Industrial time series, as a kind of data that responds to production process information, can be analyzed and predicted for effective monitoring of industrial production processes. There are problems of data shortage and algorithm cold start in industrial modeling process caused by complex working conditions, change of data acquisition environment and short running time of equipment. As a result, the accuracy of the existing data-driven industrial time series prediction algorithm is greatly limited. To address the above problems, we propose a new time series prediction method for industrial processes under limited data based on dynamic transfer learning in this work. This method aims to effectively use historical data of similar equipment or working conditions rather than discard them to help establish an industrial time series prediction model with limited target data. In this method, first, historical data are divided into multiple batches, and then a new multi-source transfer learning framework with dynamic maximum mean difference (MMD) loss is established according to the distribution distance between each batch of historical data and the limited target data at the current moment. The framework also combines multi-task learning methods to establish multi-step prediction model for online learning in industrial processes. Compared with other commonly used methods, experiments on two real-world datasets of solar power generation prediction and heating furnace temperature prediction demonstrate the effectiveness of the proposed method.
Heating temperature is very important in the process of billet production, and it directly affects the quality of billet. However, there is no direct method to measure billet temperature, so we need to accurately predict the temperature of each heating zone in the furnace in order to approximate the billet temperature. Due to the complexity of the heating process, it is difficult to accurately predict the temperature of each heating zone and each heating zone sensor datum to establish a model, which will increase the cost of calculation. To solve these two problems, a two-layer transfer learning framework based on a temporal convolution network (TL-TCN) is proposed for the first time, which transfers the knowledge learned from the source heating zone to the target heating zone. In the first layer, the TCN model is built for the source domain data, and the self-transfer learning method is used to optimize the TCN model to obtain the basic model, which improves the prediction accuracy of the source domain. In the second layer, we propose two frameworks: one is to generate the target model directly by using fine-tuning, and the other is to generate the target model by using generative adversarial networks (GAN) for domain adaption. Case studies demonstrated that the proposed TL-TCN framework achieves state-of-the-art prediction results on each dataset, and the prediction errors are significantly reduced. Consistent results applied to each dataset indicate that this framework is the most advanced method to solve the above problem under the condition of limited samples.
17 April 2023 in IEEE Transactions on Instrumentation and Measurement
Billet heating temperature directly affects the quality of the billet, but the existing technology cannot measure the billet surface temperature. Therefore, we accurately predict the temperature of the furnace by a soft sensor to approximate the billet temperature. Limited by the complexity of the heating process and the lack of computing resources in factories, the existing research pays little attention to or even cannot meet the temperature prediction requirements of multiple heating zones in multiple heating furnaces. To address the above problems, we propose a temperature prediction method for multi-heating furnaces based on transfer learning and knowledge distillation. The method establishes a multisource domain model of a source domain furnace on a cloud platform. Then, the multisource knowledge is transferred to the target domain heating furnace, and the target domain teacher model is established by fine-tuning the source domain model through the target domain data. Next, to efficiently predict the furnace temperature, a shallow multi-task student model is established at the edge server to predict multiple heating zones in the target furnace. Furthermore, a knowledge distillation method for regression prediction is proposed so that the student model can improve the prediction accuracy under the guidance of the cloud teacher model. The effectiveness of the method is verified by experiments on 20 different heating zone datasets in two heating furnaces and two wind turbine datasets. The consistency of multiple experiments shows that this method can not only improve the accuracy by transferring the source domain knowledge but also reduce the size of model parameters by knowledge distillation on the premise of meeting the requirements of prediction error.
10 February 2026 in Expert Systems with Applications
Compared with the optimization technology of retraining models, fine-tuning pretrained neural networks have been widely used in industrial process monitoring because of their high precision and low training cost. However, data privacy protection in industry makes it impossible to improve the upstream pertaining model to avoid negative transfer when the target domain and the source domain are too different or the pertaining model is maliciously attacked. To solve the above problem, we proposed a fine-tuning method assisted by target domain knowledge for industrial time-lag parameter prediction. This method uses an Auto-Encoder to learn the representation of black-box and time-lag knowledge in the target domain. Then, the black-box and time-lag knowledge are used as auxiliary information to update the high-level weights of the pretrained network. At the same time, we proposed an auxiliary learning method that can dynamically update weights without introducing new parameters and provide training methods for different neural network optimizers. The experimental results on the heating furnace temperature prediction and wind condition prediction of wind farms demonstrate that the prediction performance can be effectively improved, and the defective inheritance of the pretrained model can be effectively reduced to avoid negative transfer. Note to Practitioners —Due to the complexity of industrial processes, the detection of key parameters is time-consuming and it is difficult to obtain labeled samples. With the widespread use of large models in computer vision and natural language processing, fine-tuning pretrained models can alleviate these challenges. However, the problem that comes with it is safety. The industrial prediction model cannot be built when the publicly available pretrained model is attacked. Moreover, the pretrained model could not be modified because the pretrained data could not be obtained. Therefore, this paper used the time-lag prior knowledge of industrial data to modify the pretrained model in the fine-tuning stage, and helps improve the pretrained accuracy and model robustness by mining the knowledge of existing data. Two groups of experiments show that the auxiliary knowledge-based fine-tuning mechanism can significantly improve the prediction accuracy and alleviate the defect inheritance of the pretrained model, which is valuable for the application of large models in the field of industrial parameter prediction.
1 January 2026 in IEEE Transactions on Instrumentation and Measurement