine the summary information of the key data;

[0108] The corresponding ranging results and error evaluations in the training samples are determined based on the summary information, and the ranging results and error evaluations corresponding to each training sample are bound based on a single alignment principle.

[0109] In this embodiment, the text structure refers to the composition of the training sample, for example, it may be a text header and a data content segment.

[0110] In this embodiment, key data refers to the core data fragments of each part after the training samples are structurally split.

[0111] In this embodiment, the summary information refers to data parameters that can represent the main idea of ​​the key information.

[0112] In this embodiment, the single alignment principle means that the distance measurement result and the error evaluation of each training sample correspond one to one.

[0113] The working principle and beneficial effects of the above technical solution are: by parsing the screened training samples, the training samples are split, and the key information of each part after the split is extracted, so as to accurately and effectively determine the ranging results and error evaluation of the training samples based on the key information, thereby facilitating the effective training of the machine learning model and providing convenience and guarantee for the construction of the error correction model.

[0114] Embodiment 5:

[0115] On the basis of Example 1, this embodiment provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. In step 2, a machine learning model is trained based on the distance measurement result and the error evaluation to construct an error correction model, including:

[0116] Based on the model building requirements, the machine learning model structure is retrieved from the model library, and the parameters of the machine learning model structure are initialized to obtain the target model structure;

[0117] Determine an error measurement indicator and a regularization parameter value for the error correction model based on the model building task, and generate an initial model loss function of the target model structure based on the error measurement indicator;

[0118] Combine the regularization parameter value with the initial model loss function to obtain the model loss function;

[0119] The distance measurement results and error evaluations of different training samples are analyzed respectively to obtain the distance representation values ​​and distance error values ​​corresponding to the distance measurement results and error evaluations, and the distance representation values ​​and distance error values ​​of different training samples of the same distance measurement type are displayed differently in the same coordinate system;

[0120] Obtain the distance measurement index value of each distance measurement type, and use the distance measurement index value as an associated influencing factor to map and display the same object in the difference display result;

[0121] Based on the same-object mapping display results, a parameter control group of different training samples of the same distance measurement type is obtained, and based on the parameter control group, the distance representation value, the distance error value and the interaction weights between the associated influencing factors in each training sample are determined;

[0122] Based on the parameter control group, the interaction weight of each training sample is weighted and comprehensively analyzed to obtain the distance error correction strategy for each distance measurement type;

[0123] Iteratively train the target model structure based on the distance error correction strategy of each distance measurement type, and obtain the value range distribution characteristics of the model loss function based on the iterative training process;

[0124] The minimum value of the model loss function is determined based on the value range distribution characteristics, and the iterative training result corresponding to the minimum value is used as the error correction level of the current ranging type. The error correction levels corresponding to different ranging types are encapsulated in the target model structure to obtain the error correction model.

[0125] In this embodiment, the model construction requirements are known in advance and are used to characterize the construction requirements and limiting conditions for the error correction model.

[0126] In this embodiment, parameter initialization refers to unifying the model parameters of the machine learning model structure to facilitate targeted training of the initialized machine learning model structure, wherein the target model structure is the result obtained after parameter initialization of the machine learning model structure.

[0127] In this embodiment, the model building task is known in advance and is used to characterize the building standards and requirements for the error correction model.

[0128] In this embodiment, the error measurement index is a data parameter used to define the performance of the error correction model, including accuracy and processing efficiency.

[0129] In this embodiment, the regularization parameter value is a parameter used to prevent the model from overfitting during the training process and is used to assist in the training of the model.

[0130] In this embodiment, the initial model loss function refers to a model loss function generated according to an error measurement indicator, and no regularization parameter value is added.

[0131] In this embodiment, the model loss function is used to calculate the performance loss of the model when the target model structure is trained.

[0132] In this embodiment, the distance characterization value and the distance error value refer to specific values ​​determined according to the distance measurement result and the error evaluation result.

[0133] In this embodiment, the distance measurement index value refers to the interference factors encountered when performing distance measurement of the hazard source represented by the training sample, such as the environment and obstacles.

