ZWZiqi Wang

Ziqi Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ziqi Wang has a total of 25 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Accuracy-Preserving and Efficient AAKR for Industrial Anomaly Detection Using Explicit and Implicit Truncation Strategies

    by Wenwen Tian, Ziqi Wang, Shaokang Liu, Changliang Liu, Shuai Liu, Weiliang Liu

    In industrial anomaly detection, the traditional auto-associative kernel regression (AAKR) method suffers from low computational efficiency and interference from distant samples. To address these issues, this paper presents an AAKR anomaly detection method based on truncation strategies. Specifically, two Gaussian kernel weight sparsification strategies—explicit truncation and implicit truncation—are designed, which prune the kernel function by setting a distance threshold and limiting the number of neighbors, respectively. This effectively reduces the influence of historical samples with weak correlation to the target sample and significantly improves computational efficiency. Theoretically, the concept of a critical bandwidth is introduced. It is proven that with appropriate truncation parameters, when the bandwidth is below the critical value, truncation can greatly reduce computational burden while maintaining estimation accuracy; when the bandwidth is above the critical value, truncation helps enhance model performance. Numerical simulation validates the effectiveness of the proposed method, and further experiments on the Tennessee Eastman Process (TEP) and wind turbine (WT) datasets demonstrate that, compared with existing methods, the proposed method achieves superior performance in both fault detection and computational efficiency, exhibiting greater engineering applicability.


  • Incremental Learning Method for Wind Turbine Fault Detection Models Considering False Negatives

    by Ziqi Wang, Xiaohang Jin, Zhengguo Xu

    Fault detection (FD) algorithms based on supervisory control and data acquisition (SCADA) data have been widely used in the operation and maintenance of wind turbines (WT). However, the performance of FD models will degrade due to the time-varying operating conditions (TVOC). Incremental learning (IL) methods can be employed to update the models online to adapt to TVOC, but the error accumulation caused by false negatives will lead to a continuous decrease in the fault detection rate (FDR). A novel IL method considering false negatives is proposed to improve the performance of WT FD models. Firstly, a data buffer is built to cache some new normal data used in model updating. Secondly, a processing strategy for false negatives is proposed to block some high-risk data from being added to the buffer, thereby weakening the error accumulation. Thirdly, when the buffer is full, update the FD model using the data in it. A real-world SCADA dataset with gearbox faults and four different FD algorithms with various model updating strategies are used in the experiments. The results demonstrate that the proposed method can lower the false alarm rate (FAR) of all four FD algorithms. After processing false negatives, for example, for the FD model based on Gaussian kernel regression, its FDR increased by about 11%, and for the FD model based on multivariate state estimation technique, its FDR increased by about 14%. The results of hyperparameter experiments and early stopping experiments show that the proposed method has good potential for practical applications.


  • A Wind Turbine Condition Monitoring Method Based on FD-KNN and DLSS Active Learning Strategy

    by Shuyao Zhang, Changliang Liu, Ziqi Wang, Shuai Liu

    Condition monitoring(CM) based on Supervisory Control and Data Acquisition (SCADA) data has gained extensive attention for its ability to ensure the normal operation of a vast number of wind turbines (WTs). SCADA data exhibit significant nonlinear and non-Gaussian characteristics, making Fault Detection K-Nearest Neighbor(FD-KNN) an industrially valuable approach due to its adaptation to these features. To address the sensitivity of FD-KNN to redundant and noisy samples in SCADA data, a dual-layer sample selection (DLSS) method based on active learning (AL) strategies is proposed to create a high-quality training set. In the first layer, a comprehensive score of samples is calculated based on three AL strategies, and in the second layer, redundancy removal is performed on the candidates to enhance model performance. Subsequently, control limit (CL) of normal condition are set based on FD-KNN, and incoming samples are detected. For detected fault samples, a fault isolation method that does not require faulty samples is proposed, based on contribution decomposition and variable reconstruction, to identify anomalous variables. Finally, the reconstructed values of the anomalous variables are calculated using KNN regression, and the residuals between observations and reconstructed values can serve as a reference for the degree of anomaly. The experiment utilizes anomaly data from an actual WT in China. Results indicate that the sample selection method proposed in this paper not only increases computational efficiency by approximately 50% but also slightly improves model accuracy and is applicable to various models. Additionally, the fault isolation method introduced can precisely identify fault variables.


  • Identification and Instantaneous Frequency Estimation of Effective Vibration Signal for Pitch Gearbox

    by Changliang Liu, Shaokang Liu, Yingjie Wu, Shuai Liu, Weiliang Liu, Ziqi Wang, Zhihong Luo

    This article focuses on the little-studied pitch gearbox. Due to the intermittent operation of the pitch gearbox, there are currently no suitable methods for automatically identifying the effective vibration signal. The variable speed operation of the pitch gearbox causes the amplitude of the effective vibration signal to attenuate, rendering current instantaneous frequency estimation methods ineffective. To address these issues, the amplitude and frequency characteristics of each component in the vibration signal of the pitch gearbox are analyzed. Then, a method for identifying the effective vibration signal is proposed, which combines high-pass filtering and moving median filtering. High-pass filtering removes low-frequency components. Moving median filtering removes broadband high-amplitude components. Finally, a method for estimating the instantaneous frequency of effective vibration signals based on the short-time fractional Fourier transform is proposed. This method is accurate and not sensitive to amplitude. The experimental results show that the proposed method is superior to comparison methods in the accuracy of identification and estimation. Therefore, the proposed method lays a foundation for the monitoring and diagnosis of the pitch gearbox.


