Synthetic system fault generation

1. A system, comprising:

a memory that stores computer executable components; and

a processor that executes the computer-executable components stored in the memory, wherein the computer executable components comprise:

a generator component that employs a trained artificial intelligence (AI) model to generate a synthetic system fault, represented as a combination of discrete parameters and continuous parameters that define a system state, wherein the generator component restores complex multimodal distributions of a subset of at least one of the discrete parameters or the continuous parameters conditioned to a row of an associated dataset; and

a fault assembler component that analyzes the synthetic system fault and generates textual content corresponding to the synthetic system fault.

2. The system of claim 1 , further comprising a model trainer component that trains the AI model using a generative adversarial network architecture.

3. The system of claim 2 , wherein the generative adversarial network architecture is a conditional tabular adversarial network.

4. The system of claim 2 , wherein the model trainer component preprocesses historical system data comprising at least one of historical discrete parameters or historical continuous parameters, wherein the historical discrete parameters are represented as one-hot encoded vectors, and for the historical continuous parameters mode-specific normalization with Variational Gaussian Mixture Model is employed.

5. The system of claim 2 , wherein the model trainer component trains the AI model with Wasserstein Gradient Penalty Loss (WGAN-GP Loss).

6. The system of claim 1 , wherein the textual content comprises at least one of a system log or a description of the synthetic system fault.

7. The system of claim 1 , wherein the computer-executable components further comprise an aggregator component that generates a knowledge base of generated synthetic system faults and respective textual data.

8. A computer-implemented method comprising:

restoring, by a system operatively coupled to a processor, complex multimodal distributions of a subset of at least one of discrete parameters or continuous parameters conditioned to a row of an associated dataset;

employing, by the system, a trained artificial intelligence (AI) model to generate a synthetic system fault, represented as a combination of the discrete parameters and the continuous parameters that define a system state; and

analyzing, by the system, the synthetic system fault and generating textual content corresponding to the synthetic system fault.

9. The computer-implemented method of claim 8 , further comprising training, by the system, the AI model using a generative adversarial network architecture.

10. The computer-implemented method of claim 8 , further comprising training, by the system, the AI model using a conditional tabular adversarial network.

11. The computer-implemented method of claim 8 , further comprising preprocessing, by the system, historical system data comprising at least one of historical discrete parameters or historical continuous parameters, wherein the historical discrete parameters are represented as one-hot encoded vectors, and for the historical continuous parameters mode-specific normalization with Variational Gaussian Mixture Model is employed.

12. The computer-implemented method of claim 8 , further comprising training, by the system, the AI model with Wasserstein Gradient Penalty Loss (WGAN-GP Loss).

13. The computer-implemented method of claim 8 , further comprising generating, by the system, a knowledge base of generated synthetic system faults and respective textual data.

14. The computer implemented method of claim 13 , wherein the textual content comprises at least one of a system log or a description of the synthetic system fault.

15. A computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

restore complex multimodal distributions of a subset of at least one of discrete parameters or continuous parameters conditioned to a row of an associated dataset;

employ a trained artificial-intelligence (AI) model to generate a synthetic system fault, represented as a combination of the discrete parameters and the continuous parameters that define a system state; and

analyze the synthetic system fault and generate textual content corresponding to the synthetic system fault.

16. The computer program product of claim 15 , the program instructions further executable by the processor to cause the processor to: train the AI model using a generative adversarial network architecture.

17. The computer program product of claim 15 , the program instructions further executable by the processor to cause the processor to: train the AI model using a conditional tabular adversarial network.

18. The computer program product of claim 15 , the program instructions further executable by the processor to cause the processor to: preprocess historical system data comprising at least one of historical discrete parameters or historical continuous parameters, wherein the historical discrete parameters are represented as one-hot encoded vectors, and for the historical continuous parameters mode-specific normalization with Variational Gaussian Mixture Model is employed.

19. The computer program product of claim 15 , the program instructions further executable by the processor to cause the processor to: generate a knowledge base of generated synthetic system faults and respective textual data.

20. The computer program product of claim 15 , wherein the training the AI model is based on a Wasserstein Gradient Penalty Loss (WGAN-GP Loss).