Abstract
Fault diagnosis is critical to ensuring the safety and reliable operation of rotating machinery systems. Long short-term memory networks (LSTM) have received a great deal of attention in this field. Most of the LSTM-based fault diagnosis methods have too many parameters and calculation, resulting in large memory occupancy and high calculation delay. Thus, this paper proposes a low-delay lightweight recurrent neural network (LLRNN) model for mechanical fault diagnosis, based on a special LSTM cell structure with a forget gate. The input vibration signal is segmented into several shorter sub-signals in order to shorten the length of the time sequence. Then, these sub-signals are sent into the network directly and converted into the final diagnostic results without any manual participation. Compared with some existing methods, our experiments illustrate that the proposed method has less memory space occupancy and lower computational delay while maintaining the same level of accuracy.
Highlights
Mechanical fault diagnosis—analysing data collected by sensors and predicting the health of mechanical systems—has become a research hotspot in industry [1,2]
In order to solve the above problems, this paper proposes a low-delay lightweight recurrent neural network (LLRNN) model for mechanical fault diagnosis, making the following two main contributions: (1) The design of a lightweight network structure based on a special Long short-term memory networks (LSTM) cell with only a forget gate, reducing the parameters and calculation of the network
In order to verify whether the simplification of the cell structure had too much of a negative impact on performance, a comparison experiment was performed using Gated recurrent unit (GRU) and LSTM with the same number of hidden units and almost-calculated quantities
Summary
Mechanical fault diagnosis—analysing data collected by sensors and predicting the health of mechanical systems—has become a research hotspot in industry [1,2]. RNNs have the ability to memorise historical information, and only need to process the input data once They can detect the health status of mechanical systems in real time. In order to solve the above problems, this paper proposes a low-delay lightweight recurrent neural network (LLRNN) model for mechanical fault diagnosis, making the following two main contributions: (1) The design of a lightweight network structure based on a special LSTM cell with only a forget gate, reducing the parameters and calculation of the network. Compared with the LSTM-based methods and some CNN-based models, in our experiments the proposed algorithm took up less storage space and had shorter calculation delay under the same accuracy and noise immunity, and was more suitable for real-time fault diagnosis.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have