Sensor Compensation In MWD Instrument Based On DSP

MWD Instrument is the indispensable measurement equipment in modern directional drilling, High temperature and high pressure working environment of Inclinometer drilling leads to nonlinear sensor output, directly affects the performance of MWD Instrument. To solve this problem, we conducted a series of studies.To improve measurement precision and stability of drilling inclinometer, we created Elman neural network model used to realize compensation for the sensor. Using LM algorithm to train it, the error accuracy approach 10-7 after 10 steps iterative, increased by 4 orders of magnitude, greatly improving the drilling inclinometer measurement accuracy and stability, but the compensation effect is unstable; In order to achieve more stable compensation effect, the adaptive genetic algorithm was used as the learning algorithm of Elman network, the error precision increases 2 orders of magnitude than the Elman network, the compensation effect is stable; To further enhance the measurement precision and stability of inclinometer, the adaptive Ant colony algorithm in which pheromone evaporation factorρand pheromone update strategy adjust automaticly were adopted to optimize the Elman neural network threshold, the compensation is better than Elman neural network and the genetic network, the compensation accuracy is close to 10-8, this algorithm has good compensation effect and basically meets the needs of the scene.First we encode the colony Elman neural network on VC platform, and change the algorithm by modifying the basic data types, storage space allocation and a series of changes, and transplant it to DM642 platform successfully, then we optimize the transplanted code by the code optimization, the optimization based on hardware resources,the optimization based on compiler, the optimized program operation results show that: In this paper, the optimization algorithm has obvious effect, improved the run speed Significantly, the sensor compensation accuracy has also been further improved, meet the requirements of real-time DSP and on-site needs.