摘要本文深入讲解DolphinDB振动分析技术。从振动信号原理到数据采集从时域分析到频域分析从故障诊断到实战应用全面介绍设备振动信号处理的核心方法。通过丰富的代码示例帮助读者掌握振动分析的核心技能。一、振动分析概述1.1 振动信号原理振动分析流程振动采集信号预处理时域分析频域分析故障诊断1.2 振动参数参数说明单位位移振动幅度μm速度振动速度mm/s加速度振动加速度m/s²频率振动频率Hz1.3 振动标准标准适用范围ISO 10816机械振动评价ISO 7919转轴振动测量API 670机械保护系统二、数据采集2.1 采集配置//振动数据采集配置 vibrationConfigdict(STRING,ANY,[[sample_rate,10000],//采样率 10kHz[sample_time,1],//采样时间1秒[channels,[CH1,CH2,CH3,CH4]],//采集通道[sensitivity,100],//传感器灵敏度 mV/g[range,50]//量程 ±50g])2.2 数据存储//创建振动数据流 share streamTable(1000000:0,device_idtimestampchannelacceleration,[SYMBOL,TIMESTAMP,STRING,DOUBLE])asvibration_stream enableTablePersistence(vibration_stream,true,true,10000000)//创建振动特征表 share table(1:0,device_idtimestampchannelrmspeakcrest_factorkurtosis,[SYMBOL,TIMESTAMP,STRING,DOUBLE,DOUBLE,DOUBLE,DOUBLE])asvibration_features2.3 数据预处理//数据预处理defpreprocessVibration(data,config){//1.去除直流分量 dcavg(data)datadata-dc//2.去趋势 trendlinreg(1..data.rows(),data)[0]datadata-trend*(1..data.rows())//3.滤波简化版//实际应用中使用带通滤波器 filteredmavg(data,5)returnfiltered}三、时域分析3.1 基本统计量//时域统计特征deftimeDomainFeatures(data){//均值 meanavg(data)//均方根值 rmssqrt(avg(data*data))//峰值 peakmax(abs(data))//峰峰值 peakToPeakmax(data)-min(data)//标准差 stdstd(data)//峰值因子 crestFactorpeak/rms//峭度 kurtkurtosis(data)//偏度 skewskewness(data)returndict(STRING,ANY,[[mean,mean],[rms,rms],[peak,peak],[peak_to_peak,peakToPeak],[std,std],[crest_factor,crestFactor],[kurtosis,kurt],[skewness,skew]])}3.2 波形指标//波形指标计算defwaveformIndicators(data){//均方根值 rmssqrt(avg(data*data))//绝对平均值 absMeanavg(abs(data))//方根幅值 sqrtMeansqrt(avg(sqrt(abs(data))))^2//波形指标 shapeFactorrms/absMean//脉冲指标 impulseFactormax(abs(data))/absMean//裕度指标 clearanceFactormax(abs(data))/sqrtMeanreturndict(STRING,ANY,[[shape_factor,shapeFactor],[impulse_factor,impulseFactor],[clearance_factor,clearanceFactor]])}3.3 趋势分析//振动趋势分析defvibrationTrend(data,window100){returnselect timestamp,vibration,mavg(vibration,window)astrend,mstd(vibration,window)asvolatility,vibration-mavg(vibration,window)asdeviationfromdata context by device_id}四、频域分析4.1 FFT变换//FFT变换简化版deffftAnalysis(data,sampleRate){ndata.rows()//实际应用中使用DolphinDB的fft函数//这里展示计算过程//频率分辨率 freqResolutionsampleRate*1.0/n//频率轴 frequencies0..(n/2)*freqResolution//幅值谱简化//实际需要FFT计算 magnitudesabs(data)returndict(STRING,ANY,[[frequencies,frequencies],[magnitudes,magnitudes],[freq_resolution,freqResolution]])}4.2 频谱特征//频谱特征提取defspectrumFeatures(spectrum,freqBands){featuresdict(STRING,DOUBLE)for(bandinfreqBands.keys()){rangefreqBands[band]//计算频带能量 energysum(spectrum[(spectrum.frequenciesrange[0])(spectrum.frequenciesrange[1])].magnitudes^2)features[band]energy}//总能量 