An intelligent maintenance system and method for climbing formwork construction of the main tower

Abstract

The present invention discloses an intelligent maintenance system and method for the main tower climbing formwork construction. The method includes S1: Before the formwork of the main tower climbing formwork is removed, the temperature and humidity information is collected through the temperature and humidity monitor; S2: Based on the deep learning algorithm, perform a modeling operation on the information to obtain a state classification result; S3: Perform a maintenance operation on the main tower according to the classification result; S4: Train the model in S2 based on transfer learning; S5: After the formwork of the main tower climbing formwork is removed, perform a secondary maintenance operation on the main tower; The system includes a hydraulic climbing formwork frame, a water supply pipeline, a spray head, a temperature and humidity monitor, a control module, and a water supply module. The present application performs intelligent maintenance on the main tower climbing formwork by constructing a transfer learning model based on the deep learning algorithm.

Description

Technical Field

[0001] The present invention relates to the field of maintenance of climbing formwork for buildings, and designs an intelligent maintenance system and method for climbing formwork construction of main towers. Background Art

[0002] With the continuous development of urban construction, high-rise buildings are everywhere. Especially in the first-tier urban areas with compact land use, in order to make full use of urban space, urban construction is gradually developing towards the "ground height space" direction. In order to ensure the safety of construction workers in high-altitude building construction, a climbing formwork construction technology has emerged. It has advantages such as safety and fast construction speed, and has been widely used.

[0003] However, as a typical large-volume concrete construction, the main tower is prone to problems such as cracks during climbing formwork construction and difficult cleaning of water stains on the surface. Maintenance has always been a key and difficult point. Especially, the construction period of climbing formwork construction of the main tower is generally nearly one year and needs to experience the changes of four seasons. The problems of cracks or water stains are more serious. The traditional maintenance methods for climbing formwork construction of the main tower have problems of uneven and untimely moisture retention, resulting in the situation of wet-dry alternation, local overage or insufficient maintenance during the maintenance of the main tower. The solution to the problems existing in the main tower is poor. Therefore, the present invention proposes an intelligent maintenance system and method for climbing formwork construction of the main tower, which improves the quality of maintenance and more effectively solves the problems generated during climbing formwork construction of the main tower. Summary of the Invention

[0004] In view of the above-mentioned defects of the prior art, the present invention proposes an intelligent maintenance system and method for climbing formwork construction of the main tower. The technical solutions designed by the present invention include:

[0005] A hydraulic climbing formwork frame, a water delivery pipeline, a spray head, a temperature and humidity monitor, a control module, and a water supply module; the water delivery pipeline is fixed on the hydraulic climbing formwork frame; one end of the water delivery pipeline is connected to the water supply module, and the other end is connected to the spray head; the temperature and humidity monitor is fixed in the corresponding maintenance area for obtaining the temperature and humidity information of the maintenance area. Each maintenance area corresponds to a temperature and humidity monitor and a group of water delivery pipelines and spray heads; the control module receives the information in real time and performs calculations.

[0006] Preferably, the water delivery pipeline includes a connected hose and a hard pipe. The other end of the hose is connected to the water supply system, and the other end of the hard pipe is connected to the water delivery pipe.

[0007] Preferably, an intelligent control water valve is arranged in the hose, and the opening and closing of the intelligent control water valve are controlled by the control module.

[0008] In addition, the present invention proposes an intelligent maintenance method for climbing formwork construction of the main tower, and the steps are as follows:

[0009] S1: Before form removal, collect temperature and humidity information through the temperature and humidity monitor;

[0010] S2: Based on the deep learning algorithm, perform a modeling operation on the information to obtain a state classification result;

[0011] S3: Carry out maintenance operations on the main tower according to the classification result;

[0012] S4: Train the model in S2 based on transfer learning to improve the accuracy of result acquisition;

[0013] S5: After form removal, perform secondary maintenance operations on the main tower.