  • Comprehensive Quantitative Tuning Method for Multiple LADRCs in High Order Plus Time Delay Processes

    by Chang-Liang Liu, Chao-Jie Yang, Wei-Liang Liu, Lu-Yang Zhao, Shuai Liu, Zi-Qi Wang

    In response to the challenge posed by the limitations of current active disturbance rejection controllers (ADRCs) in parameter tuning and restricted control efficacy for $K{{e}^{-\tau s}}/{{\left( Ts+1 \right)}^{n}}$ -type high-order processes, two improved linear active disturbance rejection controllers (LADRCs) are proposed in this paper: model-aided delay LADRC (MDLADRC) and compound LADRC (CLADRC). Utilising frequency domain analysis, comprehensive quantitative parameter tuning rules for LADRCs are derived, based on principles of robust stability. The aim of the rules is to address the issue of parameter tuning for both existing LADRCs and the two newly proposed variants. The simulation and comparison results demonstrate the superiority of the improved LADRCs and the tuning rules, and reveal the relationship between the order of LADRCs and their time domain indices. Experimental results on the Peltier temperature system, water tank, and superheated steam temperature (SST) system further verify the engineering applicability of the proposed method. Furthermore, these results illustrate the promising potential of the proposed method for application in process control practice. Additionally, the quantitative parameter tuning rules are also applicable to high-order inertia and first-order plus time delay (FOPTD) processes. Note to Practitioners —In this study, an investigation is conducted into linear active disturbance rejection control (LADRC) with improved controllers for high-order plus time-delay (HOPTD) systems and the parameter tuning method. Consequently, the innovations are characterized by their improved structures and quantitative parameter tuning rules. From the perspective of the controllers’ structures, the large inertia and time delay of HOPTD process pose significant challenges for the design of many LADRCs and other types of controllers. In order to overcome the aforementioned problems, based on the existing 5 types of LADRCs for high-order processes, model-aided delay LADRC (MDLADRC) and compound LADRC (CLADRC) are designed. Compared with the existing improved LADRCs, the disturbance observation accuracy has been found to be higher, with a faster system response speed. From the perspective of the quantitative parameter tuning rules, the relationship between the initial value of the open-loop system in the frequency domain and the controllers’ parameters is determined by means of open-loop frequency domain analysis. At the same time, general parameter tuning rules based on the robust stability of the system are proposed to comprehensively solve the parameter tuning problems of the described 7 types of LADRCs. Meanwhile, the maximum sensitivity is adopted to constrain the parameter range of the controllers. Finally, the application of various experimental scenarios are applied to verify the superiority of the two proposed improved LADRCs and the universality of the parameter tuning rules. Notably, the proposed method is compatible with other types of controllers and processes, as well as backwards compatible with high-order inertial and first-order plus time delay (FOPTD) processes. In the future, further investigations should be conducted into the proposed parameter optimisation rules and the low lag filter.


  • Open Access

    A novel three-stage multi-population evolutionary algorithm for constrained multi-objective optimization problems

    by Chenli Shi, Ziqi Wang, Xiaohang Jin, Zhengguo Xu, Zhangsheng Wang, Peng Shen

    Lots of real-world optimization problems are inherently constrained multi-objective optimization problems (CMOPs), but the existing constrained multi-objective optimization evolutionary algorithms (CMOEAs) often fail to balance convergence and diversity effectively. Therefore, a novel constrained multi-objective optimization evolutionary algorithm based on three-stage multi-population coevolution (CMOEA-TMC) for complex CMOPs is proposed. CMOEA-TMC contains two populations, called mainPop and helpPop, which evolve with and without consideration of constraints, respectively. The proposed algorithm divides the search process into three stages. In the first stage, fast convergence is achieved by transforming the original multi-objective problems into multiple single-objective problems. Coarse-grained parallel evolution of subpopulations in mainPop and guidance information provided by helpPop can facilitate mainPop to quickly approach the Pareto front. In the second stage, the objective function of mainPop changes to the original problem. Coevolution of mainPop and helpPop by sharing offsprings can produce solutions with better diversity. In the third stage, the mining of the global optimal solutions is performed, discarding helpPop to save computational resources. For CMOEA-TMC, the combination of parallel evolution, coevolution, and staging strategy makes it easier for mainPop to converge and maintain good diversity. Experimental results on 33 benchmark CMOPs and a real-world boiler combustion optimization case show that CMOEA-TMC is more competitive than the other five advanced CMOEAs.


  • An Adaptive Condition Monitoring Method of Wind Turbines Based on Multivariate State Estimation Technique and Continual Learning

    by Ziqi Wang, Xiaohang Jin, Zhengguo Xu

    With the development of wind energy, the condition monitoring (CM) methods of wind turbines (WTs) based on supervisory control and data acquisition (SCADA) data have attracted much attention to detect potential faults. With the impact of complicated internal and external factors, the operation conditions of WTs are time-varying. Thus it is necessary to adaptively update CM models in long-term operation. An adaptive WT CM method based on multivariate state estimation technique (MSET) and continual learning (CL) is proposed, which is concise and suitable for practical application. MSET is used to build the non-parametric and high-accuracy normal behavior model. In the proposed CL strategy, new normal data will be temporarily stored in the data buffer to realize the adaptive update of the MSET model. And rules for missing and abnormal data are designed to stabilize update frequency and improve fault detection ability respectively. The proposed method is validated using a real-world SCADA data set with gearbox faults. The results show that the proposed method has higher estimation accuracy and lower false alarm rate than other methods, and the proposed CL strategy has popularization potential. Related hyperparameters are discussed, and when using less training data, the proposed method still has better performance.