While the storage tank is being used, the malfunction may occur because of improper operation, the machine problems or some external factors. Once the storage tank malfunction occurs, the country will suffer huge losses, not only to economy but also to environment. So it is important to diagnose the storage tank malfunction timely.As the storage tank malfunction occurs, its flow or pressure will change accordingly, so the storage tank malfunction can be diagnosed according to the change of the flow or pressure.Firstly, a net model is established that is a four-layered fuzzy system based on the characteristics of the storage tank malfunction. According to the model framework, the BP neural network method is used to solve the diagnostic model of the storage tank malfunction. Then the weights and transmitted functions of each layer in the model will be got. According to the effective analysis of the diagnostic model of the storage tank malfunction based on the neural net, it is proved that the model based on the BP network has quicker convergent speed and higher accuracy than others. On the other hand, after the simulating data, the result shows that the model can diagnose the various malfunctions of storage tank precisely.Next, compared with the linear regression, the superiority of the BP neural network is obvious. It has much quicker convergent speed and higher accuracy. Furthermore, after the simulating data, the diagnostic model of the storagetank malfunction based on the BP neural network is so effective that its results almost equals to the objects. Whereas the diagnostic system based on the linear regression has some deviation.Finally, according to the applications in Huangdao tank farm, it proves that the diagnostic system of the storage tank malfunction based on the BP network is effective and practical. And at the same time, the operational attentions have also been concluded.It is the first time to solve diagnosis of the storage tank malfunctions by using fuzzy mathematics and BP neural network in our country. Through the simulating data and analysis of the studying samples and non-studying samples, the results indicate the superiority and accuracy of the method using fuzzy mathematics and neural network to solve diagnosis of the storage tank malfunctions.
↧