CN108181028A - A kind of humble pressure sensor mushing error modification method of pressure resistance type - Google Patents

Abstract

The invention discloses a method for correcting interference errors of piezoresistive low and micro-pressure sensors, which comprises the following steps: step 1: establishing a finite element analysis model of piezoresistive low and micro-pressure sensors; step 2: performing vibration through the finite element analysis model of step 1 , temperature and its coupled loading simulation to obtain simulated data; Step 3: According to the simulated data obtained in Step 2, train through a non-parametric compensated radial basis function neural network to obtain multi-input single-output compensation for air pressure, temperature and vibration Model; Step 4: Correct the interference error of the air pressure signal according to the compensation model obtained in Step 3; the present invention can reduce the error of the test signal, obtain a real low-micro-pressure pressure signal, directly obtain the voltage output of the model, and consider more material properties and conditions.

Description

A Method for Correcting Disturbance Errors of Piezoresistive Low Micropressure Sensors

technical field

The invention relates to a piezoresistive low micropressure sensor simulation method, in particular to a piezoresistive low micropressure sensor interference error correction method.

Background technique

Micro-pressure is an important direction for the continuous improvement of pressure test sensors. Micro-pressure or even ultra-micro-pressure sensors have played an important role in the fields of pulsating pressure tests and aerodynamic tests; in actual test environments, sensors are inevitably subject to external pressure. Due to the influence of environmental interference factors, and because the range of the low micro-pressure sensor is relatively small compared with ordinary sensors; therefore, the signal-to-noise ratio is high after being interfered, and it may even distort and cover up the real signal, which brings great trouble to signal testing and extraction; Due to the advantages of high sensitivity and good mechanical properties, micro piezoresistive sensors are often the first choice for low micro pressure test sensors; they convert force signal input into electrical signal output through the piezoresistive effect of silicon, but the sensor is easily affected by temperature and vibration interference ; In the low micro piezoresistive pressure sensor, these two noises will seriously affect the real test signal and cannot be ignored.

For a long time, the error mechanism analysis and compensation method of piezoresistive low-micro-pressure sensors have been an important research topic; due to the temperature effect of single crystal silicon, its piezoresistive coefficient will change with temperature, so the current main research on sensor error interference The field is temperature; the temperature effect of silicon will lead to the zero drift and sensitivity drift of the sensor, which will bring great interference to the signal test; in terms of temperature interference mechanism compensation, at present, the temperature interference output of the sensor is mainly studied through the temperature effect test; The test sensor is placed in a closed test box, and the sensitivity output of the sensor at different temperature and air pressure levels is observed by adjusting the temperature and air pressure changes in the box; However, in this method, the accuracy and control performance of the temperature test chamber are firstly required, because the dynamic changes between temperature and pressure in the airtight container and the mutual coupling effects between the two make it difficult to measure the temperature and pressure. To achieve the expected value at the same time, multiple iterative control is required; and the establishment of models such as neural networks requires a large amount of test data, and the temperature test needs to be re-conducted for different sensors, which is too costly; and the temperature test is basically aimed at finished sensors, which is difficult to design. In the first stage, the temperature performance of the sensor is understood through experiments, so as to make corresponding improvements.

In terms of vibration interference mechanism and compensation, because the vibration interference of general sensors is small, there are few related researches in this area; Compared with low micro-pressure, it can not be directly ignored; currently, sensor vibration interference is generally tested by vibration test bench to test the vibration performance of the sensor; however, vibration interference signals are easily affected by random interference such as electromagnetic, and there is also a need to address The shortcomings of repeated tests with different sensors and the inability to know their vibration performance in advance during the design stage.