[0134] In this embodiment, the same-object mapping display refers to associating the monitoring objects corresponding to the associated influencing factors in the difference display results, so as to facilitate the determination of the influencing factors existing in the distance measurement of different hazard sources.

[0135] In this embodiment, the interaction weight is used to characterize the mutual limitation degree between the distance characterization value, the distance error value and the associated influencing factors in each training sample.

[0136] In this embodiment, the range distribution feature is used to characterize the distribution of model performance loss obtained after calculating each training result through the model loss function during the iterative training of the target model structure.

[0137] In this embodiment, the error correction level refers to using the training results corresponding to each ranging type as a part of the final error correction model.

[0138] The working principle and beneficial effects of the above technical solution are: by retrieving the corresponding machine learning model structure from the model library and analyzing the retrieved machine learning model structure, the model loss function can be accurately and effectively obtained. Secondly, the ranging results and error evaluations of different training samples are analyzed to accurately and effectively determine the distance representation values ​​and distance error values ​​corresponding to different training samples. At the same time, the associated influencing factors corresponding to each training sample are comprehensively considered to accurately and effectively determine the distance representation values, distance error values ​​and interaction weights between the associated influencing factors of different training samples. Finally, the distance error correction strategy corresponding to each ranging type is determined through the interaction weight, and the target model structure is iteratively trained according to the distance error correction strategy. A reliable and effective error correction model is obtained according to the iterative training results, which provides reliable guarantee for the distance measurement of hazardous sources at the construction site and facilitates the effective understanding of the location of hazardous sources at the construction site.

[0139] Embodiment 6:

[0140] Based on Example 1, this embodiment provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. In step 2, the error correction model is associated with an execution process of a ranging terminal, including:

[0141] Obtaining the obtained error correction model, and determining a deployment environment for the error correction model based on operation requirements;

[0142] Acquire environmental characteristics of the deployment environment, and determine environmental adaptation parameters for the error correction model based on the environmental characteristics;

[0143] Performing first parameter adaptation on the error correction model based on the environment adaptation parameter, and at the same time, obtaining device attributes of the ranging terminal, and performing second parameter adaptation on the model interface of the error correction model based on the device attributes;

[0144] The execution logic of the error correction model and the ranging terminal is determined based on the ranging execution logic, and the error correction model after the first parameter adaptation and the second parameter adaptation is associated with the ranging terminal through an execution process based on the execution logic.

[0145] In this embodiment, the operation requirements are known in advance and are used to characterize the deployment requirements of the error correction model, including deployment environment conditions and conditions for docking with other terminals.

[0146] In this embodiment, the environmental characteristics refer to the specific conditions of the environment in which the error correction model needs to be deployed, including compatibility with other application homes, etc.

[0147] In this embodiment, the environment adaptation parameters refer to the adjustment parameters that need to be performed when the error correction model is deployed in the deployment environment, so as to ensure that the error correction model can be effectively deployed in the corresponding deployment environment, wherein the first parameter adaptation is the specific parameter adjustment process of the error correction model according to the environment adaptation parameters.

[0148] In this embodiment, the device attribute refers to the terminal type of the ranging terminal and the specific operating conditions corresponding to the ranging terminal when it is working.

[0149] In this embodiment, the second parameter adaptation refers to adjusting parameters of the model interface of the error correction model according to the device properties of the ranging terminal, so as to ensure that the error correction model can be accurately and effectively connected to the ranging terminal.

[0150] In this embodiment, the execution logic is used to characterize the working sequence between the ranging terminal and the error correction model.

[0151] The working principle and beneficial effect of the above technical solution are: by adapting the environmental parameters and model interface of the error correction model, and accurately and effectively associating the adapted error correction model with the ranging terminal, the distance detected by the ranging terminal is transmitted to the error correction model for timely and effective analysis, thereby ensuring the accuracy and reliability of the final hazardous source ranging.

