Concrete material quality detection method based on multi-parameter analysis

Abstract

The invention discloses a concrete material quality detection method based on multi-parameter analysis, and relates to the technical field of concrete material quality detection.The method comprises the steps that concrete structure data are collected, and a plurality of original detection parameters representing concrete quality are obtained; performing preprocessing operation on the original detection parameters to remove abnormal values and unify dimensions, and obtaining a standardized target detection data set; constructing a weight distribution model, dynamically adjusting parameter weights in the weight distribution model according to current detection environment information, and outputting a dynamic weighted parameter set; performing training updating on historical detection data, performing fusion calculation on the dynamic weighting parameter set and a current detection result, and generating an optimized weight configuration parameter; and constructing a physical information neural network model, and embedding the mechanical equation and the heat conduction equation of the concrete material into a neural network loss function as constraint terms.

Description

Technical Field

[0001] The present invention relates to the technical field of concrete material quality detection, in particular to a concrete material quality detection method based on multi-parameter analysis. Background Art

[0002] Concrete material quality testing involves conducting a series of tests and analyses on concrete structures to assess their physical properties and durability, thereby determining whether they meet design requirements and relevant standards. Therefore, utilizing advanced technologies to improve the intelligence and safety of concrete material quality testing has become a pressing issue.

[0003] In the field of concrete material quality testing, most existing evaluation models use a fixed weight distribution method, ignoring the impact of environmental factors on the importance of different parameters, resulting in inaccurate evaluation results under different working conditions. Traditional neural network models only rely on data-driven training, ignoring the physical mechanisms within the material, making the prediction results biased and difficult to interpret. At the same time, existing methods usually directly give a deterministic quality score without considering the uncertainty caused by measurement errors and environmental noise, resulting in insufficient reliability and transparency of the evaluation results. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a concrete material quality detection method based on multi-parameter analysis to solve the problem that most existing evaluation models adopt a fixed weight distribution method, ignoring the influence of environmental factors on the importance of different parameters, resulting in inaccurate evaluation results under different working conditions, and traditional neural network models only rely on data-driven training, ignoring the physical mechanism inside the material, resulting in deviations in the prediction results and difficult to explain.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for detecting the quality of concrete materials based on multi-parameter analysis, which comprises:

[0008] Collect concrete structure data and obtain multiple original detection parameters that characterize concrete quality;

[0009] Preprocessing the original detection parameters to remove outliers and unify the dimensions to obtain a standardized target detection data set;

[0010] Constructing a weight distribution model, and dynamically adjusting the parameter weights in the weight distribution model according to the current detection environment information, and outputting a dynamic weighted parameter set;

[0011] Perform training and update on historical detection data, fuse the dynamic weighted parameter set with the current detection results, and generate optimized weight configuration parameters;

[0012] Constructing a physical information neural network model, embedding the mechanical equations and heat conduction equations of concrete materials as constraints into the neural network loss function, training the physical information neural network model using a preprocessed target detection dataset, and outputting the prediction results of the concrete internal state;

[0013] Modeling the prediction error of the physical information neural network model, calculating the influence weight of each detection parameter on the final quality score, and generating a target concrete quality score result with a confidence interval through a probability distribution function;

[0014] Based on the target concrete quality score result with confidence interval, the concrete structure is classified and rated in combination with the preset quality grading standard, and a quality assessment report including quality grade, confidence range and risk warning information is output.

[0015] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, the original detection parameters include:

[0016] The ultrasonic propagation velocity v measured by an ultrasonic detector;

[0017] Rebound value R measured by rebound hammer;

[0018] The radar reflection signal intensity A measured by the ground penetrating radar detector;

[0019] Thermal imaging temperature distribution information T(x,y) measured by an infrared thermal imager;

[0020] The above four types of parameters are combined to form the original data sample set D raw ={v,R,A,T(x,y)}.

[0021] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, the preprocessing operation includes using a sliding window filtering method to reduce the noise of the collected original signal and to unify the dimensions of each parameter to form a standardized target detection data set D norm ={R norm ,R norm ,A norm ,T norm (x,y)}.

