CN116992759A - Freeze-thawing concrete strength evaluation method and device based on interpretable neural network - Google Patents
Freeze-thawing concrete strength evaluation method and device based on interpretable neural network Download PDFInfo
- Publication number
- CN116992759A CN116992759A CN202310880333.7A CN202310880333A CN116992759A CN 116992759 A CN116992759 A CN 116992759A CN 202310880333 A CN202310880333 A CN 202310880333A CN 116992759 A CN116992759 A CN 116992759A
- Authority
- CN
- China
- Prior art keywords
- neural network
- freeze
- concrete
- interpretable
- thawing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 66
- 238000010257 thawing Methods 0.000 title claims abstract description 48
- 238000011156 evaluation Methods 0.000 title claims abstract description 29
- 238000010801 machine learning Methods 0.000 claims abstract description 42
- 238000012360 testing method Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 38
- 239000002994 raw material Substances 0.000 claims abstract description 7
- 230000003993 interaction Effects 0.000 claims description 47
- 230000000694 effects Effects 0.000 claims description 45
- 238000000034 method Methods 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 14
- 238000003860 storage Methods 0.000 claims description 11
- 238000004590 computer program Methods 0.000 claims description 8
- 244000141353 Prunus domestica Species 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 238000009966 trimming Methods 0.000 claims description 4
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 230000002068 genetic effect Effects 0.000 claims description 3
- 230000009290 primary effect Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 2
- 239000000654 additive Substances 0.000 abstract description 4
- 230000000996 additive effect Effects 0.000 abstract description 4
- 238000010998 test method Methods 0.000 abstract description 2
- 238000012545 processing Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 6
- 239000004568 cement Substances 0.000 description 4
- 238000010200 validation analysis Methods 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 239000010881 fly ash Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 239000002893 slag Substances 0.000 description 2
- 239000008030 superplasticizer Substances 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002301 combined effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a freeze-thawing concrete strength evaluation method and device based on an interpretable neural network, which are used for obtaining test parameters and evaluation parameters of freeze-thawing concrete members; dividing the freeze-thawing concrete sample data set into a training set and a testing set; establishing an interpretable neural network machine learning model, wherein the input of the model is the consumption of various raw material components in each group of freeze-thawing concrete samples, and the output is the strength of the freeze-thawing concrete; and (3) the real-time characteristic data are input into an established interpretable neural network machine learning model, and the strength of the freeze-thawing concrete is predicted. The application replaces the traditional test method to predict the strength of the freeze-thawing concrete, thereby saving a great deal of time and cost; compared with the traditional machine learning algorithm, the prediction model established by the generalized additive interpretable neural network machine learning algorithm provides an accurate and interpretable prediction result.
Description
Technical Field
The application relates to the technical field of freeze-thawing concrete strength evaluation and detection, in particular to a freeze-thawing concrete strength evaluation method and device based on an interpretable neural network.
Background
The strength of the concrete is always one of important indexes for evaluating the durability of the concrete, and the concrete member under the cold condition is often cracked due to the damage of load stress, so that the durability is reduced or even fails, the comfort and the safety of the building are affected to a certain extent, so that the data of the strength of the freeze-thawing concrete can be obtained quickly, reasonably and effectively, and the method has important guiding significance for judging the safety of the structure, thereby prolonging the service life of the concrete to the greatest extent.
In general, the production raw materials of concrete comprise cement, blast furnace slag, fly ash, water, an air entraining agent, a superplasticizer, fine aggregate, coarse aggregate and the like, and most of the conventional direct measurement of concrete strength at present requires a great deal of time and labor cost, and the related experience model also has the problems of insufficient precision and the like. Therefore, the method for rapidly predicting the strength of the freeze-thawing concrete according to the component information of the concrete raw materials and the curing conditions has great significance for optimizing the mixing ratio of the original concrete.
In the prediction and analysis of the performance of the freeze-thawing concrete at present, the traditional machine learning model is regarded as a black box due to lack of interpretability, and with the increase of the complexity of the machine learning model, people often have difficulty in understanding and accepting the prediction result, so that the neural network machine learning model based on generalized additive interpretability can obtain a better result on a prediction problem.
