Roof test lane structure real-time evaluation system and evaluation method

Abstract

The invention provides a real-time evaluation system and evaluation method for a roof test lane structure, and belongs to the technical field of roof lanes, the real-time evaluation system comprises a data acquisition module, a data processing module, a data evaluation module and an early warning module, the data acquisition module is used for acquiring various parameters of the roof test lane structure in real time; the data processing module is used for carrying out preprocessing, feature extraction, classification and fusion on collected data, the data evaluation module is used for carrying out real-time evaluation on a roof test lane structure, and the data evaluation module comprises a feature extraction unit, a state space reconstruction unit, a model coupling unit and a weight assignment unit. The early warning module sends out an early warning signal when the evaluation result shows that the roof test lane structure has potential safety hazards, and the problems that traditional manual detection is long in time, poor in real-time performance and low in accuracy are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of roof lanes, and specifically relates to a real-time evaluation system and evaluation method for the structure of a roof test lane. Background Art

[0002] With the rapid development of China's economy, applications such as roof parking lots and roof greening of various buildings are becoming more and more widespread. As an important part of the building roof, the stability of the structure of the roof test lane is directly related to people's lives and property safety. However, the existing evaluation methods for the structure of the roof test lane mainly rely on manual inspection. However, manual inspection has a long cycle, and it is impossible to grasp the health status of the roof test lane structure in real time. Moreover, the subjective factors have a great influence and the accuracy is low, making it difficult to comprehensively understand the overall condition of the roof test lane structure.

[0003] To solve the above problems, the present invention proposes a real-time evaluation system and evaluation method for the structure of a roof test lane, aiming to achieve real-time, accurate, and comprehensive evaluation of the structure of the roof test lane. Summary of the Invention

[0004] The embodiments of the present invention provide a real-time evaluation system and evaluation method for the structure of a roof test lane, which solve the problems of long time, poor real-time performance, and low accuracy of traditional manual inspection.

[0005] In view of the above problems, the technical solution proposed by the present invention is:

[0006] The present invention provides a real-time evaluation system for the structure of a roof test lane, including a data acquisition module, a data processing module, a data evaluation module, and an early warning module;

[0007] The data acquisition module is used to collect various parameters of the structure of the roof test lane in real time;

[0008] The data processing module is used to preprocess, extract features, classify, and fuse the collected data;

[0009] The data evaluation module is used to perform real-time evaluation on the structure of the roof test lane. The data evaluation module includes a feature extraction unit, a state space reconstruction unit, a model coupling unit, and a weight assignment unit;

[0010] The feature extraction unit uses the BOW method to extract the required features from the data from the data processing module;

[0011] The state space reconstruction unit divides the time series data, and applies the divided data to the PSR model for state space reconstruction operation;

[0012] The model coupling unit constructs a coupling model by coupling the feature extraction unit and the state space reconstruction unit;

[0013] The weight assignment unit processes the subjective scores of experts using intuitionistic fuzzy binary semantics and calculates the weight values of each feature using the BWM method;

[0014] When the evaluation result shows that there are potential safety hazards in the roof test lane structure, the warning module issues a warning signal.

[0015] As a preferred technical solution of the present invention, the data acquisition module monitors the roof test lane structure using sensors, including strain gauges, displacement sensors, temperature and humidity sensors, displacement sensors, electrochemical sensors, crack gauges, and pressure sensors. The strain gauges are pasted on key parts of the roof materials, the fixed displacement sensors are fixed at the fixed reference points of the structure, the temperature and humidity sensors are fixed on the surface or inside of the roof materials and at key positions of the structure, the displacement sensors are fixed at key parts of the roof to capture vibration data, the electrochemical sensors are installed on the surface or inside of metal components, the crack gauges are installed in the areas where conventional cracks appear, and the pressure sensors are installed on the support structure of the roof to directly measure the force applied to the structure.