[0014] Preferably, the modeling operation in S2 includes: First, establish a sample function, where the information is the sample, which is {(x (1) , y (1) )}, {(x (2) , y (2) )}, …, {(x (r) , y (r) )}; In the formula, r is the number of samples, and y is the class label;

[0015] y (i) ∈{1, 2, …k} = j, where j is the expression of the y value and j is the number of class labels;

[0016] Then, the hypothesis function estimates the probability value p(y = j|x) for each class j. Therefore, use the hypothesis function to output a k-dimensional vector to represent these k estimated probability values, and use the term to normalize the probability to obtain the formula:

[0017]

[0018] In the formula, θ is the Softmax regression model parameter, which is an n×k matrix, and θ is the optimal parameter.

[0019] Preferably, find the optimal parameter θ to improve the accuracy of the predicted value. The function corresponding to the parameter θ is the cross-entropy loss function, and the formula is:

[0020]

[0021] In the formula, m is the number of features of the variable x, and the value of the indicator function I{·} is: I{expression representing true} = 1, I{expression representing false} = 0.

[0022] Preferably, for the minimization problem of J(θ), use the gradient descent method in the iterative optimization algorithm to solve it. Take the derivative of the cross-entropy loss formula to obtain the gradient, and the formula is:

[0023]

[0024] In the formula, is a vector, and the l-th element is the partial derivative of J(θ) with respect to the l-th component of θ j .

[0025] Preferably, there is a redundant parameter set in the gradient formula. By adding to modify the parameter values with excessive penalty in the cost function, the cost function can be transformed into:

[0026]

[0027] In the formula, n is the number of input data;

[0028] After adding the weight decay term λ, the iterative function becomes a strict convex function. Using the gradient descent method, a unique solution can be obtained. The gradient descent method includes taking numbers for λ in ascending order from small to large, then learning the model parameters on the training set, calculating the validation error on the cross-validation set, and selecting the model with the smallest error. Finally, the optimal λ value is obtained by evaluating on the test set.

[0029] Preferably, using an optimization algorithm, the derivative of the function J(θ) can be obtained as:

[0030]

[0031] By minimizing J(θ), a regression model is obtained as a classifier to classify the state of the main tower and obtain the recognition result.

[0032] Preferably, the S3 includes: the control module triggers conditions according to the operation result, and then starts the water supply module to perform the water supply operation according to the instruction, and sprays and cures the area to be cured through the water pipeline and the spray head.

[0033] Beneficial effects:

[0034] 1. The present invention uses a transfer learning model constructed based on a deep learning algorithm to classify and detect the state of the climbing form of the main tower to improve robustness and ensure high accuracy of the results;

[0035] 2. The spraying tool of the present invention is fixed on the climbing formwork. Since there is no need to perform water spraying and watering maintenance after removing the climbing formwork, the step of additionally removing the spraying tool is omitted. Moreover, the present invention adds a secondary maintenance operation for the main tower after form removal, making the maintenance of the climbing form construction of the main tower more comprehensive;

[0036] 3. The present invention uses a control module to intelligently control the spraying operation, and combines the results of classification detection to perform different degrees of maintenance on different main tower areas, improving intelligence. Brief Description of the Drawings

[0037] Figure 1 is a schematic diagram of the system connection of a preferred embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the method flow of a preferred embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of the classification result of a preferred embodiment of the present invention;

[0040] Figure 4 is a schematic diagram of the spray curing process of a preferred embodiment of the present invention. Detailed Description of the Invention

[0041] The following is a detailed description of the embodiments of the present invention. The following embodiments are implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following embodiments.

[0042] The present invention designs an intelligent curing system and method for the main tower climbing formwork construction. The technical solution includes the following steps, as Figure 1 shown, the system specifically includes:

[0043] A hydraulic climbing formwork, a water delivery pipeline (1), a spray head (2), a temperature and humidity monitor (3), a control module (4) and a water supply module (5); the water delivery pipeline (1) is fixed on the hydraulic climbing formwork; one end of the water delivery pipeline (1) is connected to the water supply module (5), and the other end is connected to the spray head (2); the temperature and humidity monitor (3) is fixed in the corresponding curing area for obtaining the temperature and humidity information of the curing area. Each curing area corresponds to a temperature and humidity monitor (3) and a group of water delivery pipelines (1) and spray heads (2); the control module (4) controls the water supply work of the water supply module (5) and the water delivery pipeline (1) by receiving information in real time, performing arithmetic analysis, and according to the result of the arithmetic analysis.

