Research On Measuring And Control System For Automotive Detection System Based On Embedded Real-time System

In recent years, automobile is gradually accepted by private family and become one part of life with the development of automotive industry. Automobile is the most important transportion tool in modern society. However, traffic accidents occur frequently, Most of which is due to the poor quality of automotive property. All the problems cause the emergence of new challenge to improve automotive detection technique. With the increasing of driving distance, the status of automobile becomes bad which means the property of automobile decreases, stability becomes weak, failure of engine often occurs and the serious pollution. The diagnosis of automotive status is completed according to the activities such as inspection, measurement, analysis and judgment .The methods include two types: one is the traditional artificial diagnose method with experience, the other is the using modern industrial instrument in which the parameters and data is obtained with some instruments, after that the received data are processed by the procedures of filtering, analyzing, judging, saving, printing and showing results.With the innovation of modern technology, embedded real-time system is adopted in many fields such as industrial control, instrument, aviation industry et al, which has great influence on automotive industry. The development of embedded technique constructs a great space for automatic control, which make the users accomplish the efficient control system conveniently and rapidly.In this dissertation, author present a new method in the area of information extracting, data processing and resource sharing for the detection vehicle instruments that integrate embedded real-time system, CAN bus, artificial neural network and multi-sensor techniques together to realize automatic control, real-time sampling and processing for the data received of the detection system. It adopts fewer working machines, and completes the testing tasks rapidly and precisely. The hardware in this dissertation includes 32 bits RISC embedded microchip computer using ARM7TDMI-S as core. The software is theμC/OS-II real-time operate system. The author presents a scheduling algorithm for the embedded real-time testing system after analyzing the deficiency of scheduling algorithm based onμC/OS-II. In this method, two priorities are proposed. One is the basic priority, the other is schedule priority. And a scheduling optimized design method is presented based on priority successive protocol and priority conflicting protocol. Some new results are obtained by this method.Based on analyzing the problem occurred in information collection with uni-sensor, the method based on information integration with embedded multi-sensor to improve the accuracy is presented. It constructs the model of the information integration with multi-sensor in automotive detection system. An algorithm of self-adapted weight information integration is built in testing controller. Moreover, artificial neural network system is simulated in on-spot controller through transferring parameters. Using CAN bus as the bus platform, author realizes the distributed data collection which means the data showing and collection among several embedded real-time systems are built.The main research achievements in this dissertation are as following:1. Design of data collecting hardware and software. In automotive testing system, high acquirements such as collecting speed and precision for sensor, respond and feedback speed for executing equipment is asked. The general 8 bit and 16 bit microchip computer can not meet the above demands. Author applies the embedded real-time system into the data collecting system, selects the microchip computer and LPC2294 controller with ARM7 as the core. embedded real-time operate system is designed based on theμC/OS-II real-time operate system.2. Design of optimized scheduling algorithm. Through analyzing the current scheduling algorithm for real-time task and its deficiency, in order to improving the scheduling algorithm based onμC/OS-Ⅱreal-time system and solve the priority converse and locking problem, author designs a method to consider the importance of process and priority scheduling algorithm to meet the demands of real-time data processing in automotive detection system. Based on the priority real-time scheduling algorithm, author selected a real-time muilti-task scheduling algorithm to adapt automotive detection platform. The task is divided into two parts: real-time task and time-sharing task. The real-time task keeps the original priority schedule inμC/OS-Ⅱ. The two priorities are set for one task and both are based on priority success protocol and conflicting protocol to optimize the scheduling algorithm. In the time-sharing task, the originalμC/OS-Ⅱcode is modified to reach the purpose of data cycle detection.3. Transplant ofμC/OS-Ⅱ: The transplant between 32 bit ARM and 8 bit P89C58 is realized after analyzing the real-time embedded operate system, in which three integrating files OS_CPU.H, OS_CPU_A.S,OS_CPU_C.C are mainly realized. CAN bus is applied to using system to realize CAN bus protocol inμC/OS-Ⅱoperate system.4. Design the automotive detection system based on CAN. The framework for automotive detection system based on embedded real-time system is proposed. Author analyzes the protocol of CAN bus and iCan. The improvement for protocol of iCAN is accomplished to find the protocol of CAN application layer and to adapt to the bus of measuring and control system for automotive detection line. The CAN bus andμC/OS-Ⅱis used into detection system to realize the real-time operating system among several equipments.5. Study on data integrated algorithm based on multi-sensor: Several traditional data integrating algorithm is analyzed on the condition of information integration for multi-sensor. The complicated algorithm as information processing at intellectual knot on CAN bus is not suitable for the calculation capability of microchip computer system and real-time demand, so a reasonable data integrating algorithm is selected in which self-adapted weighting information integrating algorithm for sensors based on CAN bus is designed. By using the artificial neural network in on-spot controller with which the weight factors and sensor data measured in various positions are obtained. In this method, the value of minimum integrating mean error can be achieved, which make the measuring results of the system more reliable.The research works in this dissertation focus on the design of automotive detection system based on CAN with technique of combining embedded real-time system and multi-sensor data integration. Some achievements are obtained in the area of application techniques and algorithms which will be helpful for improving the real-time data collection in automotive detection system.