[0022] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, the construction of the weight distribution model to output a dynamic weighted parameter set includes:

[0023] Compare multiple detection parameters involved in concrete quality assessment and construct a judgment matrix A;

[0024] The judgment matrix A is an n×n positive reciprocal matrix, and the matrix element a ij Indicates the importance of the i-th parameter relative to the j-th parameter in quality evaluation;

[0025] Calculate the maximum eigenvalue λ of the judgment matrix max And the corresponding normalized eigenvector, which is the initial weight vector of each parameter;

[0026] Calculate the consistency index, the expression is:

[0027]

[0028] Among them, λ max is the maximum eigenvalue of the matrix, n is the number of parameters;

[0029] Look up the table to obtain the average random consistency index RI under the corresponding order and calculate the consistency ratio;

[0030] When the consistency ratio The judgment matrix is ​​considered to have acceptable consistency;

[0031] Get the initial weight vector W0={w v ,w R ,w A ,w T};

[0032] Define the factor set U = {u1,u2,u3} that affects the weight, where u1 is temperature and humidity, u2 is the age of concrete, and u3 is the construction process;

[0033] For each factor u i Establish a membership function to quantify the influence of the factor on the detection parameters;

[0034] For each detection parameter j∈{v,R,A,T}, calculate its corresponding environmental impact adjustment coefficient α j , the expression is:

[0035]

[0036] Among them, α j is the environmental impact factor of the jth parameter, a i is the weight of the i-th environmental factor, f(ui ) is the influence of the i-th environmental factor on the parameter;

[0037] Combine the adjustment coefficient with the initial weight to obtain a dynamically weighted weight vector

[0038] W dynamic ={w v α v ,w R α R ,w A α A ,w T α T};

[0039] Among them, α v is the environmental impact factor of ultrasonic propagation velocity, α R is the environmental impact factor of the rebound value, α A is the environmental impact factor of the radar reflection signal intensity, α T is the environmental impact factor of thermal imaging characteristic parameters.

[0040] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, wherein: the training and updating of historical detection data, the fusion calculation of the dynamic weighted parameter set and the current detection results to generate optimized weight configuration parameters, the specific steps are as follows:

[0041] Use database retrieval to obtain historical test data;

[0042] The historical inspection data includes a plurality of concrete structure samples with known quality status;

[0043] Each sample includes standardized target detection data and a corresponding true quality score;

[0044] Construct a complete historical data set, the expression is:

[0045]

[0046] Among them, k is the total number of historical samples, is the standardized input parameter of the i-th sample, Q (i) The true quality score of the i-th sample;

[0047] Introducing dynamic weighted parameter set W dynamic As the basis of the initial weight distribution, the prior distribution of the weight parameter is set to normal distribution, and the expression is:

[0048]

[0049] Among them, μ0 is the prior mean, is the prior variance;

[0050] Use the actual prediction error in the current detection task to correct the prior distribution;

[0051] The prediction error is defined as e=Q predicted -Q true ;

[0052] Among them, Q predicted Based on the current weight configuration, Q true Score the true quality obtained through field sampling or other high-precision methods;

[0053] In each iteration, according to Bayes' theorem, the posterior distribution is updated, the optimal weight configuration is estimated from the posterior distribution through Markov Chain Monte Carlo MCMC sampling, and the updated optimal weight configuration is obtained, which is expressed as:

[0054]

[0055] in, is the optimization weight of ultrasonic propagation velocity, is the optimization weight of the rebound value, is the optimization weight of the radar reflection signal strength, is the optimization weight of thermal imaging feature parameters;

[0056] Configure the optimized weights W opt Applied to data fusion and quality score calculation of current detection tasks.

[0057] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, the physical information neural network model is constructed to output the prediction result of the internal state of concrete, and the specific steps are as follows:

[0058] Use optimized weight configuration W opt The input data is weighted and then input into the physical information neural network model for training and prediction;

[0059] For each input data point The weighted input vector is:

[0060]

[0061] Construct a physics-guided neural network structure where the input layer is a weighted input vector The middle layer is a number of fully connected hidden layers, and the output layer is the concrete strength, density and crack depth;

[0062] Add physical equation constraints to the loss function, including the introduction of elastic mechanics equations and heat conduction equations, and the expression is:

[0063]

[0064] Where σ is the stress tensor, f is the body force, ρ is the density, and c p is the specific heat capacity, k is the thermal conductivity;

[0065] The above physical equation is added as a penalty term to the loss function to form the physical information loss function, and the total loss function L is calculated. total , the expression is:

[0066] L total =L data +λ·L PINN ;

[0067] Among them, L data is the data fitting loss, λ is the balance coefficient;

[0068] Using the target detection dataset D weigh ted The network is trained to obtain the predicted output, i.e. the predicted values ​​of concrete strength, density and crack depth.