[0016] As a preferred technical solution of the present invention, the data processing module includes a data preprocessing unit, a feature selection unit, a data fusion unit, a parameter estimation unit, and a data classification unit;

[0017] The data preprocessing unit is used to denoise and normalize the collected data and measure the linear relationship between features using the correlation coefficient;

[0018] The feature selection unit uses the Lasso method for feature selection and dimensionality reduction to reduce the computational complexity of data fusion in the data fusion unit;

[0019] The data fusion unit uses discrete models and incremental models for data fusion to provide accurate data support for subsequent data classification;

[0020] The parameter estimation unit uses the EM algorithm to optimize the classification model;

[0021] The data classification unit uses the conditional naive Bayes algorithm to establish a classification model for the fused data.

[0022] As a preferred technical solution of the present invention, the data fusion unit includes a data partitioning sub-unit, a discrete model training sub-unit, an incremental model training sub-unit, a model fusion sub-unit, and a model optimization sub-unit;

[0023] The data division sub-unit selects important features from the pre-processed data;

[0024] The discrete model training sub-unit uses labeled data to train an initial random forest model, uses the trained model to predict unlabeled data, generates pseudo-labels, calculates the confidence of the pseudo-labels, and filters out highly reliable pseudo-labels according to the confidence threshold;

[0025] The incremental model training sub-unit performs feature selection on new data, incrementally trains the existing model with new data, and uses the updated model to generate new pseudo-labels for the new data;

[0026] The model fusion sub-unit averages the prediction results of the discrete model and the incremental model to obtain the final prediction;

[0027] The model optimization sub-unit uses cross-validation to calculate the performance metrics of the model, and adjusts the model parameters or integration strategy according to the evaluation results.

[0028] As a preferred technical solution of the present invention, the feature extraction unit includes a format conversion sub-unit, a vocabulary construction sub-unit, and a feature quantization sub-unit;

[0029] The format conversion sub-unit is used to process the data into a unified format;

[0030] The vocabulary construction sub-unit is used to extract all unique words from the text in the unified format and create a vocabulary;

[0031] The feature quantization sub-unit is used to convert each text into a vector, where each element of the vector represents the number of occurrences of the corresponding word in the vocabulary, and the value of the vector is the frequency of occurrence of the word in the text.

[0032] As a preferred technical solution of the present invention, the specific reconstruction steps of the state space reconstruction unit are as follows:

[0033] Step 1, divide the time series data into multiple segments, independently perform state space reconstruction on each segment, select a suitable time window size, and slide and divide the time series according to the window size to obtain multiple subsequences;

[0034] Step 2, determine the embedding dimension Select a suitable embedding dimension and delay time, use a suitable algorithm to determine the embedding dimension, and determine the delay time through the autocorrelation function or mutual information;

[0035] Step 3, apply the PSR model to each segmented data segment for state space reconstruction;

[0036] Step 4: Use the reconstructed state space for further analysis and modeling, analyze the geometric structure of the state space, and apply machine learning models for analysis.

[0037] As a preferred technical solution of the present invention, the detailed steps of the weight calculation of the weight assignment unit are as follows:

[0038] Step A: Experts determine the key features or indicators of the roof test lane structure monitoring data based on experience as decision criteria. Experts evaluate according to the importance of each criterion to form an intuitionistic fuzzy evaluation matrix and obtain ideal and non-ideal criterion values.

[0039] Step B: Based on the intuitionistic fuzzy judgment matrix and the ideal and non-ideal criterion values, establish a linear programming model, solve the linear programming model to obtain the weight of each criterion, and normalize the calculated weights.

[0040] Step C: Convert the intuitionistic fuzzy weights into a two-tuple semantic representation to handle the uncertainty and ambiguity of the weights, and determine the final weight of each criterion according to the two-tuple semantic values.

[0041] On the other hand, an evaluation method for a real-time evaluation system of a roof test lane structure includes the following steps:

[0042] S1: Install sensors on the roof test lane structure and calibrate them. Start the data acquisition module to collect the parameters of the roof test lane structure in real time. The collected data is wirelessly transmitted to the data processing module through the data acquisition module.

[0043] S2: Preprocess the monitored data through the data processing module, extract useful features from the data, remove irrelevant or redundant features, reduce the dimension of the data, and fuse the data. Classify the processed data to provide a basis for the evaluation of the structural health status.

[0044] S3: Based on the classified data, use the feature extraction unit to extract text features by the BOW method. The state space reconstruction unit synchronously segments the time series data and reconstructs it by the PSR model. Then fuse the text features and the time series features, calculate the weights of the fused features, and output the real-time health status evaluation result of the roof test lane structure.

