CN-116432520-B - Improved temperature compensation method based on extreme learning machine algorithm

CN116432520BCN 116432520 BCN116432520 BCN 116432520BCN-116432520-B

Abstract

The invention discloses an improved temperature compensation method based on an extreme learning machine algorithm, which comprises the steps of mainly influencing the accuracy of a pressure monitoring sensor by drift through analysis of temperature compensation accuracy influence factors, so that an improved temperature compensation method is provided, a calculation model of a suitable thermistor is built, a larger gap exists between the calculation model of the thermistor and the actual use condition, so that the actual compensation effect is deteriorated, output voltages at different temperatures are converted into values at a fixed room temperature, a calibration coefficient at the room temperature is used, the compensation voltage is converted into an equivalent correction pressure value, the fitting error between the actual output voltage and the compensation voltage is minimized, and the extreme learning machine algorithm is utilized to further optimize the calculation to obtain a parameter value and compensate the linear error. Experiments prove that the low-cost piezoresistive pressure sensor with the compensation algorithm is feasible to realize the environment self-adaptation function, and provides a certain reference for the long-term unattended condition observation and monitoring under water.

Inventors

ZHAO KUI
Xiao Zunkun
CUI HAIPENG
ZHAO WEI
MA ZHIYU
LIU ZHIGANG
LIU LUXI
LI HUA
LIU ZHENGYONG
CHEN YOU

Assignees

青岛杰瑞工控技术有限公司
重庆前卫科技集团有限公司

Dates

Publication Date: 20260512
Application Date: 20230320

Claims (4)

1. An improved temperature compensation method based on an extreme learning machine algorithm is characterized by comprising the following steps: S1, analyzing the temperature characteristics of a thermistor in a traditional compensation mode, and determining temperature compensation accuracy influence factors of a pressure sensor; s2, improving a thermistor model in the traditional compensation mode in the step S1, and constructing a suitable calculation model of the thermistor; s3, obtaining a thermistor model in the step S2, and converting output voltages at different temperatures into values at a fixed room temperature through calculation of the model; S4, using the calibration coefficient at room temperature obtained in the step S3, and converting the compensation voltage into an equivalent correction pressure value; S5, carrying out further optimization calculation on the equivalent correction pressure value obtained in the step S4 by utilizing an extreme learning machine algorithm to obtain a parameter value, and compensating the linear error; the step S3 specifically comprises the following steps: S31 the output voltage V a at any pressure can be represented by a polynomial based on room temperature conditions, according to the empirical formula of a wheatstone bridge: Wherein V a0 is the output voltage of the pressure sensor at standard room temperature t 0 , 0 Is a temperature correction coefficient at room temperature t 0 , t is the actual ambient temperature, 1 And 2 Linear and nonlinear temperature correction coefficients at arbitrary pressures, respectively; s32, establishing an equivalent matrix to calculate a temperature correction coefficient according to the equation in the equation S31: Wherein t 1 、t 2 、t 3 is 3 temperatures recorded in the test, V a1 、V a2 、V a3 is t 1 、t 2 、t 3 output voltage under standard pressure, t 0 is standard room temperature, V a0 is output voltage of the pressure sensor under standard room temperature t 0 , 0、 1 、 2 Is a temperature correction coefficient at a temperature of t 1 、t 2 、t 3 ; The S4 temperature compensation calculation method specifically comprises the following steps of deriving according to a medium formula in S31, correcting each output voltage V a into a voltage at a standard room temperature t 0 by using a temperature correction coefficient in S32, and performing a temperature compensation algorithm: wherein, V b is the compensation output voltage at different temperatures, V a0 is the compensation output voltage at standard room temperature t 0 , and t is the actual ambient temperature.
2. The improved temperature compensation method based on the extreme learning machine algorithm according to claim 1, wherein the thermistor calculation model obtained in S2 is: Wherein k (R X0 ,A X ,R Y0 ,A Y ) is a proportionality coefficient related to the thermistor to be used and the reference thermistor characteristic parameter, R X0 、A X is the thermistor characteristic parameter to be used, R Y0 、A Y is the reference thermistor characteristic parameter, R Y1×N is the actual measurement value of the reference thermistor, and R 1×N 、R Y1×N are N-dimensional row vectors.
3. The improved temperature compensation method based on the extreme learning machine algorithm according to claim 1, wherein the equivalent correction pressure value obtained by S5 to S4 specifically includes: S51, inputting digital signals of pressure and temperature corresponding to two neurons of an input layer, and outputting digital signals of pressure network compensation, wherein the relation between an output z j and an input x j is as follows: Wherein the sum alpha i 、β i is a weight vector connecting the input hidden layer and the hidden output layer, y i is the deviation of the ith hidden neuron, ã is the number of hidden nodes, j is the number of samples, and f is a nonlinear activation function; S52: α i 、β i is randomly generated in the range of 0 to 1, the output weights of the i-th hidden layer unit β i , i=1, 2,..n will be derived by matrix operation: wherein H is the output of hidden layer node, H + is the molar-Peng Resi generalized inverse matrix of matrix H, and T is the desired output.
4. The improved temperature compensation method based on the extreme learning machine algorithm according to claim 1, wherein the specific step of performing temperature compensation by using the extreme learning machine algorithm in S5 comprises the following steps: s531, normalizing the sample data to the range of [ -1,1], and measuring in the pressure and temperature range of the sensor; S532, randomly dividing the normalized sample data into training data and test data according to the proportion of 2:1; s533, orderly selecting the number of hidden nodes from 1 to the number of training samples; s534, randomly initializing an input weight and hidden layer bias, inputting training data, and calculating an output weight; S535, calculating the output of the test data according to the weight and the deviation obtained in the S534; s536, repeating S532-S534 until satisfactory compensation precision is obtained; s537, writing the calculated weight and deviation into a microprocessor, and verifying an algorithm within the range of the temperature and pressure range of the sensor; s538, calculating the actual precision of the calibration sensor.

