Design Of Coal Gangue Doping Detection System Based On STM32

Coal still occupies a dominant position in our country’s energy consumption,and the detection of coal quality is related to our country’s energy security.The existing laboratory coal quality detection method is complicated and time-consuming,and the equipment for the rapid detection method is expensive.Coal gangue,coal slag,water,etc.are commonly used coal doping substances.Studies have shown that the AC impedance of coal doped with gangue and water will be significantly different from that of high-quality undoped coal samples.This paper designs the lower computer of the coal doping detection system,which mainly uses AD5933 to generate the excitation source signal and apply it to both ends of the coal sample to be tested.After the response signal passes through the conditioning circuit,it is sampled by the on-chip ADC and DFT is processed by the DSP.Finally,the impedance value is obtained after conversion calculation.After the FDR-100 obtains the moisture information,it outputs the voltage value,and uses the ADC function of the PA1 interface of the STM32 to perform conversion processing to obtain the moisture value.The impedance measurement experiment was designed,and the impedance value data of coal under different moisture content and different doping rates were obtained,and a method for constructing a mathematical prediction model was proposed.By using the moisture content and impedance values that are easily measured by sensors as input parameters,then can use mathematical models to predict coal doping rates that are not directly available.KNN,FNN and SVR algorithms were selected as the prediction model for comparative testing.SVR had the smallest root mean square error in the training set and test set,which were 0.71078 and 1.2115,so the SVR prediction model was selected;According to the mathematical principle of support vector regression,it is found that the kernel function will also affect the prediction effect of SVR,and the radial basis kernel function is selected for comparison;In order to further optimize the effect of SVR,GWO is selected to optimize SVR,and the final RMSE in the test set is 0.38764,and the relative error is within 2%,and the prediction effect is more accurate.The host computer software supporting the coal doping detection system was designed,it mainly completes the function of using the experimental impedance data as input and training to obtain the GWOSVR prediction model,using the model to input the impedance moisture data measured by the lower computer and predicting the result of the doping rate.Finally,the coal doping detection system is tested,and the results show that the relative error between the prediction result and the real value of the coal doping detection system designed in this paper is only 1.91%.