[0045] S4: According to the evaluation result of the structural health status, the warning module determines whether there are potential safety hazards based on the evaluation result. If there are potential safety hazards, a warning signal is issued. According to the warning and the evaluation result, the staff takes corresponding measures.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] (1) The present invention monitors the roof test lane structure by integrating multiple sensors, which can comprehensively collect various parameters of the structure. Through the multi-sensor fusion monitoring and data processing and evaluation methods, the system can comprehensively monitor various parameters of the roof test lane structure and accurately evaluate its health status;

[0048] (2) The present invention selects data processing methods according to data characteristics, reduces the complexity of data fusion for multivariate data, while maintaining the prediction ability of the model, and performs different feature selections on the data, which can better analyze and model the dynamic characteristics of the structure;

[0049] (3) The present invention calculates the weight value of each feature, effectively processes the uncertainty and ambiguity of the weights, provides a more accurate weight allocation for structural health monitoring and evaluation, which enables the system to dynamically monitor and evaluate the structural state, and issue timely warnings when potential safety hazards occur, improving the safety and reliability of the structure.

[0050] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention are hereinafter specifically exemplified. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a block diagram of a real-time evaluation system for a roof test lane structure disclosed by the present invention;

[0052] Figure 2 is a block diagram of the data deterioration unit of a real-time evaluation system for a roof test lane structure disclosed by the present invention;

[0053] Figure 3 is a block diagram of the feature extraction unit of a real-time evaluation system for a roof test lane structure disclosed by the present invention;

[0054] Figure 4 is a schematic flowchart of the evaluation method of a real-time evaluation system for a roof test lane structure disclosed by the present invention;

[0055] Explanation of the accompanying drawings: 100, real-time evaluation system; 101, data acquisition module; 102, data processing module; 1021, data preprocessing unit; 1022, feature selection unit; 1023, data fusion unit; 10231, data partition subunit; 10232, discrete model training subunit; 10233, incremental model training subunit; 10234, model fusion subunit; 10235, model optimization subunit; 1024, parameter estimation unit; 1025, data classification unit; 103, data evaluation module; 1031, feature extraction unit; 10311, format conversion subunit; 10312, vocabulary construction subunit; 10313, feature quantization subunit; 1032, state space reconstruction unit; 1033, model coupling unit; 1034, weight assignment unit; 104, early warning module. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0059] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0060] In addition, the terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0061] Embodiment 1

[0062] Referring to the attached Figures 1-3 As shown, the present invention provides a technical solution: a real-time evaluation system for a roof test lane structure, including a data acquisition module 101, a data processing module 102, a data evaluation module 103, and an early warning module 104;

[0063] The data acquisition module 101 is used to collect various parameters of the roof test lane structure in real time, such as stress, strain, displacement, etc.;

[0064] The data processing module 102 is used to preprocess, extract features, classify, and fuse the collected data, and provide an input for the data evaluation module 103;

[0065] The data evaluation module 103 performs real-time evaluation on the roof test lane structure based on the processed data. The data evaluation module 103 includes a feature extraction unit 1031, a state space reconstruction unit 1032, a model coupling unit 1033, and a weight assignment unit 1034;

[0066] The feature extraction unit 1031 uses the BOW method to extract the required features from the data from the data processing module 102;

[0067] The state space reconstruction unit 1032 divides the time series data, applies the divided data to the PSR model, and performs state space reconstruction operations. The specific reconstruction steps of the state space reconstruction unit 1032 are as follows:

[0068] Step 1: Divide the time series data into multiple segments, independently perform state space reconstruction on each segment, select a suitable time window size, which depends on the periodicity of the time series and the characteristics of the data, and slide and divide the time series according to the window size to obtain multiple subsequences;

[0069] Step 2: Determine the embedding dimension. Select a suitable embedding dimension and delay time, and use a suitable algorithm to determine the embedding dimension, such as the false nearest neighbor method or the mutual information method, and determine the delay time through the autocorrelation function or mutual information;

[0070] Step 3: Apply the PSR model to each segmented data fragment for state space reconstruction. For each data fragment in a time window, construct a state vector according to the selected embedding dimension and delay time. For each point in the time series, create a state vector, and repeat this process until all the data in the time window are converted into state vectors;