[0152] Embodiment 7:

[0153] Based on Example 1, this embodiment provides a method for measuring the distance of a hazardous source at a construction site based on an error correction model. In step 3, the distance measuring terminal is controlled to perform differential distance measurement on the hazardous source based on the environmental characteristics of the construction site, including:

[0154] Obtaining environmental images of the construction site, analyzing the environmental images, and determining the construction operations and environmental structure of the construction site;

[0155] Based on the construction business and environmental structure, the environmental characteristics of the construction site are obtained. At the same time, based on the environmental image analysis results, the recorded objects in the environmental image are extracted, and the object attributes of the recorded objects are analyzed based on the business execution standards to obtain the hazard sources;

[0156] The corresponding distance measurement scheme is determined based on the environmental characteristics and the hazard source, and the distance measurement terminal is controlled to perform differential distance measurement on the corresponding hazard source according to the distance measurement scheme.

[0157] In this embodiment, the construction business refers to the type of construction currently being carried out at the construction site, such as construction or firefighting.

[0158] In this embodiment, the environmental structure refers to whether there are obstacles blocking the construction site and the placement of other objects.

[0159] In this embodiment, the environmental characteristics refer to parameters that are obtained by integrating the construction operations and environmental structure of the construction site and are capable of characterizing the specific conditions of the construction site.

[0160] In this embodiment, the recorded objects refer to various types of objects contained in the environment image, including dangerous sources and other non-dangerous objects.

[0161] In this embodiment, the business execution standard is set in advance and is used to represent the safety rules that need to be followed at the construction site.

[0162] In this embodiment, the object attribute refers to the object type of each recorded object and its form in the construction site.

[0163] In this embodiment, determining the corresponding ranging scheme based on environmental characteristics and hazard sources refers to determining a suitable ranging method according to the environmental characteristics and the type of hazard source, which can be one of ultrasonic ranging, infrared ranging, laser ranging and wireless signal ranging.

[0164] The working principle and beneficial effects of the above technical solution are: by analyzing the environmental image of the construction site, the environmental characteristics and hazard sources of the construction site can be accurately and effectively determined, thereby facilitating the accurate and effective determination of the ranging scheme based on the environmental characteristics and hazard sources, and controlling the ranging terminal to perform accurate and effective ranging operations on the hazard sources according to the ranging scheme, thereby ensuring the accuracy and reliability of the ranging.

[0165] Embodiment 8:

[0166] On the basis of Example 1, this embodiment provides a method for measuring the distance of a hazard source at a construction site based on an error correction model. In step 3, error correction is performed on the difference distance measurement result based on the error correction model associated with the execution process, including:

[0167] Obtain the obtained difference ranging result, and determine the ranging solution source corresponding to the difference ranging result;

[0168] The difference ranging result and the corresponding ranging scheme source are respectively input into the error correction model, and based on the input result, the corresponding error correction layer is called according to the ranging scheme source to analyze the difference ranging result, and the correction value of the difference ranging result is determined;

[0169] The difference distance measurement result is corrected based on the correction value, and the corrected difference distance measurement result is marked based on the identity information of the hazard source;

[0170] The corrected difference ranging result is fed back to the management terminal based on the attribution marking result.

[0171] In this embodiment, the ranging solution source refers to a specific ranging method adopted by the difference ranging result.

[0172] In this embodiment, the identity information refers to a parameter that can characterize the type of the hazard source, for example, the type of the hazard source.

[0173] In this embodiment, the attribution mark refers to marking the corrected difference ranging result according to the obtained identity information, in order to identify different danger sources in the construction site.

[0174] The working principle and beneficial effects of the above technical solution are: by inputting the obtained difference ranging results and the corresponding ranging scheme sources into the error correction model for processing, the correction value corresponding to the ranging result is accurately and effectively determined; secondly, the corrected difference ranging results are marked according to the identity information of the hazardous source, and the marked results are fed back to the management terminal, so that the management terminal can accurately and effectively understand the situation of the hazardous source in the construction site, and it is also convenient to locate the hazardous source in time when an abnormal situation occurs, thereby ensuring the safe operation of the construction site.