[0069] As a preferred solution of the concrete material quality detection method based on multi-parameter analysis of the present invention, the specific steps of generating the target concrete quality score result with a confidence interval are as follows:

[0070] Based on the prediction results of the physical information neural network model and the standardized target detection dataset, a disturbance term that obeys the normal distribution is introduced into the input parameters to simulate the measurement error and environmental noise in the actual detection process.

[0071] Run the physical information neural network model multiple times, use the perturbed input data for inference, and obtain multiple prediction results of the output variable;

[0072] Count the distribution of output values ​​after each run and calculate their standard deviation;

[0073] Based on the standard deviation, the contribution weight of each input parameter to the output uncertainty is calculated to obtain the uncertainty impact weight set, which is expressed as:

[0074]

[0075] W error ={w error (v),w error (R),w error (A),w error (T)};

[0076] Among them, σ Q is the standard deviation of the comprehensive quality score, x i is the i-th input parameter;

[0077] A comprehensive quality scoring function is constructed based on the optimized weight configuration and uncertainty impact weight set. The expression is:

[0078]

[0079] Among them, w S , w D , w L The scoring weights are set based on engineering experience or historical data analysis. The inverse is because the smaller the crack depth, the higher the quality;

[0080] The Monte Carlo simulation results are Substitute into the above formula to calculate the corresponding quality score

[0081] For all The kernel density is estimated and the probability density function is fitted.

[0082] Based on the probability density function, the upper and lower limits at the 95% confidence level are determined to generate the target concrete quality score result with confidence interval.

[0083] As a preferred embodiment of the concrete material quality detection method based on multi-parameter analysis of the present invention, the concrete structure is classified and graded based on the target concrete quality score result with confidence interval and the preset quality grading standard, and the specific steps are as follows:

[0084] Prediction value based on comprehensive quality score and its confidence interval, and set the thresholds for three quality grade divisions;

[0085] The threshold can be dynamically adjusted according to different project types, usage environments or design intensity levels;

[0086] Compare the confidence interval with the quality grading threshold to determine the quality risk level of the current structure, including:

[0087] like It means that even under the best conditions, the concrete quality is still below the minimum qualified standard, which is judged as high risk;

[0088] like It means that even in the worst case, the quality of concrete is still better than the excellent standard, which is judged as low risk;

[0089] Otherwise, that is, the confidence interval crosses two or more quality level boundaries, indicating the existence of uncertainty impact, and it is judged as medium risk;

[0090] Based on the above quality grade classification and risk assessment results, a quality assessment report is generated, which includes the quality grade, confidence interval, main defect type, recommended repair measures and risk level.

[0091] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, any step of the method for detecting quality of concrete materials based on multi-parameter analysis as described in the first aspect of the present invention is implemented.

[0092] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for detecting concrete material quality based on multi-parameter analysis as described in the first aspect of the present invention is implemented.

[0093] The beneficial effects of the present invention are as follows: by constructing a physics-guided neural network structure, adding elastic mechanics equations and heat conduction equations as physical constraints in the loss function, inputting weighted detection data, training the model and outputting strength, density, and crack depth prediction values, a deep integration of mechanism modeling and data-driven is achieved, which significantly improves the physical rationality and prediction robustness of the model, and enhances the ability to identify hidden defects. By introducing normal perturbation to simulate measurement errors, running the model multiple times to obtain the output variable distribution, statistically analyzing the standard deviation and calculating the uncertainty impact weight, an uncertainty modeling and risk assessment mechanism is achieved, filling the gap in the lack of confidence description in existing detection methods, improving the credibility and transparency of the evaluation results, and providing a quantitative basis for engineering quality acceptance and maintenance decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0095] Figure 1 Flowchart of the concrete material quality detection method based on multi-parameter analysis in Example 1. DETAILED DESCRIPTION

[0096] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0097] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0098] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0099] Example, see Figure 1 , is an embodiment of the present invention, which provides a concrete material quality detection method based on multi-parameter analysis, comprising the following steps:

[0100] S1. Collect concrete structure data and obtain multiple original detection parameters representing concrete quality;

[0101] Preprocess the original detection parameters to remove outliers and unify the dimensions to obtain a standardized target detection dataset;

