| As a new type of grating displacement sensor, the time-grating sensor isdeveloping toward high speed and high precision measurement. In this paper, byanalyzing the magnetic field type time-grating sensor (only refers to field type andvariable reluctance type) measuring principle and basic structure to understand itselectromagnetic field analysis. Build a relationship between each parameter and thesensor output signal to complete the purpose of optimized parameters. In this paper,another task is to extract tiny signal of Time-Grating displacement sensor output, toensure full into the electrical measurement information processing. The front-endsignal processing circuit is the bridge connected sensors and signal processing system.⑴First, the magnetic circuit analysis method is used for two kinds oftime-grating sensor of mathematical modeling and magnetic circuit calculation toobtain magnetic field distribution and the signal output waveform. Next, through theFEA (Finite Element Analysis)to discusses and analyzes the electromagneticparameters impacting performance of time-grating sensor. These parameters includeRotor slot type, width of air gap, the rotor tooth width and the slot width. FEA allowsus to calculate time grating’s electromagnetic field accurately, not only making simple“quality†analysis but also making “quantity†analysis. The analyses results can beused to optimize the structure to achieve the purpose of improve the measurementaccuracy and stability.⑵The five structural parameters of Field-Mode Time-Grating were selected, theFEA results were collected to generate training samples by using of the orthogonaldesign. Finally, the Electromagnetic Field Optimizations was realized based on theback propagation neural networks-genetic algorithm (BPNN-GA) principle.Levenberg-Marquardt (LM) algorithm was used to train the BPNN, the network inputand output of non-linear mapping relations was established. The networkperformances were assessed by linear regression method. The BPNN was used as theobjective function solver for GA to optimize the magnetic circuit parameters. Verifiedby the new Modeling, the optimization of the output of the signal strength increase of23.15%, while the error is0.784%.⑶Front signal processing circuit was based on Modular Design，including thepreamplifier, a band-pass filter, the non-linear amplification, waveform convertingand photocopier five modules. The circuit with tiny sensor signal extraction and processing functions，Verified by using of multisim simulation and experimentaltesting，Reliability analysis and optimization on the magnetic field type Time-gratingsensor was finished by using of BPNN-GA theory, combining with the orthogonaldesign, FEA and multivariate statistical analysis method.The research results havegreat guidance in development process for the multi-varieties and small-batchproduction. The study method is more practical value of construction work, which canobviously curtail the test period, reduce design costs and significantly improve thereliability of the products. The method is also widely applicable to other projects. |
