Compressive stregnth of concerte using image processing ANM & ML
- 1. Department Of Civil Engineering Seminar Review 3 Topic Name :- Predecting Compressive Strenght Of Concrete Using Image Processing & ML. Student Name :- Mundhe Sujit S. (TECV140) Guide Name :- Prof. G.W. Rathod
- 2. 2 1) To Find Different Method Based On Dataset. 2) To Learn Working Of Each Method. 3) To Get The Final Conclusion. To Check The Applicability Of IPT & ML For Determination Of Compressive Strength Of Concrete. OBJECTIVES AIM
- 3. METHDOLOGY 3 Selection of Topic Study Through Research papers Finding different Methods Study the Outcomes of different method Literature Review & Conclusion Preparation of Report
- 4. Months January Feb March April May JUNE Weeks 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 Selection Of Topic Collection Of Research Paper Studying Different method Literature Review & Conclusion Prediction Of Results TIME SCALE CHART 4
- 5. Department Of Civil Engineering, PCCOE 7 SR. NO. NAME OF RESEARCH PAPER NAME OF JOURNAL NAME OF RESEARCHER CONCLUSIONS 1. Machine learning-based compressive strength prediction for concrete: An adaptive boosting approach Construction and Building Materials D.Cheng Feng ,Zhen-Tao Liu & X.Wang The AdaBoost algorithm is adopted to predict the compressive strength of concrete. Large Dataset is collected to train the model and reaches an accuracy of 98%. Different algorithms are compared to show the superior of the proposed model. Key factors and in the AdaBoost and influence of input variables are investigated 2. Assessment of concrete compressive strength by image processing technique CONSTRUCTION AND BUILDING MATERIALS Celalettin Basyig, Bekir Çomak & Ismail Serkan Seven different concrete specimens were produced. Concrete compressive strength values were predicted based on the Grey Scale value of image. ImageJ Software is used for proccesing on images 3. Modeling Of Strength Of High- performance Concrete Using Artificial Neural Networks CEMENT AND CONCRETE RESEARCH I.C. Yeh This paper is aimed at demonstrating the possibilities of adapting artificial neural networks (ANN) to predict the compressive strength of high-performance concrete. A set of trial batches of HPC was produced in the laboratory and demonstrated satisfactory experimental results 4 Metaheuristic-Based Machine Learning System for Prediction of Compressive Strength based on Concrete Mixture Properties and Early-Age Strength Test Results CIVIL ENGINEERING DIMENSION, VOL. 20, NO. 1 Prayogo D. This research develops an advanced machine learning method to forecast the concrete. hybrid system is proposed so called the symbiotic organisms search-least squares support vector regression (SOS–LSSVR) 5 Fuzzy logic model for the prediction of cement compressive strength CEMENT AND CONCRETE RESEARCH Sedat Akkurta, Gokmen Tayfurb, Sever Canc These rules are expressed in the If –Then format.No mathematical equations and parameters The input variables of alkali, Blaine, SO3, and C3S and the output variable of 28-day cement strength were fuzzified by the use of artificial neural networks The prediction of 50 sets of the 28-day cement strength data by the developed fuzzy model was quite satisfactory
- 6. Department Of Civil Engineering, PCCOE 8 SR. NO. NAME OF RESEARCH PAPER NAME OF JOURNAL NAME OF RESEARCHER CONCLUSIONS 6 CONSTRUCTION AND BUILDING MATERIALS Machine learning in concrete strength simulations: Multi- nation data analytics J.S. Chou ,C.F.Tsai & A.d. Pham This comprehensive study used advanced machine learning techniques to predict concrete compressive strength. Model performance is evaluated through multi-nation data simulation experiments. The prediction accuracy of ensemble technique is superior to that of single learning models. This study developed advanced learning approaches for solving civil engineering problems 7 Machine learning techniques to predict the compressive strength of concrete International Journal Of Methods Numerical For Calculation And Design In Engineering P. F. S. Silva ,G. F. Moita & V. F. Arruda Three computational methods of machine learning and artificial intelligence were used, namely Random Forest, Support Vector Machines and Artificial Neural Networks The obtained results show that the Random Forest gave the best performance. Computational intelligence models can be used to predict the compressive strength of concrete specimens, is shown in this study. suggests an approach to evaluate compressive strength against destructive testing methods 8 Machine learning in concrete strength simulations: Multi- nation data analytics CONSTRUCTION AND BUILDING MATERIALS J.S. Chou ,C.F.Tsai & A.d. Pham This comprehensive study used advanced machine learning techniques to predict concrete compressive strength. Model performance is evaluated through multi-nation data simulation experiments. The prediction accuracy of ensemble technique is superior to that of single learning models. This study developed advanced learning approaches for solving civil engineering problems. 9 Concrete compressive strength detection using image processing based new test method Measurement G.Dogan,M.H. Arslan & M. Ceylan This study uses ANN and IP together to determine the mechanical properties of concrete, such as the compressive strength, modulus of elasticity and maximum deformation. The primary objective of study is to predict the mechanical properties of concrete without causing destruction, using a new alternative method 96 cylindrical concrete samples were produced; images of the samples were taken before they were examined at the compression testing, and the training and testing procedures for ANN and IP were realized using the obtained pressure readings at the laboratory
- 7. FINDINGS 7 1030 sets of data is collected to train the model Different algorithms are compared to show the superior of the proposed model. Familiar with Softwares Such As ImageJ Adaboost & Random forest algorithm are suitable for achieving maximum accurancy than other models. The releation for training the data & Testing the data is generally 4:1 is suitable.
