基于高斯贝叶斯的建筑水暖设备故障信号诊断分析

Diagnosis and analysis of fault signal of building plumbing equipment based on Gaussian Bayesian

传统的信号稀疏表示多特征提取检测方法、多尺度排列熵故障信号检测方法只能检测低频分量,导致电压、电流检测结果不精准。根据上述问题,提出了基于高斯贝叶斯的建筑水暖设备故障信号诊断方法。该方法构建故障暂态信号模型,依据暖通设备模型分析暂态故障信号及对参数的影响。利用高斯贝叶斯进行信号诊断,以李氏指数为判断依据分类各种信号,在获得高诊断识别率(失效/故障模式正确识别率分别达到96%和86%)的同时,也识别了影响失效/故障分类的关键特征参量。仿真实验结果显示,故障信号监测方法电压暂态波形变化与实际变化情况一致,电流暂态波形变化与实际数值相差仅为0.5 A,检测精度优于传统方法,将其应用到大型建筑水暖中效果更优。

Traditional signal sparse representation multi-feature extraction detection methods and multi-scale arrangement entropy fault signal detection methods can only detect low-frequency components,resulting in inaccurate voltage and current detection results.To solve these problems,a Gaussian Bayesian-based fault signal diagnosis method for building plumbing and heating equipment was proposed.In this method,the fault transient signal model was constructed,and the transient fault signal and its influence on parameters were analyzed according to the HVAC equipment model.Gaussian Bayes was used for signal diagnosis,Li's index was used to classify various signals.The method obtained high diagnostic recognition rate(correct recognition rate of the failure/failure mode reached 96%and 86%,respectively)and also recognized the impact of key characteristic parameters for failure/fault classification.The simulation results showed that the voltage transient waveform change of the fault signal monitoring method was consistent with the actual change,and the difference between the current transient waveform change and the actual value was only 0.5 A.Therefore,the detection accuracy is better than traditional methods,and it will be more effective when applying it to large-scale building plumbing.