Hi,
I hope you are well.
There are different methods by using them it is possible to predict the response. For instance, we can use Machine Learning methods to predict seismic structural response. For this purpose, it is compulsory to have a reliable range of input data and output data. Then, using the regression analysis we can predict response. This is a functional procedure that is used in the literature. For Structural Engineering, this can be vital because it can decrease computational efforts considerably. Therefore, we won't have to use Finite Element programming (e.g., OpenSees) every single time with a huge volume of computational efforts.
I am looking for software that can predict responses in any field of expertise. If you have seen a kind of software that can predict a response within a second, and also, can decrease computational efforts exponentially in comparison with other methods, I would be grateful if you could share that with me.
Best regards,
Mohsen Masoomzadeh.
Dear Mohsen Masoomzadeh ,
Artificial neural networks (ANNs) are machine learning algorithms inspired by the human brain.
ANNs consist of interconnected nodes called neurons, which learn from data to solve problems like image recognition and speech processing. During training, the connections between neurons’ weights are adjusted to minimize errors between the desired and actual output.
https://www.linkedin.com/pulse/artificial-neural-networks-future-machine-learning-inclusioncloud/
Regards,
Shafagat
Dear Shafagat Mahmudova,
Thank you for your clarification and explanation. However, I assumed that people reading this question have a background in ANN/ML and I don't need to explain it.
We also can use "Function Approximation" (FA) or "Pattern Recognition" (PR) in conjunction with ANN to predict a response. In other words, considering input and output data (obtained from a process) and using FA or PR, we can train an ANN, and similar procedures for minimizing error and adjusting synaptic weights will be followed. Therefore, using the mathematical relationship we can predict the process without having to use the process (prediction of the response by ANN/ML). But, my question is about software that can predict the response using ANN or ML tools. For example, MATLAB is a software that I used for this purpose (FA and PR), but it is used in any field, and I am looking for professional software that is exclusively used in a field and they are not well-known in other fields. If I can find all of this software, that would be great!
By the way, this question is new at least for my field: https://www.sciencedirect.com/science/article/abs/pii/S2352012422011705
Sincerely yours,
Mohsen.
Dear Mohsen Masoomzadeh
Hope this massage finds you well. I have not seen any special software for just forecasting and predicting aims but researchers usually utilize Matlab and Python . in fact prediction of structures response can be done using forecasting methods but do not forget that these methods can not be beneficial as software like Opensees, SAP, ETABS, SeismoStruct and etc.Because all of forecasting AI tools need train and test data which is obtained by these software. in other word output of software like opensees, Abaqus, and so on is the input data for forecasting tools.
Also if you want to build an AI model that can predict structure response you should deal with very very big data from all kinds of structures with different stories and lateral load resistance frame more over you should cover all types of structure which are popular according to ASCE, AISC, ACI and so on. building such a model is like building Chat GPT. of course it is not a framework which could be done by an individual hence it needs a big team work to collect too much structures data, train and test the model with a powerful computer, minimize errors and uncertainty and so on.
But predicting structure response is common in structural health monitoring just for 2 or 3 structures to evaluate methods not to biuld a global AI model to forecast response of all types of building structures.
There are many methods for predicting structure response some of them like ARMA, ARIMA, GARCH and so on are traditional tools for predicting time series ( structure response ), and the other are AI based methods like using neural networks such as XGboost, LSTM, GRU, etc.
Good luck.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Computer Graphics and Animation
- 3D