- 8. 8 Compressive strength of the concrete is very important factor in any concrete work. Equipment's required for the testing are very expensive. Require more time for the prediction of results. Image Processing Technology (IPT). Introduction COMPRESSIVE STRENGTH Destructive Testing CTM/UTM Non-destructive Testing Rebound Hammer Test /Ultrasonic Pulse Velocity etc.
- 9. 9 IMAGE PROCESSING TECHNOLOGY A picture can be converted into digital form Adaptation of the human eye’s vision physiology to computer systems. Image processing methods include many processes such as image provision, image digitalization, segmentation, image enhancement , classification, recording and recalling. Image processing is used in medicine , radar images assessment. weather forecasting and estimation of agricultural products
- 10. 10 Self Driving Cars Image Tagging & Recognozation Optical Character Recognization OCR Why to Choose Image Processing ?
- 11. The Overall Work Can be Done With Different Ways Some Of them As Follows:- Using Grey Scale Value With Machine Learning (ML) With Artificial Neural Network (ANN) 11
- 12. In this study, seven different concrete classes were obtained by using different water/cement ratios, which are 0.79, 0.69, 0.61, 0.54, 0.47, 0.42 and 0.37. As known, with increasing cement amounts in the mixture, concrete compressive strength values also increase. Therefore, an increase is also expected in the gray color values in concrete images. 12 Using Grey Scale Value
- 13. Sample Preparation Concrete specimens, which are 100x100x100 mm in size, were marked as shown in Fig. Marked specimens were cut into pieces with water in a coring header, so that four pieces were obtained from each specimen. In total, six surfaces, one of which is from the cut surface of pieces no 1 and 4, and two of which are from pieces no 2 and 3 were obtained. Cutting was made in perpendicular direction to concrete pouring 13
- 14. 14 Setup For For Production Of Images Photographed in 4272 x 2848 pixel size and in RGB color mode, with jpg extension.
- 15. 15 Processing Sizes were turned into 2350 x 2350 pixels by means of the open source code ImageJ software The images were first transferred from RGB color mode to grayscale of 8 bytes Grayscale images are composed of different gray values (G = {0, 1, 2,..., 255}), which means that there are 256 different values in an image. The 256 gray value is defined as a byte (1 byte = 8 bit and 28 = 256). As a rule, 0 value corresponds to black, while 255 value corresponds to white. Gray tones are formed in between these value The image histogram identifies the pixels at each point in an image and shows the number of these pixels; so a great deal of information regarding an image can be provided by histograms
- 16. 16 • In the next step, histograms, which are the graphical expressions of pixel values on an image, are formed for each image turned into grayscale.
- 17. 17
- 18. 18 What Is Machine Learning? “Machine Learning is the science of getting computers to learn and act like humans do, and improve their learning over time in autonomous fashion, by feeding them data and information in the form of observations and real-world interactions.” How We Get Machines to Learn There are four basic steps for building a machine learning application (or model). These are typically performed by data scientists working closely with the business professionals for whom the model is being developed.
- 19. 19 Step 1: Select and prepare a training data set Step 2: Choose an algorithm to run on the training data set Step 3: Training the algorithm to create the model Step 4: Using and improving the model Data Collection and Analysis The output variable is the compressive concrete strength (MPa) which depends on eight input variables: 1. Cement (kg/m3 ), 2. Fly ash (kg/m3 ), 3. Blast furnace slag (kg/m3 ), 4. Water (kg/m3 ), 5. Superplasticizer (kg/m3 ), 6. Coarse aggregate (kg/m3 ), 7. Fine aggregate (kg/m3 ) 8. Age at the time of testing (days).