[0102] Furthermore, the original detection parameters include:

[0103] The ultrasonic propagation velocity v measured by an ultrasonic detector;

[0104] Rebound value R measured by rebound hammer;

[0105] The radar reflection signal intensity A measured by the ground penetrating radar detector;

[0106] Thermal imaging temperature distribution information T(x,y) measured by an infrared thermal imager;

[0107] The above four types of parameters are combined to form the original data sample set D raw ={v,R,A,T(x,y)};

[0108] The preprocessing operation includes using the sliding window filtering method to reduce the noise of the collected original signal and unify the dimensions of each parameter to form a standardized target detection data set D norm ={v norm ,R norm ,A norm ,T norm (x,y)};

[0109] It should be noted that the original detection parameters cover key information on the physical and mechanical properties and thermal response characteristics of concrete structures. Through the fusion of multi-source heterogeneous data, the internal state of concrete can be more comprehensively reflected. Among them, the ultrasonic propagation velocity can effectively reflect the material density and crack distribution; the rebound value is used to evaluate the surface hardness; the radar reflection signal intensity helps to identify internal defects and voids; infrared thermal imaging can be used to detect abnormal temperature areas and then judge potential damage. The original data sample set formed by the combination of the above parameters provides a rich information basis for subsequent quality analysis.

[0110] S2. Build a weight distribution model, dynamically adjust the parameter weights in the weight distribution model based on the current detection environment information, and output a dynamic weighted parameter set;

[0111] Furthermore, multiple detection parameters involved in concrete quality assessment are compared to construct a judgment matrix A;

[0112] The judgment matrix A is an n×n positive reciprocal matrix, and the matrix element a ij Indicates the importance of the i-th parameter relative to the j-th parameter in quality evaluation;

[0113] Calculate the maximum eigenvalue λ of the judgment matrix max And the corresponding normalized eigenvector, the eigenvector is the initial weight vector of each parameter;

[0114] Calculate the consistency index, the expression is:

[0115]

[0116] Among them, λ max is the maximum eigenvalue of the matrix, n is the number of parameters;

[0117] Look up the table to obtain the average random consistency index RI under the corresponding order and calculate the consistency ratio;

[0118] When the consistency ratio The judgment matrix is ​​considered to have acceptable consistency;

[0119] Get the initial weight vector W0={w v ,w R ,w A ,w T};

[0120] Define the factor set U = {u1,u2,u3} that affects the weight, where u1 is temperature and humidity, u2 is the age of concrete, and u3 is the construction process;

[0121] For each factor u i Establish a membership function to quantify the influence of the factor on the detection parameters;

[0122] For each detection parameter j∈{v,R,A,T}, calculate its corresponding environmental impact adjustment coefficient α j , the expression is:

[0123]

[0124] Among them, α j is the environmental impact factor of the jth parameter, a i is the weight of the i-th environmental factor, T(u i ) is the influence of the i-th environmental factor on the parameter;

[0125] Combine the adjustment coefficient with the initial weight to obtain the dynamically weighted weight vector W dynamic ={w v α v ,w R α R ,w A α A ,w T α T};

[0126] Among them, α v is the environmental impact factor of ultrasonic propagation velocity, α R is the environmental impact factor of the rebound value, α A is the environmental impact factor of the radar reflection signal intensity, α T is the environmental impact factor of thermal imaging characteristic parameters;

[0127] It should be noted that the dynamic weighting mechanism comprehensively considers the impact of expert experience and environmental factors on the test results, avoiding the problem of poor applicability of the traditional fixed weight model under different working conditions. A judgment matrix with good consistency is constructed through the hierarchical analysis method to ensure that the initial weight distribution is scientific and reasonable; further combined with the fuzzy comprehensive evaluation method, actual environmental variables such as temperature and humidity, age, and construction technology are introduced into the weight adjustment process, so that the final output dynamic weighted parameter set can adapt to the specific detection scenario, thereby improving the accuracy and stability of the evaluation results.

