Research And Application Of Automatic Pig-keeping Base On Particle Swarm Optimization BP Neural Network

With the continuous development of information industry, the data mining technology has been gradually integrated into the agriculture field. Realizing agriculture informatization and intelligentization has become the future development direction of modern agriculture. Our country is a global power in pig production field, but the feeding technology falls behind the developed countries. Applying modern information technology to improve pig production and quality has the actual perspective.In raising pig process, the choice of the pig at the right time to sell has the vital significance for the improvement of the economic efficiency of commercial enterprises. There are many factors to timely decide the pig market. The traditional market approach is the relevant expert approach using their balance of these factors to market judgment. It has a strong subjectivity. If this subjective guidance is not scientific, it is easy to result in the low rate of the pig market, enterprise losses and other serious consequences. In order to make the pig market results more scientific, this paper uses the improved PSO-BP algorithm to analyse farm data, and classifies the pig market such that the output results are more accurate. In addition, from the pig business, using the cloud platform provides third-party services. Enterprise equipment only upload data to obtain the recommendation results fast and efficiently without buying any equipments, make such recommendation service more extension.The main work of this paper is as follows:First, this paper first introduces the research background, significance, and the status of automated pig production field at home and abroad. Based on the problem of low accuracy in the current pig market, this paper proposes a timely pig market recommended model using intelligent methods to replace the traditional manual method, which provides scientific guidance for pig farmers.Second, with regard to BP neural network, it is prone to obtain local solution and slow convergence speed under the multidimensional problems, which may affect the recommendation inaccurate problem. This paper combines with the improved particle swarm algorithm to optimize the BP neural network. The improved particle swarm optimization has three improvements:changing the parameters, perturbing extremum and increasing local term. The proposed algorithm is applied to the timely market recommend model, and compares its performance with the original algorithms to find a more suitable algorithm for the pig market recommendation.Third, for the aspect of the large pig farms data and cost saving, this paper introduces a cloud platform as a third-party provider. As following, the pig data from the farm can be uploaded to cloud platform for parallel processing. The users do not buy some equipment and know the procedure of data processing. Through the display, the users can directly refer to the recommend result. It is not only save time but also save the cost.Experimental results demonstrate that the improved PSO-BP algorithm obtains better accuracy and faster running speed than original BP algorithm in the timely pig market problem. Moreover, for the shared data uploaded in each pig farm, improved PSO-BP algorithm under the cloud platform keeps the original accuracy and has a shorter processing time in comparison with the algorithm in single PC.