[0044] Preferably, the water delivery pipeline (1) includes a connected hose and a hard pipe. The other end of the hose is connected to the water supply module (5), and the other end of the hard pipe is connected to the water delivery pipe.

[0045] Preferably, an intelligent water valve is provided in the hose, and the opening and closing of the intelligent water valve are controlled by the control module (4).

[0046] In addition, as Figure 2 shown, the present invention proposes an intelligent curing method for the main tower climbing formwork construction, and the steps are as follows:

[0047] S1: Before form removal, collect temperature and humidity information through the temperature and humidity monitor;

[0048] S2: Perform a modeling operation on the information based on a deep learning algorithm to obtain a state classification result;

[0049] S3: Perform a maintenance operation on the main tower according to the classification result;

[0050] S4: Train the model in S2 based on transfer learning to improve the accuracy of result acquisition;

[0051] S5: After form removal, perform a secondary maintenance operation on the main tower.

[0052] Specifically, the operation of S2 is as follows: Learn the data of the main tower collected, obtain and distinguish different main tower concrete states, establish a deep learning model, and use the obtained data as the training set. Train the model through transfer learning to obtain the state result. r samples are {(x (1) , y (1) ), {(x (2) , y (2) ), …, {(x (r) , y (r) ), where the class label y (i) ∈ {1, 2, … k}, and the size of the class label is j; In the formula, k is 5 (as shown in Figure 3 , representing 5 categories, namely moisture level: very high, high, medium, low, very low), and r is the number of samples 250; Then assume that the function estimates the probability value p(y = j|x) for each category j. Therefore, use the hypothesis function to output a k-dimensional vector to represent these k estimated probability values, and use the term to normalize the probability to obtain the formula:

[0053]

[0054] In the formula, θ is the Softmax regression model parameter, which is an n×k matrix.

[0055] The model training process of deep learning needs to continuously iterate to find the optimal parameter θ to improve the accuracy of the predicted value. The iteration value function corresponding to the parameter θ in the present invention is the cross-entropy loss function, and the formula is:

[0056]

[0057] In the formula, m is the number of features of the variable x, and the value of the indicator function I{·} is: I{expression representing true} = 1, I{expression representing false} = 0. For the minimization problem of J(θ), the gradient descent method in the iterative optimization algorithm is used to solve it. The formula for the derivative of the above formula to obtain the gradient is:

[0058]

[0059] In the formula, is a vector, and the l-th element is the partial derivative of J(θ) with respect to the l-th component of θ. Due to the existence of redundant parameter sets, by adding j the parameter values with overly large penalties in the cost function can be modified. The cost function can be transformed into: In the formula, n is the number of input data. After adding the weight decay term (λ > 0), the iterative function becomes a strictly convex function. Using the gradient descent method can ensure obtaining a unique solution and becoming the global optimal solution. First, λ is taken in ascending order, then the model parameters are learned on the training set, the validation error is calculated on the cross-validation set, and the model with the smallest error is selected, and λ is selected; finally, it is evaluated on the test set to obtain the optimal λ value.

[0060]

[0061] Using the optimization algorithm, taking the derivative of the function J(θ) gives:

[0062] By minimizing J(θ), a regression model is obtained as a classifier to classify the state of the main tower, and a recognition result with high accuracy is obtained.

[0063]

[0064] Specifically, in the experiment, 80% of the data is used as the training set, 10% of the data is used as the validation set, and 10% is used as the test set. The model is trained and tested separately. The number of training steps is set to 5000, the learning rate is 0.01, and 50 pictures are randomly selected for training each time, and then cross-validation is performed. The correct rate is tested on the validation set every 50 iterations. Finally, the training situation of the model is evaluated by the probability of classification on the test set. (Cross-entropy is used as the loss function in the neural network to show the training effect. The smaller the value, the better the model learning effect. Each time training, the predicted value is compared with the actual value through the cross-entropy function, and then the weights of the last layer are adjusted by backpropagation. The present invention uses a transfer learning model constructed based on a deep learning algorithm to classify and detect the state of the climbing form of the main tower to improve robustness and ensure high accuracy of the results).

