CN-116415485-B - Multi-source domain migration learning residual service life prediction method based on dynamic distribution self-adaption

Abstract

The invention discloses a residual service life prediction method based on dynamic distribution self-adaption multi-source domain migration learning, which comprises the following steps of 1) giving degradation data of an existing source domain and a target domain, 2) preprocessing the degradation data, 3) extracting degradation characteristic representation of the degradation data of the source domain and the target domain, 4) aligning degradation characteristic distribution of each source domain and each target domain to obtain multiple degradation characteristic representation of the target domain, and 5) fusing RUL labels obtained by multiple degradation characteristics of the target domain through predictors of each specific domain to serve as final RUL prediction labels. The migration learning can utilize similarities between data, tasks, or models to apply models and knowledge learned from old domains to new domains. The RUL prediction method based on transfer learning utilizes the existing degradation data set to train a prediction model, and applies the learned knowledge transfer to data sets of different working conditions to realize cross-domain RUL prediction.

Inventors

• LV YI
• WEN ZHENFEI
• ZHANG QICHEN

Assignees

• 电子科技大学中山学院

Dates

Publication Date

20260512

Application Date

20221229

Claims (2)

1. 1. The residual service life prediction method based on the dynamic distribution self-adaption multi-source domain transfer learning is characterized by comprising the following steps of: 1) Given the existing active domain and target domain degradation data; 2) Preprocessing the degradation data; 3) Extracting degradation characteristic representations of degradation data of a source domain and a target domain; 4) Aligning degradation characteristic distribution of each source domain and each target domain to obtain multiple degradation characteristic representations of the target domain; 5) Fusing RUL labels obtained by multiple degradation characteristics of the target domain through predictors of each specific domain, and taking the RUL labels as final RUL prediction labels; In step 1), the source domain and target domain degradation data: Given existing multisensor degradation data As shown in the formula (1), The degradation data is represented in a matrix form, wherein Representing the number of sensors that can monitor the degradation state, Is the length of degradation data, the service life of the equipment is characterized by taking a time period as a scale, and the steps 3) to 5) comprise three modules, namely a degradation characteristic extraction module, a dynamic distribution self-adaption module and a regression prediction module, A degradation characteristic extraction module, which consists of a common characteristic extractor and a specific domain characteristic extractor, and is used for extracting degradation characteristic representation of degradation data of a source domain and a target domain, The part of the action extracts low-level feature representations of the multi-source domain and the target domain, which is a common feature extractor, the common feature extractor consists of four convolution blocks for extracting low-level feature representations of the source domain and the target domain, A domain-specific feature extractor for extracting unique features of the domain, low-level feature representations of each of the source domain and the target domain obtained through the upper layer module respectively obtaining high-level feature representations through the domain-specific feature extractor, the domain-specific feature extractor being composed of four layers of GRU units as final degradation features, the low-level feature representations obtaining high-level feature representations of the source domain and the target domain through the portion, A dynamic distribution self-adaption module used for dynamically adjusting the influence of edge distribution difference and condition distribution difference, aligning the degradation characteristic distribution of each source domain and target domain to obtain multiple degradation characteristic representations of the target domain, wherein the dynamic distribution self-adaption method provides that the edge distribution self-adaption and the condition distribution self-adaption are not equally important, the method can adaptively adjust the importance of the edge distribution and the condition distribution in the distribution self-adaption process according to the distribution situation of actual degradation data, and the dynamic distribution self-adaption adopts a balance factor To dynamically adjust the distance between the two distributions: Wherein, the , When the data is close to 0, the degradation data of the source domain and the target domain are shown to have larger difference, and the edge distribution adaptation is more important; near 1, the source domain and target domain data sets are shown to have higher similarity, condition distribution adaptation is more important, Measuring data edge distribution difference of source domain and target domain by using multi-core maximum mean difference MK-MMD method Edge distribution adaptation is achieved by minimizing MK-MMD, which is calculated in the manner shown in (5), Wherein, the Representing source domain degradation characteristics subject to p-distribution, Representing the degradation characteristics of the target domain subject to q-distribution, Is a mapping function that maps the degradation data into the renewable Hilbert space RKHS for measurement, but Because of the difficulty