features[total_energy]sum(spectrum.magnitudes^2)//重心频率 features[centroid_freq]sum(spectrum.frequencies*spectrum.magnitudes)/sum(spectrum.magnitudes)returnfeatures}4.3 特征频率//转子故障特征频率defbearingFaultFrequencies(rpm,bearingParams){//转频 f_rrpm/60.0//滚动体数量 nbearingParams[ball_count]//接触角 alphabearingParams[contact_angle]*pi/180//节径 dbearingParams[pitch_diameter]//滚动体直径 ball_dbearingParams[ball_diameter]//内圈故障频率 f_innerf_r*n/2*(1ball_d/d*cos(alpha))//外圈故障频率 f_outerf_r*n/2*(1-ball_d/d*cos(alpha))//滚动体故障频率 f_ballf_r*d/(2*ball_d)*(1-(ball_d/d*cos(alpha))^2)//保持架故障频率 f_cagef_r/2*(1-ball_d/d*cos(alpha))returndict(STRING,DOUBLE,[[rotation_freq,f_r],[inner_race_freq,f_inner],[outer_race_freq,f_outer],[ball_freq,f_ball],[cage_freq,f_cage]])}五、故障诊断5.1 故障特征库//故障特征库 faultLibrarydict(STRING,ANY,[[不平衡,dict(STRING,ANY,[[特征频率,1X转频],[时域特征,正弦波形],[频域特征,转频处有明显峰值]])],[不对中,dict(STRING,ANY,[[特征频率,2X转频],[时域特征,M形波形],[频域特征,2倍转频处有明显峰值]])],[轴承故障,dict(STRING,ANY,[[特征频率,内圈/外圈/滚动体频率],[时域特征,冲击脉冲],[频域特征,高频段能量增加]])],[松动,dict(STRING,ANY,[[特征频率,多倍频],[时域特征,不稳定波形],[频域特征,丰富的高次谐波]])]])5.2 故障识别//故障识别defidentifyFault(features,faultFreqs,rpm){faultsarray(STRING,0)//转频 f_rrpm/60.0//检查不平衡if(features[1x_amplitude]features[rms]*0.5){faults.append!(不平衡)}//检查不对中if(features[2x_amplitude]features[rms]*0.3){faults.append!(不对中)}//检查轴承故障if(features[high_freq_energy]features[total_energy]*0.2){faults.append!(轴承故障)}//检查松动if(features[harmonic_count]5){faults.append!(松动)}returnfaults}5.3 严重程度评估//故障严重程度评估defassessSeverity(features,thresholds){//ISO10816振动速度标准 rmsfeatures[rms]if(rmsthresholds[good]){severity良好level1}elseif(rmsthresholds[satisfactory]){severity满意level2}elseif(rmsthresholds[unsatisfactory]){severity不满意level3}else{severity不可接受level4}returndict(STRING,ANY,[[severity,severity],[level,level],[rms,rms]])}六、实战案例6.1 完整振动分析系统//振动分析系统//1.创建数据表 share streamTable(1000000:0,device_idtimestampchannelacceleration,[SYMBOL,TIMESTAMP,STRING,DOUBLE])asvibration_stream enableTablePersistence(vibration_stream,true,true,10000000)//2.创建分析结果表 share table(1:0,device_idtimestampchannelrmspeakcrest_factorkurtosisfault_typeseverity,[SYMBOL,TIMESTAMP,STRING,DOUBLE,DOUBLE,DOUBLE,DOUBLE,STRING,STRING])asvibration_analysis//3.振动分析函数defanalyzeVibration(deviceId,channel,data){//时域特征 timeFeaturestimeDomainFeatures(data)//故障识别 faultsidentifyFault(timeFeatures,null,1500)//严重程度 severityassessSeverity(timeFeatures,dict(STRING,DOUBLE,[[good,1.8],[satisfactory,4.5],[unsatisfactory,11.2]]))returndict(STRING,ANY,[[rms,timeFeatures[rms]],[peak,timeFeatures[peak]],[crest_factor,timeFeatures[crest_factor]],[kurtosis,timeFeatures[kurtosis]],[fault_type,faults.size()0? faults[0]:正常],[severity,severity[severity]]])}//4.