Disclosure of Invention
In order to solve the problems of long test period, complex test and lack of interpretability of a common machine learning model in the conventional concrete durability test, the application provides a freeze-thawing concrete strength evaluation method and device based on an interpretability neural network.
To solve the above technical problems, in a first aspect, the present application provides a method for evaluating strength of freeze-thawing concrete based on an interpretable neural network, comprising the steps of:
(1) Obtaining freeze-thawing concrete member test data, including concrete characteristic data and compressive strength evaluation data;
(2) Constructing a data set based on the step (1), and dividing the data set into a training set and a testing set after random arrangement;
(3) Establishing an interpretable neural network machine learning model, performing model training based on the training set in the step (2), inputting characteristic data of each group of concrete samples, namely raw material consumption and curing conditions, and outputting evaluation data of the concrete, namely compressive strength; the interpretable neural network machine learning model is an interpretable neural network xNN structure, and comprises a fully-connected multi-layer sensor which is decomposed into a projection layer and then a plurality of sub-networks, wherein each sub-network represents a nonlinear shape function, and the xNN adopts neural network parameterization on a main effect and a pairwise interaction; simulating each main effect or pair interaction with a fully connected sub-network of one or two input nodes, the sub-networks being added together to form a final output;
the machine learning model training process of the interpretable neural network is three-stage self-adaptive training: firstly, training and trimming fitting are carried out on a sub-network of a main effect; secondly, selecting, fitting and trimming important paired interactions, and fitting residual errors under genetic constraint; finally, collective fine tuning is performed on all important main effects and paired interactions;
the interpretable neural network machine learning model is expressed as:
wherein E (y|x) represents the expected compressive strength y given the characteristic x, S1, S2 represent the set of principal effects and pairwise interactions, respectively, g is a function of the model, h j And f jk Is a shape function corresponding to the principal effect and the pairwise interaction, μ is the intercept, j represents the jth principal effect, jk represents the pairwise interaction; x is x j And x k Characteristic data respectively expressed as a j-th concrete sample and a k-th concrete sample are used as input of a model;
each of the main effects and pairwise interactions is assumed to be zero-mean, i.e
Wherein F (x) j ) And F (x) j ，x k ) A cumulative distribution function representing the corresponding jth main effect and jk pairwise interactions;
(4) Adopting the test set in the step (2) to test the prediction performance of the trained model, and if the prediction precision does not meet the requirement, correcting by adjusting the parameters of the machine learning model of the interpretable neural network until the prediction precision meeting the requirement is achieved;
(5) And inputting the characteristic data of the freeze-thawing concrete acquired in real time into an interpretable neural network machine learning model meeting the requirements, and predicting the compressive strength of the freeze-thawing concrete.
Further, the training model has constraints of sparsity, genetics, and marginal clarity.
Further, model training will cease when the maximum number of training sessions or validation performance is reached without improvement over a certain number of sessions, each main effect or pair interaction is normalized to have a zero average, the bias nodes of the output layer represent the overall average, and the subnetwork prunes according to sparsity constraints.
Further, model training is amenable to marginal-definition regularization, where all the main effects and pairwise interactions of the fine-tuning stage are renormalized.
Further, the model has an intrinsic interpretability, and the contribution of each individual variable to the overall prediction is examined, and the importance ratio IR of each main effect is quantitatively measured by:
IR(j)＝D(h j )/T
IR(j，k)＝D(f jk )/T
wherein D (h j ) And D (f) jk ) To correspond to the variance of the shape function of the main effect and the pairwise interactions, T represents D (h j ) And D (f) jk ) And j represents the jth primary effect, jk represents pairwise interactions.
Further, the evaluation mode of evaluating the built interpretable neural network machine learning model is to calculate the root mean square error and the correlation coefficient of the target value of the test set and the predicted target value:
wherein N is i And P i The i test target value of the test set and the i forecast target value of the test set are respectively; n is the average test target value; n is the number of samples.