[0175] Embodiment 9:

[0176] Based on Example 8, this embodiment provides a method for measuring the distance of a hazard source at a construction site based on an error correction model, and feeds back the corrected difference distance measurement result to a management terminal based on the attribution marking result, including:

[0177] Extract multi-dimensional features of the construction site based on environmental features to obtain a three-dimensional structure of the construction site, and virtually reconstruct the construction site based on the three-dimensional structure to obtain a three-dimensional simulation model of the construction site;

[0178] Acquire the actual geographic location of the ranging terminal at the construction site, and make a first mark in the three-dimensional simulation model based on the actual geographic location;

[0179] Obtaining the corrected difference ranging result, and determining the relative position of the danger source and the ranging terminal based on the first marking result and the corrected difference ranging result;

[0180] Determining a target position of the hazard source based on the relative position, and performing a second marking in the three-dimensional simulation model based on the target position;

[0181] A three-dimensional hazard source visualization model is obtained based on the second marking result, and the three-dimensional hazard source visualization model is recorded and archived.

[0182] In this embodiment, multi-dimensional feature extraction refers to extracting features from different angles of the construction site in order to simulate and reconstruct the construction site.

[0183] In this embodiment, the first mark refers to marking in the three-dimensional virtual model according to the temporal and geographic location of the ranging terminal in the construction site.

[0184] In this embodiment, the target location refers to the specific location of the hazard source in the construction site.

[0185] In this embodiment, the second marking refers to marking the specific location of the danger source in the three-dimensional simulation model.

[0186] In this embodiment, the three-dimensional hazard source visualization model refers to the final model obtained after marking the hazard source in the obtained three-dimensional simulation model.

[0187] The working principle and beneficial effects of the above technical solution are: by extracting multi-dimensional features of the construction site, the three-dimensional structure of the construction site can be accurately and effectively determined; secondly, the construction site is virtually reconstructed according to the obtained three-dimensional structure, so that the three-dimensional simulation model of the construction site can be accurately and effectively built; finally, the positions of the ranging terminal and the hazard source are marked in the obtained three-dimensional simulation model, so as to accurately and effectively construct the three-dimensional hazard source visualization model, which is convenient for the management terminal to effectively understand the specific location distribution of the hazard source, thereby improving the safety factor of the construction site.

[0188] Embodiment 10:

[0189] This embodiment provides a construction site hazard source distance measurement system based on an error correction model. Figure 3 As shown, including:

[0190] The sample acquisition and processing module is used to retrieve training samples based on the database, screen the training samples, analyze the screened training samples, and extract the corresponding ranging results and error evaluation;

[0191] A model building module is used to train the machine learning model based on the ranging results and error evaluation, build an error correction model, and associate the error correction model with the ranging terminal for execution process;

[0192] The distance measurement and correction module is used to control the distance measurement terminal to perform differential distance measurement on the hazardous source based on the environmental characteristics of the construction work, and to perform error correction on the differential distance measurement results based on the error correction model associated with the execution process.

[0193] The working principle and beneficial effects of the above technical solution are: by retrieving corresponding training samples from the database, screening and parsing the training samples, the ranging results and error evaluations of the training samples can be accurately and effectively determined, providing data support for building an error correction model; secondly, the machine learning model is trained through the obtained ranging results and error evaluations to achieve reliable construction of the error correction model, providing a guarantee for error correction of the ranging results; finally, the ranging terminal is controlled to perform differential ranging of the hazardous source according to the environmental characteristics of the construction site, and the obtained differential ranging results are effectively corrected through the error correction model, ensuring the accuracy and reliability of the distance detection of the hazardous source, and also facilitating the accurate and effective understanding of the distribution of the hazardous source based on the ranging results, ensuring the safe operation of the construction site.

