The Study Of The Optimization Technique Application On The BBD Mill

BBD Mill is widely used in the power station as a power set.lt has a lot of strengths: Low energy consume, high working efficiency, suiting all kinds of coal and conditions. The performance of BBD mill has a very complicated relation with its structure and the material, hardly be obtained by general theory equation, empirical equation or regression analysis. A mathematic model for mill power and output of the BBD mill was built up by artificial neural network, and optimized design for the structure and work parameters of BBD mill based on genetic algorithm was proposed in this paper. Computing results show that artificial neural network has reached to a satisfactory precision and the optimization result of genetic algorithm is good.The paper detailedly introduced the theory of the artificial neural network and genetic algorithm and analysed the test result and the data of the product techno- information. The result and precision is good.The main ideas of the paper are as followed:1. Introduce the structure, principle and characteristic of the BBD Mill, the current development and application status domestic and abroad.2. Adopt the systems analysis method-assume the substance and energy in/out of the BBD mill as a changeless system. A mathematic model for mill power and output of the BBD mill was built up for the structure and work parameters.3. The paper introduces the principle of the artificial neural network, makes the learning and training program of BP algorithm and sums up the shortages in the application of it. A measure is given to improve it.4. An experiment to test the mill power and output on 40D62 BBD Mill of Shandong Laicheng power station. The result coming from the artificial nerve network method can fit the experiment data well, the modelestablished by the artificial nerve network method can be used on the engineering request.5. Introduce the basic metheod, principle and the application of the genetic algorithm. Sum up the shortages in the application and improve the optimize capability.6. Built the man-machine interactive software, It will benefit the choice of the BBD Mill, working administration and design.The mathematic model for mill power output built up by the BP network obtained a good result in actual application. The practice certificated the model is viable. However, the experiment sample of the velocity because of the roller length, the roller diameter and roller is too little, it can only be handled with the theories formula, not the independence quantity to carried for the design. This causes in the followed-up design, it designs the scope to contract, affecting whole and excellent turn the result, can be the point study from now on. Also exist the problem of a series in make use of inherit calculate way excellently turn the artificial nerve network model process, such as the selection, the chromosome length that the manhunt space really settle, grows a group of individual numbers and code the method selection and cross the rate and the variation rates take values to all need in advance according to excellent turn the problem of actual circumstance depend on personal experience enactment, then again according to excellent turn the manhunt as a result carry on appropriate of adjustment, these problems design for heredity by all means to establish to bring the certain difficulty with parameter, the direct influence heredity calculate way of excellent turn to search the result. Because the artificial nerve network power value and the limited value number many, cause inherit the calculate way excellent to turn to change to measure the number numerous, search the space huge, this aspect give heredity calculate way of excellent turned the manhunt to bring the difficulty and make excellent to turn to search the speed to become slow and refrain from rash action the result become bad, still increased theprocedure at the same time time of the movement.The problem that enumerate above improves to inherit the calculate way exactly excellent turn the main aspect of the artificial nerve network model, also can be the point to study from now on.