给出了一种基于人工神经网络的传感器非线线估计新方法。该方法用幂级数多项式拟合传感器的非线性模型，多项式的系数可由神经网络学习算法得到。当环境条件发生变化时，只要给出几组测量数据对，该方法可自动重新训练网络，获得新的多项式系数，实现传感器的在线标定。该方法保证了纸浆浆量计量仪的计量精度。 A new method of sensor nonlinear line estimation based on artificial neural network is given. This method uses power series polynomials to fit the nonlinear model of the sensor. The polynomial coefficients can be obtained by the neural network learning algorithm. When the environmental conditions change, as long as several pairs of measurement data are given, the method can automatically retrain the network to obtain new polynomial coefficients to realize the online calibration of the sensor. This method ensures the measurement accuracy of pulp slurry measuring instrument.