In terms of accuracy, and training requirement for different operating conditions.
The major drawback is in design phase i.e to.determine the optimal number of nodes, hidden layers, sigmoid function, etc
hello friend
Greetings
There are plenty of issues with ANN like listed below
you don't know which is the best configuration of the ANN like nodes, sigmoid function, etc
Its considered a black box model.
Well there are better models than that
See my papers for more prediction models better than ANN.
All the best.
Best regards
Dr.Indrajit Mandal
Hello dear friend.
That is great dear Dr. Indrajit. Your contribution is highly appreciated. I will go throw the papers and return to you ..
Great thanks and best regards.
The main drawbacks with Neural Networks (NNs) are complexity (i.e. "black-box" model as Dr. Mandal has mentioned), and also that a large dataset is often needed to train and initialise the model properly. NN models, if correctly implemented, can have excellent performance for pattern recognition. However they are not widely used in power applications outside of academia, and I don't know of any utilities that are currently using NN's for voltage stability assessment.
As Dr. Mandal says NN is black box and we don't know which is the best configuration of the ANN like nodes, sigmoid function, etc He is correct.
If you are using NN then you must take care of following points....
1. train NN up to regration factor more than 0.93 (train NN by maximum possible targets. say more than 500 set)
2. output layer must have appropriate TF to generate exact activation function (i.e., o/p)
3. W matrix and B matrix must be designed as per requirement.
- Badges
- Science topic