As finite element analysis technology continues to mature and is widely used in the industrial field, it has become an important means in the process of sensor design and processing to analyze the relevant performance of sensors through numerical simulation methods; the simulation results can directly provide a basis for the design and improvement of sensor performance and structure. Reference: At present, the relatively mature method for numerical simulation of piezoresistive low-micro-pressure sensors is to establish the sensitive conversion element structure of the sensor, solve the stress distribution on the structure, and then obtain the voltage output through the piezoresistive effect and the Wheatstone bridge model; Although this method is simple and easy to understand, it ignores the vertical stress distribution, and it is difficult to fully consider the anisotropic material properties; it is difficult to fully extract the stress in the resistance distribution area, so there is a certain error, and currently the piezoresistive low micro pressure sensor There are few studies on finite element simulation of vibration and temperature error interference.

Low micro-pressure has small amplitude and is susceptible to vibration, temperature and other interference. To obtain real pressure signals, it is necessary to compensate for vibration and temperature interference; compensation methods can currently be divided into hardware compensation and software compensation. Hardware compensation is through compensation circuits combined with sensors. The compensation performed by its own circuit has good real-time performance; however, the effect of error compensation for unknown factors is not ideal; software compensation includes parametric compensation and non-parametric compensation, and parametric compensation needs to understand the interference mechanism in advance to establish a correct analytical expression ;Non-parametric compensation is neural network training through a large number of experimental data; software compensation is not as real-time as hardware compensation, but its applicability is higher; and for low micro-pressure test sensors, their internal structure is complex, and it is difficult to establish mathematical theory interference of sensors Analysis model.

Contents of the invention

Aiming at the problems existing in the prior art, the present invention provides a piezoresistive low micro-pressure sensor interference error correction method for compensating air pressure, vibration, and temperature interference to obtain real pressure signals.

The technical solution adopted by the present invention is: a piezoresistive low micro-pressure sensor interference error correction method, comprising the following steps:

Step 1: Establish the finite element analysis model of the piezoresistive low and micro pressure sensor;

Step 2: Carry out the simulation of air pressure, vibration, temperature and their coupling loading through the finite element analysis model of step 1, and obtain the simulated data;

Step 3: According to the simulated data obtained in step 2, train through a non-parametric compensation radial basis function neural network to obtain a multi-input single-output compensation model for air pressure, temperature and vibration;

Step 4: Perform interference error correction on the air pressure signal according to the compensation model obtained in Step 3.

Further, the structural-electrical joint finite element simulation method is adopted in the air pressure simulation process in the step 2, and the specific process is as follows:

S1: Divide the resistance of the actual size in the stress concentration area on the model, and assign anisotropic material properties respectively;

S2: set the wire unit to couple the resistance into a Wheatstone bridge;

S3: Set the Wheatstone bridge bridge voltage, coupling wire and resistance cross-sectional current by the method of coupling degrees of freedom;

S4: converting the input signal into a voltage output through steps S1-S3.

Further, the structure-electricity joint finite element simulation method is adopted in the vibration simulation process in the step 2.

Further, in the temperature simulation process in step 2, a structure-electricity joint finite element simulation method is used.

The beneficial effects of the present invention are:

(1) The present invention can compensate vibration and temperature interference by establishing a multi-input single-output interference model, reduce test signal errors, obtain real low micro-pressure pressure signals, and effectively measure low micro-pressure pressures;

(2) The present invention simulates the vibration through finite elements, and can obtain the vibration interference output of the sensor structure through analysis, avoiding many shortcomings such as high cost investment brought by the vibration test, susceptibility to interference, and unfavorable model structure design improvement;

(3) The present invention simulates the temperature through finite elements, and can obtain the sensor temperature zero drift and temperature sensitivity drift through analysis; thus improving the sensor structure can reduce the sensor design and development cycle, and avoid the difficulties of temperature test pressure, temperature coupling, etc. control issues;

(4) The present invention adopts the structure-electricity joint finite element simulation method, can directly obtain the voltage output of the model, and considers more material properties and conditions.

Description of drawings

Fig. 1 is the piezoresistive low micro pressure sensor silicon cup film model structure adopted in the embodiment of the present invention, a is a bottom view, and b is a cross-sectional view of the A-A plane.