[0071] Step 4: Use the reconstructed state space for further analysis and modeling. Analyze the geometric structure of the state space, such as calculating attractor trajectories, Lyapunov exponents, etc., and apply machine learning models for analysis. Use validation techniques to evaluate the model, and adjust the embedding dimension and delay time to optimize the model;

[0072] The model coupling unit 1033 couples the feature extraction unit 1031 and the state space reconstruction unit 1032 to construct a coupling model. The feature fusion steps of the model coupling unit 1033 are as follows:

[0073] a. The bag-of-words model extracts text features to obtain a feature matrix. Convert the time series data of the state space reconstruction model into state vectors to obtain a feature matrix. Merge the feature vectors obtained by the BOW method with the state space vectors reconstructed by the PSR model, and perform fusion through concatenation, feature concatenation, or other fusion techniques;

[0074] b. Use the fused features to train machine learning models, such as support vector machines (SVM), neural networks, etc., and use evaluation metrics to evaluate the model, such as accuracy, F1 score, ROC-AUC, etc.;

[0075] The weight assignment unit 1034 processes the subjective scores of experts using intuitionistic fuzzy binary semantics and calculates the weight values of each feature using the BWM method. The detailed steps of the weight calculation of the weight assignment unit 1034 are as follows:

[0076] Step A: Experts determine the key features or indicators of the roof test lane structure monitoring data based on experience as decision criteria. Experts evaluate according to the importance of each criterion to form an intuitionistic fuzzy evaluation matrix, and obtain ideal and non-ideal criterion values;

[0077] Step B: According to the intuitionistic fuzzy judgment matrix and the ideal and non-ideal criterion values, establish a linear programming model, solve the linear programming model to obtain the weight of each criterion, and normalize the calculated weights to ensure that the sum of all weights is 1;

[0078] Step C: Convert the intuitionistic fuzzy weights into binary semantic representations to handle the uncertainty and fuzziness of the weights, and determine the final weights of each criterion according to the binary semantic values. The specific conversion steps are:

[0079] Calculate the intuitionistic fuzzy scores for each criterion;

[0080] Convert the intuitionistic fuzzy scores into two-tuple semantic values;

[0081] Sort and compare the two-tuple semantic values to determine the final weights;

[0082] The assigned weights represent the relative importance of each feature or index in the roof test lane structure monitoring data in the decision-making process. These weights can help the staff identify key factors in structural health monitoring and assessment;

[0083] When the evaluation result shows that there are potential safety hazards in the roof test lane structure, the early warning module 104 issues an early warning signal.

[0084] The embodiments of the present invention are also implemented through the following technical solutions.

[0085] In the embodiments of the present invention, the data acquisition module 101 monitors the roof test lane structure using sensors, including strain gauges, displacement sensors, temperature and humidity sensors, displacement sensors, electrochemical sensors, crack gauges, and pressure sensors. The strain gauges are pasted on the key parts of the roof materials, such as the surfaces of beams, columns, and slabs, to ensure that the sensors are closely attached to the structure surface. The displacement sensors are fixed at the fixed reference points of the structure, aligned with the parts to be monitored, such as specific points on the roof or support structures. The temperature and humidity sensors are fixed on the surface or inside of the roof materials, as well as at the key positions of the structure, to ensure that the sensors can accurately reflect temperature changes. The displacement sensors are fixed at the key parts of the roof, such as the ends of beams or the centers of slabs, to capture vibration data. The electrochemical sensors are installed on the surface or inside of metal components, especially in areas vulnerable to corrosion. The crack gauges are installed in the areas where cracks usually appear, and the pressure sensors are installed on the support structures of the roof to directly measure the forces applied to the structure.

