Human Motion Recognition Based On Knitted Strain Sensor

With the widespread application of interdisciplinary and artificial intelligence,intelligent wearable products continue to emerge.Clothing products that involve both comfort and intelligence are popular among consumers.The combination of intelligent sensing equipment and textiles and clothing can maximize the use of intelligent transmission.The sensing device's electrical conductivity and the comfort and lightness of clothing make smart wearable devices a daily wearable product.Wearable intelligent products have a wide range of applications and are suitable for human health,medical treatment,and sports competitions.For example,intelligent human monitoring clothing can prevent hemiplegia,Parkinson's early diagnosis,as well as genetic joint gait instability.The transmission of early warning or diagnostic signals to medical staff can effectively reduce the risk of illness;during daily training,sports athletes can judge the standard of posture or the degree of movement specification through limb movement monitoring or movement bending amplitude,thereby significantly improving sports performance or efficiency.Elastic knitted fabric is a material with enormous strain and high tensile recovery.As a clothing fabric,it deforms significantly with the movement of the human body.The addition of conductive fibers can convert the fabric's deformation into electrical signal changes to prepare an elastic knitting strain sensor.Clothing can realize the daily motion detection of the human body.Therefore,it is essential to seek a knitted smart garment with good daily wear experience and high recognition accuracy.High-performance knitted strain sensors based on knitted materials are expected to provide excellent wearing comfort,three-dimensional surface fit,high sensitivity and stability in use,and have important theoretical significance and use value for real-time motion detection.In this thesis,stainless steel/polyester and silver-plated nylon knitted strain sensors were taken as the research objects,revealing the inherent relationships among knitted fabric deformation,strain sensing,and human movement.The two-dimensional and three-dimensional curved surface strain sensing performance was analyzed and two-dimensional and three-dimensional dynamic equivalent resistance models were established to clarify elastic conductive knitted fabrics' sensing mechanism.The design and preparation of multi-area strain sensing sports pants with high sensitivity and stability were realized.Human gait signals were collected,motion characteristics were extracted,recognition methods and accuracy were studied,and a wearable recognition system was finally realized.The main content of this thesis includes the following parts.(1)Two knitted strain sensors of stainless steel/polyester type and silver-plated nylon type were prepared.The influence of the structure,density,stretching rate,and other parameters of the two strain sensors on the sensing performance was compared,and the sensing mechanism of the knitted strain sensor was explored.Sensitivity,hysteresis,repeatability,stability,and other sensing performance indicators were also explored to look for a knitted strain sensor structure suitable for human motion detection.The test results show that compared with the stainless steel/polyester knitted strain sensor,the silver-plated nylon knitted strain sensor has higher strain sensitivity when longitudinally stretched.The sensor's strain sensing range can reach 140%,and the strain rate is 0-70%.The weft-plain stitch strain sensor has better sensing performance than rib 1+1 and rib 2+2 weaves.At different speeds,the silver-plated nylon strain sensor can still maintain stable sensing performance.Simultaneously,after 3,000 cycles of repeated extension tests,the strain sensor has good reproducibility,small resistance offset change,and specific long-term stability sensing performance.(2)The three-dimensional curved surface sensing performance of knitted strain sensors was studied,the three-dimensional curved surface test methods and the force of knitted strain sensors were analyzed,the three-dimensional curved surface strain rate,and the effect of the three-dimensional curved surface strain-resistance change,burst rate and repeatability on the three-dimensional curved surface strain were calculated.The effect of the two-dimensional strain stretch and three-dimensional curved surface strain sensing performance was compared.The test results show that the three-dimensional curved surface sensor's strain sensing range can reach 140%.When the sensing range is 0-120% strain rate,the sensor resistance change rate and strain are positively correlated.By comparing the two-dimensional tensile test and the three-dimensional burst test,the three-dimensional curved surface strain's effective sensing range is twice that of the two-dimensional tensile test.The sensitivity of the two-dimensional longitudinal tensile sensing is relatively higher,and the response time of the three-dimensional sensing is shorter than that of the two-dimensional strain.Both sensing strain methods have good cycle repeatability and continuous working stability.The three-dimensional curved surface sensing has fewer restrictions on the sensing direction and surface shape of the object to be measured and has a broader application range,which can meet more sensing scenes and objects..(3)The macro-micro and topological equivalent resistance models of the knitted strain sensor based on the two-dimensional strain were established,and the advantages and disadvantages of the two equivalent resistance models were compared.They established the equivalent resistance model based on the three-dimensional curved surface strain.Three equivalent resistance models were compared with the test results.The test results show that the knitted strain sensor's resistance change is directly related to the length resistance of the sensor fabric and the transfer of the loop.The prediction and fitting accuracy of the equivalent resistance model of the topological structure is high,but the calculation is complicated,and it is suitable for the resistance prediction of the knitted strain sensor with a small number of courses and wales.The macro-micro equivalent resistance model is quick and straightforward to calculate.It is suitable for the resistance prediction of knitted strain sensors with a large area.The three-dimensional curved surface equivalent resistance model directly relates to the burst depth and can be used to predict the resistance change of the knitted strain sensor when the three-dimensional curved surface is strained.(4)The strain of human knee joint skin was measured through the 3D human body scanning system,and highly elastic sports pants integrated with multi-zone knitted strain sensors were prepared to test the knee joint flexion angle and gait movement of human limb movement.The test results show that the area with the largest knee skin strain rate is located in the front middle bisector and knee girth area.The front middle bisector area sensor can better reflect the movement posture of running,walking,climbing,and descending steps,suitable for high elastic sports pants.In the main sensing area,highly elastic sports pants with integrated knitted strain sensors have good motion sensing performance and sensitivity.A wearable motion recognition system was established to recognize human body's gait movement by using resistance value characteristics,walking cycle division,multi-area sensing,and machine learning.The test results show that the identification method of resistance value characteristic is relatively simple.There is a one-to-one correspondence between the walking cycle and the resistance curve.The use of multi-region sensing and support vector machine algorithms can better improve the accuracy of motion recognition.The four sports' complete recognition accuracy is 84.38%,and the recognition rate of running is the highest.This research system comprehensively analyzed the two-dimensional and three-dimensional strain sensing performance of knitted strain sensors.The dynamic equivalent resistance model was used to establish the relationships among knitted fabric deformation,strain sensing,and human body motion and prepare a high-speed multi-region strain sensor.Elastic sports pants realized the corresponding motion state's output while sensing the strain sensor,and a wearable motion recognition system was built,providing a reference for smart wearable devices and human motion recognition.