Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/jancervenka/turbofan_failure

Aircraft engine failure prediction model
https://github.com/jancervenka/turbofan_failure

lstm prediction-model python scikit-learn svm tensorflow

Last synced: about 6 hours ago
JSON representation

Aircraft engine failure prediction model

Awesome Lists containing this project

README 

        
# Aircraft Engine Failure prediction Model



[0]: https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/#turbofan

[1]: https://ieeexplore.ieee.org/document/4711423/

[2]: https://ieeexplore.ieee.org/document/6678166/

[3]: https://ieeexplore.ieee.org/document/4711414/



This problem requires to accurectly predict remaining useful life (RUL) of aircraft

turbofan engines based on various sensor measurements (multivariate time series).

The RUL is defined as the number of engine cycles before failure.



I tried to predict the RUL values for engine units in the FD004 dataset from

[Turbofan Engine Degradation Simulation Data Set][0] using two different models

(LSTM network and support vector machine). My code can be found in the

`turbofan.ipynb` file.



## Data Preparation and Feature Engineering



First, I computed the RUL value for each row in the dataset to get a dataframe in the

following form:



| unit | cycle | sensor_1 | sensor_2 | sensor_n | RUL |

|------|-------|----------|----------|----------|-----|

| 1    | 1     | 0.2      | 30       | 0.9      | 192 |

| 1    | 2     | 0.3      | 29       | 0.2      | 191 |



Each row can be used as a model training sample where the `sensor_k` columns are

the features and the `RUL` is the model target. The rows are treated as independend

observations and the measurement trends from the previous cycles are ignored.



As recommended in [(1)][1], the features are normalized to `μ = 0, σ = 1`

and PCA is applied.



This simplified approach is used to train the support vector machine model.



### Samples as Time Series



For the LSTM model, I opted for more advanced feature engineering and chose to

incorporate the trends from the previous cycles. In this case, each training sample

consists of masurements at cycle `i` as well as `i-5, i-10, i-20, i-30, i-40`.



The model input is a 3D tensor with shape `(n, 6, 24)` where `n` is the number of

training samples, `6` is the number of cycles (timesteps), and `24` is the number

of principal components (features).  



## LSTM Regressor



After running random search to optimize the hyperparameters and some experimentation,

I settled on the following architecture:



```

Model: "rlu_estimator"

_________________________________________________________________

Layer (type)                 Output Shape              Param #   

=================================================================

components (InputLayer)      [(None, 6, 24)]           0         

_________________________________________________________________

lstm (LSTM)                  (None, 64)                22784     

_________________________________________________________________

dropout_lstm (Dropout)       (None, 64)                0         

_________________________________________________________________

hidden_0 (Dense)             (None, 64)                4160      

_________________________________________________________________

dropout_0 (Dropout)          (None, 64)                0         

_________________________________________________________________

hidden_1 (Dense)             (None, 64)                4160      

_________________________________________________________________

dropout_1 (Dropout)          (None, 64)                0         

_________________________________________________________________

hidden_2 (Dense)             (None, 64)                4160      

_________________________________________________________________

dropout_2 (Dropout)          (None, 64)                0         

_________________________________________________________________

rul_prediction (Dense)       (None, 1)                 65        

=================================================================

Total params: 35,329

Trainable params: 35,329

Non-trainable params: 0

```



The model is using L1L2 regularization and dropout layers to mitigate overfitting.



I trained the model 25 epochs and used annealing scheduler to decrease the

learning rate over time.



![LSTM History](img/lstm_history.svg "LSTM Training")



## Support Vector Machine



Use of a Support vector machine (SVM) model is suggested in [(2)][2]. The authors

recommend to use non-linear radial basis (RBF) function.



## Results



I chose three different metrics to assess the performance of the models. Mean square

error (MSE), median absolute error (MAE) and the Score as defined in [(3)][3].

I modified the Score formula (11) in [(3)][3] by dividing the overall value by the

number of testing samples (the definition in the paper contains an error, `a_1`

and `a_2` should be switched in order to penalize RUL overshooting more heavily).



| metric | LSTM    | SVM     |

|--------|---------|---------|

| MSE    | 4627    | 5894    |

| MAE    | 35      | 45      |

| Score  | 8.58e10 | 1.65e14 |



![Comparison](img/comparison.svg "Model Comparison")



## Conclusions



It is clear that the advanced features together with the LSTM outperform the SVM model. 

The high Score values are caused by few outliers with significant errors and the exponential

nature of the formula [(3)][3].



## References

* [(1) Data driven prognostics using a Kalman filter ensemble of neural network models][1]

* [(2) PHM-Oriented Integrated Fusion Prognostics for Aircraft Engines Based on Sensor Data][2]

* [(3) Damage Propagation Modeling for Aircraft Engine Run-to-Failure Simulation][3]