Fig. 2 is a piezoresistive low micro-pressure sensor silicon cup film model structure division resistance model adopted in the embodiment of the present invention.

Fig. 3 is a grid division diagram of the silicon cup film model of the piezoresistive low micro pressure sensor used in the embodiment of the present invention.

Fig. 4 is a voltage output diagram in the embodiment of the present invention.

Fig. 5 is a comparison diagram of vibration interference output in the embodiment of the present invention.

Fig. 6 is an output diagram of the coupled interference voltage of air pressure, vibration and temperature in the embodiment of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

A piezoresistive low micro pressure sensor interference error correction method, comprising the following steps:

Step 1: Establish the finite element analysis model of the piezoresistive low and micro pressure sensor;

Step 2: Carry out the simulation of air pressure, vibration, temperature and their coupling loading through the finite element analysis model of step 1, and obtain the simulated data;

Step 3: According to the simulated data obtained in step 2, the non-parametric compensation radial basis function neural network is used for training to obtain a multi-input single-output compensation model for air pressure, temperature and vibration, where the input signals are air pressure, vibration, temperature, and the output The signal is coupled interference output;

Step 4: Perform interference error correction on the air pressure signal according to the compensation model obtained in step 3, and pass the measured air pressure signal, vibration and temperature signals through the compensation model in step 3 to obtain the real low and micro pressure signal output by the sensor.

Further, the structural-electrical joint finite element simulation method is adopted in the air pressure simulation process in the step 2, and the specific process is as follows:

S1: Divide the resistance of the actual size in the stress concentration area on the model, and assign anisotropic material properties respectively;

S2: set the wire unit to couple the resistance into a Wheatstone bridge;

S3: Set the Wheatstone bridge bridge voltage, coupling wire and resistance cross-sectional current by the method of coupling degrees of freedom;

S4: converting the input signal into a voltage output through steps S1-S3.

The core conversion structure of the piezoresistive low-micro-pressure sensor is the silicon cup film, as shown in Figure 1. The traditional finite element numerical simulation method obtains the stress distribution on the film surface by imposing constraints on the bottom of the silicon cup and applying a load on the film surface; extracting the film The stress distribution of the surrounding edge high-stress area (that is, the resistance layout area) is converted into a resistance change through the piezoresistive effect, and then converted into a voltage output through the Wheatstone bridge; the finite element numerical simulation method of the present invention is a structural-electrical combination method , by dividing the actual resistance in the stress concentration area, and setting anisotropic material properties, and applying constraints and loads in the same way; the stress on the film will be directly converted into resistance changes through the set resistivity and other properties, and then according to the set The set coupling bridge voltage is converted into a voltage output; by dividing the actual size of the resistance on the model, giving different material properties, setting the wire unit to couple the resistance into a Wheatstone bridge, and setting the Wheatstone through the coupling degree of freedom Bridge voltage, coupling wires and resistance cross section current; in this way the air pressure load is converted into a voltage output.

Further, in the vibration simulation process in step 2, the structure-electricity joint finite element simulation method is adopted; the vibration disturbance affects the stress distribution of the film through the inertial force. When the sensor is arranged on the surface of the train, the bottom of the silicon cup passes through the substrate and the metal The package is bonded to the car body, and the vibration of the car body is also transmitted to the film through the metal package, the substrate, and the bottom of the silicon cup, so that inertial force is generated on the film and affects the stress distribution of the film; the finite element numerical simulation can be done through the silicon cup bottom The vibration output of the sensor can be obtained by applying the vibration acceleration signal in a constrained manner; the damping ratio of the model needs to be set in the vibration analysis, and the modal analysis of the model can be performed first to obtain the main operating modal frequency and the modal damping ratio; the main The damping ratio of the working mode is approximately considered as the global damping ratio; taking the finite element software Ansys as an example, which uses Rayleigh damping, the formula is as follows:

ξ _i =α/2ω _i +βω _i /2

In the formula: α is the mass damping matrix, β is the stiffness damping matrix, ξ _i is the damping ratio of the i-th vibration mode, ω _i is the natural angular frequency of the i-th vibration mode; the influence of the mass damping matrix can be ignored in general analysis, and the damping The ratio is obtained from the sensor test. For the low micro-pressure pressure sensor, its first-order mode is its main working mode, so the global damping ratio can be approximately considered as the first-order damping ratio; after the vibration signal is loaded, the vibration interference passes through the structure-electrical coupling method can be directly converted to voltage output.

