The Research Of Vehicle Fault Diagnosis Method Based On Fuzzy Neural Network

With electronic control technology in a wide range of automotive applications, the overall performance of the automobile has been greatly improved, resulting vehicle faults are becoming increasingly complex, or a simple rule of thumb to determine the detection apparatus has been unable to requirements for vehicle fault diagnosis. Currently, OBD (On-Board Diagnostics) board diagnostic system has been used in automotive electronic control system fault diagnosis, fault code diagnosis mainly in the form of presentation and associated data flow, but ultimately fault to identify and resolve need the aid of other instruments (such as the fault diagnostic scanner) testing and maintenance of professional knowledge. In order to expand the scope of this diagnosis and improve diagnostic accuracy. Therefore, to find a more intelligent and efficient diagnosis of automotive ECU for automotive fault diagnosis is an important direction of research.To overcome these shortcomings, the introduction of this system in the diagnosis of artificial intelligence, fuzzy logic and neural network organically integrated, together with their respective advantages to get a more effective vehicle fault diagnosis method, called fuzzy neural network.First of all, in order to collect relevant sensor information on the dynamic parameters of the automotive electronic control system, developing a PHVFDA(Personal Hand-held Vehicle Fault Diagnostic Apparatus) and vehicle fault diagnosis software which supports multiple protocols and OBDâ…¡standard. This PHVFDA can be used to collect vehicle fault data for subsequent data processing.Secondly, we study the fault diagnosis of automobile anti-lock brake system ABS (Antilock Braking System), and establish the T-S model based fuzzy neural network model, then ant colony optimization algorithm to train the neural network. By comparing the experimental results show that the diagnostic method has the advantages of more efficient intelligence. ABS based fuzzy neural network fault diagnosis expert system, with fuzzy membership function to describe the variety of symptoms, with the distribution of neurons and connection weights to express the distribution of faults, with the ACO-BP algorithm which combines ACO algorithm and BP algorithm to acquire knowledge from training examples to achieve knowledge representation, storage and reasoning. This shows a clear superiority in knowledge acquisition, parallel reasoning, adaptive learning, associative reasoning and fault tolerance, etc. T-S model based fuzzy neural did a good job of the ABS fault diagnosis tasks.Finally, the application of vehicle fault diagnosis method based on fuzzy neural network in PHVFDA, making the diagnostic accuracy is further improved. T-S model based fuzzy neural network training is completed, can be used to quickly and accurately diagnose the corresponding failure to achieve the desired objectives. To firm to a diagnostic apparatus, not only to enhance the precision of the diagnosis, and can attain more vehicle fault data flow in the case of no diagnostic prompt, so that expanding the diagnosis scope of diagnostic equipment.