Portable detection device of soil moisture for precision irrigation decision system

Precision irrigation by applying state-of-the-art techniques is an unavoidable trend for modern agriculture, but the dilemma is faced for predicting the soil moisture of a given crop area: Arranging limited number of soil detection sensors may not able to capture the whole information well, but adding sensors means increased investment costs. Hence, in this paper, a new soil moisture decision system by using portable detection device was designed and put into field-testing. The detection system can be decomposed into the portable soil moisture detection device and the host computer decision software. And the portable soil moisture detection device was designed with the soil moisture sensor MS-10 which follows the FDR principle, the C8051F410 low power consumption Micro-controller, the Bluetooth wireless transmission module, the data display module and the peripheral circuit. This hardware owns several useful functions such as time recording, data storage, data transmission and real-time display by LED. All these units are well organized to reduce the volume and weight of the device, which makes it easier to operate in the field. After experimental calibration and the field tests, the maximum allowable error and the design accuracy of this portable detection device were 2.2% and 95%, respectively. Meanwhile the software in PC is composed of the data receiving module, the distributed binary consensus algorithm module and the system operation interface. These sub-modules are developed by Visual Basic, Matlab and Matlab GUI, respectively. For the data collected by the portable device, firstly, it can be transmitted from the portable device to the upper computer through the data receiving module with wireless Bluetooth transmission mode; Then, it can be normalized as two states (state "0" and "1") by the normalization method through compared with the set irrigation low limit; The third, the distributed binary consensus algorithm module is used to compute the total and the local decision results according to these data which comes from at the same time; and finally, the decision results can be displayed on the system operation interface as pictures and symbols. In addition, this software can obtain multi-scale and adjustable decision results with higher estimation accuracy if several key parameters in the distributed binary consensus algorithm are optimized by the genetic algorithm (GA). By using the bio-inspired operators such as mutation, crossover and selection, high-quality solutions can be generated efficiently for the optimization and searching problems encountered in our application. Several field tests for the designed decision system are implemented on a 30 m×30 m lawn, the results showed that the system had several advantages, such as high accurate predication (the average accuracy rate of simulation test is about 98% when data is more than 63 and the average accuracy rate of experiment tests is above 90%), lower power cost and strong extendibility. In addition, this system not only can be used independently, but also can be used as an effective complement for the classical fixed detection method. Consequently, this system improves the utilization rate of water resources effectively by means of measuring the soil moisture content accurately, reliable data wireless transmission and fast parallel computation.