分析任务defanalysisTask(){while(true){nownow()//获取最新数据 dataselect*fromvibration_stream where timestampnow-60000if(data.rows()0){//按设备和通道分组分析for(deviceIdinexecdistinct device_idfromdata){for(channelinexecdistinct channelfromdata where device_iddeviceId){//获取通道数据 channelDataexecaccelerationfromdata where device_iddeviceIdandchannelchannel order by timestampif(channelData.size()100){//分析 resultanalyzeVibration(deviceId,channel,channelData)insert into vibration_analysis values(deviceId,now,channel,result[rms],result[peak],result[crest_factor],result[kurtosis],result[fault_type],result[severity])}}}}sleep(60000)}}//5.启动系统 submitJob(vibration_analysis,振动分析,analysisTask)//6.查询接口defgetVibrationAnalysis(deviceId){returnselect*fromvibration_analysis where device_iddeviceId order by timestamp desc limit10}defgetFaultDevices(){returnselect device_id,fault_type,severityfromvibration_analysis where timestampnow()-300000andfault_type!正常}addFunctionView(getVibrationAnalysis)addFunctionView(getFaultDevices)print(振动分析系统启动完成)七、总结本文详细介绍了DolphinDB振动分析振动信号原理振动参数、振动标准数据采集采集配置、数据存储、数据预处理时域分析统计量、波形指标、趋势分析频域分析FFT变换、频谱特征、特征频率故障诊断故障特征库、故障识别、严重程度思考题如何选择合适的振动分析参数如何提高故障诊断的准确性如何实现振动分析的实时处理参考资料DolphinDB信号处理ISO 10816振动标准
摘要本文深入讲解DolphinDB振动分析技术。从振动信号原理到数据采集从时域分析到频域分析从故障诊断到实战应用全面介绍设备振动信号处理的核心方法。通过丰富的代码示例帮助读者掌握振动分析的核心技能。一、振动分析概述1.1 振动信号原理振动分析流程振动采集信号预处理时域分析频域分析故障诊断1.2 振动参数参数说明单位位移振动幅度μm速度振动速度mm/s加速度振动加速度m/s²频率振动频率Hz1.3 振动标准标准适用范围ISO 10816机械振动评价ISO 7919转轴振动测量API 670机械保护系统二、数据采集2.1 采集配置//振动数据采集配置 vibrationConfigdict(STRING,ANY,[[sample_rate,10000],//采样率 10kHz[sample_time,1],//采样时间1秒[channels,[CH1,CH2,CH3,CH4]],//采集通道[sensitivity,100],//传感器灵敏度 mV/g[range,50]//量程 ±50g])2.2 数据存储//创建振动数据流 share streamTable(1000000:0,device_idtimestampchannelacceleration,[SYMBOL,TIMESTAMP,STRING,DOUBLE])asvibration_stream enableTablePersistence(vibration_stream,true,true,10000000)//创建振动特征表 share table(1:0,device_idtimestampchannelrmspeakcrest_factorkurtosis,[SYMBOL,TIMESTAMP,STRING,DOUBLE,DOUBLE,DOUBLE,DOUBLE])asvibration_features2.3 数据预处理//数据预处理defpreprocessVibration(data,config){//1.去除直流分量 dcavg(data)datadata-dc//2.去趋势 trendlinreg(1..data.rows(),data)[0]datadata-trend*(1..data.rows())//3.