[0086] In the embodiments of the present invention, the data processing module 102 includes a data preprocessing unit 1021, a feature selection unit 1022, a data fusion unit 1023, a parameter estimation unit 1024, and a data classification unit 1025;

[0087] The data preprocessing unit 1021 is used to denoise and normalize the collected data, and use the correlation coefficient to measure the linear relationship between features;

[0088] The feature selection unit 1022 uses the Lasso method for feature selection and dimensionality reduction to reduce the computational complexity of the data fusion unit 1023 for data fusion;

[0089] Specifically: a. Select a Lasso regression model and set the regularization parameter. Use the standardized data to fit the Lasso regression model. Through cross-validation, such as K-fold cross-validation, select the value of the regularization parameter to balance the complexity and fitting degree of the model. Achieve feature selection and dimensionality reduction through the regularization penalty term, and at the same time prevent overfitting;

[0090] b. According to the results of the Lasso regression, select the features with non-zero coefficients. These features are considered important features in the model. Sort the features according to the magnitude of the coefficients. The larger the coefficient, the more important the feature. Identify the features important for predicting the target variable and simplify the model;

[0091] c. Retrain the model using the selected features and evaluate the performance of the model. For example, use metrics such as mean squared error and coefficient of determination. Compare the performance of the model after using Lasso feature selection with the performance of the original model using all features to ensure that the feature selection process improves the prediction ability of the model;

[0092] d. According to the results of the model evaluation, adjust the parameters of the Lasso regression, such as the value of the regularization parameter, or further process the features to further improve the model performance and ensure the generalization ability of the model on unknown data;

[0093] Repeat the above steps until a better feature set and model parameters are found;

[0094] If the number of features in the roof test lane structure monitoring data is large and there is multicollinearity, use the Elastic Net regression model for feature selection. Specifically: Set the regularization parameter to control the intensity of the penalty term. Set L1, where L1 is the ratio of the Lasso and Ridge penalty terms. Use cross-validation to evaluate the performance of the model under different parameter combinations, such as K-fold cross-validation. Calculate the mean squared error or other performance metrics for each parameter combination. Use the parameter combination determined by cross-validation to fit the Elastic Net model. The model will output the coefficients of each feature. Select the features with non-zero coefficients. These features are considered important. Sort the features according to the magnitude of the coefficients to identify important features. Retrain the model using the selected features and use an independent test set to evaluate the model performance;

[0095] The data fusion unit 1023 uses the discrete model and the incremental model for data fusion, providing accurate data support for subsequent data classification. The data fusion unit 1023 includes a data division sub-unit 10231, a discrete model training sub-unit 10232, an incremental model training sub-unit 10233, a model fusion sub-unit 10234, and a model optimization sub-unit 10235;

[0096] The data division sub-unit 10231 selects important features from the preprocessed data;

[0097] The discrete model training sub-unit 10232 trains an initial random forest model using labeled data, predicts unlabeled data using the trained model, generates pseudo-labels, calculates the confidence of the pseudo-labels, filters out highly reliable pseudo-labels according to the confidence threshold, and assigns lower weights to data points with lower prediction confidence to reduce their impact on model training;

[0098] The incremental model training sub-unit 10233 performs feature selection on new data, performs incremental training on the existing model using new data, and generates new pseudo-labels for the new data using the updated model;

[0099] The model fusion sub-unit 10234 averages the prediction results of the discrete model and the incremental model, assigns weights to the prediction results of each model according to model performance, and then calculates the weighted average to obtain the final prediction, providing a reference basis for the data evaluation module 103;

[0100] The model optimization sub-unit 10235 calculates the performance metrics of the model using cross-validation or other evaluation methods, and adjusts the model parameters or integration strategy according to the evaluation results;

[0101] The parameter estimation unit 1024 optimizes the classification model using the EM algorithm;

[0102] Specifically: a, assign initial values to the model parameters to provide a starting point for the EM algorithm;

[0103] b, calculate the posterior probability that the observed data comes from each latent variable under the given parameters, and update the distribution of the latent variables.

[0104] c, maximize the likelihood function, update the model parameters, and optimize the model parameters according to the posterior probability obtained in the E step.

[0105] D, repeat steps a and b until the parameters converge, continuously optimize the model parameters until the model meets the stopping conditions;

[0106] The data classification unit 1025 establishes a classification model for the fused data using the conditional naive Bayes algorithm;

[0107] Specifically: a, split the preprocessed data set into features and target variables, calculate the prior probability of each class, count the number of samples in each class and the prior probability of each class, prior probability = number of samples in the class / total number of samples, separate the input features and target variables in the data set, and divide the data set into a training set and a test set to prepare for subsequent model training;

[0108] b. Calculate the conditional probability of each feature under each category. Count the occurrences of each feature value under each category, and then calculate the conditional probability. For continuous features, calculate the mean and standard deviation of the features under each category. For discrete features, calculate the relative frequency of each feature value in the category.