In a second aspect, the application also provides a freeze-thawing concrete strength evaluation device based on the interpretable neural network, which comprises a processor and a memory, wherein the memory is stored with a computer program, and the computer program realizes the freeze-thawing concrete strength evaluation method based on the interpretable neural network when being executed by the processor.
In a third aspect, the present application also provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the method for evaluating the strength of freeze-thaw concrete based on an interpretable neural network.
The application has the beneficial effects that: the method and the device for evaluating the strength of the freeze-thawing concrete based on the interpretable neural network replace the traditional test method to predict the strength of the freeze-thawing concrete, and save a great deal of time and cost; compared with the traditional machine learning algorithm, the prediction model established by the generalized additive interpretable neural network machine learning algorithm provides an accurate and interpretable prediction result.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a freeze-thaw concrete strength evaluation method based on an interpretable neural network;
FIG. 2 is a diagram illustrating model parameter changes during a training phase of machine learning according to an embodiment of the present application;
FIG. 3 is a global interpretation of the prediction result in the first embodiment of the present application;
fig. 4 is a partial explanation of the prediction result in the first embodiment of the present application.
Fig. 5 is a block diagram of a freeze-thaw concrete strength evaluation apparatus based on an interpretable neural network of the present application.
Detailed Description
The following detailed description of preferred embodiments of the application is made in connection with the accompanying drawings, which form a part hereof, and together with the description of the embodiments of the application, are used to explain the principles of the application and are not intended to limit the scope of the application.
The application provides a freeze-thawing concrete strength evaluation method based on an interpretable neural network, which is shown in fig. 1 and is a flow diagram of the freeze-thawing concrete strength evaluation method based on the interpretable neural network, and specifically comprises the following steps:
s1: obtaining freeze-thawing concrete member test data, wherein the freeze-thawing concrete feature data and strength evaluation data comprise information such as cement, blast furnace slag, fly ash, water, an air entraining agent, a superplasticizer, fine aggregate, coarse aggregate, a maintenance period, freeze-thawing times and the like;
s2: the data set is constructed, the data set is divided into a training set and a test set after being arranged randomly, and the test set accounts for 20% of the number of samples when all experimental data are divided into the training set and the test set. In the first embodiment, the test set accounts for 20% of the total number of samples;
s3: establishing an interpretable neural network machine learning model, wherein the input of the model is the characteristic data of each group of freeze-thawing concrete samples, namely the raw material consumption and the test conditions, and the input of the model is the evaluation data of the freeze-thawing concrete;
s4: training an interpretable neural network machine learning model, and storing the trained interpretable neural network machine learning model on the device;
s5: the prediction performance of the trained model is checked by adopting a test set, and if the prediction precision does not meet the requirement, the model is corrected by adjusting the parameters of the machine learning model of the interpretable neural network until the proper prediction precision is achieved;
s6: and inputting the real-time characteristic data, namely the consumption of various raw materials of the freeze-thawing concrete and the maintenance conditions, into an established interpretable neural network machine learning model to predict the strength of the freeze-thawing concrete.
Specifically, when the interpretable neural network machine learning model is built in step S3, the machine learning framework adopts a TensorFlow.
The built interpretable neural network machine learning model has the constraints of sparsity, genetics and marginal definition.
The training of the interpretive neural network machine learning model in step S4 includes:
s41: a three-stage adaptive training algorithm first trains and prunes the sub-network for the main effect. Second, important pairwise interactions are selected, fitted, and trimmed, and residuals are fitted under genetic constraints. Finally, collective fine tuning is performed for all important main effects and pairs of interactions.
