A Data-Driven Method for Fault Detection and Isolation of the Integrated Energy-Based District Heating System

Abstract

Effective fault detection and isolation can improve the safety, reliability and efficiency of the district heating system. In order to detect and locate the sensor, actuator and component faults in the district heating system with faster response speed and higher accuracy, a two-level fault detection and isolation scheme, consisting of upper-level classifier for system faults and lower-level classifier for subfaults, is developed based on convolutional neural networks. In consideration of the difficulty of obtaining the operation data of a real district heating system under various faulty states, a test benchmark model of an integrated energy based district heating system is built from the system level, which contains the renewable energy based water boilers, transmission networks, and heat load substations to examine the effectiveness of the proposed scheme. To improve the model reality, the dynamics of sensors and actuators, models of heat exchangers, and thermal inertia are considered in the district heating system, and Gaussian noises are added in the raw data signals. Nine kinds of system faults including sensor, actuator, component faults, along with three sub-fault types of bias, drift, complete failure, are investigated in the benchmark system. The performance of the proposed two-level fault detection and isolation scheme is evaluated under different data windows and Gaussian noises in the district heating system, and is compared with other data-driven methods including k-nearest neighbor, random forest and back propagation neural networks. Experimental results show that the two-level fault detection and isolation scheme can detect the faults in the district heating system accurately and robustly, and the proposed scheme has the potential to become an effective solution to real-time monitoring of faults in the district heating system. INDEX TERMS District heating system, fault detection and isolation, thermal inertia, data-driven, convolutional neural networks.

Deep learning in fault detection and diagnosis of building HVAC systems: A systematic review with meta analysis

“…Methods used to formulate the input data for a 2D CNN include converting numeric features of the original data into the corresponding images with the y-axis representing the time and the x-axis representing the features [42] [13] [85] [16] or with the y-axis representing the features and the x-axis representing the time [50], placing each measured feature value row-wise and column-wise in a matrix at each timestep [45] [46]. However, the data conversion process causes extra computational cost, and information embedded in the original 1D data can be distorted during the converting procedure.…”

Intelligent Approaches to Fault Detection and Diagnosis in District Heating: Current Trends, Challenges, and Opportunities

“…Li et al [95] propose an FDD approach using a DH network simulation with kNN, RF, ANN, and CNN. The authors can classify nine types of faults, such as sensor, actuator, component faults, bias, drift, and complete failure faults.…”

“…Overall, it was not the choice of ML algorithm, but primarily data quality, that impacted the performance the most, which is specifically challenging in fault diagnosis, as there is a severe lack of labeled data. GBR, TPOT Sensor failure Guelpa et al [87] Analytical Fouling Cadei et al [88] ARIMA, RIDGE, one-class SVM Fouling Kim et al [89] kM, MLP Fouling Park et al [90] RF Valves Langroudi et al [91] LR, DTR, RIDGE, kNN, PLS, SVM, RF, LASSO, XGBoost, ANN Pipes Bahlawan et al [92] Analytical Pipes Manservigi et al [93] Analytical Pipes Bode et al [94] LR, kNN, CART, RF, NB, SVM, ANN Multi-label Choi et al [96] AE, MLP Multi-label Li et al [95] kNN, RF, ANN, CNN Multi-label…”

“…What stands out in Figure 1 is that for fault diagnosis, the most popular methods are: kNN, RF, and ANN. Multiple studies compare several algorithms, such as in [85,91,94,95]. In [85], the authors suggest that using ERT results in the best performance.…”

Opportunities for Machine Learning in District Heating

“…They applied their method on consumption data from the primary side of a DH system located in China. On the other hand, the work in [29] proposes a two-level fault detection and isolation scheme with Convolutional Neural Networks (CNNs). They use simulated data labelled with different fault scenarios to evaluate the performance of their model.…”