of selection, the kernel function calculation is not explicitly defined but is introduced The MMD is indirectly calculated, the Gaussian kernel function is adopted, the calculation formula is shown as (6), Wherein, the Taking a plurality of sigma values in MK-MMD to calculate to obtain a plurality of kernel matrixes for the width of the kernel function, summing to obtain a final Gaussian kernel matrix, For conditional distribution differences In the method, a conditional maximum mean difference CMMD based on MK-MMD is designed, firstly, RUL labels of degraded data samples are divided into four classes, label classification modes are shown in (7), Wherein, the And And respectively representing a classification label and an RUL label, wherein Y is the maximum life cycle, the classification method takes a degradation data sample of a health state as a class, the data sample of the degradation stage is divided into three stages, the degradation degree is gradually increased from front to back, and as the degradation data of a target domain is not marked, in the training process, the model firstly uses a source domain pre-training label classifier, then target domain data obtains a classification pseudo label through the classifier, and along with iterative training, the precision of the classifier is gradually improved, so that an accurate classification label is obtained, the invention uses a cross entropy loss function to calculate label classification loss of the source domain, as shown in (8), Wherein, the Representing the number of source domain samples, And The real label and the classification label of the ith source domain sample are respectively represented, and according to the classification results of the source domain and the target domain, the CMMD calculation method is as follows (9): Wherein, the Representing the class of degraded data samples, And Representing samples of degraded data belonging to class c in the source domain and the target domain respectively, Dynamically distributing adaptive factors during model training The calculation method of (2) is shown in (10), Wherein, the A metric representing the feature variance of the source domain and target domain degradation data samples, may measure the edge distribution alignment of the source domain and target domain features during training, The measurement of the feature difference of the degradation data samples representing the source domain and the target domain belonging to the class c can measure the condition distribution alignment condition of the features of the source domain and the target domain in the training process, Representing the error of the support vector based classifier in distinguishing source domain and target domain data samples, As can be derived from equations (5) (9) (10), the objective function of the dynamically distributed adaptive module is: by minimizing the objective function, the degradation features of each source domain and target domain are mapped to the same feature space reduction feature distribution difference, and finally the target domain can get multiple degradation feature representations, Regression prediction module firstly, the degradation characteristics of the source domain and the target domain obtained by the dynamic distribution self-adaption module are used for obtaining RUL prediction labels of the source domain and the target domain through a specific domain regression predictor, the invention uses the RMSE performance evaluation index as a prediction loss function, as shown in (13), Finally, the module fuses RUL labels obtained by various degradation characteristics of the target domains through predictors of each specific domain, and uses a mean fusion method as a final RUL prediction label, so that the module ensures that decision boundaries of each domain pair are aligned, identical target samples predicted by different regressors should obtain identical predictions, therefore, the model needs to minimize differences among regressors of all specific domains, an objective function for aligning prediction results of each regressor is shown as (14), Wherein S and Representing the number of source fields and the number of samples per source field, And Representing the predicted results of the regressors m and n i-th data samples respectively, Joint loss function the joint loss function of MDDAN model consists of four parts, regression prediction loss error Label classification loss error Dynamic distributed adaptive objective function Objective function aligned with predicted outcome Thus, the joint loss function of the model can be expressed as: Wherein, the Is a trade-off coefficient for controlling The proportion of the loss is calculated by the method, Is a time-varying coefficient, which varies with each training iteration, i is the current iteration number, epochs is the total iteration number.
2. 2. The method for predicting the remaining service life of the multi-source domain transfer learning based on dynamic distribution self-adaption according to claim 1, wherein the method comprises the following steps: in the step 2), since the magnitude and size of the monitoring data of the plurality of sensors are greatly different, normalization is needed before the monitoring data are applied to a model, the degradation data are preprocessed by adopting a maximum and minimum normalization method, a calculation formula is shown in the step (2), Wherein, the Representing the ith characteristic signal in data sample x Maximum and minimum values of (a), normalized data Thereafter, a sliding time window method is used to convert the degraded data into a time series input, the size of the input time window being The time step is The input data may be expressed as: 。