S42: the adaptive training algorithm comprises:
the interpretable neural network machine learning model is expressed as:
wherein E (y|x) represents the expected compressive strength y given the characteristic x, S1, S2 represent the set of active principal effects and pairwise interactions, respectively, μ is the intercept, g is a function of the model, h j And d jk Is a shape function corresponding to the principal effect and the pairwise interaction, j represents the jth principal effect, jk represents the pairwise interaction; x is x j And x k Respectively denoted asCharacteristic data of the jth and kth concrete samples are used as input of a model;
each of the main effects and pair interactions is assumed to be zero average, i.e.,
wherein F (x) j ) And F (x) j ，x k ) Representing the cumulative distribution function of the corresponding jth main effect and jk pairwise interactions.
S43: a novel interpretable neural network (xNN) structure, comprising:
a fully connected multi-layer perceptron is decomposed into a projection layer, followed by a plurality of sub-networks, wherein each sub-network represents a nonlinear shape function, xNN employs neural network parameterization for both the main effect and the pairwise interactions; ,
the maximum number of pairwise interactions is set to k=20, each subnetwork is configured with 5 ReLU hidden layers, 40 nodes per layer, the weights of the subnetworks are initialized using gaussian orthogonal initializers, the initial learning rate of adam optimizer is set to 0.0001, the training cycles of the three training phases are set to 5000, 5000 and 500, respectively, 20% of the validation sets are used for early stops, the early stop threshold is set to 50 epochs, and the tolerance threshold η is set to 1% of the minimum validation loss. The marginal sharpness regularization strength may be empirically selected from 0.0001 to 1;
each main effect or pair interaction is simulated with a fully connected subnetwork consisting of one or two input nodes, which subnetworks are then added together to form the final output.
S44: model training will stop when the maximum number of training epochs or validation performance is reached without improvement over a certain number of epochs, each main effect or pairwise interaction is normalized to have a zero average, so that the bias nodes of the output layer represent the overall average, and the trivial subnetwork prunes according to sparsity constraints.
Model training is subject to marginal sharpness regularization, in which all the main effects and pairwise interactions of the fine-tuning stage are renormalized.
The model has an intrinsic interpretability, and examining the contribution of each individual variable to the overall prediction, the Importance Ratio (IR) of each main effect can be quantitatively measured by:
IR(j)＝D(h j )/T
IR(j，k)＝D(f jk )/T
wherein D (h j ) And D (f) jk ) To correspond to the variance of the shape function of the main effect and the pairwise interactions, T represents D (n j ) And D (f) jk ) And j represents the jth primary effect, jk represents pairwise interactions.
In step S5, the evaluation mode of the built machine learning model of the interpretable neural network is to calculate the root mean square error and correlation coefficient of the test target value and the predicted target value of the test set,
wherein Ti and Pi are respectively an ith test target value of the test set and an ith predicted target value of the test set; n is the average test target value; n is the number of samples.
The freeze-thawing concrete data collected in the embodiment of the application are derived from relevant articles of freeze-thawing concrete member tests at home and abroad. After training of the prediction model, the RMSE of the predicted value and the true value of the sample data in the test set is 0.045, and the correlation coefficient R 2 0.952.
The application provides a schematic diagram of model parameter change in machine learning training stage, referring to fig. 2, the application provides the change of training loss and verification loss in the three stages of model training with training time, and the change can be obtained from the diagram: with the addition of the main effect and the paired interaction in the model, the loss is reduced continuously, and the prediction precision is improved.
The present application provides a global interpretation of machine learning predictions, see fig. 3, using a line graph for the main effect to represent the input-output relationship of the individual variables, and a two-dimensional graph for the pairwise interaction to visualize the combined effect of two underlying variables, the percentage of each characteristic variable representing the importance ratio of that variable. The vertical axis in the one-dimensional graph represents the contribution value of the characteristic variable to the prediction, the horizontal axis represents the input value of the characteristic variable, the histogram shows the distribution condition of the characteristic variable, and the influence of the main effect on the prediction result can be judged through the trend of the line graph. The horizontal axis in the two-dimensional graph represents the input value of the first feature in the pairwise interaction, the vertical axis represents the input value of the second feature in the pairwise interaction, and the influence of the pairwise interaction on the prediction result can be judged through the depth of the graph. In the figure, the water content and the concrete strength are in negative correlation, and the cement content and the concrete strength are in positive correlation.
