Preparation Of Flexible Graphene Sensors And Its Application In Sound Detection

With the rapid development of automatic speech recognition(ASR)systems and devices,ASR technologies are proliferating,but most of these new interactive systems deal with normal human speech and the application environment is mostly quiet near-field.The accessibility of these systems for patients with speech impairments is low and the performance of these systems and devices is substantially reduced in noisy environments.As a result,research into silent speech recognition technology is beginning to emerge.The pronunciation of people with dysarthria is usually characterized by poor phonetic and breath sounds,so the clarity of speech is reduced in proportion to the severity of the dysarthria and people with dysarthria have difficulty communicating with others.In addition,accurate recognition of speech based on microphone device radios in noisy environments is one of the most pressing challenges in far-field speech recognition.To address the above problems,this topic proposes a flexible graphene sensor to detect vocal fold vibration signals based on the excellent electromechanical properties of graphene structures,which can help patients with a speech impairment to interact with each other while effectively preventing the interference of environmental noise.In this paper,the preparation of graphene and the structure of a flexible substrate for graphene sensors are first investigated.Graphene films are prepared by chemical vapor deposition,where the number of layers of graphene is controlled by changing the time and temperature of the different stages of growth.A stencil embossing technique is proposed for the preparation of flexible substrates with cylindrical microstructures.At the same time,the proposed columnar microsurface structure acts as both a support for the graphene film and reduces the probability of random fracture of the graphene layer.The structural mechanics of the graphene sensor with the improved substrate is investigated through finite element simulation.And the effects of the sensor’s substrate structure on the sensing performance are discussed and tested.Second,a variety of properties of the sensor applied to the new flexible substrate are analyzed,including resolving power,stability,response time,and the frequency response characteristics of its acoustic detection.The sensor has a large response range and a short response time with a load.In the range of 187～2615Hz,the average voltage gain of the sensor is about 48 d B,and this frequency range basically covers the human speech frequency,which provides the basis for the implementation of speech detection.Thanks to the microcrack connection sensing mechanism of graphene sheets,flexible graphene speech detection sensor(FGSDS)can be used for the detection of laryngeal vocal cord vibration.All the obtained data shows that the graphene sensor has sufficient sensitivity to extract the characteristics of acoustic waves.Finally,by combining artificial intelligence with digital signal processing and applying the knowledge of transfer learning,an extreme gradient boosting model with principal component analysis fusion is proposed for efficient recognition of vocal fold vibration signals,which are based on the source model XGBoost for bearing fault detection and combined with feature engineering techniques.The model reduces the data dimension,saves operation time,and can automatically extract the characteristic parameters with a large contribution to the vocal cord vibration signal at the same time.It is expected that in the future,this flexible graphene sensor will be widely used in applications such as interaction for people with a speech impairment,and wearable rehabilitation medical devices.