Description

Improved temperature compensation method based on extreme learning machine algorithm Technical Field The invention relates to the field of oil and gas resource exploration, in particular to an improved temperature compensation method based on an extreme learning machine algorithm. Background With the economic development of China, land resources are not capable of meeting the requirements of social development on petroleum and natural gas, and exploration and development of oil and gas resources are being strategically shifted from land to ocean. The pressure monitoring sensor is used for monitoring the pressure of the underwater petroleum and is a basic instrument of equipment in ocean development. In recent years, the pressure monitoring sensing and design and verification difficulties of underwater petroleum are high, the manufacturing process is complex, the manufacturing precision is high, and the product cannot be replaced in China. The problems of foreign technical blockade, high monopoly price, long delivery cycle and difficult guarantee of delivery schedule are faced, and the development of localization and industrialization of the oil gas monitoring equipment in China is severely restricted. Secondly, the contradiction between the accurate measurement of the stability of the pressure sensor for a very long time and the fact that the circuit cannot guarantee the precision requirement under the long-time working condition is solved. The improvement of the signal detection instantaneity and accuracy and the signal intelligent processing reliability has become the mainstream demand in the industry at present. Disclosure of Invention In order to overcome the problems in the prior art, the invention provides an improved temperature compensation method based on an extreme learning machine algorithm. The invention solves the technical problems by adopting the technical scheme that the improved temperature compensation method based on the extreme learning machine algorithm comprises the following steps: S1, analyzing the temperature characteristics of a thermistor in a traditional compensation mode, and determining temperature compensation accuracy influence factors of a pressure sensor; s2, improving a thermistor model in the traditional compensation mode in the step S1, and constructing a suitable calculation model of the thermistor; s3, obtaining a thermistor model in the step S2, and converting output voltages at different temperatures into values at a fixed room temperature through calculation of the model; S4, using the calibration coefficient at room temperature obtained in the step S3, and converting the compensation voltage into an equivalent correction pressure value; s5, carrying out further optimization calculation on the equivalent correction pressure value obtained in the step S4 by utilizing an extreme learning machine algorithm to obtain a parameter value, and compensating the linear error. The improved temperature compensation method based on the extreme learning machine algorithm, wherein the thermistor calculation model obtained in the step S2 is as follows: R1×N＝k(RX0,AX,RY0,AY)RY1×N Wherein k (R X0,AX,RY0,AY) is a proportionality coefficient related to the thermistor to be used and the reference thermistor characteristic parameter, R X0、AX is the thermistor characteristic parameter to be used, R Y0、AY is the reference thermistor characteristic parameter, R Y1×N is the actual measurement value of the reference thermistor, and R 1×N、RY1×N are N-dimensional row vectors. The above-mentioned improved temperature compensation method based on the extreme learning machine algorithm, the S3 specifically includes: S31 the output voltage V a at any pressure can be represented by a polynomial based on room temperature conditions, according to the empirical formula of a wheatstone bridge: Va＝Va0+(λ0+λ1Va0)(t-t0)+λ2Va0(t-t0)2 Wherein V a0 is the output voltage of the pressure sensor at a standard room temperature t 0, lambda 0 is the temperature correction coefficient at a room temperature t 0, t is the actual ambient temperature, and lambda 1 and lambda 2 are the linear and nonlinear temperature correction coefficients at any pressure respectively; s32, establishing an equivalent matrix to calculate a temperature correction coefficient according to the equation in the equation S31: wherein, t 1、t2、t3 is 3 temperatures recorded in the test, V a1、Va2、Va3 is t 1、t2、t3 output voltage under standard pressure, t 0 is standard room temperature, V a0 is output voltage of the pressure sensor under standard room temperature t 0, and λ 0、λ1、λ2 is temperature correction coefficient under t 1、t2、t3 temperature. The above-mentioned improved temperature compensation method based on the extreme learning machine algorithm, the S4 temperature compensation calculation method specifically comprises deriving according to the medium formula in S31, and correcting each output voltage V a to the voltage at the standard ro