Smart Rechargeable Ni-mh Battery Based On Artificial Neural Network Research

With the rapid development of electronic technology, electricity storage plays a pivotal role in our daily life. We can not only solve the energy problems of automobile, but also can reduce the exhaust emissions and retard the greenhouse effect, if we apply the electricity storage to automobile. Therefore, automobiles that consume the electricity will become the future development of automobile industry from all over the world. Lead-acid battery, lithium ion battery and MH-Ni battery are the main storage medium that can store electrical energy. By comparing these batteries, we find out that nickel metal hydride(MH-Ni battery)is the first choice of the industry for its excellent combinational properties, such as high specific energy, long circular life, adaptable to large current discharge, none-polluted and less explosive. Based on current references, this paper introduces neural network and fuzzy control technology to electricity consuming automobile to lucubrate the problem of Ni-MH battery fast and loseless charging.Pointing out the current problem that electricity consuming automobiles have a long charging time, we come up with the idea of combine the fast speed charging station, middle speed charging station and family slow speed charging for the electric automobile. The middle speed charging station is the combination of charging station and public parking areas , which mainly solves the problem as decrease of battery capacity caused by unfully fast charging, while the fuzzy control is used in the middle speed charging station, achieving 30% to 70% power charging in 0.5 to 1.5 hours. If the parking time is longer enough, the charged can be finished in 5 hours.Secondly, the charging character of MH-Ni battery is studied. Based on the slow convergence speed of BP neural network and some problems of local minimum, this paper takes an approximate charge model of MH-Ni battery, and proposed an effective neural network optimization algorithm based on the RBF design difficulty of appointing the number of the hidden layer nodes and center dynamically: First adjusted the number of the hidden layer nodes and initialized the center using nearest neighborhood clustering learning algorithm, then optimized the center numbers of the hidden layer and resolve the output weights and threshold based on the generalization inverting matrix. The stimulation results indicate that this method can realize precise prediction. Finally, the neural network and fuzzy control are combined to design a fuzzy neural network controller which can be used for recharging the MH-Ni battery intellectually, the stimulation results show that this method can short the recharge time of MH-Ni battery and realize intelligent recharge of MH-Ni battery. Finally,we combined neural network and fuzzy control and designed a fuzzy neural network controller to automatically charge the MH-Ni storage battery model. It was proved in a simulation that this method can complete the later stage of charging in a short time during the whole charging process.According to the fuzzy control method that mentioned above, we designed a single phase charging device based on microcontroller MSP430F169 to sampling the terminal voltage,charging current information of MH-Ni in the fourth stage where the voltage information would be processed through a 2-dimensional fuzzy controller. By querying the fuzzy control table, the corresponding current changing level can be obtained to calculate the current of next stage. Then, the MSP430F169 delivers the adjusted PWM wave to output to control the on-off of transistor and changes direct voltage to control the charging current.We evaluate the whole device after finished relevant hardware and software design. We find out that, without harming the battery, the charge can be accomplished for 4 hours and 20 minutes saver than constant current/constant voltage charging method, which we consider with higher intellectuality.