[0065] Preferably, as

[0066] shown, S3 includes: The control module triggers conditions according to the operation result, and then starts the water supply module to perform water supply operations according to the instruction, and sprays and cures the area to be cured through the intelligent control of the water valve to link the spraying device. Figure 4

[0067] ​Maintenance method after dismantling the climbing formwork: Spraying and watering maintenance are not carried out because cracks are likely to occur on the surface of the main tower during the operation process or the surface cracks after formwork removal may develop into deep cracks. Therefore, after formwork removal, a curing agent is sprayed on the main tower to form an emulsion or polymer solution of a continuous waterproof and airtight curing film on the surface of the main tower concrete, reduce water loss, and perform surface heat preservation on the main tower to maintain the main tower after formwork removal.

Claims

1. An intelligent maintenance method for climbing formwork construction of a main tower, characterized in that the steps Including: S1: Before removing the formwork of the main tower climbing formwork, collect temperature and humidity information through a temperature and humidity monitor; S2: Perform a modeling operation on the said information based on a deep learning algorithm to obtain a state classification result; S3: Carry out a maintenance operation on the main tower according to the classification result; S4: Train the model in S2 based on transfer learning; S5: After removing the formwork of the main tower climbing formwork, perform a secondary maintenance operation on the main tower; The modeling operation in S2 includes: First, establish a sample function. The information is the sample, which is {(x (1) , y (1) ), {(x (2) , y (2) ), …, {(x (r) , y (r) )}; where r is the number of the samples, y is the class label, and x is the sample function; y (i) ∈ {1, 2, … k} = j, where j is the expression of the y value and k is the number of class labels; Then, estimate the probability value p(y = j|x) for each category j, and then output a k-dimensional vector representing the k estimated probability values, and use item to normalize the probability values to obtain the formula: In the formula, θ is the optimal parameter, e is the exponential function, i is the i-th sample, and T represents the test of θ.

2. The intelligent maintenance method for the climbing formwork construction of the main tower according to claim 1, characterized in that, Including: Find the optimal parameter θ, the function corresponding to the parameter θ is the cross-entropy loss function, and the formula is: In the formula, m is the number of features of the variable x, and the value of the indicator function I{·} is: I{expression representing true} = 1, I{expression representing false} = 0.

3. The intelligent maintenance method for the climbing formwork construction of the main tower according to claim 2, characterized in that, Including: For the minimization problem of J(θ), use the gradient descent method in the iterative optimization algorithm to solve it. The formula for the gradient obtained by taking the derivative of the cross-entropy loss formula is: In the formula, is a vector.

4. The intelligent maintenance method for the climbing formwork construction of the main tower according to claim 3, characterized in that, Including: The gradient formula has a redundant parameter set. By adding modifying the parameter values with excessive penalty in the cost function, the cost function can be transformed into: where n is the number of input data, and θ ij is the l-th quantity of θ j ; After adding the weight decay term λ, the iterative function becomes a strictly convex function. The gradient descent method can be used to obtain a unique solution. The gradient descent method includes taking numbers for λ from small to large in turn, then learning the model parameters on the training set, calculating the verification error on the cross-validation set, and selecting the model with the smallest error. Finally, evaluate on the test set to obtain the optimal value of λ.

5. The intelligent curing method for the climbing formwork construction of the main tower according to claim 4, characterized in that, Including: Using the optimization algorithm, taking the derivative of the function J(θ) gives: By minimizing J(θ), obtain a regression model as a classifier to classify the state of the main tower and get an identification result.

6. The intelligent maintenance method for the main tower climbing formwork construction according to claim 1, characterized in that, S3 includes: The control module triggers conditions according to the operation result, and then starts the water supply module to perform a water supply operation according to the instruction. Spraying and curing operations are carried out on the area to be maintained through the water pipeline and the spray head.

Images