Method for detecting three-dimensional force of a fiber-integrated monolithic clamp guided by metal additive manufacturing

A method for detecting a three-dimensional force of a fiber-integrated monolithic clamp guided by metal additive manufacturing, comprising the following steps:

• S1:
• pre-preparing a three-dimensional force sensor, wherein the three-dimensional force sensor comprises a clamp head (1), an elastomer (2), and a transmission component, one end of the elastomer (2) in an axial direction is connected with the clamp head (1), and the clamp head (1) is configured to clamp tissues;
• another end of the elastomer (2) in the axial direction is provided with the transmission component, and the transmission component passes through the elastomer (2) to be in a transmission connection with the clamp head (1);
• and three groups of etched stepped reduced-diameter fiber gratings (7) are provided on the elastomer (2) in a centrosymmetric distribution;
• S2:
• constructing, when the clamp head (1) of the three-dimensional force sensor clamps the tissues, a mechanical model of the elastomer (2), obtaining a strain of the elastomer (2) under action of the three-dimensional force, establishing, in combination with temperature sensitivity coefficients of the three groups of etched stepped reduced-diameter fiber gratings (7), a relationship between central wavelength drift amounts of the etched stepped reduced-diameter fiber gratings (7) and a temperature as well as the three-dimensional force, and deriving a force and temperature sensitivity matrix;
• S3:
• decoupling, by solving a generalized inverse matrix of the force and temperature sensitivity matrix as a calibration matrix, central wavelength values of the three groups of etched stepped reduced-diameter fiber gratings (7) to measure the three-dimensional force and the temperature;
• S4:
• filtering three-dimensional force time series data, constructing a training sample, and using a long short-term memory (LSTM) neural network to train network parameters;
• and predicting three-dimensional force data at a current moment and recovering subsequent interaction force measurement data after training is completed, according to a sample corresponding to current input three-dimensional force data and in combination with historical fault-free three-dimensional force data and fault-free three-dimensional force data prior to the current moment, and realizing a fault-tolerant output of three-dimensional force decoupling under a fault of the etched stepped reduced-diameter fiber gratings (7);
• and
• S5:
• outputting types of the tissues clamped by the clamp in a classified manner through a random forest algorithm in combination with central wavelength drift amount information of the three groups of etched stepped reduced-diameter fiber gratings (7).