[0109] c. Use the prior probability and conditional probability to construct a conditional naive Bayes classifier and verify the accuracy of the model.

[0110] d. Use the trained model to classify unknown data and extract features from the new data.

[0111] e. Use the probability model constructed in step c to calculate the posterior probability of each category, and classify the new data point as the category with the maximum posterior probability as the prediction result.

[0112] In an embodiment of the present invention, the feature extraction unit 1031 includes a format conversion sub-unit 10311, a vocabulary construction sub-unit 10312, and a feature quantization sub-unit 10313.

[0113] The format conversion sub-unit 10311 is used to process the data in a unified format to ensure the consistency of the text data and prepare for subsequent processing.

[0114] The vocabulary construction sub-unit 10312 is used to extract all unique words from the text after unified format and create a vocabulary. The vocabulary contains all the words that have appeared in the text and is the basis for constructing the bag-of-words model.

[0115] The feature quantization sub-unit 10313 is used to convert each text into a vector. Each element of the vector represents the occurrence times of the corresponding word in the vocabulary, and the value of the vector is the occurrence frequency of the word in the text. This vector is the bag-of-words model. Each dimension in the bag-of-words model is a feature, which represents the frequency information of a specific word in the vocabulary. These features together constitute the feature representation of the text.

[0116] Embodiment II

[0117] Refer to the appendix Figure 4 As shown, another evaluation method for the real-time evaluation system of the roof test lane structure provided by the embodiment of the present invention includes the following steps:

[0118] S1. Install sensors on the roof test lane structure and calibrate them. Start the data acquisition module 101 to collect the parameters of the roof test lane structure in real time. The collected data is wirelessly transmitted to the data processing module 102 through the data acquisition module 101.

[0119] S2. The data processing module 102 preprocesses the monitored data, extracts useful features from the data, removes irrelevant or redundant features, reduces the data dimension, improves the training efficiency and accuracy of the subsequent model, fuses the data, combines different types of data to provide more comprehensive structural state information, classifies the processed data, and provides a basis for structural health condition assessment;

[0120] S3. Based on the classified data, the feature extraction unit 1031 uses the BOW method to extract text features. The synchronous state space reconstruction unit 1032 segments and reconstructs the time series data using the PSR model, analyzes the dynamic characteristics of the time series data, then fuses the text features and time series features, calculates the weights of the fused features to determine the relative importance of each feature, compares the implementation data with the threshold, and scores according to the weights of the features, and outputs the real-time health status assessment result of the roof test lane structure;

[0121] Example of health assessment: Suppose the assessment model outputs a health score between 0 and 100, where 100 represents the best structural condition and 0 represents the worst structural condition. After model assessment, the health score of the roof test lane is 85 points. This score is based on the text data indicating that there are no obvious cracks or leakage problems in the structure, and the time series data shows that the structural vibration mode is stable without abnormal fluctuations. A score of 85 means that the structural condition of the roof test lane is generally good, but there may be some minor problems or areas that need attention. It is recommended to conduct regular inspections and maintenance. If the score is lower than the threshold, such as 70 points, further detailed inspections and repair measures are required;

[0122] The specific basis for scoring includes converting each indicator into a score according to its value and predefined rules, multiplying the score of each indicator by its corresponding weight to obtain a weighted score, and summing up all the weighted scores to obtain a comprehensive score.

[0123] S4. According to the structural health condition assessment result, the warning module 104 determines whether there are potential safety hazards based on the assessment result. If there are potential safety hazards, a warning signal is issued. According to the warning and assessment result, the staff takes corresponding actions;

[0124] S5. Regularly check and maintain the sensors to ensure the accuracy of data collection, update the data processing and assessment models to adapt to structural changes or new monitoring requirements, and upgrade the system software and hardware to improve the system performance and reliability;

[0125] S6. Record all monitored data, warning events, and treatment measures to generate a regular report, summarizing the structural health condition and system operation status.

[0126] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0127] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the protection scope of the present disclosure. The appended method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy.