- 20. 20 List of Common Machine Learning Algorithms Here is the list of commonly used machine learning algorithms. These algorithms can be applied to almost any data problem: 1.Linear Regression 2.Logistic Regression 3.Decision Tree 4.Support Vector Machine(SVM) 5.Random Forest 6.Dimensionality Reduction Algorithms 7.Gradient Boosting algorithms 8. Adaptive boosting Out Of Which the Adaptive boosting (Adaboost) is most used for the concrete & it Founded that it is More Accurate Than Others,
- 21. 21 Adaptive boosting ( Adaboost)) How AdaBoost Works? It makes n number of decision trees during the training period of data. As the first decision tree/model is made, the record which is incorrectly classified during the first model is given more priority. Only these records are sent as input for the second model. The process will go on until we specify a number of base learners we want to create. The repetition of records is allowed with all boosting techniques. The characteristic of AdaBoost is to use the initial training data to generate a weak learner, then adjust the distribution of the training data according to the predicting performance for the next round weak learner training. The training samples with low predicting accuracy in the previous step will get more attention in the next step. Finally the weak learners are integrated together with different weights to a strong learner
- 22. 22 This figure shows that when the first model is made and the errors from the first model are noted by the algorithm, the record which is incorrectly classified is given as the input for the next model. This process is repeated until the specified condition is met. As you can see in the figure, there are n number of models made by taking the errors from the previous model. This is how boosting works. The models 1,2, 3,…, N are individual models that can be known as decision trees. All types of boosting models work on the same principle
- 23. 23 Implementation process The implementation of AdaBoost can be performed easily. In general it has four stages: (1) collection of the experimental data; (2) generation of the strong learner; (3) test or validation of the learner; (4) application of the learner to the engineering problems.
- 24. 24 There are many different algorithms used to train neural networks with too many variants. there are three layers in the neural network. •The input layer •Hidden Layer •The output layer Artificial Neural Network (ANN) Artificial neural networks (ANN) is a Biologically inspired computing code with the number of simpl highly interconnected processing elements for simulating human brain working & to process information model What Is Meant by ANN ?
- 25. 25
- 26. 26 Steps Involved In ANN In ANN model the data set was divided into three subsets 1.Training which contains 70% of total data 2.Validation which contains 15% of total data 3.Test data which contains 15% of total data
- 27. 27 Training stage The training state for the artificial neural network model. As it is illustrated in the Figure, the errors are repeated 6 times after epoch 9 and the test is stopped at epoch 15. This error repeats starting at epoch 10 demonstrated over-fitting of the data. Therefore, the epoch 9 is selected as the base and its weights are chosen as the final weights. Moreover, the validation check is equal to 6, due to the fact that the errors are repeated 6 times before stopping the process.
- 28. 28 Figure presents the validation performance and mean squared error of the network starting at a large value and reducing to a small value. The plot consists of three lines for three different steps of training, validation, and test. Training process on the training vectors continues until the model gets to the point that the training reduces the error of network on the validation vectors which would lead to avoiding the over-fitting of the data sets. As it is shown in the Figure, the best validation performance is happened at epoch 9, and after 6 error repetitions, the process is stopped at epoch 15. Validation stage
- 29. 29 Figure shows the error histogram for the three steps of training, validation, and test in artificial neural network modeling. As it is shown in the Figure, the zero error is illustrated with a yellow line in the middle with 9 instances in the training set. Validation stage
- 30. 30 Comparative Results Obtained From ML & ANN
- 31. 31 CONCLUSION Image Processing & ML Helps to getting Easily the Compressive Strength of Concrete Data must be well trained to get the maximum Accurancy With IPT & ANN the cost required can be reduced upto a great extend The Results Obtained From these techniques are very close to the actual values At Initial Stages For the data preparation & data training requires much algorithms knowledge . With very intelligently using these technique we can by-pass the costly traditional instruments Such As CTM/UTM. This technique also helps to nullified the 28 days wait for the results . Ultimately Saves Time & hence Speed of work increase.
- 32. 32 D.Feng , & Zhen-tao Liu (2019), “Machine Learning-based Compressive Strength Prediction For Concrete: An Adaptive Boosting Approach”, Construction And Building Materials. Celalettin, Bekir Çomak & Ismail Serkan (2012), “Assessment Of Concrete Compressive Strength By Image Processing Technique” , Construction And Building Materials. Prayogo, D (2018) “Metaheuristic-based Machine Learning System For Prediction Of Compressive Strength”, Civil Engineering Dimension, Vol. 20. I.C. Yeh (1998), “Modeling Of Strength Of High-performance Concrete” Cement And Concrete Research, vol. 28. G.Dogan &Musa H. Arslan,(2017) “Concrete Compressive Strength Detection Using Image Processing Based New Test Method”, Measurement. References
- 33. J.Chou, C.F. Tsai & A.D. Pham ,(2014), “Machine Learning In Concrete Strength Simulations: Multi-nation Data Analytics”, Construction And Building Materials. Sedat Akkurta, Gokmen Tayfurb, & Sever Canc,(2004), “Fuzzy Logic Model For The Prediction Of Cement Compressive Strength”, Cement And Concrete Research. Priscila F. S. Silva G. F.Moita & V.F. Arruda (2020), “Machine learning techniques to predict the compressive strength of concrete” Revista International. Shivaraj. M ,Ravi Kumar H , Prema Kumar W & Preetham. S (2015), “Prediction of Compressive, Flexural and Splitting Tensile Strengths of Concrete using Machine Learning Tools”, International Journal Of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol.4 Susom Dutta, P. Samuiand & D. Kim,(2018), “Comparison of machine learning techniques to predict compressive strength of concrete”, Computers and Concrete. 33 References
- 34. 34