The present application provides a local interpretation of machine learning predictions, see fig. 4, which provides a decision-making case of a machine learning model for a single sample prediction, with the horizontal axis representing the principal effect or pairwise interactions, where the intercept represents the average of all sample predictions and the vertical axis represents the contribution of each principal effect or pairing interaction to the sample prediction. The data in the sample, in which water has a positive effect on the predicted result and cement has a negative effect on the predicted result, is regularized, see fig. 4, with an actual value of 0.5052 and a predicted value of 0.5139.
Compared with the traditional machine learning algorithm, the prediction model established by the generalized additive interpretable neural network machine learning algorithm provides an accurate and interpretable prediction result.
Corresponding to the embodiment of the freeze-thaw concrete strength evaluation method based on the interpretable neural network, the application also provides an embodiment of a freeze-thaw concrete strength evaluation device based on the interpretable neural network.
Referring to fig. 5, the device for evaluating the strength of the freeze-thawing concrete based on the interpretable neural network provided by the embodiment of the application comprises a memory and one or more processors, wherein executable codes are stored in the memory, and the processors are used for realizing the method for evaluating the strength of the freeze-thawing concrete based on the interpretable neural network in the embodiment when executing the executable codes.
The embodiment of the freeze-thawing concrete strength evaluation device based on the interpretable neural network can be applied to any equipment with data processing capability, and the equipment with data processing capability can be equipment or a device such as a computer. The apparatus embodiments may be implemented by software, or may be implemented by hardware or a combination of hardware and software. Taking software implementation as an example, the device in a logic sense is formed by reading corresponding computer program instructions in a nonvolatile memory into a memory by a processor of any device with data processing capability. In terms of hardware, as shown in fig. 5, a hardware structure diagram of an apparatus with optional data processing capability, where a freeze-thawing concrete strength evaluation device based on an interpretable neural network is located, is provided in the present application, and in addition to a processor, a memory, a network interface, and a nonvolatile memory shown in fig. 5, the apparatus with optional data processing capability in the embodiment generally includes other hardware according to an actual function of the apparatus with optional data processing capability, which is not described herein.
The implementation process of the functions and roles of each unit in the above device is specifically shown in the implementation process of the corresponding steps in the above method, and will not be described herein again.
For the device embodiments, reference is made to the description of the method embodiments for the relevant points, since they essentially correspond to the method embodiments. The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purposes of the present application. Those of ordinary skill in the art will understand and implement the present application without undue burden.
The embodiment of the application also provides a computer readable storage medium, on which a program is stored, which when executed by a processor, implements a freeze-thaw concrete strength evaluation method based on an interpretable neural network in the above embodiment.
The computer readable storage medium may be an internal storage unit, such as a hard disk or a memory, of any of the data processing enabled devices described in any of the previous embodiments. The computer readable storage medium may be any external storage device that has data processing capability, such as a plug-in hard disk, a Smart Media Card (SMC), an SD Card, a Flash memory Card (Flash Card), or the like, which are provided on the device. Further, the computer readable storage medium may include both internal storage units and external storage devices of any data processing device. The computer readable storage medium is used for storing the computer program and other programs and data required by the arbitrary data processing apparatus, and may also be used for temporarily storing data that has been output or is to be output.
The above-described embodiments are intended to illustrate the present application, not to limit it, and any modifications and variations made thereto are within the spirit of the application and the scope of the appended claims.