滤波简化版//实际应用中使用带通滤波器 filteredmavg(data,5)returnfiltered}三、时域分析3.1 基本统计量//时域统计特征deftimeDomainFeatures(data){//均值 meanavg(data)//均方根值 rmssqrt(avg(data*data))//峰值 peakmax(abs(data))//峰峰值 peakToPeakmax(data)-min(data)//标准差 stdstd(data)//峰值因子 crestFactorpeak/rms//峭度 kurtkurtosis(data)//偏度 skewskewness(data)returndict(STRING,ANY,[[mean,mean],[rms,rms],[peak,peak],[peak_to_peak,peakToPeak],[std,std],[crest_factor,crestFactor],[kurtosis,kurt],[skewness,skew]])}3.2 波形指标//波形指标计算defwaveformIndicators(data){//均方根值 rmssqrt(avg(data*data))//绝对平均值 absMeanavg(abs(data))//方根幅值 sqrtMeansqrt(avg(sqrt(abs(data))))^2//波形指标 shapeFactorrms/absMean//脉冲指标 impulseFactormax(abs(data))/absMean//裕度指标 clearanceFactormax(abs(data))/sqrtMeanreturndict(STRING,ANY,[[shape_factor,shapeFactor],[impulse_factor,impulseFactor],[clearance_factor,clearanceFactor]])}3.3 趋势分析//振动趋势分析defvibrationTrend(data,window100){returnselect timestamp,vibration,mavg(vibration,window)astrend,mstd(vibration,window)asvolatility,vibration-mavg(vibration,window)asdeviationfromdata context by device_id}四、频域分析4.1 FFT变换//FFT变换简化版deffftAnalysis(data,sampleRate){ndata.rows()//实际应用中使用DolphinDB的fft函数//这里展示计算过程//频率分辨率 freqResolutionsampleRate*1.0/n//频率轴 frequencies0..(n/2)*freqResolution//幅值谱简化//实际需要FFT计算 magnitudesabs(data)returndict(STRING,ANY,[[frequencies,frequencies],[magnitudes,magnitudes],[freq_resolution,freqResolution]])}4.2 频谱特征//频谱特征提取defspectrumFeatures(spectrum,freqBands){featuresdict(STRING,DOUBLE)for(bandinfreqBands.keys()){rangefreqBands[band]//计算频带能量 energysum(spectrum[(spectrum.frequenciesrange[0])(spectrum.frequenciesrange[1])].magnitudes^2)features[band]energy}//总能量 features[total_energy]sum(spectrum.magnitudes^2)//重心频率 features[centroid_freq]sum(spectrum.frequencies*spectrum.magnitudes)/sum(spectrum.magnitudes)returnfeatures}4.3 特征频率//转子故障特征频率defbearingFaultFrequencies(rpm,bearingParams){//转频 f_rrpm/60.0//滚动体数量 nbearingParams[ball_count]//接触角 alphabearingParams[contact_angle]*pi/180//节径 dbearingParams[pitch_diameter]//滚动体直径 ball_dbearingParams[ball_diameter]//内圈故障频率 f_innerf_r*n/2*(1ball_d/d*cos(alpha))//外圈故障频率 f_outerf_r*n/2*(1-ball_d/d*cos(alpha))//滚动体故障频率 f_ballf_r*d/(2*ball_d)*(1-(ball_d/d*cos(alpha))^2)//保持架故障频率 f_cagef_r/2*(1-ball_d/d*cos(alpha))returndict(STRING,DOUBLE,[[rotation_freq,f_r],[inner_race_freq,f_inner],[outer_race_freq,f_outer],[ball_freq,f_ball],[cage_freq,f_cage]])}五、故障诊断5.1 故障特征库//故障特征库 faultLibrarydict(STRING,ANY,[[不平衡,dict(STRING,ANY,[[特征频率,1X转频],[时域特征,正弦波形],[频域特征,转频处有明显峰值]])],[不对中,dict(STRING,ANY,[[特征频率,2X转频],[时域特征,M形波形],[频域特征,2倍转频处有明显峰值]])],[轴承故障,dict(STRING,ANY,[[特征频率,内圈/外圈/滚动体频率],[时域特征,冲击脉冲],[频域特征,高频段能量增加]])],[松动,dict(STRING,ANY,[[特征频率,多倍频],[时域特征,不稳定波形],[频域特征,丰富的高次谐波]])]])5.2 故障识别//故障识别defidentifyFault(features,faultFreqs,rpm){faultsarray(STRING,0)//转频 f_rrpm/60.0//检查不平衡if(features[1x_amplitude]features[rms]*0.5){faults.append!(不平衡)}//检查不对中if(features[2x_amplitude]features[rms]*0.3){faults.append!