Application Research Of Adaptive Sampling And Quantization Technology In Transmission Of Fault Diagnosis Signal

As the special road section composed of highways,long tunnels urgently need to be upgraded intelligently,so as to realize the all-round perception of tunnel equipment.Due to the narrow and complex terrain environment and serious multipath interference of the tunnel,it is a good choice to use Long Range Radio(LoRa)technology to transmit device status data.But the significant disadvantage of LoRa is the narrow bandwith in strong noise environment,when a large amount of data is transmitted,narrowband channels cannot meet the transmission requirements.For example,when monitoring the status of ventilation equipment in a long tunnel,we need to upload the sensor data of the axial flow fan to the fault diagnosis terminal at all times,and then perform time-frequency analysis by the fault diagnosis terminal.When the channel environment is not ideal,the transmitting rate of LoRa technology is extremely low,the data cannot be transmitted to the receiving end in time,which resulting in a serious delay in the diagnosis results of the fan status.In order to solve the above problems,this paper mainly researches the contradiction between a large amount of fault diagnosis data and the narrow bandwidth transmission.Improving the A/D conversion process of sensor output data.Focusing on sampling and quantization to reduce the faulty data to be transmitted.The main work is as follows:(1)Downsampling the vibration data.The fault diagnosis of axial fans is based on the frequency components contained in the signal.When compressing the sampling rate of vibration data,it is necessary to consider that downsampling will cause the loss of high frequency components of the signal.Therefore,this paper studies the spectrum characteristics of axial fans under different fault types,and obtains the upper limit of the sampling rate compression parameter D without affecting the judgment of the fault type at the receiving end.In order to verify the correctness of the theoretical analysis,the frequency spectrum of the signals under different working conditions and the normal signal under the maximum sampling rate compression parameter D is compared through experiments.(2)Performing non-uniform quantization processing on the vibration sampling points.After the number of sampling points cannot be further reduced,the transmission rate of LoRa in harsh environments only supports low-resolution quantization(Quantization Bits N is from3 bits to 5 bits).Therefore,this paper proposes a non-uniform quantization technology based on Kernel Density Estimation(KDE),studies how to select the KDE kernel function and bandwidth,and verifies the feasibility of the scheme through experiments.Compared with the existing parameter estimation methods,the KDE-based quantization scheme shows better quantization performance and lower quantization error at low bit resolution,and is more suitable for data transmission under narrow bandwidth.(3)According to the link transmission characteristics of LoRa,the sampling rate compression parameter D and the quantization parameter N are adaptively adjusted.As the transmission signal-to-noise ratio changes,the LoRa transmission rate will be adjusted.Because at some LoRa transmission rates,there are many D and N combinations that satisfy the number of sampling points per second multiplied by the quantization resolution not greater than the current LoRa transmission rate.Therefore,through experimental simulation obtain the optimal parameter combination at each LoRa rate,so as to perform adaptive parameter setting.The adaptive sampling and quantization method proposed in this paper can realize the downsampling of data and low-level non-uniform quantization to ensure the accuracy of fault diagnosis at the receiving end when the signal-to-noise ratio deteriorates,and can transmit relatively complete vibration data when the channel environment is good And improve the quantization resolution,so as to make full use of channel resources.