Description

Multi-source domain migration learning residual service life prediction method based on dynamic distribution self-adaption Technical Field The invention relates to the technical field of fault processing, in particular to a residual service life prediction method based on dynamic distribution self-adaption multi-source domain migration learning. Background With the rapid development of the intelligent industrial age, the artificial intelligent technology is widely applied to the field of mechanical equipment fault and predictive management (PHM), and greatly improves the operation reliability of mechanical equipment while reducing manpower and material resources. As one of key technologies of PHM, the residual service life prediction utilizes condition monitoring data and fault mechanism of equipment to build a degradation model, analyzes degradation trend of the equipment to judge the fault time of the equipment, and has wide prospect in the fields of manufacturing industry, aerospace and the like. The development of the technology aims at giving an early warning to equipment which is about to fail, preventing the equipment from suddenly failing to cause huge loss and safety problems, reducing the maintenance cost of start to write and improving the operation reliability of the equipment. The accuracy of RUL predictions is susceptible to various factors such as uncertainty factors, e.g., operating conditions, operating environment, and noise in the monitored data. Therefore, RUL prediction of mechanical devices has been a challenging task. In recent years, many RUL prediction methods have been proposed, which can be classified into three types, a model-based method, a data-driven method, and a hybrid method of the former two. The model-based approach is to build a mathematical or physical degradation model to predict the RUL of the mechanical component based on the system failure mechanism. The data driving method is to use a large amount of historical data to extract degradation characteristics of equipment, establish a mapping relation between the degradation characteristics and the residual service life, and fit a degradation curve to achieve the purpose of predicting RUL. The mixing method is to utilize the historical operation data and fault mechanism of the equipment at the same time, and fully combine the advantages of the two methods to carry out RUL prediction. Because model-based methods require a lot of a priori knowledge and for complex devices it is very difficult to build corresponding degradation models, data-driven methods have been a research hotspot for RUL prediction. The deep learning has been developed in the field of residual service life prediction due to its strong feature extraction capability and accurate regression analysis capability, but a key problem still exists. In RUL prediction for most devices, it is generally assumed that the test set and the training set are from the same operating conditions, follow the same distribution, and therefore the model only has accurate prediction results under the same operating conditions. However, in the actual operation process of the equipment, the working conditions of most of the equipment are different, and certain difference exists in the distribution of the data collected by the sensors, so that the accuracy of RUL prediction is drastically reduced. Disclosure of Invention Aiming at the problems existing in the prior art, the invention aims to provide a residual service life prediction method based on dynamic distribution self-adaption multi-source domain transfer learning. In order to solve the problems, the invention adopts the following technical scheme. A residual service life prediction method based on dynamic distribution self-adaption multi-source domain transfer learning comprises the following steps: 1) Given the existing active domain and target domain degradation data; 2) Preprocessing the degradation data; 3) Extracting degradation characteristic representations of degradation data of a source domain and a target domain; 4) Aligning degradation characteristic distribution of each source domain and each target domain to obtain multiple degradation characteristic representations of the target domain; 5) And merging the RUL labels obtained by the multiple degradation characteristics of the target domain through the predictors of the specific domains to serve as final RUL prediction labels. As a further improvement of the present invention, In step 1), the source domain and target domain degradation data: Given the existing multisensor degradation data { X } s×b, as shown in equation (1). The degradation data is represented in a matrix form, where s represents the number of sensors that can monitor the degradation state and n is the length of the degradation data, typically on a time period scale, to characterize the lifetime of the device. As a further improvement of the present invention, In the step 2), since the magnitud