(不对中)}//检查轴承故障if(features[high_freq_energy]features[total_energy]*0.2){faults.append!(轴承故障)}//检查松动if(features[harmonic_count]5){faults.append!(松动)}returnfaults}5.3 严重程度评估//故障严重程度评估defassessSeverity(features,thresholds){//ISO10816振动速度标准 rmsfeatures[rms]if(rmsthresholds[good]){severity良好level1}elseif(rmsthresholds[satisfactory]){severity满意level2}elseif(rmsthresholds[unsatisfactory]){severity不满意level3}else{severity不可接受level4}returndict(STRING,ANY,[[severity,severity],[level,level],[rms,rms]])}六、实战案例6.1 完整振动分析系统//振动分析系统//1.创建数据表 share streamTable(1000000:0,device_idtimestampchannelacceleration,[SYMBOL,TIMESTAMP,STRING,DOUBLE])asvibration_stream enableTablePersistence(vibration_stream,true,true,10000000)//2.创建分析结果表 share table(1:0,device_idtimestampchannelrmspeakcrest_factorkurtosisfault_typeseverity,[SYMBOL,TIMESTAMP,STRING,DOUBLE,DOUBLE,DOUBLE,DOUBLE,STRING,STRING])asvibration_analysis//3.振动分析函数defanalyzeVibration(deviceId,channel,data){//时域特征 timeFeaturestimeDomainFeatures(data)//故障识别 faultsidentifyFault(timeFeatures,null,1500)//严重程度 severityassessSeverity(timeFeatures,dict(STRING,DOUBLE,[[good,1.8],[satisfactory,4.5],[unsatisfactory,11.2]]))returndict(STRING,ANY,[[rms,timeFeatures[rms]],[peak,timeFeatures[peak]],[crest_factor,timeFeatures[crest_factor]],[kurtosis,timeFeatures[kurtosis]],[fault_type,faults.size()0? faults[0]:正常],[severity,severity[severity]]])}//4.分析任务defanalysisTask(){while(true){nownow()//获取最新数据 dataselect*fromvibration_stream where timestampnow-60000if(data.rows()0){//按设备和通道分组分析for(deviceIdinexecdistinct device_idfromdata){for(channelinexecdistinct channelfromdata where device_iddeviceId){//获取通道数据 channelDataexecaccelerationfromdata where device_iddeviceIdandchannelchannel order by timestampif(channelData.size()100){//分析 resultanalyzeVibration(deviceId,channel,channelData)insert into vibration_analysis values(deviceId,now,channel,result[rms],result[peak],result[crest_factor],result[kurtosis],result[fault_type],result[severity])}}}}sleep(60000)}}//5.启动系统 submitJob(vibration_analysis,振动分析,analysisTask)//6.查询接口defgetVibrationAnalysis(deviceId){returnselect*fromvibration_analysis where device_iddeviceId order by timestamp desc limit10}defgetFaultDevices(){returnselect device_id,fault_type,severityfromvibration_analysis where timestampnow()-300000andfault_type!正常}addFunctionView(getVibrationAnalysis)addFunctionView(getFaultDevices)print(振动分析系统启动完成)七、总结本文详细介绍了DolphinDB振动分析振动信号原理振动参数、振动标准数据采集采集配置、数据存储、数据预处理时域分析统计量、波形指标、趋势分析频域分析FFT变换、频谱特征、特征频率故障诊断故障特征库、故障识别、严重程度思考题如何选择合适的振动分析参数如何提高故障诊断的准确性如何实现振动分析的实时处理参考资料DolphinDB信号处理ISO 10816振动标准