[0128] S3. Train and update historical detection data, fuse the dynamic weighted parameter set with the current detection results, and generate optimized weight configuration parameters;

[0129] Furthermore, a database search method is used to obtain historical test data;

[0130] The historical inspection data includes multiple concrete structure samples with known quality status;

[0131] Each sample contains the normalized target detection data and the corresponding true quality score;

[0132] Construct a complete historical data set, the expression is:

[0133]

[0134] Among them, k is the total number of historical samples, is the standardized input parameter of the i-th sample, Q (i) The true quality score of the i-th sample;

[0135] Introducing dynamic weighted parameter set W dynamic As the basis of the initial weight distribution, the prior distribution of the weight parameter is set to normal distribution, and the expression is:

[0136]

[0137] Among them, μ0 is the prior mean, is the prior variance;

[0138] Use the actual prediction error in the current detection task to correct the prior distribution;

[0139] The prediction error is defined as e = Q predicted -Q true ;

[0140] Among them, Q predicted Based on the current weight configuration, Q true Score the true quality obtained through field sampling or other high-precision methods;

[0141] In each iteration, according to Bayes' theorem, the posterior distribution is updated, the optimal weight configuration is estimated from the posterior distribution through Markov Chain Monte Carlo MCMC sampling, and the updated optimal weight configuration is obtained, which is expressed as:

[0142]

[0143] in, is the optimization weight of ultrasonic propagation velocity, is the optimization weight of the rebound value, is the optimization weight of the radar reflection signal strength, is the optimization weight of thermal imaging feature parameters;

[0144] Configure the optimized weights W opt Applied to data fusion and quality score calculation of current detection tasks;

[0145] It should be noted that by introducing the Bayesian learning algorithm, the organic integration of historical detection data and current detection results is achieved, giving the system the ability to continuously optimize. Unlike the traditional static modeling method, this method uses prior knowledge to guide model updates and continuously corrects the posterior distribution based on the prediction error, thereby gradually approaching the optimal weight configuration. The dynamic update mechanism based on probabilistic reasoning not only improves the generalization ability of the model in new tasks, but also enhances the system's adaptability to complex engineering scenarios, enabling it to maintain a high evaluation accuracy during long-term use.

[0146] S4. Construct a physical information neural network model, embed the mechanical equation and heat conduction equation of concrete materials as constraints into the neural network loss function, train the physical information neural network model using the preprocessed target detection dataset, and output the prediction results of the internal state of concrete;

[0147] Furthermore, we use the optimized weight configuration W opt The input data is weighted and then input into the physical information neural network model for training and prediction;

[0148] For each input data point The weighted input vector is:

[0149]

[0150] Construct a physics-guided neural network structure where the input layer is a weighted input vector The middle layer is a number of fully connected hidden layers, and the output layer is the concrete strength, density and crack depth;

[0151] Add physical equation constraints to the loss function, including the introduction of elastic mechanics equations and heat conduction equations, and the expression is:

[0152]

[0153] Where σ is the stress tensor, f is the body force, ρ is the density, and c p is the specific heat capacity, k is the thermal conductivity;

[0154] The above physical equation is added as a penalty term to the loss function to form the physical information loss function, and the total loss function L is calculated. total , the expression is:

[0155] L total =L data +λ·L PINN ;

[0156] Among them, L data is the data fitting loss, λ is the balance coefficient;

[0157] Using the target detection dataset D weigh ted The network is trained to obtain the predicted output, i.e. the predicted values ​​of concrete strength, density and crack depth;

[0158] It should be noted that the physical information neural network model constructed in this step combines the advantages of physical mechanisms and data-driven, breaking through the limitations of the "black box" characteristics of traditional deep learning models. By embedding the elastic mechanics equation and the heat conduction equation as constraints into the loss function, the model not only fits the data during training, but also satisfies the basic physical laws, thereby significantly improving the credibility and interpretability of the prediction results. In addition, by weighting the input data, the model's attention to key features is further strengthened, thereby improving its recognition ability and robustness in complex concrete structures.

[0159] S5. Modeling the prediction error of the physical information neural network model, calculating the influence weight of each detection parameter on the final quality score, and generating the target concrete quality score result with a confidence interval through a probability distribution function;

[0160] Furthermore, based on the prediction results of the physical information neural network model and the standardized target detection dataset, a disturbance term that follows a normal distribution is introduced into the input parameters to simulate the measurement errors and environmental noise in the actual detection process.

