Abstract: Method for predicting a stability of a steel cylinder including: in response to the steel cylinder tilting toward a seaward side at an arbitrary rotation point, obtaining a safety factor for the steel cylinder tilting toward the seaward side by determining an anti-tilt moment and a tilt moment when the steel cylinder is tilted to the seaward side; under a same rotation point, in response to determining that the steel cylinder is rotated toward a land side, determining the safety factor for the steel cylinder tilting toward the land side; taking a smaller safety factor as the safety factor under the rotation point; re-selecting a new rotation point, determining a safety factor corresponding to the new rotation point, and taking a safety factor with a smallest value among all rotation points as a final safety factor; and generating, based on the final safety factor, an anti-tilt instruction.

Abstract: A full-ocean-depth fidelity enzymological measurement device for microbial extracellular enzyme is provided, comprising a pressure-maintaining sampling bottle, pressure-maintaining transfer equipment, a pressure-maintaining enzymological reactor, and heat preservation equipment and enzyme activity detection equipment. The pressure-maintaining enzymological reactor comprises a barrel body, a plug, polytetrafluoroethylene gaskets, an O-ring, a piston, a high-pressure straight-through valve, and a high-pressure connector. The pressure-maintaining enzymological reactor is in a closed barrel body shape and is internally provided with the piston, and the plug and the valve are arranged at each of two ends of the pressure-maintaining enzymological reactor; and the valve is connected to the pressure-maintaining transfer equipment through the high-pressure connector.

Abstract: The invention provides a non-negative matrix factorization face recognition method and system based on kernel machine learning, which comprises five steps. The invention has the following beneficial effects: the invention avoids the learning of the inaccurate pre-image matrix by directly learning two kernel matrices, Kwx and Kww, and avoids the derivation of the kernel function in the iterative formula by changing the learning object, so that there is no limit to the selection of kernel function and a general algorithm for any kernel function is obtained.

Abstract: The invention provides a non-negative matrix factorization face recognition method and system based on kernel machine learning, which comprises five steps. The invention has the following beneficial effects: the invention avoids the learning of the inaccurate pre-image matrix by directly learning two kernel matrices, Kwx and Kww, and avoids the derivation of the kernel function in the iterative formula by changing the learning object, so that there is no limit to the selection of kernel function and a general algorithm for any kernel function is obtained.