Prediction Of Moisture Content In Wood Drying Process Based On Neural Network And Ant Colony Algorithm

Moisture Content(MC),as a technical index for the long-term stable use of wood,and as a benchmark for measuring dry quality,is an important indicator to check the quality of the drying process.In the detection of wood moisture content,the electrical measurement method is widely used in the actual production of wood drying because of its convenience and high efficiency.However,because the wood moisture content changes near the fiber saturation point due to the physical characteristics of the wood,the wood is driven to contain water.The rate presents a non-linear trend of change,leading to deviations in moisture content measurements near this point,even to large distortions.Once data deviation occurs,it may lead to mistakes in the drying process.In the face of high economic value timber,if there are mistakes,not only the loss of raw materials but also lost processing time,seriously affecting the quality and efficiency of production.From this it can be seen that it is a very necessary task to strengthen the detection of moisture content at the saturation point of wood fiber.In order to predict the value of moisture content near the saturation point of wood fibers,beech wood and poplar wood were selected as the research object,drying process experiments were conducted according to their drying benchmarks,and the experimental parameters affecting wood moisture content were collected and analyzed(dry-bulb temperature,wet-bulb Temperature,core temperature,wind speed.).Then use BP neural network,ant colony algorithm and improved ant colony algorithm respectively to optimize the parameters and forecast the wood moisture content.Compare the prediction results with the actual moisture content of the weighing method and analyze the error results and running time.In order to discuss the prediction effect of the three algorithms on wood moisture content.The ant colony classification algorithm Ant-Miner,which is the main classification prediction algorithm selected in this paper,is a classification technology that can be compared with classical classification algorithms.It is based on ant colony algorithm in group intelligence.However,Ant-Miner did not fully utilize the idea of ant colony,and its heuristic strategy contained local information.Different from the ant colony algorithm,the heuristic function value of Ant-Miner constantly changes with the running of the algorithm,which increases the computational complexity of the algorithm.In order to improve Ant-Miner's efficiency in solving data classification problems,an improved ant colony classification algorithm,mAnt-Miner+ was used.The algorithm draws on the idea of multi-ant colony constructing ant colony of mAnt-Miner and uses a new heuristic strategy.The simulation results of wood moisture content prediction show that mAnt-Miner+ improves the operating efficiency without affecting the accuracy of prediction and the simplicity of the rules.Compared with the traditional Ant-Miner,it improves the error of prediction classification and greatly improves the algorithm.The prediction accuracy has a high engineering practical value.