Research On Anti-vibration Method Of Dynamic Checkweigher Based On NARX Neural Network

Dynamic checkweigher is a kind of dynamic weighing apparatus,which is widely used in modern factory flow production line.In recent years,the increasing level of social production automation requires the dynamic checkweigher to carry out more accurate online weighing in a shorter weighing time.However,the impact caused by the non-static weighing environment and the rapid movement of the weighed object on the weighing platform of the dynamic checkweigher introduces serious vibration interference to the dynamic checkweigher,so that the output signal of the weighing sensor is not a stable value at this time,Moreover,the frequency characteristics of the vibration noise and the output signal of the weighing sensor in the frequency domain are very close,so that the useful signals in the weighing process are completely submerged in the vibration interference noise,and the accurate weighing results cannot be obtained.In view of this situation,this thesis designs a multi-sensor signal acquisition system based on arm and FPGA to study the vibration interference of the weighing scale.Firstly,this thesis introduces the research background and application value of dynamic weighing technology,and briefly expounds the research basis of dynamic weighing technology studied in this thesis by summarizing the research status of dynamic weighing technology at home and abroad.Combined with the research on the structure and weighing process of the checkweigher,the source of vibration interference and the relationship between vibration interference and the output signal of the weighing sensor are analyzed.Based on the mechanical analysis of the checkweigher,the principle of vibration compensation of the checkweigher is sketched.Secondly,this thesis expounds the overall design of multi-sensor signal system,gives the hardware design of the system and the software design flow chart of each module,and introduces in detail the design ideas of device selection of signal acquisition circuit,synchronous acquisition circuit,dual DAC output circuit,main control module,communication interface and power supply.Then,this thesis making a breakthrough that limitations of traditional methods based on digital filtering,this thesis proposes to use NARX neural network to model and analyze the vibration interference of checkweigher.In chapter 4,some parameter settings,optimal learning algorithm and model training process of the network are introduced,and the actual sampling data is used to train and verify the performance of the model,The verification results show that this method can establish the vibration interference model of the checkweigher with strong anti-interference and high accuracy for vibration interference compensation.Finally,In the resonance state,the vibration compensation method based on NARX neural network proposed in this thesis is compared with the traditional method.It is verified that the processing result of this method is more accurate than that of the traditional method.In addition,the effectiveness and practicability of this method are verified by testing standards of different quality at different speeds.The operation speeds of 2.0m/s and maximum weighing of 200.0g are achieved,which meets the requirements of establishing the national standard“GB /T 27739-2011 Automatic Catchweighing Instruments”category XIII scale.