Misfire Detection Using Crank Speed and Long Short-Term Memory Recurrent Neural Network

In this work, a new approach was developed for the detection of engine misfire based on the long short-term memory recurrent neural network (LSTM RNN) using crank speed signal. The datasets are acquired from a six-cylinder-inline, turbo-charged diesel engine. Previous works investigated misfire detection in a limited range of engine running speed, running load or misfire types. In this work, the misfire patterns consist of normal condition, six types of one-cylinder misfire faults and fifteen types of two-cylinder misfire faults. All the misfire patterns are tested under wide range of running conditions of the tested engine. The traditional misfire detection method is tested on the datasets first, and the result show its limitation on high-speed low-load conditions. The LSTM RNN is a type of artificial neural network which has the ability of considering both the current input in-formation and the previous input information; hence it is helpful in extracting features of crank speed in which the misfire-induced speed fluctuation will last one or a few cycles. In order to select the engine operating conditions for network training properly, five data division strategies are attempted. For the sake of acquiring high performance of designed network, four types of network structure are tested. The results show that, utilizing the datasets in this work, the LSTM RNN based algorithm can overcome the limitation at high-speed low-load conditions of traditional misfire detection method. Moreover, the network which takes fixed segment of raw speed signal as input and takes misfire or fault-free labels as output achieves the best performance with the misfire diagnosis accuracy not less than 99.90%.

Misfire detection of diesel engine based on convolutional neural networks

With the ever-stringent vehicles exhaust emission standard and higher requirements on on-board diagnostic technology, the importance of misfire detection in vehicle emission control is emerging. The performance of a traditional misfire detection algorithm predominantly depends on the features and classifier selected. Fixed and handcrafted features require either a reliable dynamic model of an engine or a large number of experiment data to define the threshold, and then, form a map. Since convolutional neural networks (CNNs) have an inherent adaptive design and integrate the feature extraction with classification functions into a compact learning framework, the misfire fault-sensitive features can be auto-discovered from raw speed signals. Furthermore, CNNs can detect the fault features of the misfire through network training with fewer engine operating conditions. In this paper, the theory and method of the misfire diagnosis based on CNNs are presented. The experimental data for network training and testing are sampled on a six-cylinder inline diesel engine. The misfire patterns containing every one-cylinder and two-cylinder misfire are tested under the wide speed and load conditions of the engine. The results show that when the engine operates under steady-state conditions, one-cylinder or two-cylinder complete misfires can be detected accurately by CNNs. In addition, one-cylinder partial misfire is employed to examine the adaptability of trained 1-D CNN. It turns out that when the partial misfire reaches the same level as half amount of the normal fuel injection quantity, one-cylinder partial misfire can be detected with accuracy more than 96%. At last, the misfire detection under the non-stationary conditions, such as acceleration or deceleration, is conducted. The results show the 1-D CNN performed well in a limited acceleration range, and network failure occurs when the absolute acceleration of the engine speed is more than 100 r/min/s.