[0161] Run the physical information neural network model multiple times, use the perturbed input data for inference, and obtain multiple prediction results of the output variable;

[0162] Count the distribution of output values ​​after each run and calculate their standard deviation;

[0163] Based on the standard deviation, the contribution weight of each input parameter to the output uncertainty is calculated to obtain the uncertainty impact weight set, which is expressed as:

[0164]

[0165] W error ={w error (v),w error (R),w error (A),w error (T)};

[0166] Among them, σ Q is the standard deviation of the comprehensive quality score, x i is the i-th input parameter;

[0167] A comprehensive quality scoring function is constructed based on the optimized weight configuration and uncertainty impact weight set. The expression is:

[0168]

[0169] Among them, w S , w D , w L The scoring weights are set based on engineering experience or historical data analysis. The inverse is because the smaller the crack depth, the higher the quality;

[0170] The Monte Carlo simulation results are Substitute into the above formula to calculate the corresponding quality score

[0171] For all The kernel density is estimated and the probability density function is fitted.

[0172] Based on the probability density function, the upper and lower limits at the 95% confidence level are determined to generate the target concrete quality score result with confidence interval;

[0173] It should be noted that by introducing the uncertainty modeling method, the quantitative analysis of prediction errors and the generation of confidence intervals are realized, filling the gap in the lack of risk assessment mechanism in existing detection methods. The statistical distribution of the prediction output is obtained through Monte Carlo simulation, and the probability density function is constructed in combination with the kernel density estimation method, which can intuitively present the uncertainty range of the quality score. The method not only improves the transparency of the evaluation results, but also provides a scientific basis for engineering quality acceptance and maintenance decision-making, enhancing the practicality and engineering application value of the system.

[0174] S6. Based on the target concrete quality score result with confidence interval, the concrete structure is classified and rated in combination with the preset quality grading standard, and a quality assessment report including quality grade, confidence range and risk warning information is output;

[0175] Furthermore, based on the comprehensive quality score prediction value and its confidence interval, and set the thresholds for three quality grade divisions;

[0176] The threshold can be dynamically adjusted according to different project types, usage environments or design intensity levels;

[0177] Compare the confidence interval with the quality grading threshold to determine the quality risk level of the current structure, including:

[0178] like It means that even under the best conditions, the concrete quality is still below the minimum qualified standard, which is judged as high risk;

[0179] like It means that even in the worst case, the quality of concrete is still better than the excellent standard, which is judged as low risk;

[0180] Otherwise, that is, the confidence interval crosses two or more quality level boundaries, indicating the existence of uncertainty impact, and it is judged as medium risk;

[0181] Based on the above quality grading and risk assessment results, a quality assessment report is generated. The quality assessment report includes the quality grade, confidence interval, main defect types, recommended remediation measures, and risk level;

[0182] It should be noted that this step realizes the quality classification and risk level division of concrete structures based on the comparison of confidence intervals and preset quality grading standards, providing clear risk warnings and repair suggestions for engineering personnel. Compared with the traditional evaluation method that only relies on a single scoring value, this method can more accurately judge the safety status and potential risk level of the structure by introducing comparative analysis of confidence interval boundaries and thresholds, supporting priority management and scientific decision-making, and greatly enhancing the guiding significance and promotion potential of the system in actual engineering applications.

[0183] This embodiment further provides a computer device suitable for the concrete material quality detection method based on multi-parameter analysis, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the concrete material quality detection method based on multi-parameter analysis proposed in the above embodiment.

[0184] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0185] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the method for detecting the quality of concrete materials based on multi-parameter analysis as proposed in the above embodiment is implemented. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0186] In summary, the present invention constructs a physics-guided neural network structure, adds the elastic mechanics equation and the heat conduction equation as physical constraints in the loss function, inputs weighted detection data, trains the model and outputs the predicted values ​​of strength, density and crack depth, thereby realizing the deep integration of mechanism modeling and data-driven, significantly improving the physical rationality and prediction robustness of the model, and enhancing the ability to identify hidden defects. By introducing normal perturbation to simulate measurement errors, running the model multiple times to obtain the output variable distribution, statistically analyzing the standard deviation and calculating the uncertainty impact weight, the uncertainty modeling and risk assessment mechanism are realized, filling the gap in the lack of confidence description in existing detection methods, improving the credibility and transparency of the evaluation results, and providing a quantitative basis for engineering quality acceptance and maintenance decision-making.