[0128] Those skilled in the art should also understand that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments herein can be implemented as electronic hardware, computer software, or a combination thereof. To clearly illustrate the interchangeability of hardware and software, the above description of the various illustrative components, blocks, modules, circuits, and steps has been generally described in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the particular application and the design constraints imposed on the overall system. Skilled artisans can implement the described functionality in a flexible manner for each particular application, but such implementation decisions should not be construed as departing from the protection scope of the present disclosure.

[0129] The steps of the methods or algorithms described in connection with the embodiments herein can be directly embodied as hardware, software modules executed by a processor, or a combination thereof. The software modules can be located in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is connected to the processor so that the processor can read information from and write information to the storage medium. Of course, the storage medium can also be a part of the processor. The processor and the storage medium can be located in an ASIC. The ASIC can be located in a user terminal. Of course, the processor and the storage medium can also exist as discrete components in the user terminal.

[0130] For software implementation, the techniques described in this application can be implemented using modules (e.g., procedures, functions, etc.) that perform the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented within the processor or outside the processor. In the latter case, it is communicatively coupled to the processor via various means, which are well known in the art.

[0131] The foregoing description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but those of ordinary skill in the art should recognize that the various embodiments can be further combined and arranged. Accordingly, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. In addition, with respect to the term "comprising" as used in the specification or claims, that term is inclusive in a manner similar to the term "including," as is explained when "including" is used as a transitional word in a claim. Further, any use of the term "or" in a claim or the specification is to mean "non-exclusive or."

Claims

1. A real-time evaluation system for roof test lane structure, characterized in that: It includes a data collection module (101), a data processing module (102), a data evaluation module (103) and an early warning module (104); The data acquisition module (101) is used to collect various parameters of the roof test lane structure in real time; The data processing module (102) is used to pre-process, extract features, classify and fuse the collected data; The data evaluation module (103) is used to perform real-time evaluation on the roof test lane structure, and the data evaluation module (103) includes a feature extraction unit (1031), a state space reconstruction unit (1032), a model coupling unit (1033) and a weight assignment unit (1034); The feature extraction unit (1031) extracts required features from the data from the data processing module (102) using the BOW method; The state space reconstruction unit (1032) segments the time series data, applies the segmented data to the PSR model, and performs a state space reconstruction operation; The model coupling unit (1033) constructs a coupling model between the feature extraction unit (1031) and the state space reconstruction unit (1032); The weight assignment unit (1034) processes the subjective scores of the experts using intuitive fuzzy binary semantics, and calculates the weight value of each feature using the BWM method; The early warning module (104) sends out an early warning signal when the evaluation result shows that there is a safety hazard in the roof test lane structure.

2. A real-time evaluation system for roof test lane structure according to claim 1, characterized in that: The data acquisition module (101) monitors the roof test track structure using sensors, including strain gauges, displacement sensors, temperature and humidity sensors, displacement sensors, electrochemical sensors, crack meters and pressure sensors. The strain gauges are pasted on key parts of the roof material, the displacement sensors are fixed to fixed reference points of the structure, the temperature and humidity sensors are fixed on the surface or inside of the roof material, and on key positions of the structure, the displacement sensors are fixed on key parts of the roof to capture vibration data, the electrochemical sensors are installed on the surface or inside of the metal components, the crack meters are installed in conventional crack occurrence areas, and the pressure sensors are installed on the supporting structure of the roof to directly measure the force applied to the structure.

3. A real-time evaluation system for roof test lane structure according to claim 2, characterized in that: The data processing module (102) includes a data preprocessing unit (1021), a feature selection unit (1022), a data fusion unit (1023), a parameter estimation unit (1024) and a data classification unit (1025); The data preprocessing unit (1021) is used to perform denoising and normalization processing on the collected data, and use correlation coefficients to measure the linear relationship between features; The feature selection unit (1022) uses the Lasso method to perform feature selection and dimension reduction processing, thereby reducing the computational complexity of data fusion performed by the data fusion unit (1023); The data fusion unit (1023) uses a discrete model and an incremental model to perform data fusion, thereby providing accurate data support for subsequent data classification; The parameter estimation unit (1024) optimizes the classification model using an EM algorithm; The data classification unit (1025) uses the conditional naive Bayes algorithm to establish a classification model for the fused data.