Further, in the temperature simulation process in step 2, the structure-electricity joint finite element simulation method is adopted; the influence of temperature on the sensor structure includes two aspects, temperature zero drift and temperature sensitivity drift; the main reason for the temperature zero drift is due to the silicon cup film The thermal expansion of the model leads to thermal stress in the film; the thermal sensitivity drift is caused by the piezoresistive effect of silicon changing with temperature, and the coupling relationship between the film’s compression deformation and temperature deformation; During the simulation, it is necessary to set the temperature properties such as the thermal expansion coefficient of each material; generally, the thin film structure of the silicon cup is packaged on the substrate, and the constraint conditions are set according to the difference in the thermal expansion coefficient of the silicon cup material and the substrate material, and the normal temperature is set as the thermal expansion zero point; for each grid The expected temperature load value is applied to the nodes, and different pressures are applied at different temperature levels, so as to obtain the temperature zero point output and temperature sensitivity drift at each temperature level.

Example

Figure 1 is a silicon cup thin-film structure sensor designed for low-microvoltage testing. Four resistors are installed at the place where the stress on the lower surface of the film is most concentrated; the resistor layout is shown in Figure 2. Divide the grid in different ways, divide the hexahedral grid by resistance, and divide the tetrahedral grid around it, and reduce the calculation amount as much as possible while ensuring the calculation accuracy. The grid division results are shown in Figure 3; through the above settings and grid division , The voltage output result is shown in Figure 4, the sensor range is 20kPa, the sensitivity is 15.32mv/kPa, and the linearity is 0.27%.

The vibration acceleration signal measured in a vibration test was intercepted, and the silicon cup film model was subjected to vibration loading. Fixed constraints were imposed on the silicon cup bottom surface of the silicon cup film structure in the x and y directions, and the vibration acceleration signal was applied in the z direction. The results are shown in Figure 5; It can be seen from the figure that the acceleration sensitivity of the silicon cup film structure sensor is roughly 5Pa/g.

Select five temperature measuring points of -20°C, 0°C, 20°C, 40°C, and 60°C, and select five pressure measuring points of 0kPa, 5kPa, 10kPa, 15kPa, and 20kPa at each temperature level to measure the temperature of the silicon cup film structure sensor. Interference analysis; assuming that the thermal expansion coefficient of the substrate is small, a fixed constraint is imposed on the bottom of the silicon cup, temperature loads are applied to all nodes, and 22°C is set as normal temperature and zero point of thermal expansion. The analysis results are shown in Table 1 below.

Table 1 Sensor output of each temperature level and pressure measuring point

After calculation, the performance index of the sensor is shown in Table 2. It can be seen from the analysis that the sensitivity of the sensor is greatly affected by the temperature change, and it can be improved by considering the relevant heat treatment of the subsequent processing technology.

Table 2 Temperature interference performance index

Considering the coupling interference of vibration, temperature and air pressure at the same time, temperature loads of -20°C, 22°C and 60°C are applied to the thin film structure of the silicon cup, and a vibration signal with an amplitude of 10g and an angular frequency of 2π is applied at the same time. The voltage output is shown in Figure 6 It can be seen from the analysis that the voltage output is obviously affected by vibration and temperature interference, and the vibration is also affected by temperature. Not only its performance parameters are affected by temperature, but also the vibration variation is also affected by temperature.