[0187] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting the quality of concrete materials based on multi-parameter analysis, characterized in that: include: Collect concrete structure data and obtain multiple original detection parameters that characterize concrete quality; Preprocessing the original detection parameters to remove outliers and unify the dimensions to obtain a standardized target detection data set; Constructing a weight distribution model, and dynamically adjusting the parameter weights in the weight distribution model according to the current detection environment information, and outputting a dynamic weighted parameter set; Perform training and update on historical detection data, fuse the dynamic weighted parameter set with the current detection results, and generate optimized weight configuration parameters; Constructing a physical information neural network model, embedding the mechanical equations and heat conduction equations of concrete materials as constraints into the neural network loss function, training the physical information neural network model using a preprocessed target detection dataset, and outputting the prediction results of the concrete internal state; Modeling the prediction error of the physical information neural network model, calculating the influence weight of each detection parameter on the final quality score, and generating a target concrete quality score result with a confidence interval through a probability distribution function; Based on the target concrete quality score result with confidence interval, the concrete structure is classified and rated in combination with the preset quality grading standard, and a quality assessment report including quality grade, confidence range and risk warning information is output.

2. The method for detecting concrete material quality based on multi-parameter analysis according to claim 1, wherein: The original detection parameters include: The ultrasonic propagation velocity v measured by an ultrasonic detector; Rebound value R measured by rebound hammer; The radar reflection signal intensity A measured by the ground penetrating radar detector; Thermal imaging temperature distribution information T(x,y) measured by an infrared thermal imager; The above four types of parameters are combined to form the original data sample set D raw ={v,R,A,T(x,y)}.

3. The method for detecting concrete material quality based on multi-parameter analysis according to claim 2, wherein: The preprocessing operation includes using the sliding window filtering method to reduce the noise of the collected original signal and to unify the dimensions of each parameter to form a standardized target detection data set D norm ={v norm ,R norm ,A norm ,T norm (x,y)}.

4. The method for detecting the quality of concrete materials based on multi-parameter analysis according to claim 3, wherein: The step of constructing a weight distribution model to output a dynamic weighted parameter set includes: Compare multiple detection parameters involved in concrete quality assessment and construct a judgment matrix A; The judgment matrix A is an n×n positive reciprocal matrix, and the matrix element a ij Indicates the importance of the i-th parameter relative to the j-th parameter in quality evaluation; Calculate the maximum eigenvalue λ of the judgment matrix max And the corresponding normalized eigenvector, which is the initial weight vector of each parameter; Calculate the consistency index, the expression is: Among them, λ max is the maximum eigenvalue of the matrix, n is the number of parameters; Look up the table to obtain the average random consistency index RI under the corresponding order and calculate the consistency ratio; When the consistency ratio The judgment matrix is ​​considered to have acceptable consistency; Get the initial weight vector W0={w v ,w R ,w A ,w T }; Define the factor set U = {u1, u2, u3} that affects the weight, where u1 is temperature and humidity, u2 is the age of concrete, and u3 is the construction process; For each factor u i Establish a membership function to quantify the influence of the factor on the detection parameters; For each detection parameter j∈{v,R,A,T}, calculate its corresponding environmental impact adjustment coefficient α j , the expression is: Among them, α j is the environmental impact factor of the jth parameter, a i is the weight of the i-th environmental factor, f(u i ) is the influence of the i-th environmental factor on the parameter; Combine the adjustment coefficient with the initial weight to obtain a dynamically weighted weight vector W dynamic ={w v ·a v ,w R ·a R ,w A ·a A ,w T ·a T }; Among them, α v is the environmental impact factor of ultrasonic propagation velocity, α R is the environmental impact factor of the rebound value, α A is the environmental impact factor of the radar reflection signal intensity, α T is the environmental impact factor of thermal imaging characteristic parameters.

5. The method for detecting concrete material quality based on multi-parameter analysis according to claim 4, characterized in that: The training and updating of historical detection data, the fusion calculation of the dynamic weighted parameter set and the current detection results, and the generation of optimized weight configuration parameters are specifically performed as follows: Use database retrieval to obtain historical test data; The historical inspection data includes a plurality of concrete structure samples with known quality status; Each sample includes standardized target detection data and a corresponding true quality score; Construct a complete historical data set, the expression is: Among them, k is the total number of historical samples, is the standardized input parameter of the i-th sample, Q (i) The true quality score of the i-th sample; Introducing dynamic weighted parameter set W dynamic As the basis of the initial weight distribution, the prior distribution of the weight parameter is set to normal distribution, and the expression is: Among them, μ0 is the prior mean, is the prior variance; Use the actual prediction error in the current detection task to correct the prior distribution; The prediction error is defined as e=Q predicted -Q true ; Among them, Q predicted Based on the current weight configuration, Q true Score the true quality obtained through field sampling or other high-precision methods; In each iteration, according to Bayes' theorem, the posterior distribution is updated, the optimal weight configuration is estimated from the posterior distribution through Markov Chain Monte Carlo MCMC sampling, and the updated optimal weight configuration is obtained, which is expressed as: in, is the optimization weight of ultrasonic propagation velocity, is the optimization weight of the rebound value, is the optimization weight of the radar reflection signal strength, is the optimization weight of thermal imaging feature parameters; Configure the optimized weights W opt Applied to data fusion and quality score calculation of current detection tasks.