4. A real-time evaluation system for roof test lane structure according to claim 3, characterized in that: The data fusion unit (1023) includes a data partitioning subunit (10231), a discrete model training subunit (10232), an incremental model training subunit (10233), a model fusion subunit (10234) and a model optimization subunit (10235); The data partitioning subunit (10231) selects important features from the preprocessed data; The discrete model training subunit (10232) uses the labeled data to train the initial random forest model, uses the trained model to predict the unlabeled data, generates pseudo labels, calculates the confidence of the pseudo labels, and selects pseudo labels with high reliability according to the confidence threshold; The incremental model training subunit (10233) performs feature selection on the new data, performs incremental training on the existing model using the new data, and generates new pseudo labels for the new data using the updated model; The model fusion subunit (10234) averages the prediction results of the discrete model and the incremental model to obtain a final prediction; The model optimization subunit (10235) uses cross-validation to calculate the performance indicators of the model and adjusts the model parameters or integration strategy based on the evaluation results.

5. A real-time evaluation system for roof test lane structure according to claim 4, characterized in that: The feature extraction unit (1031) includes a format conversion subunit (10311), a vocabulary construction subunit (10312) and a feature quantization subunit (10313); The format conversion subunit (10311) is used to process the data into a unified format; The vocabulary building subunit (10312) is used to extract all unique words from the text after the unified format and create a vocabulary; The feature quantization subunit (10313) is used to convert each text into a vector, where each element of the vector represents the number of occurrences of the corresponding word in the vocabulary, and the value of the vector is the frequency of occurrence of the word in the text.

6. A real-time evaluation system for roof test lane structure according to claim 5, characterized in that: The specific reconstruction steps of the state space reconstruction unit (1032) are as follows: Step 1: Split the time series data into multiple segments, reconstruct the state space independently on each segment, select a suitable time window size, and slide the time series according to the window size to obtain multiple subsequences. Step 2: Determine the embedding dimension. Select the appropriate embedding dimension and delay time. Use a suitable algorithm to determine the embedding dimension. Determine the delay time through the autocorrelation function or mutual information. Step 3: Apply the PSR model to each segmented data segment to reconstruct the state space; Step 4: Use the reconstructed state space for further analysis and modeling, analyze the geometric structure of the state space, and apply machine learning models for analysis.

7. A real-time evaluation system for roof test lane structure according to claim 6, characterized in that: The detailed steps of weight calculation of the weight assignment unit (1034) are as follows: Step A, experts determine the key features or indicators of the roof test lane structure monitoring data based on experience as decision criteria, and the experts evaluate the importance of each criterion to form an intuitive fuzzy evaluation matrix to obtain ideal and unideal criterion values; Step B, establishing a linear programming model according to the intuitive fuzzy judgment matrix and the ideal and unideal criterion values, solving the linear programming model, obtaining the weight of each criterion, and normalizing the calculated weight; In step C, the intuitionistic fuzzy weights are converted into binary semantic representations to deal with the uncertainty and ambiguity of the weights, and the final weight of each criterion is determined according to the binary semantic value.

8. An evaluation method for a real-time evaluation system for a roof test lane structure, applied to a real-time evaluation system for a roof test lane structure as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: S1, installing sensors on the roof test track structure and calibrating them, starting the data acquisition module (101), collecting parameters of the roof test track structure in real time, and wirelessly transmitting the collected data to the data processing module (102) through the data acquisition module (101); S2, preprocessing the monitored data through the data processing module (102), extracting useful features from the data, removing irrelevant or redundant features, reducing the dimension of the data, fusing the data, and classifying the processed data to provide a basis for structural health status assessment; S3, based on the classified data, the feature extraction unit (1031) uses the BOW method to extract text features, and the synchronous state space reconstruction unit (1032) segments the time series data and reconstructs the PSR model, and then fuses the text features and the time series features, and calculates the weights of the fused features to output the real-time health status assessment results of the roof test lane structure; S4, based on the structural health status assessment result, the early warning module (104) determines whether there is a safety hazard based on the assessment result. If there is a safety hazard, an early warning signal is issued, and the staff performs corresponding processing based on the early warning and assessment results.

Images