In order to reduce the impact of vibration and temperature interference on the test pressure, the obtained numerical analysis results of vibration, temperature and coupling interference are trained through the RBF neural network to obtain a multi-input single-output compensation model for air pressure, vibration, and temperature, and the measured air pressure The data and vibration and temperature data are compensated by the compensation model obtained through neural network training, and a real and effective air pressure output signal can be obtained.

The present invention proposes a structural-electrical joint simulation method, which can directly convert the pressure load into a voltage output, and analyze vibration interference and temperature interference through the finite element structure-electrical joint finite element simulation method; deeply explore sensor vibration and temperature interference Mechanism and compensation method; the established model is universal for micro piezoresistive pressure sensors; compared with traditional test methods, the finite element analysis model has the advantages of ease of operation, economy, and accuracy; Provide a reference for interference design, and provide a basis for sensor error interference compensation; the air pressure, vibration, temperature and corresponding coupling interference output values obtained by finite element numerical analysis are trained through the non-parametric compensation method RBF neural network to establish a multi-input single-output Compensation model, the input signal is air pressure, vibration, temperature signal, the output is coupling interference output pressure, including real pressure and coupling interference pressure; for a set of measured signals, the measured air pressure signal and vibration and temperature signals are known, and the measured air pressure signal includes Real low and micro pressure signals and interference signals, input the measured air pressure signal and vibration and temperature signals into the RBF neural network compensation model that has been trained, and the corresponding real low and micro pressure air pressure signals can be obtained.

The low and micro pressure test is generally only tens to hundreds of Pascals, the signal-to-noise ratio is low, vibration and temperature interference have a great influence on it, it is difficult to extract the real pressure, and the vibration and temperature interference compensation model is established to correct the test signal; the sensor The establishment of the interference analysis model not only plays an important role in the analysis and extraction of low and micro-pressure pressure noise, but also has significant significance for the structural design of the sensor itself and other signal testing fields; The structure-electrical coupling numerical simulation method establishes the vibration interference analysis and temperature interference analysis models of the sensor; it can accurately simulate the vibration interference output and temperature drift interference of the sensor structure, which is of great significance for the extraction and separation of test signal noise and the improvement of sensor structure performance The vibration and temperature interference compensation model established by the present invention can correct the measured signal, reduce the influence of external interference, and thus obtain a real and effective low micro-pressure pressure signal.

Claims (4)

1. a kind of humble pressure sensor mushing error modification method of pressure resistance type, which is characterized in that include the following steps：

Step 1：Establish the finite element analysis model of the humble pressure sensor of pressure resistance type；

Step 2：The simulation of air pressure, vibration, temperature and its coupling loading is carried out by the finite element analysis model of step 1, is obtained Analogue data；

Step 3：Radial basis function neural network is compensated according to the analogue data that step 2 obtains by imparametrization to be trained, Obtain the multiple input single output compensation model about air pressure, temperature and vibration；

Step 4：Mushing error amendment is carried out to air pressure signal according to the compensation model that step 3 obtains.

A kind of 2. humble pressure sensor mushing error modification method of pressure resistance type according to claim 1, which is characterized in that institute It states in step 2 and Finite Element Method is closed using structure-Electricity Federation in air pressure simulation process, detailed process is as follows：

S1：The resistance of actual size is divided, and assign anisotropic material properties respectively in model upper stress concentrated area；

S2：Resistance is coupled into Wheatstone bridge by setting lead unit；

S3：Wheatstone bridge bridge pressure, coupling conducting wire and resistance sections electric current are set by the method for coupling degree of freedom；

S4：Input signal is converted by voltage output by step S1-S3.

A kind of 3. humble pressure sensor mushing error modification method of pressure resistance type according to claim 2, which is characterized in that institute It states in step 2 and Finite Element Method is closed using structure-Electricity Federation during vibration simulation.

A kind of 4. humble pressure sensor mushing error modification method of pressure resistance type according to claim 2, which is characterized in that institute It states in step 2 and Finite Element Method is closed using structure-Electricity Federation in temperature simulation process.