6. The method for detecting concrete material quality based on multi-parameter analysis according to claim 5, characterized in that: The physical information neural network model is constructed to output the concrete internal state prediction result, and the specific steps are as follows: Use optimized weight configuration W opt The input data is weighted and then input into the physical information neural network model for training and prediction; Construct a physics-guided neural network structure where the input layer is a weighted input vector The middle layer is a number of fully connected hidden layers, and the output layer is the concrete strength, density and crack depth; Add physical equation constraints to the loss function, including the introduction of elastic mechanics equations and heat conduction equations, and the expression is: Where σ is the stress tensor, f is the body force, ρ is the density, and c p is the specific heat capacity, k is the thermal conductivity; The above physical equation is added as a penalty term to the loss function to form the physical information loss function, and the total loss function L is calculated. total , the expression is: L total =L data +λ·L PINN ; Among them, L data is the data fitting loss, λ is the balance coefficient; Using the target detection dataset D weigh ted The network is trained to obtain the predicted output, i.e. the predicted values ​​of concrete strength, density and crack depth.

7. The method for detecting concrete material quality based on multi-parameter analysis according to claim 6, wherein: The specific steps of generating the target concrete quality score result with a confidence interval are as follows: Based on the prediction results of the physical information neural network model and the standardized target detection dataset, a disturbance term that obeys the normal distribution is introduced into the input parameters to simulate the measurement error and environmental noise in the actual detection process. Run the physical information neural network model multiple times, use the perturbed input data for inference, and obtain multiple prediction results of the output variable; Count the distribution of output values ​​after each run and calculate their standard deviation; Based on the standard deviation, the contribution weight of each input parameter to the output uncertainty is calculated to obtain the uncertainty impact weight set, which is expressed as: W error ={w error (v),w error (R),w error (A),w error (T)}; Among them, σ Q is the standard deviation of the comprehensive quality score, x i is the i-th input parameter; A comprehensive quality scoring function is constructed based on the optimized weight configuration and uncertainty impact weight set. The expression is: Among them, w S , w D , w L The scoring weights are set based on engineering experience or historical data analysis. The inverse is because the smaller the crack depth, the higher the quality; The Monte Carlo simulation results are Substitute into the above formula to calculate the corresponding quality score For all The kernel density is estimated and the probability density function is fitted. Based on the probability density function, the upper and lower limits at the 95% confidence level are determined to generate the target concrete quality score result with confidence interval.

8. The method for detecting concrete material quality based on multi-parameter analysis according to claim 7, characterized in that: The concrete structure is classified and graded based on the target concrete quality score result with the confidence interval and in combination with the preset quality grading standard. The specific steps are as follows: Prediction value based on comprehensive quality score and its confidence interval, and set the thresholds for three quality grade divisions; The threshold can be dynamically adjusted according to different project types, usage environments or design intensity levels; Compare the confidence interval with the quality grading threshold to determine the quality risk level of the current structure, including: like It means that even under the best conditions, the concrete quality is still below the minimum qualified standard, which is judged as high risk; like It means that even in the worst case, the quality of concrete is still better than the excellent standard, which is judged as low risk; Otherwise, that is, the confidence interval crosses two or more quality level boundaries, indicating the existence of uncertainty impact, and it is judged as medium risk; Based on the above quality grade classification and risk assessment results, a quality assessment report is generated, which includes the quality grade, confidence interval, main defect type, recommended repair measures and risk level.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the concrete material quality detection method based on multi-parameter analysis according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the concrete material quality detection method based on multi-parameter analysis according to any one of claims 1 to 8 are implemented.

Images