Claims (8)
1. The freeze-thawing concrete strength evaluation method based on the interpretable neural network is characterized by comprising the following steps of:
(1) Obtaining freeze-thawing concrete member test data, including concrete characteristic data and compressive strength evaluation data;
(2) Constructing a data set based on the step (1), and dividing the data set into a training set and a testing set after random arrangement;
(3) Establishing an interpretable neural network machine learning model, performing model training based on the training set in the step (2), inputting characteristic data of each group of concrete samples, namely raw material consumption and curing conditions, and outputting evaluation data of the concrete, namely compressive strength; the interpretable neural network machine learning model is an interpretable neural network xNN structure, and comprises a fully-connected multi-layer sensor which is decomposed into a projection layer and then a plurality of sub-networks, wherein each sub-network represents a nonlinear shape function, and the xNN adopts neural network parameterization on a main effect and a pairwise interaction; simulating each main effect or pair interaction with a fully connected sub-network of one or two input nodes, the sub-networks being added together to form a final output;
the machine learning model training process of the interpretable neural network is three-stage self-adaptive training: firstly, training and trimming fitting are carried out on a sub-network of a main effect; secondly, selecting, fitting and trimming important paired interactions, and fitting residual errors under genetic constraint; finally, collective fine tuning is performed on all important main effects and paired interactions;
the interpretable neural network machine learning model is expressed as:
wherein E (y|x) represents the expected compressive strength y given the characteristic x, S1, S2 represent the set of principal effects and pairwise interactions, respectively, g is a function of the model, h j And f jk Is a shape function corresponding to the principal effect and the pairwise interaction, μ is the intercept, j represents the jth principal effect, jk represents the pairwise interaction; x is x j And x k Characteristic data respectively expressed as a j-th concrete sample and a k-th concrete sample are used as input of a model;
each of the main effects and pairwise interactions is assumed to be zero-mean, i.e
Wherein F (x) j ) And F (x) j ，x k ) A cumulative distribution function representing the corresponding jth main effect and jk pairwise interactions;
(4) Adopting the test set in the step (2) to test the prediction performance of the trained model, and if the prediction precision does not meet the requirement, correcting by adjusting the parameters of the machine learning model of the interpretable neural network until the prediction precision meeting the requirement is achieved;
(5) And inputting the characteristic data of the freeze-thawing concrete acquired in real time into an interpretable neural network machine learning model meeting the requirements, and predicting the compressive strength of the freeze-thawing concrete.
2. The method for evaluating the strength of freeze-thawing concrete based on an interpretable neural network according to claim 1, wherein the method comprises the following steps: the training model has constraints of sparsity, genetics, and marginal clarity.
3. The method for evaluating the strength of freeze-thawing concrete based on the interpretable neural network according to claim 2, wherein the method comprises the following steps of: model training will stop when the maximum number of training durations or verification performance is not improved for a certain number of durations, each main effect or pair interaction is normalized to have zero mean, the bias nodes of the output layer represent the overall mean, and the subnetwork prunes according to sparsity constraints.
4. The method for evaluating the strength of freeze-thawing concrete based on an interpretable neural network according to claim 1, wherein the method comprises the following steps: model training is subject to marginal sharpness regularization, in which all the main effects and pairwise interactions of the fine-tuning stage are renormalized.
5. The method for evaluating the strength of freeze-thawing concrete based on an interpretable neural network according to claim 1, wherein the method comprises the following steps: the model has an intrinsic interpretability, and the contribution of each individual variable to the overall prediction is examined, and the importance ratio IR of each main effect is quantitatively measured by:
IR(j)＝D(h j )/T
IR(j，k)＝D(f jk )/T
wherein D (h j ) And D (f) jk ) To correspond to the variance of the shape function of the main effect and the pairwise interactions, T represents D (h j ) And D (f) jk ) And j represents the jth primary effect, jk represents pairwise interactions.
6. The method for evaluating the strength of freeze-thawing concrete based on an interpretable neural network according to claim 1, wherein the method comprises the following steps: the evaluation mode of evaluating the built interpretable neural network machine learning model is to calculate the root mean square error and the correlation coefficient of the target value and the predicted target value of the test set:
wherein N is i And P i The i test target value of the test set and the i forecast target value of the test set are respectively; n is the average test target value; n is the number of samples.
7. The freeze-thawing concrete strength evaluation device based on the interpretable neural network is characterized in that: comprising a processor and a memory, said memory having stored thereon a computer program which, when executed by said processor, implements a method for evaluating the strength of freeze-thaw concrete based on an interpretable neural network according to any one of claims 1 to 6.
8. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program, when executed by a processor, implements the method for evaluating the strength of freeze-thaw concrete based on an interpretable neural network according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310880333.7A CN116992759A (en) | 2023-07-18 | 2023-07-18 | Freeze-thawing concrete strength evaluation method and device based on interpretable neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310880333.7A CN116992759A (en) | 2023-07-18 | 2023-07-18 | Freeze-thawing concrete strength evaluation method and device based on interpretable neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116992759A true CN116992759A (en) | 2023-11-03 |
Family
ID=88533100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310880333.7A Pending CN116992759A (en) | 2023-07-18 | 2023-07-18 | Freeze-thawing concrete strength evaluation method and device based on interpretable neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116992759A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117388374A (en) * | 2023-12-13 | 2024-01-12 | 南京建正建设工程质量检测有限责任公司 | Method and system for detecting strength of concrete for building |
| CN117763361A (en) * | 2024-02-22 | 2024-03-26 | 泰山学院 | Student score prediction method and system based on artificial intelligence |
-
2023
- 2023-07-18 CN CN202310880333.7A patent/CN116992759A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117388374A (en) * | 2023-12-13 | 2024-01-12 | 南京建正建设工程质量检测有限责任公司 | Method and system for detecting strength of concrete for building |
| CN117388374B (en) * | 2023-12-13 | 2024-02-20 | 南京建正建设工程质量检测有限责任公司 | Method and system for detecting strength of concrete for building |
| CN117763361A (en) * | 2024-02-22 | 2024-03-26 | 泰山学院 | Student score prediction method and system based on artificial intelligence |
| CN117763361B (en) * | 2024-02-22 | 2024-04-30 | 泰山学院 | Student score prediction method and system based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116992759A (en) | Freeze-thawing concrete strength evaluation method and device based on interpretable neural network | |
| CN116757534B (en) | Intelligent refrigerator reliability analysis method based on neural training network | |
| CN108445752B (en) | Random weight neural network integrated modeling method for self-adaptively selecting depth features | |
| CN113962314B (en) | Non-invasive enterprise load decomposition method based on federal learning | |
| CN111257934B (en) | Seismic oscillation peak acceleration prediction method based on second-order neuron deep neural network | |
| CN113822499B (en) | Train spare part loss prediction method based on model fusion | |
| CN111310722A (en) | Power equipment image fault identification method based on improved neural network | |
| Jia et al. | Poisoning attacks on data-driven utility learning in games | |
| CN110334478A (en) | Machinery equipment abnormality detection model building method, detection method and model | |
| Buragohain | Adaptive network based fuzzy inference system (ANFIS) as a tool for system identification with special emphasis on training data minimization | |
| CN115482877A (en) | Fermentation process soft measurement modeling method based on time sequence diagram network | |
| CN117370771A (en) | Knowledge embedding filling soft measurement method based on conditional fractional diffusion | |
| US20140052425A1 (en) | Method and apparatus for evaluating a model of an industrial plant process | |
| CN115217152B (en) | Method and device for predicting open-close deformation of immersed tunnel pipe joint | |
| CN116304940A (en) | Analog circuit fault diagnosis method based on long-short-term memory neural network | |
| Kotenko et al. | Formation of Indicators for Assessing Technical Reliability of Information Security Systems | |
| CN115545446A (en) | Multi-parameter evaluation method and system for reform enterprises | |
| CN113779858B (en) | Combustion optimization method and system, storage medium and electronic equipment | |
| CN107025497A (en) | A kind of electric load method for early warning and device based on Elman neutral nets | |
| Andersson et al. | Data-driven impulse response regularization via deep learning | |
| Liu et al. | Research on software reliability evaluation technology based on BP neural network | |
| CN117391895A (en) | Judgment result prediction method and system based on causal learning | |
| CN117093896A (en) | Heat pump system fault diagnosis method based on combined data driving model | |
| Greenberg | Validation of decision support systems | |
| CN118521189A (en) | Raw water turbidity control method based on convolution-long-short-term memory-attention |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |