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.
In recent years, heuristic intelligent algorithms have achieved rapid development in solving combinatorial optimization problems. Travelling salesman problem(TSP) is one of the classical NP-hard problems in combinatorial optimization, and it needs for more accurate and faster algorithms. Combining heuristic algorithm and cell-like P system, this paper proposes a membrane evolutionary algorithm for solving TSP(MEATSP). MEATSP abstracts the behaviors of biological cells such as fusion, division, cytolysis and selection into operators, and obtains the optimal solution of the problem through the evolution of membrane structure and objects within the membrane. Experiments on TSPLIB data sets show that the proposed algorithm performs well in both optimal solution and average residual, and it has strong stability when solving TSP with different scales.
Membrane computing is a kind of distributed and parallel computing model inspired by biological cell mechanism. The maximum parallelism of membrane computing improve the computational efficiency of its computational model. In this paper, a tissue P system named ΠKP is proposed to solve the 0-1 knapsack problem, which is one of classic NP hard problems. ΠKP could obtain accurate solutions of knapsack problem and point out the amount of accurate solutions, which mainly consists of three stages: first, generate all the solutions of knapsack problem by cell division; then calculate the weights and total values in all the candidate membranes, which will be kept or dissolved according to the restriction of knapsack problem; check out the final solutions. The instances are executed on a membrane simulator named UPSimulator, and the result of the experiments shows the whole searching procedure by the rules and proves the correctness and efficiency of the system.
1 January 2016 in Communications in Computer and Information Science