可见一近红外光谱法异位发酵床垫料水分快速检测

打开文本图片集

中图分类号：X713；X76 文献标识码：A 文章编号：2095-5553（2025）07-0281-

DOI：10.13733/j.jcam.issn.2095-5553.2025.07.040

Abstract：Toaddresstheneedforrapid moisturedetectioninectopicfermentationbedding materials，thisstudyexploredthe feasibilityofdevelopingamoisturepredictionmodelusingvisible-nearinfraredspectroscopy（Vis-NIR）technology.Overaperiod of 4-5 months，samplesofbedding materials werecolected，and spectraldata ranging from40Onm to990nm wereobtained using aspectrometer.Key characteristicwavelengths wereselectedusing the Competitive AdaptiveReweighted Sampling （CARS）algorithm.Subsequently，a Backpropagation（BP） neural network modelwas thenconstructedandoptimizedusing three optimization algorithms：Grey Wolf Algorithm（GWO），Haris Hawk Algorithm（HHO），and the Guanhao Pig Algorithm （CPO）.Among these，theCPOalgorithm demonstrated the bestoptimization performance.To further enhance modelacuracy， theParticle Swarm Optimization（PSO）algorithmwas improved by integrating Chebyshevchaoticmapping，resulting inthe CARS—ICPO model． The final model achieved R2 values of O.993 5 and O.995 6 on the validation and prediction sets， respectively，withcorresponding RMSEvaluesofO.O1landO.O09，indicating excellentpredictiveaccuracyand generalizationcapability.ThesefindingsconfirmedthefeasibilityofVis-NIRspectroscopycombinedwithadvancedmachine learningtechniquesformoisturecontrolpredictioninectopicfermentationbeddingmaterials，oferinganovelaproachand technical support for rapid detection and intelligent management of fermentation systems.

Keywords：ectopic fermentation bed；padding material；visible near infraredspectroscopy；moisture content detection;neural network;algorithm optimization

0 引言

随着我国畜禽养殖规模不断扩大，畜禽粪污类废弃物大量产生[]。（剩余11659字）