Research On Performance Reliability Analysis And Assessment Of Automotive Engine ECU Based On Degradation Data

The reliability assessments of automobile engine ECU are applied to its development,testing and maintenance phases,and which can provide a favorable guarantee for ECU optimization design and production test.Therefore,the study on reliability evaluation methods of ECU has an important theoretical significance and practical value.At present,the reliability evaluation of ECU is mainly based on life data,which takes the failure time or the number of failed products as statistical objects.However,with the continuous improvement of material properties and device performance,it is very difficult for ECU to appear failure data or even zero failure in a specific time,which poses a great challenge to the reliability evaluation of ECU.Therefore,it is urgent to explore a new method of ECU reliability analysis.A large number of degradation data during the processing of ECU working and testing are used to build models by the author,and the ECU reliabilities of module level and system level are studied,respectively.It includes ECU feature parameters extraction and failure threshold determination method,module level reliability evaluation based on multivariate degradation parameters distribution,module level reliability assessment under the condition of stress accelerated degradation,as well as ECU system level reliability evaluation based on information fusion and weighted fault tree,and so on.The study can provide a new idea for ECU reliability evaluation.The working principle and failure mechanism of ECU are studied deeply.The results show that it is difficult to establish the degradation trajectory of each sample under the complex and changeable working environment and degradation process.Moreover,it is hard to characterize the degradation process with a single performance degradation parameter for every ECU module.Therefore,it is suitable to make reliability evaluation of ECU by using the method based on multivariate degradation parameters distribution.However,existing reliability assessment methods based on multiple degradation parameters don't consider the contribution degree of each parameter degradation to the reliability,which leads to low reliability evaluation accuracy,and it has certain deviation from the actual working state of the product.Therefore,the correlation features among multivariate degradation parameters and contribution of degradation quantity to product performance reliability are taken as a starting point,a performance weighted reliability assessment method based on multivariate degradation characteristic parameters distribution is proposed by the author.Firstly,the correlation coefficient between multivariate degradation parameters are calculated in this algorithm,and the correlation characteristic among them is determined according to the set correlation coefficient threshold.Then,the determination methods of weight coefficients based on contribution and correlation characteristic are given respectively for the relevant and irrelevant two cases among multivariate characteristic parameters.Finally,performance weighted reliability evaluation method based on weighted coefficients is given.On this basis,the feature extraction method and failure threshold determination method for each module of ECU are studied,and the performance weighted reliability evaluation method proposed in this paper is applied to the ECU module level reliability evaluation.The study on ECU reliability evaluation under the condition of accelerated stress shows that temperature stress and electrical stress have a great influence on the degradation failure of ECU,and the higher the temperature or the greater the electrical stress,the faster the degradation rate of the ECU.Moreover,it is difficult to determine the degradation values distribution type at each measurement time because of ECU complex degradation characteristic under electric stress condition.Based on the above analysis,a temperature stress accelerated degradation model is put forward in the light of temperature stress accelerated condition,and performance reliability evaluation methods are given respectively for degradation values obeying Weibull distribution and the normalized distribution which is applied to the ECU module evaluation.Aiming at the electrical stress accelerated degradation situation,a performance reliability evaluation model based on support vector machine regression is established.The model needn't know the distribution type of degradation values in advance at each testing time,and it only need to train and establish SVR model for degradation data under various electrical stress conditions.Based on this,the degradation data in the normal working stress condition can be predicted and it is used to carry out reliability assessment of ECU.On the basis of performance reliability assessment of ECU modules,a system performance reliability evaluation algorithm based on information fusion and weighted fault tree is presented by the author.The multi-layer information fusion method is used to carry out the three levels fusion for the prior information about ECU similar products and ECU products,and the weighted factors of the performance reliability for every module of ECU are obtained.These weighting factors are used as the influence factor of the fault tree,and then the performance reliability of ECU system is evaluated by using the weighted fault tree.In addition,based on the fault tree analysis method,the qualitative analysis is carried out for the common faults in the fuel injection module and the power management module of ECU,which can provide technical guidance for its rapid diagnosis and fault location.Based on the above research,the performance testing and analysis evaluation system of engine ECU is developed by the author.The signal is simulated by using ECU hardware-in-loop system based on closed-loop-feedback,and the degradation data under the condition of stress acceleration for each module are acquired accurately by using the electrical performance data acquisition algorithm based on adaptive sampling rate.The module reliability and system reliability of ECU are evaluated respectively according to the proposed performance reliability algorithms,and system reliability is compared with the actual system reliability.The results show that the accuracy and efficiency of the method can meet the actual needs of the project,and it can be used as a useful supplement to the traditional ECU reliability evaluation,which can provide theoretical guidance for ECU reliability design and optimization.