Machine vibration wireless sensor networks node often works at the wireless radio frequency phase where the transmitted energy is the largest to ensure the reliability of transmission, but resulting in a large transmission power consumption of sensor nodes, especially in mass raw data transmission requirements, In the light of this problem, a least squares adaptive transmission power control(LSATPC) method for wireless sensor networks nodes is proposed. First, the discrete relationship between transmission power and link quality indicator (LQI) is studied through inner-cluster sensor nodes communication in the mechanical vibration monitoring in laboratory environment. Then, the least squares method is utilized to fit the discrete data, thus to construct the least squares adaptive transmission power mathematical model for reliable and energy-efficient data transmission. Finally, combined with the threshold value of LQI, the optimal wireless transmission power of the nodes at different communication conditions is calculated, satisfied with the result of the reliable data transmission. Experimental results indicate that the transmission energy consumption of mechanical vibration wireless sensor network nodes can be effectively reduced by this method.