[0194] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for measuring the distance of hazardous sources at a construction site based on an error correction model, characterized in that: include: Step 1: Retrieve training samples based on the database, screen the training samples, parse the screened training samples, and extract the corresponding ranging results and error evaluation; Step 2: Train the machine learning model based on the ranging results and error evaluation, build the error correction model, and associate the error correction model with the ranging terminal for execution process; Step 3: Based on the environmental characteristics of the construction site, the distance measurement terminal is controlled to perform differential distance measurement on the hazardous source, and the error correction of the differential distance measurement result is performed based on the error correction model associated with the execution process; In step 2, the machine learning model is trained based on the ranging results and error evaluation to build an error correction model, including: Based on the model building requirements, the machine learning model structure is retrieved from the model library, and the parameters of the machine learning model structure are initialized to obtain the target model structure; Determine an error measurement indicator and a regularization parameter value for the error correction model based on the model building task, and generate an initial model loss function of the target model structure based on the error measurement indicator; Combine the regularization parameter value with the initial model loss function to obtain the model loss function; The distance measurement results and error evaluations of different training samples are analyzed respectively to obtain the distance representation values ​​and distance error values ​​corresponding to the distance measurement results and error evaluations, and the distance representation values ​​and distance error values ​​of different training samples of the same distance measurement type are displayed differently in the same coordinate system; Obtain the distance measurement index value of each distance measurement type, and use the distance measurement index value as an associated influencing factor to perform same-object mapping display in the difference display result; Based on the same-object mapping display results, a parameter control group of different training samples of the same distance measurement type is obtained, and based on the parameter control group, the distance representation value, the distance error value and the interaction weights between the associated influencing factors in each training sample are determined; Based on the parameter control group, the interaction weight of each training sample is weighted and comprehensively analyzed to obtain the distance error correction strategy for each distance measurement type; Iteratively train the target model structure based on the distance error correction strategy of each distance measurement type, and obtain the value range distribution characteristics of the model loss function based on the iterative training process; The minimum value of the model loss function is determined based on the value range distribution characteristics, and the iterative training result corresponding to the minimum value is used as the error correction level of the current ranging type. The error correction levels corresponding to different ranging types are encapsulated in the target model structure to obtain the error correction model.

2. A construction site hazard source distance measurement method based on an error correction model according to claim 1, characterized in that: In step 1, training samples are retrieved based on the database, including: Obtain sample retrieval tasks based on the management terminal, analyze the sample retrieval tasks, determine the retrieval dimensions of the training samples and the retrieval indicators under each retrieval dimension; Determine the database corresponding to each dimension based on the retrieved dimension, and traverse the cached data in the corresponding database based on the retrieved index; Based on the traversal result, the cache data to be retrieved in each database is locked, and the cache data to be retrieved is compressed, and dimension tags are generated based on the retrieved dimension to dimensionally mark the data compression results; Based on the dimension labeling results, the cached data to be retrieved in each database is retrieved in a distributed manner to obtain training samples.

3. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 1 is characterized in that: In step 1, the training samples are screened, including: Acquire the obtained training samples, and acquire data screening events for different types of training samples from the management terminal based on sample attributes of the training samples; Determine event center features and screening indicators based on data screening events, and analyze corresponding types of training samples based on the screening indicators to obtain target distances between sample centers and event center features of different training samples; The attribution degree of each training sample in different types of training samples relative to the event center feature is determined based on the target distance, and each training sample in different types of training samples is screened based on the relative size relationship between the attribution degree and a preset threshold.

4. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 1, characterized in that: In step 1, the screened training samples are parsed to extract the corresponding ranging results and error evaluation, including: Obtain the screened training samples, and perform structural splitting on the training samples based on the text structure of the training samples; Extract key data of each part based on the structural splitting results, and perform semantic analysis on the key data based on the neural network to determine the summary information of the key data; The corresponding ranging results and error evaluations in the training samples are determined based on the summary information, and the ranging results and error evaluations corresponding to each training sample are bound based on a single alignment principle.

5. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 1, characterized in that: In step 2, the error correction model is associated with the ranging terminal to perform an execution process, including: Obtaining the obtained error correction model, and determining a deployment environment for the error correction model based on operation requirements; Acquire environmental characteristics of the deployment environment, and determine environmental adaptation parameters for the error correction model based on the environmental characteristics; Performing first parameter adaptation on the error correction model based on the environment adaptation parameter, and at the same time, obtaining device attributes of the ranging terminal, and performing second parameter adaptation on the model interface of the error correction model based on the device attributes; The execution logic of the error correction model and the ranging terminal is determined based on the ranging execution logic, and the error correction model after the first parameter adaptation and the second parameter adaptation is associated with the ranging terminal through an execution process based on the execution logic.

6. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 1, characterized in that: In step 3, the distance measurement terminal is controlled to perform differential distance measurement on the hazard source based on the environmental characteristics of the construction site, including: Obtaining environmental images of the construction site, analyzing the environmental images, and determining the construction operations and environmental structure of the construction site; Based on the construction business and environmental structure, the environmental characteristics of the construction site are obtained. At the same time, based on the environmental image analysis results, the recorded objects in the environmental image are extracted, and the object attributes of the recorded objects are analyzed based on the business execution standards to obtain the hazard sources; The corresponding distance measurement scheme is determined based on the environmental characteristics and the hazard source, and the distance measurement terminal is controlled to perform differential distance measurement on the corresponding hazard source according to the distance measurement scheme.

7. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 1, characterized in that: In step 3, error correction is performed on the difference ranging result based on the error correction model associated with the execution process, including: Obtain the obtained difference ranging result, and determine the ranging solution source corresponding to the difference ranging result; The difference ranging result and the corresponding ranging scheme source are respectively input into the error correction model, and based on the input result, the corresponding error correction layer is called according to the ranging scheme source to analyze the difference ranging result, and the correction value of the difference ranging result is determined; The difference distance measurement result is corrected based on the correction value, and the corrected difference distance measurement result is marked based on the identity information of the hazard source; The corrected difference ranging result is fed back to the management terminal based on the attribution marking result.

8. The method for measuring the distance between dangerous sources at a construction site based on an error correction model according to claim 7, characterized in that: Feedback the corrected difference ranging result to the management terminal based on the attribution marking result, including: Extract multi-dimensional features of the construction site based on environmental features to obtain a three-dimensional structure of the construction site, and virtually reconstruct the construction site based on the three-dimensional structure to obtain a three-dimensional simulation model of the construction site; Acquire the actual geographic location of the ranging terminal at the construction site, and make a first mark in the three-dimensional simulation model based on the actual geographic location; Obtaining the corrected difference ranging result, and determining the relative position of the danger source and the ranging terminal based on the first marking result and the corrected difference ranging result; Determining a target position of the hazard source based on the relative position, and performing a second marking in the three-dimensional simulation model based on the target position; A three-dimensional hazard source visualization model is obtained based on the second marking result, and the three-dimensional hazard source visualization model is recorded and archived.

9. A construction site hazard source distance measurement system based on an error correction model, characterized in that: include: The sample acquisition and processing module is used to retrieve training samples based on the database, screen the training samples, analyze the screened training samples, and extract the corresponding ranging results and error evaluation; A model building module is used to train the machine learning model based on the ranging results and error evaluation, build an error correction model, and associate the error correction model with the ranging terminal for execution process; The distance measurement and correction module is used to control the distance measurement terminal to perform differential distance measurement on the hazardous source based on the environmental characteristics of the construction work, and to perform error correction on the differential distance measurement results based on the error correction model associated with the execution process; Among them, the model building module includes: Based on the model building requirements, the machine learning model structure is retrieved from the model library, and the parameters of the machine learning model structure are initialized to obtain the target model structure; Determine an error measurement indicator and a regularization parameter value for the error correction model based on the model building task, and generate an initial model loss function of the target model structure based on the error measurement indicator; Combine the regularization parameter value with the initial model loss function to obtain the model loss function; The distance measurement results and error evaluations of different training samples are analyzed respectively to obtain the distance representation values ​​and distance error values ​​corresponding to the distance measurement results and error evaluations, and the distance representation values ​​and distance error values ​​of different training samples of the same distance measurement type are displayed differently in the same coordinate system; Obtain the distance measurement index value of each distance measurement type, and use the distance measurement index value as an associated influencing factor to perform same-object mapping display in the difference display result; Based on the same-object mapping display results, a parameter control group of different training samples of the same distance measurement type is obtained, and based on the parameter control group, the distance representation value, the distance error value and the interaction weights between the associated influencing factors in each training sample are determined; Based on the parameter control group, the interaction weight of each training sample is weighted and comprehensively analyzed to obtain the distance error correction strategy for each distance measurement type; Iteratively train the target model structure based on the distance error correction strategy of each distance measurement type, and obtain the value range distribution characteristics of the model loss function based on the iterative training process; The minimum value of the model loss function is determined based on the value range distribution characteristics, and the iterative training result corresponding to the minimum value is used as the error correction level of the current ranging type. The error correction levels corresponding to different ranging types are encapsulated in the target model structure to obtain the error correction model.

Images