Pressure sensor temperature compensation method combining polynomial fitting and neural network algorithm

Abstract

The invention discloses a pressure sensor temperature compensation method combining polynomial fitting and a neural network algorithm, and belongs to the technical field of sensors, and the method comprises the following steps: collecting output data of a pressure sensor in a wide temperature and pressure range, carrying out the cleaning and normalization processing of the collected data, and removing abnormal values; according to the collected data, polynomial fitting is carried out by using a least square method and the like, initial compensation is carried out on the original data, and a pressure value after initial compensation is calculated; constructing a neural network structure, training the neural network by using the preliminarily compensated data, and adjusting the weight and bias of the neural network to minimize the error between the output of the neural network and the actual pressure value; a temperature value measured in real time and a sensor output value are input into a trained system, primary processing is carried out through a polynomial fitting model, then a result is input into a neural network, and the neural network outputs a final pressure value after temperature compensation.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature compensation of pressure sensors, and in particular to a temperature compensation method of a pressure sensor combining polynomial fitting and a neural network algorithm. Background Art

[0002] In complex environments, temperature changes can seriously affect the test accuracy of pressure sensors. Existing single compensation methods, such as using only polynomial fitting or neural network algorithms, have certain limitations. Polynomial fitting has advantages in simple relationship processing, but its ability to process complex nonlinear relationships is limited; although neural network algorithms can handle complex nonlinear problems, their training is complex and prone to overfitting. Combining the two can effectively make up for their respective shortcomings. This method of combining polynomial fitting and neural network algorithms has both the simplicity and efficiency of polynomial fitting and the powerful nonlinear processing capabilities of neural networks. It can more accurately compensate for the impact of temperature on pressure sensors, improve the measurement accuracy of pressure sensors in different temperature environments, and ensure reliable use in high-precision application scenarios such as aerospace and automotive industries. Summary of the invention

[0003] The technical problem solved by the present invention is as follows: a temperature compensation method for a pressure sensor combining polynomial fitting and a neural network algorithm is proposed. The method can provide a high-precision and computationally simple temperature compensation algorithm for a pressure sensor. The output values ​​of the pressure sensor at different temperatures and pressures are obtained by collecting a large amount of experimental data. A mathematical model between temperature, pressure and output values ​​is established using a polynomial fitting algorithm. The pressure value after preliminary compensation is calculated. The neural network is trained using the data after preliminary compensation. The weight and bias of the neural network are adjusted to minimize the error between the output of the neural network and the actual pressure value. The real-time measured temperature value and sensor output value are input into the trained system. Preliminary processing is first performed using a polynomial fitting model. Then the result is input into the neural network. The output of the neural network is the accurate pressure value after temperature compensation.

[0004] The technical solution of the present invention:

[0005] 1. A temperature compensation method for a pressure sensor combining polynomial fitting and neural network algorithm, characterized in that the method comprises the following steps:

[0006] Step 1: Use a high-precision experimental device to collect the output data of the pressure sensor in a wide range of temperature and pressure, clean and normalize the collected data, remove outliers, and make the data within a suitable range to facilitate subsequent operations;

[0007] Step 2: Based on the collected data, use the least square method to perform polynomial fitting. Establish a relationship model between the output value Y, pressure P and temperature T, obtain a polynomial model through fitting, perform preliminary compensation on the original data, and calculate the pressure value after preliminary compensation;

[0008] Step 3: Construct a neural network structure, use the preliminarily compensated data to train the neural network, use a suitable training algorithm, adjust the weights and bias of the neural network, and minimize the error between the neural network output and the actual pressure value;

[0009] Step 4: Input the real-time measured temperature value and sensor output value into the trained system, first perform preliminary processing through the polynomial fitting model, and then input the result into the neural network. The output of the neural network is the final pressure value after temperature compensation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The diagram is a schematic diagram of a temperature compensation method for a pressure sensor that combines polynomial fitting with a neural network algorithm. DETAILED DESCRIPTION

[0011] 1. A temperature compensation method for a pressure sensor combining polynomial fitting and neural network algorithm, characterized in that the method comprises the following steps:

[0012] Step 1: Use a high-precision temperature control box to control the ambient temperature range of the sensor to -40°C to 150°C, use a standard pressure source to input a pressure range of 0 to 10MPa to the pressure sensor, collect the output data of the pressure sensor, clean and normalize the collected data, remove outliers, and keep the data in a suitable range for subsequent operations;

[0013] Step 2: According to the collected data, the least square method is used to perform polynomial fitting to establish the relationship model between the output value Y, pressure P and temperature T. The polynomial model is obtained by fitting, Y = a0 + a1P + a2T + a3p 2 +a4T 2 +a5PT, where a0~a5 are fitting coefficients; obtain the current ambient temperature T1 and the pressure sensor output value Y1, substitute T1 and Y1 into the above model, and calculate the pressure value after preliminary compensation;

[0014] Step 3: Construct a neural network structure, use the preliminarily compensated data P1 and temperature T1 to train the neural network, and use the Adam algorithm to adjust the weight and bias of the neural network to minimize the error between the neural network output and the actual pressure value;

[0015] Step 5: Input the real-time measured temperature value and sensor output value into the trained system, first perform preliminary processing through the polynomial fitting model, and then input the result into the neural network. The output of the neural network is the final pressure value after temperature compensation.

Claims

1. A temperature compensation method for a pressure sensor combining polynomial fitting and neural network algorithm, characterized in that: The method comprises the following steps: Step 1: Use a high-precision experimental device to collect the output data of the pressure sensor in a wide range of temperature and pressure, clean and normalize the collected data, remove outliers, and make the data within a suitable range to facilitate subsequent operations; Step 2: Based on the collected data, use the least square method to perform polynomial fitting. Establish a relationship model between the output value Y, pressure P and temperature T, obtain a polynomial model through fitting, perform preliminary compensation on the original data, and calculate the pressure value after preliminary compensation; Step 3: Construct a neural network structure, use the preliminarily compensated data to train the neural network, use a suitable training algorithm, adjust the weights and bias of the neural network, and minimize the error between the neural network output and the actual pressure value; Step 4: Input the real-time measured temperature value and sensor output value into the trained system, first perform preliminary processing through the polynomial fitting model, and then input the result into the neural network. The output of the neural network is the final pressure value after temperature compensation.

Images